Paper Digest: Recent Papers on Speech Recognition
Paper Digest Team extracted all recent Speech Recognition related papers on our radar, and generated highlight sentences for them. The results are then sorted by relevance & date. In addition to this ‘static’ page, we also provide a real-time version of this article, which has more coverage and is updated in real time to include the most recent updates on this topic.
Since 2018, Paper Digest has built a foundation of data spanning decades of conferences, journals, and research topics. The platform features a daily digest service that sifts through tens of thousands of new papers, clinical trials, news articles, and community posts, filtering the noise to highlight what matters most to specific interests. Beyond daily updates, dozens of built-in research tools streamline the academic workflow, supporting efficient reading and writing, comprehensive literature reviews, and automated research report generation.
Paper Digest Team
New York City, New York, 10017
team@paperdigest.org
TABLE 1: Paper Digest: Recent Papers on Speech Recognition
| Paper | Author(s) | Source | Date | |
|---|---|---|---|---|
| 1 | InterCorrect: Intersection-Aware Correction of Demographic Model Merging for Fair ASR Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Automatic Speech Recognition (ASR) systems often show uneven performance across demographic groups, and errors can be especially difficult to address for speakers belonging to … |
Ashley E. Bravo-Bravo; Yuchen Zhang; Haralambos Mouratidis; Ravi Shekhar; Monorama Swain; | arxiv-cs.CL | 2026-10-06 |
| 2 | DirectSpeech2LLM: A Simple End-to-End Framework to Mitigate Prompt Overfitting in Speech-LLMs Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose DirectSpeech2LLM, a simple end-to-end framework that preserves the instruction-following ability of the LLM on unseen tasks when conditioned on speech. |
Hemant Yadav; Sunayana Sitaram; Roger Zimmermann; Rajiv Ratn Shah; | arxiv-cs.CL | 2026-10-06 |
| 3 | Pronunciation-Oriented Reinforcement Learning for Japanese Text-to-Speech with Kana-Domain ASR Rewards Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We instead compute CER in the kana domain using a kana-transcribing ASR model and reference readings (Kana-CER). |
Shiao Zhu; Lianbo Liu; Kai Washizaki; Koki Nikaido; Yui Sudo; | arxiv-cs.SD | 2026-10-05 |
| 4 | Mind The Accent Gap: British Accent Robustness in Speech-Driven Financial Voice Assistants Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce CavaBench, the first internally collected benchmark of spoken financial queries, and use it to evaluate a range of ASR models and their end-to-end ASR-LLM pipeline behaviour across self-reported British accents. |
AADAM HAQ et. al. | arxiv-cs.SD | 2026-10-05 |
| 5 | Hiding Tool Latency in On-Device Cascaded Voice Agent Through Speculative Execution Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present speculative tool execution for on-device cascaded voice agents, which predicts tool requests from partial ASR hypotheses and initiates tool execution while speech is still being received, thereby reducing end-to-end response latency. |
KYUDAN JUNG et. al. | arxiv-cs.SD | 2026-10-05 |
| 6 | AraYoungVoices: A Diverse L1/L2 Corpus of Arabic Child and Adolescent Speech Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce AraYoungVoices, a 151.72-hour Arabic read-speech corpus from 286 speakers aged 7–18, comprising AraKids (7–12) and AraTeens (13–18). |
Shammur Absar Chowdhury; Zien Sheikh Ali; Houssam Eddine-Othman Lachemat; Hamdy Mubarak; | arxiv-cs.CL | 2026-10-04 |
| 7 | Automatic Speech Recognition for Low-Resource Sinhala: A Critical Review of Methods, Challenges, and Future Directions Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Sinhala, the primary language of Sri Lanka with about 16 million speakers, illustrates the difficulty: agglutinative morphology, a 54-phoneme inventory, subject-object-verb (SOV) syntax and scarce annotated speech data limit both conventional and modern ASR systems. This paper presents the first critical review of Sinhala ASR research, tracing its development from Hidden Markov Models (HMMs) through deep neural networks to self-supervised pre-trained models such as wav2vec 2.0, XLS-R, Whisper and Massively Multilingual Speech (MMS). |
Chanuka Dinuwan; Sanath Jayasena; Buddhika Karunarathne; | arxiv-cs.CL | 2026-10-04 |
| 8 | Steering Speech-Language Models: Training-Free Task Specialization Via Contrastive Activation Addition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose a training-free Contrastive Activation Addition (CAA) protocol that derives steering vectors for common speech tasks (e.g. transcription) in SpeechLLMs from a small number of labeled utterances. |
SÉVERIN BAROUDI et. al. | arxiv-cs.CL | 2026-10-03 |
| 9 | Personalized Automatic Speech Recognition for A Dysarthric and Tracheostomic Speaker Using Artificial Conversations Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose a multi-stage training pipeline based on Whisper Base: fine-tuning on standard Czech speech, acoustically simulated tracheostomic speech, and the speaker’s data. |
David Nadrchal; Monorama Swain; Florian Schmid; Gerhard Widmer; Paul Primus; | arxiv-cs.AI | 2026-10-02 |
| 10 | Toward Elastic Speech Inference: Training-Free Wake-Word Detection from Pretrained ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we explore the use of a shared pretrained ASR backbone for WuW detection without gradient-based fine-tuning and examine whether a compact encoder can be extracted using the PCA-based structured pruning approach of SliceGPT. |
Hwayeon Kim; Youngwon Choi; Hyeonyu Kim; | arxiv-cs.SD | 2026-10-01 |
| 11 | Benchmarking Automatic Speech Recognition Tools for Iberian Languages Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We observe consistent sex disparities across most systems, highlighting fairness challenges in multilingual ASR. |
Fernando López; Pablo Gómez; David Solans; Paulo Villegas; Jordi Luque; | arxiv-cs.CL | 2026-09-29 |
| 12 | BaLEEN: Biasing with Latent Encoded Entities for Context-Aware ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we propose BaLEEN (Biasing with Latent Encoded Entities), a lightweight, hypernetwork-based framework for dynamic contextual adaptation without fine-tuning the underlying ASR model. |
Chihiro Taguchi; Yotaro Kubo; Rujikorn Charakorn; | arxiv-cs.CL | 2026-09-29 |
| 13 | FFASR: Benchmarking Far-Field Automatic Speech Recognition Using High-Fidelity Simulated RIRs Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present FFASR, a held-out corpus of 15,637 utterances and an open leaderboard spanning nine conditions, each varying a single acoustic factor: anechoic near-field speech, a measured-versus-simulated office-lab pair, static far-field mixtures at high/mid/low signal-to-noise ratio(SNR), and moving-talker variants at matched SNR. |
SHIVAM SAINI et. al. | arxiv-cs.SD | 2026-09-29 |
| 14 | CARDAMOM: A Micro-Dialectal Arabic Speech Dataset for ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present Cardamom, a micro-dialectal Arabic speech dataset designed to support fine-grained evaluation and adaptation of automatic speech recognition (ASR) systems. |
BASHAR TALAFHA et. al. | arxiv-cs.CL | 2026-09-28 |
| 15 | Almieyar: A Culturally Grounded Benchmark for Multi-Dialect Arabic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce ALMIEYAR, a culturally grounded ASR benchmark covering 17 Arabic dialects across six families, built entirely from newly recorded speech unseen by existing models. |
OMID GHAHROODI et. al. | arxiv-cs.CL | 2026-09-28 |
| 16 | CoSE-E: A Benchmark for Code-switched Speech Evaluation in Enterprise Settings Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we propose (1) a CS-ASR synthetic benchmark and multidimensional evaluation framework tailored to enterprise domains, (2) systematic evaluation of frontier ASR systems across 5 language pairs, (3) diagnostic analysis of the additional transcription errors that code-switching introduces across language pairs and models. |
Shama Gupta; Hoang H Nguyen; Chelsea Huang; Lindsay Devon Brin; Fanny Riols; | arxiv-cs.SD | 2026-09-28 |
| 17 | FuseAlign: Forced Alignment in The Wild Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Additionally, forced aligners often operate on error-prone automatic speech recognition (ASR) output. We address these gaps with improved evaluation metrics, a scoring protocol for real ASR transcripts, and AlignBench, a benchmark spanning diverse speaker, acoustic, and text conditions. |
Mithilesh Vaidya; Stephen Bailey; Sumukh Badam; Matthew Bendel; Xingzhe He; | arxiv-cs.LG | 2026-09-27 |
| 18 | Evaluating Machine Unlearning in ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we investigate whether existing MU algorithms and evaluation tools are suitable for ASR. |
Diogo Dinis; Francisco Teixeira; Bhiksha Raj; Alberto Abad; Isabel Trancoso; | arxiv-cs.CL | 2026-09-27 |
| 19 | In-Context Adaptation of Encoder-Decoder Models in Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we ask whether in-context adaptation is an inherent ability for all encoder-decoder models. |
Yen Meng; Sharon Goldwater; Hao Tang; | arxiv-cs.CL | 2026-09-27 |
| 20 | Automatic Speech Recognition for The Basaà Language: A Low-Resource Approach Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: The rapid advancement of Artificial Intelligence (AI) and Natural Language Processing (NLP) has revolutionized the way humans interact with machines. Among the most impactful … |
Sophie Gertrude Ngo Mock; Charles Moudina Varmantchaonala; Paul Dayang; Jean Michel Nlong II; Christopher Gies; | arxiv-cs.CL | 2026-09-26 |
| 21 | FA-Bench: A Benchmark for Word-Level and Phone-Level Forced-Alignment and ASR Timestamps Under Clean and Noisy Conditions Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present FA-Bench, an open framework that fixes those choices once and releases the code, splits, phone mapping, text normalization and scoring script, with results published periodically. |
WEI CHU et. al. | arxiv-cs.CL | 2026-09-26 |
| 22 | NVAlign: Direct-Gradient Optimization for Non-Verbal Control in Continuous Autoregressive Flow Matching Text-to-Speech Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To this end, we present NVAlign, a direct-gradient post-training framework for NVV tag-following in this architecture. |
Qiaolin Wang; Pedro Sandoval-Segura; Anunaya Joshi; Edvardas Jurkonis; Jake Downie; | arxiv-cs.SD | 2026-09-25 |
| 23 | Mu-bench: A Multilingual Utterance Transcription Benchmark Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce mu-bench, a dataset of 4,270 caller utterances from 250 phone calls to an AI banking agent in English, Spanish, Turkish, Vietnamese, and Mandarin, centered on form-field inputs such as names, email addresses, and confirmation codes. |
Andrea Li; Soham Ray; | arxiv-cs.CL | 2026-09-25 |
| 24 | STAM-ASR: Speaker-Temporal Anchoring with Memory for Multi-Speaker ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose STAM-ASR, Speaker-Temporal Anchoring with Memory, a lightweight framework that extends an already pretrained AudioLLM for multi-speaker ASR. |
Victor Tolulope Olufemi; Syeda Faiza Ahmed Sara; Shammur Absar Chowdhury; | arxiv-cs.SD | 2026-09-24 |
| 25 | Inference-Time Target Speaker Unlearning in LLM-Based Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce target-speaker unlearning ASR (TSU-ASR) task in a fully end-to-end framework for multi-speaker ASR and diarization. |
Bo Su; Yueru Yan; Thai Le; | arxiv-cs.CL | 2026-09-24 |
| 26 | TS-OPD: Reconciling ASR and QA in Speech Language Models Via Task-Specific On-Policy Distillation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Speech Language Models (SLMs) inherit strong instruction-following capabilities from pretrained language models, yet ASR specialization can substantially degrade them. To address this ASR–QA trade-off, we propose Task-Specific On-Policy Distillation (TS-OPD), which leverages models before and after ASR specialization as complementary QA and ASR teachers. |
YUJIE GUO et. al. | arxiv-cs.SD | 2026-09-24 |
| 27 | Adaptive Fisher-Whitened Cross-Covariance for Low-Resource Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To investigate whether task-informed PEFT can better support low-resource ASR, we apply Fisher-Whitened Cross-Covariance Analysis (FCCA) to Whisper and Qwen3-ASR, and introduce two complementary extensions: Asymmetric-Coupled FCCA (AC-FCCA), which exploits structured cross-layer sharing, and Adaptive-Rank FCCA (AR-FCCA), which reallocates adaptation capacity across projection matrices under a fixed parameter budget. |
Asmee Mishra; Mengjie Qian; Brechtje Post; Kate Knill; | arxiv-cs.CL | 2026-09-24 |
| 28 | Joint Analysis of Latent Dimensionality and Frame Rate in Continuous Audio Encoders Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We train sixteen encoders spanning four widths and four frame rates, with downstream adapters and probes, using matched training protocols. |
Kyudan Jung; Sehyun Lee; Son-ha Jo; Jaegul Choo; Sanghyuk Shoi; | arxiv-cs.SD | 2026-09-24 |
| 29 | Training-Free Contextual ASR Via SpeechLLM-Based Error-Aware Selective Retrieval Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose a training-free contextual ASR framework in which a pretrained speech large language model (SpeechLLM) jointly generates an ASR hypothesis and localizes error spans likely to involve domain-specific terms. |
Natsuo Yamashita; Ai Nemoto; Ryosuke Koichi; Masaaki Yamamoto; | arxiv-cs.SD | 2026-09-24 |
| 30 | CLEAR: Online Speech Content Leakage Estimation Through Cross-ASR Disagreement Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present CLEAR, a reference-free approach for estimating speech-content leakage at runtime using disagreement among heterogeneous ASR systems. |
Bhawana Chhaglani; Tanvi Kandepuneni; Jeremy Gummeson; Prashant Shenoy; | arxiv-cs.SD | 2026-09-24 |
| 31 | Agentic-ger: Terminology Recovery in Long-form Speech Using Global Context Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Motivated by the world knowledge and contextual capability of large language models (LLMs), we propose Agentic-GER, an LLM-based agent for terminology correction in long-form speech. |
Yanqiao Zhu; Wupeng Wang; Zhifu Gao; Xiangang Li; Xie Chen; | arxiv-cs.CL | 2026-09-24 |
| 32 | Benchmarking and Domain Adaptation of Automatic Speech Recognition (ASR) for Adolescent Health Communication in Ghanaian Languages Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents an end-to-end study of automatic speech recognition (ASR) for adolescent health communication in three Ghanaian languages (Twi, Dagbani, and Ewe). |
STEPHEN E. MOORE et. al. | arxiv-cs.CL | 2026-09-24 |
| 33 | Whispered Speech Recognition Enhancement Using Teager Energy Operator with PLP Features Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: The proposed method exploits instantaneous energy variations to characterize the nonlinear aerodynamic behavior inherent in whispered speech. |
Branko R. Marković; Jovan Galić; Asutosh Kar; | Circuits, Systems, and Signal Processing | 2026-09-23 |
| 34 | ASR Ensembling for Phoneme Intelligibility Evaluation of Speech Anonymizers Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present the first phoneme-level intelligibility evaluation of speech anonymizers, assessing the performance of ASR-ensemble-based metrics against measured intelligibility from a crowdsourced listening test. |
Victor Ménestrel; Sebastian Möller; Slim Ouni; Dorothea Kolossa; | arxiv-cs.SD | 2026-09-23 |
| 35 | Automatic Estimation of Verbal Fluency Index in People with Motor Neuron Disease Using ASR Alignment and Pause Modelling Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Building on recent advances in automated speech analysis, this study proposes a system for estimating VFI. |
BAHMAN MIRHEIDARI et. al. | arxiv-cs.CL | 2026-09-22 |
| 36 | NADI 2026: The Second Multidialectal Arabic Speech Processing Shared Task Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: NADI 2026 is the seventh edition of the Nuanced Arabic Dialect Identification (NADI) shared task series and the second dedicated to multidialectal Arabic speech processing. This … |
PETER SULLIVAN et. al. | arxiv-cs.CL | 2026-09-22 |
| 37 | Ruby-ASR: Evidence-Preserving Supervision for Joint Orthographic and Lexical-Reading Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present Ruby-ASR, which refines the conventional target into a span-bound orthographic–lexical-reading sequence. |
HAO SHI et. al. | arxiv-cs.CL | 2026-09-22 |
| 38 | From Reliable Text to Real Voices: Trust-Aware Progressive Adaptation for Low-Resource TTS Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose trust-aware progressive adaptation: synthetic-to-real adaptation first establishes text-speech correspondences, then restores reference-speaker control using real speech. |
JIAYI LU et. al. | arxiv-cs.SD | 2026-09-22 |
| 39 | Quieter Than The Room: Representation Drift and Task Robustness in Speech Encoders Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Non-speech interference can change a speech representation without causing comparable task loss. We test eight frozen encoders on four tasks, adding non-speech sounds throughout … |
Vsevolod Kovalev; Pranay Manocha; | arxiv-cs.SD | 2026-09-22 |
| 40 | Vimarsha: Faithful ASR Evaluation for Indian Languages with Demographic Diversity, In-the-Wild Audio and Spelling Variations Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce Vimarsha, a 100-hour benchmark spanning all 22 scheduled Indian languages, designed to address both distortions. |
KAUSHAL SANTOSH BHOGALE et. al. | arxiv-cs.CL | 2026-09-21 |
| 41 | HaikuS2S: A Cascaded System For Responding In Verse Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present a cascaded system, HaikuS2S, combining ASR (automatic speech recognition), LLM (large language model)-generated haiku, and TTS fine-tuning on both prose and custom haiku datasets. |
Devangi Sharma; Sophia Judicke; Glenda Tan; Conrad Schaumburg; Shinji Watanabe; | arxiv-cs.CL | 2026-09-20 |
| 42 | Federated Multilingual Speech-LLMs: Architecture and Aggregation Strategy Benchmarking Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present a comprehensive benchmark of Federated Learning (FL) for multilingual Automatic Speech Recognition (ASR), evaluating four Speech-LLM architectures on the Multilingual LibriSpeech dataset. |
Jordi Luque; Aleix Sant; Fernando López; | arxiv-cs.CL | 2026-09-20 |
| 43 | Beyond Encoder Fusion: Multi-View Discrete Token Augmentation for LLM-Based ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose multi-view discrete token augmentation, a simple strategy that augments each training utterance by generating alternative token sequences from fixed SSL encoders, such as HuBERT, WavLM, and MMS-300M. |
Paul Moïse Gangbadja; Mickael Rouvier; Fabrice Lefèvre; | arxiv-cs.SD | 2026-09-20 |
| 44 | Rethinking Human-Aligned Evaluation: An Analysis of Semantic Metrics Beyond WER Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce HATS-en, an English dataset for human-centered ASR evaluation. |
HRITIKA SHARMA et. al. | arxiv-cs.CL | 2026-09-18 |
| 45 | HearInContext: A Benchmark for Implicit Context in Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce HearInContext, a Mandarin–English benchmark that pairs shared synthetic speech with assistant replies supporting different interpretations. |
Yifan Gao; Yao Tian; Hongbin Suo; | arxiv-cs.CL | 2026-09-16 |
| 46 | FD-VAD: Semantic Endpoint Detection for Streaming Full-Duplex Speech Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We formulate semantic endpoint detection as a causal audio-language reasoning task and introduce FD-VAD, an ASR-free streaming endpointer that maps bounded causal audio windows directly to Continue/Stop decisions. |
Puneet Mathur; Dinesh Manocha; | arxiv-cs.CL | 2026-09-16 |
| 47 | Look Less, Hear Better: Jointly Rewarded GRPO for Streaming ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce AWED, a word-level emission-delay metric defined relative to the acoustic end of each word, and post-train a DSM recognizer with GRPO under a reward that scores transcription accuracy and measured delay jointly. |
Xiuwen Zheng; | arxiv-cs.SD | 2026-09-16 |
| 48 | T-SANDHI: Tone Sandhi-aware Adaptive Network with Decoupled Hybrid Injection for Low-resource Taiwanese Hokkien Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: However, our experiments on Taiwanese Hokkien reveal that speech foundation models actually handle tone sandhi variations effectively, and the real performance bottleneck stems from a localized confusion between these variations and retained citation tones. To address this, we propose T-SANDHI to explicitly decouple surface acoustics from underlying lexical intent on top of a frozen Whisper backbone. |
HUNG-YANG SUNG et. al. | arxiv-cs.CL | 2026-09-16 |
| 49 | A Probe Shift Is Not A Fairness Fix: The Limits of Representation Steering in Speech Models Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Automatic speech recognition (ASR) systems exhibit unequal error rates across speaker groups, motivating interventions on their internal representations. We ask whether … |
Nicolas Bourrel; Abderrahmane Issam; Gerasimos Spanakis; | arxiv-cs.CL | 2026-09-16 |
| 50 | DUPAR: Dual-Path Conversational Retrieval Via Speech Retriever with Cross-Turn Evidence Caching Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This cascade adds latency and propagates recognition errors, whereas direct speech retrieval is vulnerable to cross-modal misalignment. To address these limitations, we propose DUPAR, a conversational retrieval framework with complementary slow and fast paths. |
YUANJUN LI et. al. | arxiv-cs.IR | 2026-09-15 |
| 51 | Merging The Knowledge of LLMs for Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this study, we propose incorporating external LMs via model merging. |
Hayato Futami; Tatsuya Kawahara; | arxiv-cs.CL | 2026-09-14 |
| 52 | Enabling Streaming User Transcription in Full-Duplex Speech-to-Speech Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we propose an efficient method to add streaming ASR capabilities to an existing duplex S2S model by introducing a lightweight ASR head in parallel to the agent text head. |
KE HU et. al. | arxiv-cs.CL | 2026-09-14 |
| 53 | Sequential Adapter Stacking for Cross-Lingual Low-Resource ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Alongside warm initialization and attention-based fusion, we propose Sequential Adapter Stacking, which places a trainable target-language adapter on top of a frozen source-language adapter. |
Thai Thi Thanh Thao Dang; Mengjie Qian; Kate Knill; | arxiv-cs.CL | 2026-09-14 |
| 54 | Typhoon ASR Streaming: Steerable Low-Latency Thai Speech Recognition with Real-Time Shallow Fusion Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Open Thai automatic speech recognition (ASR) is dominated by offline, Whisper-based models that read the whole utterance before transcribing, ruling out low-latency uses such as live captioning and voice agents. We present a deployable system for streaming Thai ASR that lets a user steer its vocabulary at decode time, without retraining. |
Warit Sirichotedumrong; Tanawin Samutsin; Shah Faisal Wani; Sittipong Sripaisarnmongkol; Kunat Pipatanakul; | arxiv-cs.CL | 2026-09-13 |
| 55 | Grounded in Sound: Reinforcement Learning with A Frozen Acoustic Judge to Curb ASR Insertion Hallucinations Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose an acoustic-fidelity reward: a GRPO reward augmented with a separately pretrained, permanently frozen, non-autoregressive character-level wav2vec2-CTC acoustic judge, used strictly at training and absent at inference, where a single model decodes greedily. |
Tingzhen Xiong; Rilin Chen; Weiwei Li; Wentao Zhang; Qicong Xie; | arxiv-cs.SD | 2026-09-13 |
| 56 | Are Caption Metrics Broken? Latency, Deaf and Hard of Hearing User Ratings, and Bias Across Technologies Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To this end, we describe a U.S.-based large-scale online survey with 216 validated participants, who provided 302 responses containing a cumulative 4,832 data points. |
BERNARD THOMPSON et. al. | arxiv-cs.HC | 2026-09-10 |
| 57 | How Robust Are Neural Audio Codecs for African Speech? A Multi-Task Benchmark and The Limits of Perceptual Quality Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Neural audio codecs enable low-bitrate speech compression and tokenization, yet their robustness on accented and multilingual speech remains under-evaluated. We benchmark seven … |
Chibuzor Okocha; Christan Earl Grant; | arxiv-cs.SD | 2026-09-10 |
| 58 | Assessing The Use of Automatic Speech Recognition and Large Language Models for Individuals with Language Impairments Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Here, we systematically evaluate edge-oriented ASR-LLM pipelines for individuals with language impairments using comparison studies and ablation experiments across aphasia, child language impairment, and dementia datasets. |
GELEI XU et. al. | Nature Communications | 2026-09-10 |
| 59 | Quantifying Consonant Contributions to Word Intelligibility Via Acoustic Masking Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents a scalable method that measures consonant contribution using acoustic masking. |
EUNJUNG YEO et. al. | arxiv-cs.CL | 2026-09-10 |
| 60 | Xiaomi-CocktailASR-1 Technical Report Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we propose Xiaomi-CocktailASR-1, an LLM-based end-to-end TS-ASR architecture. |
YIRU ZHANG et. al. | arxiv-cs.SD | 2026-09-10 |
| 61 | Beyond Word Error Rate: A Switch Aware Evaluation of ASR and Audio Language Models on English Yoruba Code-Switched Speech Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present a switch aware evaluation of eleven modern systems (six ASR models and five audio LMs) on English Yoruba code-switched speech, using a deterministic 2000 utterance evaluation set and a shared scoring pipeline. |
Chibuzor Okocha; Christan Earl Grant; | arxiv-cs.CL | 2026-09-10 |
| 62 | Leveraging Fine-grained Error Correction in Korean Speech Recognition for Consultation Services Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Derived from genuine call center interactions, it comprises 1,974 dialogues with 115,460 utterances. Leveraging this resource, we propose Detector-Gated Contextual Span Correction (DCSC), a text-only post-editing framework for error-sparse Korean speech recognition transcripts. |
YONGHYUN JUN et. al. | arxiv-cs.CL | 2026-09-09 |
| 63 | StreamAlign: Streaming Text-Aligned Speech Tokenization Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce StreamAlign, a text-aligned speech tokenization framework that enables streaming tokenization for real-time speech-text joint modeling. |
KANG-WOOK KIM et. al. | arxiv-cs.CL | 2026-09-09 |
| 64 | BuzzASR: A Swarm of 100+ Monolingual Speech Recognition Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce BuzzASR, a collection of language-specialized fine-tuned Whisper models adapted for automatic speech recognition (ASR) in 102 languages. |
Shivam Singh; Aditya Yadavalli; Catherine Arnett; Alex Warstadt; | arxiv-cs.CL | 2026-09-08 |
| 65 | X2Streaming-ASR: Wait When Uncertain, Emit When Ready for Streaming ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose X2Streaming-ASR, which decomposes streaming recognition into when to commit and what to commit. |
ZHIWEI LIN et. al. | arxiv-cs.SD | 2026-09-08 |
| 66 | Qwen-Audio-3.0-ASR Technical Report Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: However, bridging the gap between academic benchmark performance and real-world production utility remains a persistent challenge, particularly in handling diverse regional dialects, dynamic entities and hotwords, long-range contextual information, and disfluent spontaneous speech. In this report, we present Qwen-Audio-3.0-ASR, a Mixture-of-Experts (MoE) LLM-based ASR system designed to address these production demands through a unified, instruction-following framework. |
CHUANMENG BIAN et. al. | arxiv-cs.CL | 2026-09-07 |
| 67 | Adaptive Artificial Intelligence for Personalized Pronunciation and Grammar Instruction in Multilingual African Classrooms: A QuasiExperimental Study of Automatic Speech Recognition and Natural Language Processing Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This study investigated the effectiveness of an adaptive AI framework integrating Automatic Speech Recognition , Natural Language Processing , adaptive learning algorithms, and peer-supported learning for improving language proficiency among multilingual learners. |
Onyekewen Mercy Monday; | International Journal of Education and Evaluation | 2026-09-07 |
| 68 | A Systems Approach Toward AI-Enabled Radiology Dictation for Workflow Efficiency in India Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Radiology reporting remains a major workflow bottleneck, with inefficiencies in dictation, editing, and report finalization leading to delays and an increased documentation … |
Satvik Tripathi; Tessa S. Cook; | Indian Journal of Radiology and Imaging | 2026-09-03 |
| 69 | Reducing Hallucinated Transcripts in Whisper Via Hallucination Space Projection Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Whisper is a widely used foundation model for automatic speech recognition (ASR), but its generative decoder can produce fluent hallucinated transcripts for inputs containing little or no speech. We propose a training-free, inference-time method to reduce these hallucinations using low-rank projection of decoder activations. |
Maryam Abbasihafshejani; Murtuza Jadliwala; | arxiv-cs.AI | 2026-09-03 |
| 70 | TRILOGUE: A Trilingual Spoken Dialogue Fact-Checking Benchmark with Evidence and Paired Audio Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce TRILOGUE (TRIlingual spoken diaLOGUE fact-checking), a large-scale trilingual benchmark of source-grounded spoken dialogues in English, Russian, and Kazakh. |
CHAEWAN CHUN et. al. | arxiv-cs.CL | 2026-09-03 |
| 71 | Choosing A PEFT Variant for Per-Patient Dysarthric ASR: A Single-Speaker Case Study on Two ASR Bases Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present a single-speaker case study comparing seven LoRA-family methods (LoRA, QLoRA, AdaLoRA, DoRA, LoHA, VeRA, VB-LoRA) on two production bases (Whisper-large-v3 with Hungarian fine-tuning, and a multilingual Qwen3-ASR-1.7B checkpoint) for one post-stroke Hungarian male speaker (S1, 409 utterances; severe dysarthria on auditory-perceptual clinical assessment). |
Bernard Muller; László Tóth; LaVonne Roberts; | arxiv-cs.CL | 2026-09-02 |
| 72 | CACTUS: Mask-Guided Semantic Clean-Label Backdoors in Decentralized Federated Learning Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Sensitivity analyses show that ASR varies with network topology and increases with the malicious-node ratio. |
Chao Feng; Burkhard Stiller; | arxiv-cs.LG | 2026-09-02 |
| 73 | Soft Posterior Speaker Injection for Multi-Talker Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Hard diarization-based segmentation introduces irreversible errors, whereas serialized output training (SOT) avoids explicit segmentation but does not condition a pretrained encoder on speaker activity. We propose Soft Posterior Speaker Injection (SPSI): a lightweight head predicts frame-level speaker posteriors $\hat{\mathbf{P}}$ and injects them into Whisper through multi-layer feature-wise linear modulation (FiLM) and decoder speaker-memory prompts. |
Jian Zhu; Cheng Luo; | arxiv-cs.SD | 2026-09-01 |
| 74 | Multilevel Contextual Prompting for Conversational ASR: Unifying Conversation History and Hotwords with Speech LLM Related Papers Related Patents Related Grants Related Venues Related Experts View Save |
Jiewen Zheng; Gaofeng Cheng; Xuyang Wang; Qingwei Zhao; Yonghong Yan; | Speech Commun. | 2026-09-01 |
| 75 | Listen to The Latents: Self-Correcting Speech Recognition in Large Audio Language Models Through Hidden-State Interactions Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we refine warm-initialized LLM-based ASR models by leveraging their own pre-adaptation base LLMs, focusing on LoRA-adapted settings where the base LLM is preserved. |
CHAN-JAN HSU et. al. | arxiv-cs.CL | 2026-08-31 |
| 76 | Parallel Time-Band Mixing with Learned Observation-Adding for Robust ASR Front-Ends Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we present a sequence-parallel band-split enhancement front-end built on a Parallel Time-Band Mixer (PTBM) block that eliminates within-block recurrent unrolling. |
Xingyu Shen; Runze Wang; Wei-Ping Zhu; Benoit Champagne; | arxiv-cs.SD | 2026-08-31 |
| 77 | Linguistic Distance Segregates Latent Representations in Automatic Speech Recognition Systems Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper investigates the relationship between first language background and English ASR performance. Through empirical analysis, we observe that the correlation between speakers’ L1 distance and ASR error rates yields a systematic effect on English Speech, with its strength varying across datasets and models. |
Ting-Hui Cheng; Line Katrine Harder Clemmensen; Sneha Das; | arxiv-cs.CL | 2026-08-31 |
| 78 | Anchoring Speech with Semantics: A Multimodal Adapter Mechanism for Automatic Speech Recognition in Low-Resource Languages Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Low-resource ASR remains difficult because scarce transcripts provide limited supervised evidence for target-side generation. To address this gap, we propose SAMA-ASR, a lightweight adapter mechanism that augments the decoder with semantic anchors from auxiliary translations and an acoustic anchor from speech; in principle, the mechanism can be applied to similar encoder–decoder multitask speech models. |
KUAN-TANG HUANG et. al. | arxiv-cs.CL | 2026-08-29 |
| 79 | VoiceCodeBench: Evaluating Exact Structured-Token Recovery in Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce VoiceCodeBench, a benchmark for evaluating exact structured-token recovery in English ASR. |
TYLER BAUMGARTNER et. al. | arxiv-cs.CL | 2026-08-28 |
| 80 | When Robots Mishear Us: Mapping The Safety Risks of Voice-Controlled Embodied AI Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We investigate whether automatic speech recognition (ASR) errors in user input can lead to unsafe outputs from Embodied AI (EAI) models. |
Sihan Jia; Oliver Lemon; | arxiv-cs.AI | 2026-08-28 |
| 81 | Topic Matching in The Wild: Benchmark and Lessons from Real-World ASR Transcripts Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We compare three types of matchers: a regex-based baseline, zero-shot sentence embedding encoders, and Gemini-based LLM matchers. |
Saman Rahbar; Xiliang Zhu; Irvin Cardoza; David Rossouw; | arxiv-cs.CL | 2026-08-27 |
| 82 | Scaling Phoneme-based TTS Augmentation for ASR: A Unified Pipeline and Controlled Study Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present a unified phoneme-based TTS-to-ASR augmentation pipeline built around a multilingual TTS model trained from scratch using the F5-TTS architecture with language-ID conditioning. |
Zhen Wang; TianRui Wu; RongQi Han; Hao Wu; Wei Liang; | arxiv-cs.CL | 2026-08-27 |
| 83 | Generative Vs. Encoder Large Language Models for ASR Evaluation: A Comparative Study Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents a comparative study of both families for ASR evaluation. |
THIBAULT BAÑERAS-ROUX et. al. | arxiv-cs.CL | 2026-08-26 |
| 84 | Speech Recognition Application for Surgical Safety Checklist Verification: A Systematic Review Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: The aim of the review was to examine VAs designed to record speech in the operating room, check SCs items, and interact with the surgical team. |
Nodari M. Kakabadze; Elena B. Kleymenova; Anton G. Nazarenko; Luibov Yashina; | N.N. Priorov Journal of Traumatology and Orthopedics | 2026-08-26 |
| 85 | AfriSwitch: A Benchmark for In-the-Wild African Code-Switched Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present AfriSwitch, a 61.36-hour human-transcribed benchmark of in-the-wild code-switched speech spanning 16 African languages and language varieties, released with switch-level English span tags, perutterance Code-Mixing Index (CMI), and switch-point counts. |
Gabrial Zencha Ashungafac; Busayo Awobade; Tobi Olatunji; | arxiv-cs.CL | 2026-08-26 |
| 86 | Fine-Tuning Whisper for Automatic Speech Recognition in Baniwa: A Preliminary Study Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This work presents a preliminary study on the adaptation of Whisper for Automatic Speech Recognition in Baniwa, an indigenous Arawakan language spoken in Brazil, Colombia, and Venezuela. |
Leonardo Duart; Tiago Fonseca; Thiago Chacón; | arxiv-cs.CL | 2026-08-26 |
| 87 | Better Retrieval, Worse Robustness:How Multi-hop RAG Amplifies Upstream ASR Errors Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Using four English accents synthesized through neural TTS, we evaluate four RAG configurations on three multi-hop QA benchmarks (HotpotQA, 2WikiMultiHopQA and MuSiQue) against a clean-text oracle. |
Zhenghua Bao; | arxiv-cs.CL | 2026-08-24 |
| 88 | Bulbul: A Dataset for Dialectal Arabic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present BULBUL, a multi-dialect Arabic ASR dataset collected from 275 speakers in 11 Arab countries. |
AHMED ASHRAF et. al. | arxiv-cs.CL | 2026-08-22 |
| 89 | A Modular Multimodal Framework for Automated Misinformation Investigation: Integrating OCR, NLP, Vision-Language Captioning, and Speech Analysis for Evidence-Based Trust Scoring Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents the design and implementation of a modular, multimodal misinformation investigation platform that combines optical character recognition (OCR), natural language processing (NLP), vision-language image captioning, and automatic speech recognition (ASR) within a single evidence-fusion pipeline. |
Shivani Tiwari; | International Journal of Creative and Open Research in … | 2026-08-19 |
| 90 | A Speech Corpus for Mizo Automatic Speech Recognition: Whisper and SraVaani 1.0 Fine-Tuning with Morphology-Aware Evaluation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This study reports the development of an Automatic Speech Recognition (ASR) system in Mizo, a low-resource language. |
Priyankoo Sarmah; Sanasam Ranbir Singh; | arxiv-cs.CL | 2026-08-19 |
| 91 | Unadapted Multilingual ASR on A Garrusi Kurdish Evaluation Set: A Common-Reference Staged Normalization Analysis Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Evaluating speech recognition for a Kurdish variety written in a Latin field orthography, using a model that outputs Arabic script, creates a measurement problem before a modelling one: direct scoring treats writing-system differences as recognition errors. |
Hiwa Asadpour; | arxiv-cs.CL | 2026-08-17 |
| 92 | The Null Token Knows: Reducing Message-Free Hallucination in ASR and NMT Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Modern encoder-decoder systems can produce fluent text even when their input contains no recoverable message. We study this failure in ASR and NMT through the models’ reserved null tokens, asking whether the score for ending generation already carries a usable abstention signal. |
Kirill Borodin; Vasiliy Kudryavtsev; Ivan Viakhirev; | arxiv-cs.CL | 2026-08-16 |
| 93 | DeepL Voice: Real-Time Speech-to-Speech Translation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: DeepL Voice is a real-time speech-to-speech translation system for global business communication, following a pragmatic incremental approach: developing a production-grade cascaded speech-to-speech-translation (S2ST) system, while exploring end-to-end solutions in parallel.The production system (launched November 2024) achieves competitive transcription quality through proprietary real-time ASR models and eliminates translation "flickering" via stable text streaming while maintaining low latency.Supporting 18 input languages and 30+ target languages, it offers DeepL Voice for Meetings (Microsoft Teams/Zoom integration), DeepL Voice for Conversations (mobile apps), as well as the DeepL API for Voice.Key features include customizable formality and glossary support for business-appropriate communication, with voice cloning TTS under development. |
JOHANNES ERNESTI et. al. | ijcai | 2026-08-15 |
| 94 | MDwAIstScheduler: Bringing On-Device Voice Documentation Into Clinical Practice Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present MDwAIstScheduler, a low- cost, belt-worn pipeline that lets a physician speak naturally dur- ing the encounter and have the resulting medications, allergies, labs/orders/referrals, follow-up scheduling, vitals, and problems land in the EHR as review-ready drafts. |
Diego Mardian; Frank Liu; | arxiv-cs.HC | 2026-08-15 |
| 95 | Towards Fine-Grained Code-Switch Speech Translation with Semantic Space Alignment Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Previous studies mainly rely on the models themselves to implicitly learn semantic representations and resort to costly manual annotations. To mitigate these limitations, we propose enhancing Large Language Models (LLMs) with a Mixture-of-Experts (MoE) speech projector composed of language expert groups, where each group specializes in the semantic space of a specific language for fine-grained speech feature modeling. |
YAN GAO et. al. | ijcai | 2026-08-15 |
| 96 | Persona-ASR: Bilingual Target-Speaker Speech Recognition for Kazakh–English Overlapping Speech Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To evaluate the system, we introduce KazMix3, a Kazakh overlap dataset for TS-ASR training, and PersonaMix, a controlled bilingual benchmark spanning same- and cross-language enrollment across varying interferer counts (1–3) and signal-to-noise ratios (−3 to +3 dB). |
Rakhat Meiramov; Tomiris Rakhimzhanova; Adil Taibassarov; Zhanat Makhataeva; Huseyin Atakan Varol; | Machine Learning and Knowledge Extraction | 2026-08-14 |
| 97 | StreamHear: Domain-Adapted Pseudo-Labeling for Semi-Supervised Streaming Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present StreamHear, a semi-supervised pipeline that adapts a pretrained streaming student by fine-tuning an offline transducer teacher on the labeled training set, generating pseudo-labels on the unlabeled portion, and fine-tuning the student on the mixture. |
ZEFANG LIU et. al. | arxiv-cs.CL | 2026-08-13 |
| 98 | Machine Unlearning for Speech Question Answering in Large Audio-Language Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we study machine unlearning for speech QA in LALMs, a setting that is more challenging than prior work on text-based Large Language Models (LLMs) or Automatic Speech Recognition (ASR) due to the tight coupling between acoustic perception and factual knowledge. |
Zhe Liu; | arxiv-cs.LG | 2026-08-12 |
| 99 | Easper: An Accessible ASR Pipeline for Language Documentation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present Easper, an open-source, no-code workflow enabling linguists to iteratively fine-tune ASR models via cloud resources directly from ELAN annotations. |
Aso Mahmudi; Ting Dang; Ekaterina Vylomova; Nick Thieberger; | arxiv-cs.CL | 2026-08-12 |
| 100 | ASR-Roundtrip Evaluation Can Mask Context- and Convention-Dependent Reading Errors in Chinese News TTS Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We study Chinese news TTS spans whose correct reading depends on context or domain conventions, such as sports scores, aircraft models, technical units, and membership names. |
Shijun Luo; Lizhi Wan; | arxiv-cs.CL | 2026-08-11 |
| 101 | Whisper-Aware LLM: Self-Supervised Uncertainty Learning for Robust Whispered Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: The signal ambiguity of whispered speech drives ASR systems toward two opposing failure modes: failing to capture whispered speech or hallucinatory transcription of noise. This paper introduces the Whisper-Aware LLM, a framework that teaches an Audio-LLM to perceive and react to this uncertainty. |
Gaopeng Xu; Zhenyu Wang; Zheng Xue; Yinfeng Xia; Haitao Yao; | arxiv-cs.SD | 2026-08-11 |
| 102 | DonorRank: Donor Language Selection for Low-Resource Cross-Lingual Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present DonorRank, a learning-to-rank framework for predicting effective donor languages for zero-shot ASR. |
Akriti Dhasmana; Aarohi Srivastava; David Chiang; | arxiv-cs.CL | 2026-08-11 |
| 103 | MyMediWhisper: Construction of Burmese Medical Speech Corpus and Whisper Fine-Tuning for Clinical Dialogue ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This work presents a Burmese medical speech recognition framework built on a high-quality 28-hour corpus recorded and validated by native speakers. |
YE KYAW THU et. al. | arxiv-cs.CL | 2026-08-11 |
| 104 | Edge Phoneme Recognition for Children’s Speech Through Age-Aware Training Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: During a phoneme detection competition, we found that training a lightweight model to predict the age of the learner, as well as the phoneme sequence, enabled a 94M-parameter model to outperform WavLM Large models (317M) on the target DrivenData distribution, and fall within approximately 0.04 CER of competition ensembles with 90 times the parameters. |
MATTHEW ARBOLEDA et. al. | arxiv-cs.AI | 2026-08-10 |
| 105 | VoxZip: Semantic-Anchored Temporal KV Cache Compression for Long-Context Audio Inference Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Existing text-centric compression methods struggle here, often disrupting speech continuity or discarding crucial semantic cues. To address this, we propose VoxZip, a train-free, two-stage semantic-anchored KV cache compression framework. |
WENXU JIA et. al. | arxiv-cs.AI | 2026-08-09 |
| 106 | LSAR: Sparse Lexical Representation Learning for Efficient and Interpretable Audio Retrieval Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Dense retrievers bypass transcription, yet they compress each clip into a single embedding, making relevance difficult to inspect and incurring non-trivial indexing and query-time overhead as collections grow. To address these limitations, we propose LSAR (Learned Sparse Audio Retrieval), the first learned sparse retrieval framework for audio that maps audio directly into a sparse lexical space compatible with inverted-index search. |
Haoyue Li; Yuzhe Bai; Li Niu; | kdd | 2026-08-06 |
| 107 | Decolonizing Linguistic Policies in Automated Speech Recognition: A Framework for Cross-Culturally Competent Speech AI Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce the Three Harms (3M) taxonomy—Misrecognition, Misalignment, and Mistrust—and a seven-layer situatedness model for linguistic diversity in ASR and ASR-mediated voice interfaces. |
Jay L. Cunningham; Mark Atta Mensah; Richard Martinez; Joao Vieira da Silva Neto; Efi Dawodu; | arxiv-cs.CL | 2026-08-06 |
| 108 | How to Recognize New Words: A Comparison Between Context Biasing Methods and Speech LLMs Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Recognizing new and rare words – named entities, acronyms, domain specific special words, and other items scarce in training data – remains a key challenge for automatic speech recognition (ASR). We compare two strategies for this: context biasing methods, where an ASR model is extended such that during inference a word list can be supplied, and speech large language models (LLMs) prompted with context directly. |
Christian Huber; Alexander Waibel; | arxiv-cs.CL | 2026-08-06 |
| 109 | Voxlect: A Speech Foundation Model Benchmark for Modeling Dialects and Regional Languages Around The Globe IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present Voxlect, a novel benchmark for modeling dialects and regional languages worldwide using speech foundation models. |
TIANTIAN FENG et. al. | kdd | 2026-08-06 |
| 110 | PolySpeech-100: A Large-Scale Benchmark for Speech Understanding Across 100+ Languages and Dialects Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Existing benchmarks suffer from three critical limitations: a pronounced bias towards high-resource languages, a focus on low-level recognition (ASR) rather than semantic reasoning, and a neglect of regional dialects. To bridge this gap, we introduce PolySpeech-100, a massive-scale benchmark designed to assess ‘native-level’ speech comprehension across 110 linguistic variants. |
SICHENG YANG et. al. | kdd | 2026-08-06 |
| 111 | SD-DVQ: Semantic-Driven Data Visualization Query Generation from Speech Queries Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Yet, existing methods still face issues such as speech recognition errors and insufficient model generalization. We propose SD-DVQ, a Semantic-Driven Data Visualization Query Generation from Speech Queries framework, to address these issues. |
Haodi Zhang; Xiaohui Tang; Yuanfeng Song; | kdd | 2026-08-06 |
| 112 | Analyzing Speech Condition Effects in Dysarthric ASR: A Layer-wise Probing Study Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present a layer-wise probing analysis of a transformer ASR encoder on Mandarin dysarthric speech under three transcript-matched conditions: original dysarthric speech, speaker conditioned zero-shot TTS resynthesis, and unconditioned TTS. |
DARWIN JELESTIN MUTHU et. al. | arxiv-cs.CL | 2026-08-03 |
| 113 | ParaASR: Multi-Token Prediction for Fast and Long-Context LLM-Based Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We argue this trade-off is not inherent: unlike open-ended text generation, ASR outputs are strongly anchored to the input speech signal, providing a natural inductive bias toward high-parallelism decoding. Building on this, we introduce ParaASR, an ASR system that leverages Multi-Token Prediction (MTP) to let a 4B LLM decoder emit multiple tokens per forward step. |
QINGJIAN LIN et. al. | arxiv-cs.SD | 2026-07-31 |
| 114 | Curriculum-Based Noise Adaptation for Phoneme-to-Text Reconstruction in Visual Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Existing reconstruction approaches are commonly trained on clean phoneme sequences or synthetically corrupted inputs, leading to a mismatch between training conditions and the realistic phoneme prediction errors encountered during inference. To address this limitation, this paper proposes progressive error curriculum training (PECT). |
Matthew Kit Khinn Teng; Haibo Zhang; Takeshi Saitoh; | arxiv-cs.CL | 2026-07-30 |
| 115 | AgenticASR: Refining Speech Recognition in Real-World Scenarios Via An Agentic Approach Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Existing spoken-to-written methods process completed audio or transcripts but cannot revise emitted text when later speech changes how preceding content should be interpreted. We therefore formulate Agentic Speech Recognition (AgenticSR), an audio-to-clean-text task that removes disfluencies, resolves self-corrections, and normalizes written form while preserving the speaker’s final intent. |
ZIXUAN JIANG et. al. | arxiv-cs.AI | 2026-07-30 |
| 116 | A Comparative Analysis of Pretrained Wav2Vec XLSR-53 and Whisper-Small Models for Automatic Speech Recognition in The Telugu Language Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This study presents the development of an automatic speech recognition (ASR) system tailored for Telugu, one of the widely spoken Indian languages. |
JAGALINGAM PUSHPARAJ et. al. | Frontiers in Artificial Intelligence | 2026-07-30 |
| 117 | CAFAD: Common Acoustic Features for Adversarial Audio Detection Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In addition, few studies have explored the intrinsic characteristics of audio adversarial examples or how these characteristics can be leveraged for robust detection. To address this limitation, this paper proposes CAFAD, a plug-and-play detection framework that combines multi-domain acoustic feature fusion and temporal pyramid matching for variable-length adversarial audio detection. |
Wenjie Li; Pengyu Wei; Xuejing Yuan; Zizhuang Deng; Yuxuan Chen; | Cybersecurity | 2026-07-30 |
| 118 | Noise-Invariant Agentic Human–Robot Interaction GenAI System Using A Dual-Encoder Contrastive ASR Architecture and VLMs for Robot Control and Navigation in Acoustically Challenging Jobsites Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: The primary contribution of this research is the construction-tailored, noise-robust deep learning–based ASR architecture and data set curation, with the VLM control stack demonstrating the feasibility of d |
Oscar Poudel; Rayan H. Assaad; Mohamad Awada; | Journal of Computing in Civil Engineering | 2026-07-29 |
| 119 | Voice Memory for Agentic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present Voice Memory, a inference-only scheme for agentic speech recognition: at stream time, a frozen corrector reads a single per-domain memory.md and decides per utterance whether to act on the hypothesis or abstain and keep the 1-best. |
CHAO-HAN HUCK YANG et. al. | arxiv-cs.CL | 2026-07-28 |
| 120 | SpeechLLM Meets Federated Learning for End-to-End ASR: English and Italian Case Studies Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents the first systematic study of federated training for SpeechLLM-based end-to-end ASR systems. |
Mohamed Nabih Ali; Daniele Falavigna; Alessio Brutti; | arxiv-cs.CL | 2026-07-28 |
| 121 | The Genealogy of Large Language Models: From Auxiliary Tools in ASR to Foundational Transformers and Back Again Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: This paper traces the evolutionary trajectory of Large Language Models (LLMs), arguing that their origins lie in the practical need to correct transcription errors in Automatic … |
José Luciano Maldonado; | WSEAS TRANSACTIONS ON COMPUTERS | 2026-07-27 |
| 122 | MoLGE: Mixture of Language Group Experts for Efficient Scaling of Massively Multilingual Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: However, these models often encounter the curse of multilinguality, where model capacity is diluted across languages. To address this challenge, we propose Mixture of Language Group Experts (MoLGE), built upon speech self-supervised models (S3Ms). |
Sangmin Lee; Woojin Chung; Woongjib Choi; Hong-Goo Kang; | arxiv-cs.CL | 2026-07-27 |
| 123 | Indic DiarBench: A Multilingual Joint Diarization and ASR Benchmark for Indian Languages Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we introduce Indic DiarBench, a speaker diarization and ASR benchmark dataset spanning all 22 scheduled languages of India. |
Deovrat Mehendale; Aditya Mehndiratta; Dhruv Rathi; Kaushal Bhogale; Mitesh M. Khapra; | arxiv-cs.CL | 2026-07-26 |
| 124 | Earnings25: A Comprehensive 500-Hour Speech Benchmark for Finance Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce Earnings25, a finance-domain benchmark for evaluating automatic speech recognition (ASR) on English-language earnings calls under realistic conditions. |
DENGLIN JIANG et. al. | arxiv-cs.CL | 2026-07-26 |
| 125 | Real-Time Subtitling in The Streaming Era Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: The aim of the study was to identify the characteristics of real-time automatic subtitling systems in the context of global streaming platforms. |
Muratbek Toktagazin; Gulmira Amangeldiyeva; Lazzat Adilbekova; Rakhmet Kulaikhan; Sholpan Sagatova; | The International Journal of Communication and Linguistic … | 2026-07-24 |
| 126 | Transsion’s Speaker-Attributed Multilingual ASR System for The MLC-SLM 2026 Challenge Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose a cascaded framework consisting of three components: a speaker diarization module, a long-form multilingual ASR module, and a speaker-transcription fusion module. |
ZHECHENG REN et. al. | arxiv-cs.CL | 2026-07-24 |
| 127 | MEUSLI: A Multilingual Projector for LLM-based ASR and Beyond Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce MEUSLI, the first open-science multilingual projector family that links a Whisper encoder with open-source multilingual LLMs, enabling fully open-source end-to-end ASR in 28 European languages. |
Lorenzo Concina; Seraphina Fong; Marco Matassoni; Alessio Brutti; | arxiv-cs.CL | 2026-07-24 |
| 128 | Synthetic Speech, Real Signal: Paralinguistic Preservation and Cross-Lingual Augmentation Via Voice Cloning Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We then clone English clinical speech into Japanese and find that training on cloned data outperforms raw cross-lingual transfer for depression and anxiety detection on real Japanese speech, suggesting voice cloning is a promising direction for augmenting clinical speech data in low-resource languages. |
ROSELINE POLLE et. al. | arxiv-cs.LG | 2026-07-24 |
| 129 | DONDO: Open W2v-BERT Speech-Recognition Base Models for African Languages Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present DONDO, a family of open, permissively licensed automatic speech recognition (ASR) base models for African languages, built on the w2v-BERT 2.0 self-supervised speech encoder. |
Paul Azunre; | arxiv-cs.CL | 2026-07-23 |
| 130 | Transcription Policy As A Latent Variable: Activating Controllable Verbatim ASR with Word-Level Timing Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Modern ASR models trained on heterogeneously annotated data treat transcription style (verbatim vs. intended) as an uncontrolled latent variable, causing measurable decoding instability, evaluation confounding (up to 60% of reported WER attributable to style mismatch), and unreliable word-level timing. |
Laurin Wagner; Mario Zusag; Bernhard Thallinger; | arxiv-cs.CL | 2026-07-21 |
| 131 | Speech-GAN: A Black-Box Generative Adversarial Network Attack Against Automatic Speech Recognition Systems Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we present Speech-GAN, a generative adversarial network designed to perform black-box untargeted attacks against deep learning-based ASR systems by generating adversarial audio samples.We validate the effectiveness of Speech-GAN through 10,000 attack runs conducted on 1,000 audio samples spanning 10 command words, targeting theWav2Vec 2.0 ASR model. |
FARAZ MOHAMMAD MUSHTAK MOGAL et. al. | Vietnam Journal of Computer Science | 2026-07-21 |
| 132 | From A Multilingual Streaming ASR Backbone to Kenyan-Language Systems: Data-Centric Adaptation of Nemotron 3.5 for Kikuyu, Dholuo, and Kalenjin Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present an end-to-end engineering study adapting NVIDIA Nemotron 3.5 ASR Streaming 0.6B to Kikuyu, Dholuo, and Kalenjin. |
Mark Gatere; | arxiv-cs.CL | 2026-07-21 |
| 133 | Benchmarking Human and Automatic Speech Recognition of Diverse Speech: Initial Results Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present a preliminary comparison of the performances of state-of-the-art ASR systems and Dutch native listeners on the recognition of diverse speech, specifically Dutch child and older adults’ speech and Flemish. |
Ilse Huisman; Rares Popa; Yuanyuan Zhang; Odette Scharenborg; | arxiv-cs.CL | 2026-07-21 |
| 134 | Re-Sonance: A Dysarthric Asynchronous Real-Time Speech Conversion System Based on A Three-Stage Cascaded ASR-LLM-TTS Architecture Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents Re-Sonance, a novel LLM-enhanced speech-driven AAC system designed for real-time professional speaking scenarios. |
YUXUAN WU et. al. | arxiv-cs.SD | 2026-07-20 |
| 135 | Robust Assamese Speech Recognition Through Controlled Fine-Tuning of Whisper Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents a controlled, fine-tuned Whisper-based Assamese ASR system trained on the Mozilla Common Voice 24.0-Assamese corpus. |
GANAPATI DAS et. al. | arxiv-cs.LG | 2026-07-19 |
| 136 | Mixed Approach Speech-to-text Translation for Endangered Language Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Purpose This study aims to address the technological marginalization of endangered regional languages by evaluating speech-to-text translation for Dayak Ma’anyan, an extremely low-resource Austronesian language. |
BENYAMIN LANGGU SINAGA et. al. | Data Technologies and Applications | 2026-07-18 |
| 137 | RW-Voice-EQ Bench: A Real World Benchmark for Evaluating Voice AI Systems Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To this end, we introduce the Real World Voice EQ Bench, a multidimensional benchmark for evaluating voice AI across text-to-speech (TTS), speech-to-speech (STS), speech understanding (SU), and automatic speech recognition (ASR). |
DAVID AYLLON et. al. | arxiv-cs.SD | 2026-07-16 |
| 138 | An Omnilingual-ASR-Based Speech-LLM System for The 2nd MLC-SLM Challenge Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We describe our submission to Task 1 of the 2nd MLCSLM Challenge: a cascaded diarization-then-recognition system that combines DiariZen-Large-s80 (WavLM-Large) segmentation, CAM++ embedding-based two-speaker clustering, and a LoRA-adapted omniASR LLM 7B v2 recognizer, with no oracle segmentation or speaker labels at test time. |
Shuming Fang; Shuifei Zeng; | arxiv-cs.SD | 2026-07-14 |
| 139 | The High-Volume OPD Problem: Why Indian Clinical Documentation Requires A Purpose-Built Artificial Intelligence Model Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: However, existing commercial platforms are engineered for Western clinical ecosystems featuring 15-to-30-minute monolingual consultations. This narrative review evaluates the structural mismatches encountered when translating these architectures directly into the high-volume, short-duration, multilingual realities of Indian outpatient departments (OPDs). |
Anushtha Rakesh Chillure; | Journal of Medical Science and Clinical Research | 2026-07-13 |
| 140 | Casting Everything to Online API Services? A Survey of Integrating Localized Speech Recognition Models in Robotic Systems Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This article provides an overview of how ASR technologies are integrated into various intelligent robots and machines. |
Sheng Li; Jing Li; Felix Schijve; Jun Hu; Emilia Barakova; | arxiv-cs.RO | 2026-07-13 |
| 141 | Which Languages Transfer Best to Warlpiri? A Similarity-Based Study for Low-Resource ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose a framework combining acoustic similarity from pre-trained speech models with linguistic similarity based on typology, phoneme inventories, grammatical, and syntactic features to rank high-resource source languages and evaluate their effectiveness for ASR transfer to Warlpiri. |
Pravina Mylvaganam; Eliathamby Ambikairajah; Ting Dang; Vidhyasaharan Sethu; Tuende Szalay; | arxiv-cs.CL | 2026-07-11 |
| 142 | Generative Testing of Automated Speech Recognition Systems Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This work introduces GATAS, a black-box testing approach that generates failure inducing inputs by operating in the phoneme-level latent space of a text- to-speech model. |
Yanis Xabier Wilbrand Peña; Oliver Weißl; Andrea Stocco; | arxiv-cs.CR | 2026-07-10 |
| 143 | Best-of-$N$ TTS Evaluation Is Confounded By ASR Family Alignment Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose two \textbf{cross-family rank ensembles} (rank-averaging and conjunctive max-rank) that attain the lowest mean WER across three independent evaluators — $1.61\%$ at $N{=}10$ ($-12\%$ relative to F5-TTS) — with no measurable degradation under automatic SIM-o/UTMOS metrics; the best single verifier drives WER from $2.06\%$ to $1.72\%$ ($-16.5\%$) under the official F5-TTS evaluator. |
Taehyung Yu; Seongjae Kang; | arxiv-cs.CL | 2026-07-09 |
| 144 | Diarization-Guided Qwen-ASR Adaptation for Multilingual Two-Speaker Conversational Speech Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper describes our self-designed system for Task 1 of the MLC-SLM 2026 Challenge for multilingual two-speaker conversational speech. |
Hao Wu; RongQi Han; Zhen Wang; Wei Liang; Wei Xu; | arxiv-cs.CL | 2026-07-09 |
| 145 | VSRo-200: A Romanian Visual Speech Recognition Dataset for Studying Supervision and Multimodal Robustness Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce VSRo-200, the first large-scale dataset for visual speech recognition (lip reading) in Romanian, comprising 200 hours of real-world podcast videos. |
Iulia-Maria Udrea; Alexandra Diaconu; Bogdan Alexe; | arxiv-cs.CV | 2026-07-09 |
| 146 | COALA: Robust Contextualized Speech-augmented Language Modeling for ASR Via Contrastive Regularizer and Biasing Score Estimation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we propose COALA (Contextualized ASR Leveraging Biasing Scoring), a robust framework designed to enhance speech-augmented language models (SLMs) in complex multi-entity scenarios. |
Jhih-Rong Guo; Bi-Cheng Yan; Tien-Hong Lo; Berlin Chen; | arxiv-cs.CL | 2026-07-09 |
| 147 | Best-of-N TTS Evaluation Is Confounded By ASR Family Alignment Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Best-of-$N$ (BoN) inference improves content consistency in zero-shot text-to-speech by selecting among multiple candidates with an automatic speech recognition (ASR) verifier. We … |
Taehyung Yu; Seongjae Kang; | ArXiv | 2026-07-09 |
| 148 | Gradient-Based Speech-to-Text Alignment for Any ASR Model: From CTC to Speech LLMs Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We evaluate it on sixteen models from four families, on read (TIMIT) and spontaneous (Buckeye) speech, each against the model’s own native or attention-based alignment. |
Albert Zeyer; Ralf Schlüter; Hermann Ney; | arxiv-cs.CL | 2026-07-07 |
| 149 | From Sinhala to Dhivehi: Cross-Lingual Transfer Learning for Low-Resource Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This study investigates whether cross-lingual transfer learning from Sinhala, a linguistically related, relatively well-resourced Insular Indo-Aryan language, can improve Dhivehi ASR. |
Lukmal Ilyas; Nevidu Jayatilleke; | arxiv-cs.CL | 2026-07-07 |
| 150 | REDDIT: Correcting Model-Generated Timestamp Drift in ASR Without Forgetting Via Replay-Based Distribution Editing Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Naive timestamp-corrected fine-tuning improves alignment but can severely degrade non-target ASR behavior, exposing a forgetting problem. We propose REDDIT(REplay-based Distribution eDITing), a lightweight two-stage post-training framework that corrects timestamps while avoiding this catastrophic forgetting: it first edits timestamp targets under the model’s own replayed decoder context while matching the frozen base distribution on non-timestamp tokens, then applies a short edited-prefix refinement stage. |
Cheng-Kang Chou; Ming-To Chuang; Ke-Han Lu; Chan-Jan Hsu; Hung-yi Lee; | arxiv-cs.CL | 2026-07-06 |
| 151 | SPEARBench: A Benchmark for Naturalness Evaluation in Streaming Speech-to-Speech Language Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce SPEARBench, a benchmark for evaluating naturalness in speech-to-speech language models from question-answer interactions. |
THOMAS THEBAUD et. al. | arxiv-cs.CL | 2026-07-06 |
| 152 | Revisiting The Relation Between Language Model Perplexity and ASR Word Error Rate for Modern End-to-End Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper revisits the relation between PPL and WER for modern ASR systems. |
MOHAMMAD ZEINELDEEN et. al. | arxiv-cs.CL | 2026-07-06 |
| 153 | Evaluating The Effect of Linguistic Relatedness on Cross-Lingual Transfer in Large Multilingual Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Although this strategy has shown meaningful improvements in small ASR models, its effectiveness in large ASR remains unclear. We extend this framework to large multilingual ASR through a systematic controlled experimental design spanning six factors, two Africa-centric corpora, and four large ASR models, isolating whether linguistic relatedness reliably predicts cross-lingual transfer gains in this setting. |
ANDREI FLORIAN et. al. | arxiv-cs.CL | 2026-07-06 |
| 154 | Progressive Refinement: An Iterative Pseudo-Labeling Approach for Mandarin-English Code-Switching ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Code-switching (CS), alternating languages within the same utterance, poses significant challenges for automatic speech recognition (ASR) due to limited CS training data. This paper applies an iterative pseudo-labeling training approach to CS-ASR for the first time, demonstrating its effectiveness in leveraging unlabeled data to improve CS-ASR performance. |
Qu Yang; Cakra Wardhana; Tim Ng; | arxiv-cs.CL | 2026-07-06 |
| 155 | Context-Aware ASR for Mandarin Technical Lectures Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Character error rate (CER) therefore hides their recognition failures. We study whether lecture context helps recognize these terms. |
Ho-Lam Chung; Yiming Chen; Hung-yi Lee; | arxiv-cs.SD | 2026-07-06 |
| 156 | Dual-Form ASR: Semantics-Aware Inverse Text Normalization for Chinese Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we propose Dual-Form ASR (DF-ASR), a framework that extends spoken-form ASR capability to semantics-aware written-form ITN through paired spoken-form and written-form supervision while retaining prompt-level selection between transcript forms. |
FENGRUN ZHANG et. al. | arxiv-cs.CL | 2026-07-05 |
| 157 | S-DiverSe: Spanish Diverse Speech Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present S-DiverSe (Spanish Diverse Speech), a corpus of 3.2 hours of in-the-wild Spanish speech from 22 speakers with amyotrophic lateral sclerosis, Parkinson’s disease, and stroke. |
FERNANDO LÓPEZ et. al. | arxiv-cs.CL | 2026-07-03 |
| 158 | Jointly Improving Dialect Identification and ASR in Indian Languages Using Multimodal Feature Fusion Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we propose a multimodal framework that jointly improves ASR and DID. |
Saurabh Kumar; Prasanta Kumar Ghosh; | arxiv-cs.CL | 2026-07-02 |
| 159 | NAVER LABS Europe Submission to The Instruction-following 2026 Short Track Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we describe NAVER LABS Europe’s submission to the instruction-following speech processing short track at IWSLT 2026. |
Marcely Zanon Boito; Hemant Yadav; Jean-Luc Meunier; Ioan Calapodescu; | arxiv-cs.CL | 2026-07-02 |
| 160 | Rethinking Speech-LLM Integration for ASR: Effective Joint Speech-Text Training By Interleaving Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We observe that as supervised ASR training data increases, the contribution of LLM priors becomes less evident, and simple speech-text joint training under-utilizes textual knowledge. We therefore propose Joint Speech-Text Interleaved Pretraining (JSTIP), an ASR-oriented pretraining strategy that constructs word-level and segment-level interleaved speech-text sequences within aligned pairs for speech-LLM architectures that accept continuous inputs. |
RUCHAO FAN et. al. | arxiv-cs.CL | 2026-07-02 |
| 161 | Reinforcement Learning for Data-Efficient Code-Switched ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose a practical reinforcement learning with verifiable rewards recipe for data-efficient adaptation of audio-language models to code-switched ASR using group relative policy optimization, combining an error rate reward with a script fidelity reward that penalizes wrong writing systems and a two-pass draft-and-refinement procedure. |
Ziwei Ye; Peter Vickers; | arxiv-cs.CL | 2026-07-02 |
| 162 | From Monolingual to Multilingual: Evaluating Mamba for ASR in South African Languages Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we evaluate Mamba for ASR on seven South African languages. |
Jesujoba O. Alabi; Julian Herreilers; Badr M. Abdullah; Dietrich Klakow; | arxiv-cs.CL | 2026-07-01 |
| 163 | H-SAGE: Holistic Speaker-Aware Guided Experts for MoE-based Multi-Talker ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: While recent Mixture-of-Experts (MoE) approaches have shown promise, they typically rely on frame-independent routing that leads to temporal myopia, and depend solely on the downstream ASR objective, which results in implicit and ungrounded representation learning. To address these limitations, we propose Holistic Speaker-Aware Guided Experts (H-SAGE) for MoE-based MTASR. |
Yujie Guo; Jiaming Zhou; Yuhang Jia; Yang chen; Yong Qin; | arxiv-cs.SD | 2026-07-01 |
| 164 | Gated Multi-Graph Fusion Via Graph Attention Networks for Alzheimer’s Disease Detection Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose a Multi-View Gated Graph Attention Network that transcribes audio via Automatic Speech Recognition (ASR) to construct semantic, dependency, and co-occurrence graphs, characterizing speech through a content-structure-flow framework. |
JINYU LI et. al. | arxiv-cs.CL | 2026-06-30 |
| 165 | Adapting Foundation ASR Models to Dysarthric Speech: A Case Study Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents a personalized ASR system for a dysarthric speaker, built by adapting a foundation ASR model to speaker-specific data. |
Christian Huber; Laura Kernahan; Alexander Waibel; | arxiv-cs.CL | 2026-06-30 |
| 166 | What Counts As An Error? Dual-Reference Benchmarking for Atypical ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Through this analysis, we highlight the importance of selecting a suitable transcription reference for valid model selection depending on the use-case, particularly for atypical ASR. |
Hawau Olamide Toyin; Srinivasan Umesh; Hanan Aldarmaki; | arxiv-cs.CL | 2026-06-30 |
| 167 | Building An ASR Solution for Training and Assessing Children’s Reading Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present an open-source system for assessing children’s reading in Bambara, developed through an end-to-end process linking field data collection, benchmark construction, model adaptation, a reading application, and classroom validation. |
Yacouba Diarra; Nouhoum Souleymane Coulibaly; Mamadou Dembele; Aymane Dembele; Michael Leventhal; | arxiv-cs.CL | 2026-06-30 |
| 168 | Comparing Human and Automatic Recognition of Dutch Dysarthric Continuous Speech: A Case Study Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In our goal to develop personalised dysarthric speech recognition (DSR) models, this study compared the recognition performances of human listeners and those of three state-of-the-art, off-the-shelf ASR systems (Whisper-large-V3, Google Chirp 3, and Omnilingual) on the recognition of Dutch continuous read and spontaneous speech from a single speaker with severe dysarthria. |
Yuanyuan Zhang; Dimme de Groot; Jorge Martinez; Odette Scharenborg; | arxiv-cs.CL | 2026-06-29 |
| 169 | Position-Aware Target Speaker Extraction for Long-Form Multi-Party Conversations: A Diarization-Free Framework for ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Motivated by the stability of speakers’ directions of arrival (DOAs) in meetings, we propose PATSE, a multi-channel Position-Aware Target Speaker Extraction front-end that uses DOA as a spatial prior to directly extract the speech of each target speaker. |
Yichi Wang; Junzhe Chen; Wangjin Zhou; Tatsuya Kawahara; | arxiv-cs.SD | 2026-06-28 |
| 170 | Preference-ASR: A Preference-Aware Test Set for Benchmarking ASR in The Era of Speech LLMs Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce PreferenceASR, a test set evaluating ASR systems on their ability to follow natural-language preference instructions across four categories: normalization, entities, disfluencies, and case. |
NITHIN RAO KOLUGURI et. al. | arxiv-cs.CL | 2026-06-28 |
| 171 | How to Leverage Synthetic Speech for LLM-Based ASR Systems? Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Yet a persistent distributional gap between synthetic and real data limits how far it can replace genuine recordings. Prior work largely treats this gap as a black box to be engineered around, but in our work, we instead examine its origin directly by probing a SLAM-ASR architecture. |
YANIS LABRAK et. al. | arxiv-cs.CL | 2026-06-27 |
| 172 | LCMA-SRT: Language-Conditional Mixture-of-Experts Adapters for Joint Multilingual Speech Recognition and Translation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose LCMA-SRT (Language-Conditional Mixture-of-Experts Adapters for Speech Recognition and Translation), which augments a hierarchical transducer with language-conditional Mixture-of-Experts (MoE) adapters. |
Nanjie Li; Xiaoyong Guo; Hao Huang; Xu Haihua; Wei Shi; | acl | 2026-06-27 |
| 173 | Difference in Task Performance on Sparse Speech Representations Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Mainly motivated by the no-free-lunch theorem and speech production, this work investigates changes in task performance in sparse speech representations, providing interpretability analysis under the Information Bottleneck (IB) framework. |
Wenjie Peng; Chen Chen; Thomas Hain; | acl | 2026-06-27 |
| 174 | Graph-Based Phonetic Error Correction of Noisy ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose a structured ASR correction framework, that we call G-SPIN, that combines phonetic graph modeling with contextual language understanding. |
Pratik Rakesh Singh; Mohammadi Zaki; Aneesh Mukkamla; Pankaj Wasnik; | acl | 2026-06-27 |
| 175 | An Exploration of Mamba for Speech Self-Supervised Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: While Mamba has demonstrated strong performance in language modeling, its potential as a speech self-supervised learning (SSL) model remains underexplored, with prior studies limited to isolated tasks. To address this, we explore Mamba-based HuBERT models as alternatives to Transformer-based SSL architectures. |
TZU-QUAN LIN et. al. | acl | 2026-06-27 |
| 176 | Listening Like Humans: Semantics-Guided Noise-Robust Multimodal Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Inspired by human speech comprehension, we propose Speech-MLM, a novel multimodal framework that reframes ASR as semantics-guided speech reconstruction. |
YAN FANG et. al. | acl | 2026-06-27 |
| 177 | Mixture-of-Experts with Intermediate CTC Supervision for Accented Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce MoE-CTC, a Mixture-of-Experts architecture with intermediate CTC supervision that jointly promotes expert specialization and generalization. |
Wonjun Lee; Hyounghun Kim; Gary Lee; | acl | 2026-06-27 |
| 178 | POWSM: A Phonetic Open Whisper-Style Speech Foundation Model IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we introduce POWSM (Phonetic Open Whisper-style Speech Model), the first unified framework capable of jointly performing multiple phone-related tasks. |
CHIN-JOU LI et. al. | acl | 2026-06-27 |
| 179 | LoRA-Tuned Large Language Models for Dementia Detection Via Multi-View Speech-Derived Features Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose a low-rank adaptation (LoRA)-tuned large language model (LLM) that performs structured multi-view reasoning over four complementary speech-derived signals: ASR transcripts with pause markers, discourse-level topic cues, temporal fluency statistics, and phonological sequences. |
Jonghyeon Park; Olivier Jiyoun Jung; Myungwoo Oh; | arxiv-cs.SD | 2026-06-26 |
| 180 | AI Based Doctor-Patient Diarization System Using Speech Recognition and NLP with SOAP Notes Generation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper proposes an AI-based Automated Clinical Documentation System that converts doctor-patient conversations into structured medical records. |
Prof.Shruti Hatwar Prof.Shruti Hatwar; Atharv Borhade Atharv Borhade; Yash Dhole Yash Dhole; Vishal Wagh Vishal Wagh; Sanskar Darekar Sanskar Darekar; | International Scientific Journal of Engineering and … | 2026-06-26 |
| 181 | SamaVaani: Auditing and Debiasing Multilingual Clinical ASR for Indian Languages Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To this end, we propose SamaVaani, a unified debiasing technique that simultaneously improves ASR performance and improves fairness across demographic groups. |
SUBHAM KUMAR et. al. | arxiv-cs.CL | 2026-06-25 |
| 182 | Enhancing BEST-RQ Pseudo-Label Quality Through Online Refinement for Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we improve online pseudo-label generation while preserving simplicity. |
JINGJING XU et. al. | arxiv-cs.SD | 2026-06-24 |
| 183 | HALAS: A Human-Annotated Dataset of Hallucinations of Modern ASR Systems Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce HALAS, the first human-annotated dataset of naturally occurring hallucinations from seven state-of-the-art ASR models on real unprocessed earnings call recordings. |
Mateusz Barański; Jan Jasiński; Julitta Bartolewska; Marcin Witkowski; Konrad Kowalczyk; | arxiv-cs.SD | 2026-06-22 |
| 184 | EmotionAI: A Privacy-Preserving Computational Intelligence Pipeline for Speech-Emotion-Grounded Conversational Analysis Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Reviewing recorded interviews for affective cues such as composure, hesitation and agitation is slow and subjective, and cloud services that could automate it require sensitive … |
Wai Laam Mak; Isibor Kennedy Ihianle; Pedro Machado; | arxiv-cs.SD | 2026-06-22 |
| 185 | Layer-wise Probing of Wav2vec 2.0 and Whisper for Consonant Cluster Reduction in African American English Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Self-supervised and supervised speech models are increasingly used to investigate which linguistic information their internal representations encode, and at what level of abstraction they encode it. |
Hamid Mojarad; Kevin Tang; | arxiv-cs.CL | 2026-06-22 |
| 186 | ATCCaps: A Call-Sign-Aware Speech Dataset for Air Traffic Control Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents ATCCaps, a call-sign-aware ATC speech dataset with caption-level audio-text supervision. |
Dongdong Li; Jianwei Song; Jianwei Wang; Zhe Wang; | arxiv-cs.SD | 2026-06-21 |
| 187 | Scaling Audio Models Efficiently: A Joint Study of Compute Constraints and Optimization Behavior Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we investigate the tradeoffs between compute allocation and model performance for two speech processing tasks: Automatic Speech Recognition (ASR) and Speech Emotion Recognition (SER). |
Vyom Agarwal; Mokshda Gangrade; Siddharth Pal; Jerry Wu; | arxiv-cs.SD | 2026-06-21 |
| 188 | Synthetic Audio Generation Framework for Air Traffic Control Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Automatic Speech Recognition (ASR) systems, despite achieving remarkable accuracy in general-purpose domains with native speech (L1), struggle in domains like Air Traffic Control (ATC) due to strong channel noise, a presence of non-native (L2) English accents, and data scarcity. We propose a synthetic data generation pipeline with acoustical properties simulations specifically designed to address this lack of real data to improve recognition accuracy in the ATC domain. |
Raphaël Bagat; Zhe Zhang; Junichi Yamagishi; Irina Illina; Emmanuel Vincent; | arxiv-cs.CL | 2026-06-19 |
| 189 | DisSpeech: Low-Resource Controllable Mandarin Stuttered Speech Synthesis for ASR Augmentation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This problem is further aggravated in Mandarin scenarios by the limited availability of stuttered speech data, which makes it difficult to train robust ASR models for diverse disfluency patterns. To address this problem, this paper proposes DisSpeech, a discrete speech token-based framework for low-resource controllable Mandarin stuttered speech synthesis and ASR data augmentation. |
Yao Lu; | arxiv-cs.SD | 2026-06-19 |
| 190 | CORTIS: Text-Only Adaptation of Spoken Language Models for Task-Oriented Voice Agents Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Spoken language models (SLMs) offer a direct speech-based alternative, yet adapting them to new tasks typically requires paired speech-target annotations. Motivated by this gap, we present CORTIS, a text-only adaptation framework for task-oriented voice agents. |
Youngwon Choi; Hyeonyu Kim; Taeyoun Kwon; Donghyuk Jung; Myeongkyun Cho; | arxiv-cs.HC | 2026-06-19 |
| 191 | OpenWER: Improving Cross-Lingual ASR Evaluation and Enabling Token-Based Accuracy Metrics Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present OpenWER, an open-source implementation that improves WER robustness through language-specific normalisation and compound word detection. |
Korbinian Kuhn; Gottfried Zimmermann; | arxiv-cs.CL | 2026-06-19 |
| 192 | Self-Supervised Learning for Speech Recognition: A Review Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We included each aspect that could affect an ASR’s performance in our investigation. |
Humam Khalid Jameel; Assad H. Thary Al Ghrairi; Mohammed M. Neamah; | Dijlah Journal of Engineering Sciences ISSN: 3078-9664, … | 2026-06-17 |
| 193 | Responsible ASR: Overcoming Challenges of Foundational Models in Narrow-Band and Low-Resource Settings Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper evaluates the performance of widely used foundational automatic speech recognition (ASR) models — both open-source and commercial — on narrow-band conversations in Hindi, a low-resource language, and Indian-accented English, a low-resource accent. |
Tejas Godambe; Nutan Choudhary; Sanket Shah; Nagaraj Adiga; Sharath Adavanne; | arxiv-cs.SD | 2026-06-17 |
| 194 | A Comparative Study of Pretrained Transformer Models for Quranic ASR: Speech Representations, Label Formats, and Dataset Composition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents a systematic empirical study of domain-specific fine-tuning of pretrained Transformer-based models for Quranic ASR, using advanced speech feature extraction methods: Wav2Vec2.0, HuBERT, and XLS-R. |
Nabil Mosharraf Hossain; Riasat Islam; Unaizah Obaidellah; | arxiv-cs.AI | 2026-06-17 |
| 195 | A Hybrid RWHISYMP Speech-to-Text Noise Suppression Model: Integration of The Whisper Base Model, RNNoise, and SympSpell Algorithms Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This study introduces the RWhiSymp, a hybrid speech-to-text noise suppression model that integrates three components: RNNoise for noise suppression, Whisper Base Model for ASR, and SymSpell for spelling correction. |
Mariann F. Bragas; Laurence D. Ganda; Leonila R. Juanatas MIT; Charisse S. Ronquillo MIT; | American Journal of Smart Technology and Solutions | 2026-06-16 |
| 196 | ASTRA: A Scalable Next-Generation ATCO Training Simulator with Autonomous Simpilots Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Existing automated solutions rely on Western-centric speech models that perform poorly in Singaporean operational contexts, with off-the-shelf systems exhibiting Word Error Rates (WER) of up to 107.80% on Singaporean-accented aviation speech. We introduce ASTRA, an end-to-end training simulator that automates these simpilot roles through a pipeline that transcribes ATCO speech, interprets instructions, and generates appropriate pilot and ATCO responses using locally adapted voice models. |
ETHAN CHEW et. al. | arxiv-cs.LG | 2026-06-16 |
| 197 | Low-resource Language Discrimination Towards Chinese Dialects with Transfer Learning and Data Augmentation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this article, we develop a novel Chinese dialects discrimination framework with transfer learning and data augmentation (CDDTLDA) in order to overcome the shortage of resources. |
Fan Xu; Yangjie Dan; Keyu Yan; Yong Ma; Mingwen Wang; | arxiv-cs.CL | 2026-06-16 |
| 198 | When Multiple Scripts Matter: Evaluating ASR in Clinical Settings Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Conventional string-matching evaluation metrics often underestimate ASR performance by treating orthographic variants as errors. To address this issue, we introduce MultiClin, a clinical ASR benchmark designed to evaluate robustness to multiscript variability. |
JEAN SEO et. al. | arxiv-cs.CL | 2026-06-16 |
| 199 | Improving Low-resource ASR Using Bilingual Fine-tuning with Language Identification: A Cross-linguistic Evaluation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We evaluate this method across nine linguistically and geographically diverse language pairs, covering a range of language families and writing systems. |
REIHANEH AMOOIE et. al. | arxiv-cs.CL | 2026-06-16 |
| 200 | Are You Speaking My Languages? On Spoken Language Adherence in Multimodal LLMs Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To preserve flexibility and code-switching capabilities, we propose a soft prompting approach that hints at potential spoken languages without strictly constraining the output. |
Hyungwon Kim; Kandarp Joshi; Lillian Zhou; Pavel Golik; Petar Aleksic; | arxiv-cs.CL | 2026-06-15 |
| 201 | Building and Evaluating A Synthetic Bengali Speech Resource for Telecom Customer Care Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present a synthetic Bengali speech dataset for telecom customer-care scenarios. |
Kawshik Kumar Paul; Md. Nafiul Alam Fuji; | arxiv-cs.CL | 2026-06-14 |
| 202 | Improving Code-Switching ASR with Code-Mixing Guided Synthetic Speech Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper proposes a code-mixing guided preference-learning framework that steers synthetic speech generation toward improved code-switching fidelity using the Code Mixing Index (CMI). |
YUE HENG YEO et. al. | arxiv-cs.SD | 2026-06-14 |
| 203 | ROMPAR: Morphological Completion and Demographic Unlearning for Romanian-Accented Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To build a robust ASR system, we propose a multi-task adversarial training framework that enforces demographic invariance across age, gender, and dialect. |
ANDREI-MARIUS AVRAM et. al. | arxiv-cs.CL | 2026-06-14 |
| 204 | A Practical Evaluation Method for Long-Form Simultaneous Speech-to-Speech Translation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present a practical evaluation method for long-form SimulS2ST. |
Yulin Xue; Siqi Ouyang; Lei Li; | arxiv-cs.CL | 2026-06-12 |
| 205 | Learning to Hear Hesitation: Continual Learning for Disfluency-Aware ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Through a detailed analysis of model dynamics during training, we identify a trade-off between marker learning and ASR performance, and a consistent cross-attention head mechanism shared across CL methods. |
Henri-Leon Kordt; Theresa Pekarek Rosin; Jae Hee Lee; Stefan Wermter; | arxiv-cs.CL | 2026-06-12 |
| 206 | MoDiCoL: A Modular Diagnostic Continual Learning Dataset for Robust Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we argue that model robustness can be treated as a dynamic capability that continually develops, and we introduce MoDiCoL, a Modular Diagnostic Continual Learning dataset designed for controlled analysis of linguistic content, speaker characteristics, and acoustic environments. |
Theresa Pekarek Rosin; Matthias Kerzel; Stefan Wermter; | arxiv-cs.CL | 2026-06-12 |
| 207 | Listening with Attention: Entropy-Guided Explainability for Transformer-Based Audio Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose Listening with Entropy-guided Attention for Faithful explainability (LEAF-X), a model-intrinsic XAI framework for transformer-based ASR. |
Ravi Ranjan; Utkarsh Grover; Xiaomin Lin; Agoritsa Polyzou; | arxiv-cs.SD | 2026-06-12 |
| 208 | Ontology Memory-Augmented ASR Correction for Long Text-Speech Interleaved Conversations Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In such contexts, sparse correction evidence can be difficult to locate amid redundancy and noise. Addressing these challenges, we propose an ontology memory-augmented ASR correction framework for long text-speech interleaved conversations. |
XINXIN LI et. al. | arxiv-cs.CL | 2026-06-11 |
| 209 | Towards Personalized Federated Learning for Dysarthric Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To this end, this paper explores two aggregation strategies to achieve personalization, including the parameter-based averaging strategy and the embedding-based averaging strategy. |
Tao Zhong; Mengzhe Geng; Jiajun Deng; Shujie Hu; Xunying Liu; | arxiv-cs.SD | 2026-06-11 |
| 210 | PiDA: Phonetically-Informed Data Augmentation for Robust Vietnamese Speech Translation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We confirm that most ASR substitution errors arise from phonetic confusions rather than random noise, and that these phonetic errors significantly degrade ST quality. Motivated by this finding, we propose Phonetically-Informed Data Augmentation (PiDA), which generates ASR-like corruptions by substituting words with phonetically similar alternatives using phonetic word embeddings. |
GIANG SON NGUYEN et. al. | arxiv-cs.CL | 2026-06-11 |
| 211 | Rethinking Text-based Extractive Speech Summarization in Noisy ASR Settings for Low-resource Language Related Papers Related Patents Related Grants Related Venues Related Experts View Save |
Priyanjana Chowdhury; Sanghamitra Nath; Utpal Sharma; | Multimedia Tools and Applications | 2026-06-10 |
| 212 | Evaluating Bias in Phoneme-Based Automatic Speech Recognition Systems: An Analysis of IPA Transcription Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this study, we evaluate the performance of two state-of-the-art open-source ASR systems, WhisperIPA and ZIPA, that generate IPA transcriptions across diverse accents and language sources. |
Catherine Bao; Maneesha Rani Saha; Neal Patwari; | arxiv-cs.CL | 2026-06-10 |
| 213 | Speaker Group Encoding in Self-supervised Speech Recognition Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We investigate what self-supervised speech recognition models (S3Ms) learn about speaker groups (SGs). |
Felix Herron; Solange Rossato Alexandre Allauzen; Benoit Favre; François Portet; | arxiv-cs.CL | 2026-06-09 |
| 214 | Enhancing Multilingual LLM-based ASR with Mixture of Experts and Dynamic Downsampling Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To this end, this work proposes a projector-based LLM-ASR framework targeting the key challenges of multilingual generalization and modality alignment. |
Guodong Lin; Ziqi Chen; Yuxiang Fu; Ke Li; Wei-Qiang Zhang; | arxiv-cs.SD | 2026-06-09 |
| 215 | AuRA: Internalizing Audio Understanding Into LLMs As LoRA Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: While these routes respectively reuse strong pretrained components, enable native speech-language interaction, or offer lightweight adaptation, they often suffer from transcript-interface latency, costly multimodal training, or sequential speech-language coupling. To address these limitations, we present AuRA, a method that distills audio encoding capability into the LLM. |
BO CHENG et. al. | arxiv-cs.LG | 2026-06-09 |
| 216 | Speech Meets ELF: Audio Conditional Continuous-Target Diffusion for Speech Recognition and Translation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In contrast, continuous-target language modelling performs generation in a continuous space, yet its potential for S2T remains unexplored. To bridge this gap, we propose ELF-S2T, an audio-conditioned continuous-target generative model for S2T. |
XUANCHEN LI et. al. | arxiv-cs.SD | 2026-06-08 |
| 217 | Is Text All You Need? Text As A Universal Information Bottleneck for Speech LLMs Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we propose Convex Gate (C-Gate), a speech-to-LLM bridge that constrains all speech representations to lie within the LLM’s input embedding manifold with an architectural convex-hull constraint. |
MING-HAO HSU et. al. | arxiv-cs.CL | 2026-06-08 |
| 218 | TRADE: Transducer-Augmented Decoder for Speech LLM Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose TRADE TRansducer-Augmented DEcoder, which augments a multimodal LLM with a transducer branch that shares the audio encoder and uses the LLM’s hidden states directly as the prediction network — coupling frame-synchronous acoustic alignment with the LLM’s linguistic reasoning. |
Yun Tang; Shanil Puri; Shinji Watanabe; Subhabrata Mukherjee; | arxiv-cs.CL | 2026-06-07 |
| 219 | Contrastive Training with LLM-generated Near-Misses for Robust Code-Switching Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Code-switching (CS), the alternation between multiple languages within a single utterance, remains challenging for Automatic Speech Recognition (ASR). To address this issue, we propose a Point-of-Interest (POI)-aware contrastive training framework that improves recognition at CS-critical regions. |
TUNG X. NGUYEN et. al. | arxiv-cs.CL | 2026-06-05 |
| 220 | Enhancing Multilingual Automatic Speech Recognition for Low-resource Code-switched Languages: A Scalable Data Augmentation Strategy Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: The proposed system aims to generate natural-sounding Egyptian Colloquial Arabic (ECA) speech and improve Automatic Speech Recognition (ASR) performance, particularly for code-switched utterances. To achieve this, we collected data from transcribed online videos to fine-tune a text-to-speech (TTS) model (XTTSv2) for ECA. |
Mohab Mostafa Morsi; Radwa Fathalla; Sherif Abdou; Mohamed Waleed Fakhr; | PeerJ Computer Science | 2026-06-04 |
| 221 | Hearing The Unspoken: Language Model Priors for Acoustic Adversarial Attacks Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Ultimately, this work reveals how common, low-latency LLM tooling can be exploited to systematically subvert real-time ASR pipelines. |
Jiani Xie; Andrew C. Cullen; Paul Montague; Benjamin I. P. Rubinstein; | arxiv-cs.LG | 2026-06-04 |
| 222 | FiLM-Based Speaker Conditioning of A SpeechLLM for Pathological Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We investigate speaker conditioning via Feature-wise Linear Modulation (FiLM), injecting x-vector-derived information into each transformer layer of a frozen ASR encoder to adapt internal representations to individual pathological speakers without modifying base model weights. |
Fernando López; Santosh Kesiraju; Jordi Luque; | arxiv-cs.CL | 2026-06-04 |
| 223 | Beyond Waveform Robustness: Robust Feature-Vocoder Adversarial Attacks on Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we propose a Clean-Referenced Feature-Vocoder Attack, a surrogate-based black-box attack that moves the adversarial search space from raw waveforms to self-supervised learning (SSL) representations. |
YIFAN LIAO et. al. | arxiv-cs.SD | 2026-06-04 |
| 224 | Leveraging Soft Distributions of SSL-Derived Discrete Speech Tokens for Downstream Inference Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we propose to apply soft token assignment only during downstream inference. |
Kentaro Onda; Satoru Fukayama; Daisuke Saito; Nobuaki Minematsu; | arxiv-cs.SD | 2026-06-04 |
| 225 | Towards Truly Multilingual ASR: Generalizing Code-Switching ASR to Unseen Language Pairs Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we investigate whether CS capabilities learned from a limited set of seen language pairs can generalize to unseen language pairs through model merging and domain generalization methods. |
Gio Paik; Hyunseo Shin; Soungmin Lee; | arxiv-cs.CL | 2026-06-04 |
| 226 | Test-Time Compute Scaling for ASR with Depth-Conditioned Looped Transformers Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce LARM, a depth-conditioned looped Transformer that turns recurrent encoder depth into a controllable test-time compute axis. |
YACOUBA KALOGA et. al. | arxiv-cs.LG | 2026-06-03 |
| 227 | Efficient ASR Training with Conversations That Never Happened Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose an augmentation pipeline that generates scenario-level dialogues with participant metadata, maps speaker attributes to TTS voice profiles, and assembles synthesized utterances into speaker-aware simulated conversations. |
Máté Gedeon; Péter Mihajlik; | arxiv-cs.CL | 2026-06-02 |
| 228 | BaltiVoice: A Speech Corpus and Fine-tuned Whisper ASR System for The Balti Language Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present BaltiVoice, a 16.8-hour read-speech corpus for Balti (ISO 639-3: bft), a Tibetic language spoken in Gilgit-Baltistan, Pakistan, with no prior publicly available ASR resources. |
Muhammad Ali; | arxiv-cs.CL | 2026-06-02 |
| 229 | WAXAL-NET: Finetuned Edge ASR Across 19 African Languages Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We evaluate whether compact domain-specialized ASR models can outperform massively multilingual foundation models for conversational African speech across 19 languages in the WAXAL corpus. |
VICTOR TOLULOPE OLUFEMI et. al. | arxiv-cs.CL | 2026-06-01 |
| 230 | SN-WER: Script-Normalized WER for Multi-Script Indic ASR Evaluation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose Script-Normalized WER (SN-WER), a training-free, evaluation-only scoring method that transliterates both reference and hypothesis text into a language-specific canonical script before computing WER. |
Priyaranjan Pattnayak; | arxiv-cs.CL | 2026-06-01 |
| 231 | MURMUR: An Efficient Inference System for Long-Form ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Long-context ASR models resolve everything in a single pass for better accuracy, but are an order of magnitude slower. We propose Murmur, an inference system that overcomes this trade-off by operating at two levels. |
Wei-Tzu Lee; Keisuke Kamahori; Baris Kasikci; | arxiv-cs.LG | 2026-05-31 |
| 232 | Spiking and Event-driven Neuromorphic Mamba Models for Efficient Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce an event-driven SpeechMamba with FATReLU activation, achieving over 60% activation sparsity with less than 1% accuracy degradation on LibriSpeech. |
Tauseef Ahmed; Tao Sun; Jeronimo Castrillon; Kanishkan Vadivel; Guangzhi Tang; | arxiv-cs.NE | 2026-05-31 |
| 233 | A Deep Learning Framework for Speech Emotion Recognition: A Gender-Aware Hierarchical Pipeline with Optimized 18-Layer Convolutional Neural Network Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Real-world acoustic signals are heavily influenced by environmental noise, speaker idiosyncrasies, and physical variability across genders. This paper introduces a high-performance, structurally optimized hierarchical framework that addresses these limitations through three primary contributions: (1) a dense 182-feature extraction pipeline unifying spectral, linear predictive, dynamic energy, prosodic, and statistical shape profiles; (2) an early-stage, gender-aware hierarchical pipeline driven by a Gender Recognition (GR) circuit that splits the processing stream based on fundamental frequency distribution to eliminate cross-gender acoustic overlaps; and (3) a customized 18-layer Deep Convolutional Neural Network (CNN) integrated with meta-heuristic hyper-parameter optimization. |
Savita Jain; | International Journal For Multidisciplinary Research | 2026-05-30 |
| 234 | SALSA: Speech Aware LLM Adaptation Via Learned Steering Activation Vectors Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose SALSA (Speech-Aware LLM Adaptation via Learned Steering Activations), a lightweight adaptation method that learns layer-wise steering vectors. |
YEKATERINA YEGOROVA et. al. | arxiv-cs.CL | 2026-05-29 |
| 235 | Speech Recognition Systems Using Deep Learning Accuracy Vs. Efficiency Trade-offs Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper explores the balance between recognition accuracy and computational efficiency in deep learning–based speech recognition systems, analyzing major architectures, optimization techniques, performance challenges, and emerging approaches for scalable, efficient, and intelligent ASR deployment across modern digital ecosystems. |
Dandao Xuebao/Journal of Ballistics | 2026-05-29 | |
| 236 | Towards Human-Like Interactive Speech Recognition With Agentic Correction and Semantic Evaluation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Meanwhile, token-level metrics such as WER or CER cannot adequately reflect such a problem. To address these limitations, we formulate \emph{Interactive ASR} as a multi-turn refinement task and propose \textbf{Agentic ASR}, a closed-loop framework that combines a single-pass ASR front-end with semantic correction, intent routing, and reasoning-based editing. |
ZIXUAN JIANG et. al. | arxiv-cs.AI | 2026-05-28 |
| 237 | KVoiceBench, KOpenAudioBench, and KMMAU: Agent-Driven Korean Speech Benchmarks for Evaluating SpeechLMs Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Straightforward benchmark transfer through ASR, translation, normalization, and TTS can corrupt language-specific instructions, answer constraints, and spoken forms; for audio understanding, transferring source-language audio also fails to preserve target-language speaker attributes, accents, and paralinguistic properties. To address these limitations, we propose two human-agent benchmark-construction frameworks: one transfers source-language SpokenQA benchmarks into target-language SpokenQA benchmarks, and the other converts target-language ASR corpora into audio understanding benchmarks using transcriptions and speaker metadata. |
Haechan Kim; Seungjun Chung; Inkyu Park; Jihoo Lee; Jonghyun Lee; | arxiv-cs.CL | 2026-05-27 |
| 238 | Data-Efficient On-Policy Distillation for Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We study Ark-ASR, a 0.6B- parameter audio-conditioned language model trained with 100k hours of speech, and examine whether a strong Qwen-ASR teacher can transfer additional recognition capability through on-policy distillation. |
Yu Lin; Yiming Wang; Runyuan Cai; Xiaodong Zeng; | arxiv-cs.AI | 2026-05-27 |
| 239 | Building Community-Centred NLP Resources for Puno Quechua Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present the first dedicated ASR resources for Puno Quechua (ISO 639-3: qxp): (1) the largest speech corpus for any single Quechua variety, consisting in 66 hours of recordings for scripted and spontaneous speech (including 36 hours of manually transcribed and validated data), collected via a participatory design campaign; (2) the first systematic ASR benchmark for Puno Quechua, evaluating state-of-the-art models and fine-tuning Whisper-base, wav2vec2-base, and XLS-R-300M, with and without continued pre-training (CPT); (3) an open release of all datasets and fine-tuned models. |
Elwin Huaman; Adrian Gamarra Lafuente; Johanna Cordova; Anna Korhonen; | arxiv-cs.CL | 2026-05-27 |
| 240 | Breaking The Script Barrier: Enabling Automatic Alignment for PoS-based ASR Error Analysis in Non-Latin Scripts Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: However, existing alignment tools are often unreliable for languages written in non-Latin scripts. In this work, we address this gap by proposing a robust, automated, language-agnostic alignment mechanism applicable across ASR architectures and across languages written in both Latin and non-Latin scripts. |
Prasenjit K Mudi; Dahlia Devapriya; Sheetal Kalyani; | arxiv-cs.CL | 2026-05-27 |
| 241 | A Study of ASR Adaptation and Representation Dimensionality Reduction in Persian Speech Emotion Recognition Using Whisper Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we study the use of Whisper for Persian SER with a particular focus on representation dimensionality reduction and language-specific model adaptation. |
Ali Shendabadi; Parnia Izadirad; Mostafa Salehi; | arxiv-cs.CL | 2026-05-26 |
| 242 | Syllabic-Structure Decoder for Automatic Speech Recognition in Vietnamese Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we are motivated from the phonemic features of Vietnamese to propose a Syllabic-Structure Decoder for ASR, which models speech at the phoneme level instead of the orthographic level. |
Nghia Hieu Nguyen; Quan Ngoc Hoang; Long Hoang Huu Nguyen; Kiet Van Nguyen; Ngan Luu-Thuy Nguyen; | arxiv-cs.CL | 2026-05-26 |
| 243 | PashtoTTS-Bench: Automated Screening for Low-resource Non-Latin-script Text-to-speech Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: A system may produce no audio, speak a neighbouring language, preserve target script text only in an ASR transcript, or sound unnatural to native listeners. We introduce INSV (Intelligibility, Naturalness, Script fidelity, and Verification), a reporting framework that separates these cases. |
Hanif Rahman; | arxiv-cs.CL | 2026-05-26 |
| 244 | How to Evaluate Speech Translation with Source-Aware Neural MT Metrics Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we conduct the first systematic study of source-aware metrics for ST, with a particular focus on real-world operating conditions where source transcripts are not available. |
Mauro Cettolo; Marco Gaido; Matteo Negri; Sara Papi; Luisa Bentivogli; | Computational Linguistics | 2026-05-26 |
| 245 | TARQ: Tail-Aware Reconstruction Quantization for Rare-Word Robust Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose \textbf{Tail-Aware Reconstruction Quantization} (\TARQ), a label-free PTQ framework that shifts calibration toward the lexical tail via \textbf{\rareBAL}, a closed-form per-Linear-layer rule equalizing common/tail mass, paired with a metric-consistent residual correction. |
XINYU WANG et. al. | arxiv-cs.CL | 2026-05-26 |
| 246 | FalAR: A Large-scale Speaker-Annotated European Portuguese Speech Corpus of Parliamentary Sessions Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we describe the data collection process, together with the main characteristics of the FalAR corpus. |
FRANCISCO TEIXEIRA et. al. | arxiv-cs.CL | 2026-05-26 |
| 247 | Proactive for Uncertainty: Cause-Aware Error Diagnosis and Interactive Clarification for Spoken Dialogue Systems Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we propose a cause-aware error recovery paradigm that fundamentally rethinks robustness in SDS. |
YIZHOU PENG et. al. | arxiv-cs.CL | 2026-05-24 |
| 248 | Enhancing Arabic Dialectical Speech Recognition with Large Language Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this study, we present a comparative evaluation of state-of-the-art ASR models, Whisper (medium and large) and Wav2Vec2, applied to five Arabic varieties: Egyptian, Hijazi, Khaliji, Najdi, and MSA. |
Rehab Albeladi; | Islamic University Journal of Applied Sciences | 2026-05-23 |
| 249 | How Can Deep Learning Technique Be Leveraged to Enhance ASR Systems for Dysarthric Speech Recognition? Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This research aims to expand the available data and explore novel approaches in developing robust ASR systems tailored for dysarthric speech. |
Leon Starr; | New Vistas | 2026-05-21 |
| 250 | Mega-ASR: Towards In-the-wild^2 Speech Recognition Via Scaling Up Real-world Acoustic Simulation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose Mega-ASR, a unified ASR-in-the-wild framework that combines scalable compound-data construction with progressive acoustic-to-semantic optimization. |
ZHIFEI XIE et. al. | arxiv-cs.SD | 2026-05-19 |
| 251 | Can Large Language Models Reliably Correct Errors in Low-Resource ASR? A Contamination-Aware Case Study on West Frisian Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Large language models (LLMs) have shown promise for improving ASR through generative error correction (GER), but their effectiveness in low-resource settings remains underexplored. |
Yun Hao; Reihaneh Amooie; Wietse de Vries; Rik van Noord; Martijn Wieling; | arxiv-cs.CL | 2026-05-19 |
| 252 | FormalASR: End-to-End Spoken Chinese to Formal Text Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present FormalASR, two compact end-to-end models (0.6B and 1.7B) that directly transcribe spoken Chinese into formal written text. |
WANYI NING et. al. | arxiv-cs.CL | 2026-05-18 |
| 253 | Benchmarking Commercial ASR Systems on Code-Switching Speech: Arabic, Persian, and German Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present a benchmark evaluating five commercial ASR providers across four language pairs: Egyptian Arabic–English, Saudi Arabic (Najdi/Hijazi)–English, Persian (Farsi)–English, and German–English. |
Sajjad Abdoli; Ghassan Al-Sumaidaee; Clayton W. Taylor; Ahmed Rashad; | arxiv-cs.CL | 2026-05-18 |
| 254 | PAREDA: A Multi-Accent Speech Dataset of Natural Language Processing Research Discussions Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In particular, we introduce PAper REading DAtaset (PAREDA), a first-of-its-kind multi-accent speech dataset consisting of discussions on academic Natural Language Processing (NLP) papers between speakers with Australian, Indian-English, and Chinese English accents. |
Sicheng Jin; Dipankar Srirag; Aditya Joshi; | arxiv-cs.CL | 2026-05-18 |
| 255 | A Context Aware Igbo Language Voice Assistant Using Natural Language Processing Tools Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: The aim of this research is to develop a prototypecontext aware voice assistant system for the Igbo language using natural language processing tools. |
Emmanuel Nwabueze Ekwonwune; Leticia E. Elebiri; Abraham Oghenemega Ovwonuri; Chibundu D. Asiegbu; Netochukwu Nwokafor; | Scholarly Journal of Science and Technology Research and … | 2026-05-17 |
| 256 | Sometin Beta Pass Notin (SBPN): Improving Multilingual ASR for Nigerian Languages Via Knowledge Distillation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: SBPN is released in two sizes: SBPN-Base (120 M parameters) and SBPN-Large (600 M parameters). By releasing these as open foundation models, we aim to provide ASR resources for further research into the rich phonetic and cultural landscape of the region. |
Sewade Ogun; | arxiv-cs.CL | 2026-05-17 |
| 257 | Analyzing Error Propagation in Korean Spoken QA with ASR-LLM Cascades Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We analyze how automatic speech recognition (ASR) errors propagate through ASR-LLM cascades in Korean spoken question answering (SQA), focusing on downstream semantic failures that conventional ASR metrics cannot fully capture. |
Donghyuk Jung; Youngwon Choi; | arxiv-cs.CL | 2026-05-17 |
| 258 | JSPG: Dynamic Dictionary Filtering Via Joint Semantic-Pinyin-Glyph Retrieval for Chinese Contextual ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Base models often produce homophonic or near-homophonic errors that preserve the phonetic cues of the target keywords but severely distort their semantic meaning, rendering standard semantic retrievers ineffective. To resolve this, we propose a filtering framework that jointly integrates Semantic, Pinyin, and Glyph features (JSPG). |
Shilin Zhou; Zhenghua Li; | arxiv-cs.CL | 2026-05-16 |
| 259 | A Calculus-Based Framework for Determining Vocabulary Size in End-to-End ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: The main contribution of this paper in formalizing an approach to identify the vocabulary size best suited for training an end-to-end ASR system. |
Sunil Kumar Kopparapu; | arxiv-cs.CL | 2026-05-14 |
| 260 | JODAL: Joint Domain Adversarial Learning for TTS-Augmented Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Although we expect that training with both real and synthetic datasets together can mitigate the data scarcity issue while maintaining ASR performance, we observe that the distributional discrepancy between real and synthetic data introduces a domain inconsistency problem, which negatively impacts ASR performance. To mitigate this, we propose a Joint Domain Adversarial Learning (JODAL) method using real and TTS-generated samples that mitigates domain mismatches through domain adversarial training. |
June-Woo Kim; Ho-Young Jung; | Mathematics | 2026-05-14 |
| 261 | Refining Pseudo-Audio Prompts with Speech-Text Alignment for Text-Only Domain Adaptation in LLM-Based ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: The former neglects essential acoustic context, while the latter either suffers from limited scalability in data-scarce conditions, or yields inexpressive prompts by leveraging only textual features, ignoring audio modality. To address this, we propose an enhanced framework that explicitly models speech-text alignment. |
Ryo Magoshi; Takashi Maekaku; Yusuke Shinohara; | arxiv-cs.SD | 2026-05-14 |
| 262 | Automatic Speech Recognition in Healthcare in The Post-LLM Era: A Scoping Review Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: While traditional ASR focused on transcription fidelity, LLM-based systems extend this capability to intelligently reason, summarize, and structure clinical data. This scoping review maps the emerging landscape of LLM-based ASR in healthcare, examining its applications, technical foundations, evaluation practices, and reported challenges. |
Maram Alabbad; Waad Alhoshan; | Healthcare | 2026-05-13 |
| 263 | Vividh-ASR: A Complexity-Tiered Benchmark and Optimization Dynamics for Robust Indic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Fine-tuning multilingual ASR models like Whisper for low-resource languages often improves read speech but degrades spontaneous audio performance, a phenomenon we term studio-bias. To diagnose this mismatch, we introduce Vividh-ASR, a complexity-stratified benchmark for Hindi and Malayalam across four tiers: studio, broadcast, spontaneous, and synthetic noise. |
Kush Juvekar; Kavya Manohar; Aditya Srinivas Menon; Arghya Bhattacharya; Kumarmanas Nethil; | arxiv-cs.CL | 2026-05-13 |
| 264 | Automatic Speech Recognition and Large Language Models for Multilingual Pathology Report Generation: Proof-of-Concept Study Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Methods We conducted a controlled proof-of-concept study using 125 simulated mixed Chinese-English pathology gross examination audio recordings created by physicians or pathologists. |
KUAN-HSUN LIN et. al. | JMIR Formative Research | 2026-05-13 |
| 265 | Responsible Benchmarking of Fairness for Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We find that evaluating fairnessbased on single heterogeneous SG’s, such as they are defined in fairness benchmarks, can lead to misidentifyingwhich SG’s are actually being mistreated by ASR systems. We advocate for as fine-grained an analysis as possibleof the intersectionality of as many demographic variables as are available in the metadata of fairness corpora in orderto tease out such spurious correlations |
Felix Herron; Ange Richard; François Portet; Alexandre Allauzen; Solange Rossato; | arxiv-cs.CL | 2026-05-11 |
| 266 | Meri Awaaz Hi Pehchan Hai: A Survey on Multilingual Podcast Processing System Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: The survey reviews multilingual podcast processing and summarization techniques, including Automatic Speech Recog-nition (ASR), Natural Language Processing (NLP), extractive and abstractive summarization, machine translation, and Text-to-Speech systems. |
Gayana V; Gagana Poojari; Bhagyashree Bhagyashree; Arpita Rathod; Dr. Vijayalaxmi Mekali,; | International Scientific Journal of Engineering and … | 2026-05-10 |
| 267 | Can We Trust LLMs for Mental Health Screening? Consistency, ASR Robustness, and Evidence Faithfulness Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We find that (i) Phi-4 and Gemma-2-9B achieve excellent intra-model consistency (ICC > 0.89) with minimal degradation under ASR; (ii) Llama-3.1-8B shows ASR-fragile consistency, with ICC dropping from 0.82 to 0.36 at 10% WER; (iii) predictive validity is largely preserved under ASR for robust models; and (iv) keyword groundedness exceeds 93% for Phi-4 and Gemma-2-9B but falls to 77-81% for Llama-3.1-8B. |
Erfan Loweimi; Sofia de la Fuente Garcia; Samira Loveymi; Hadi Daneshvar; Saturnino Luz; | arxiv-cs.CL | 2026-05-10 |
| 268 | AI-Based Real-Time Speech-to-Sign Language Translation System for Assistive Communication Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents the design and development of an Artificial Intelligence (AI)-based, real-time Speech-to- Indian Sign Language (ISL) Translation System that bridges the communication gap between the hearing population and the deaf and hard-of-hearing community. |
Y. Dayanand Kumar Y. Dayanand Kumar; Bhukya Siddhu Bhukya Siddhu; | International Journal of Creative and Open Research in … | 2026-05-09 |
| 269 | WorldSpeech: A Multilingual Speech Corpus from Around The World Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To this end, we introduce WorldSpeech, a 24 kHz multilingual speech corpus comprising 65k hours of aligned audio-transcript data across 76 languages, collected from diverse public sources including parliamentary proceedings, international broadcasts, and public-domain audiobooks. |
Antonis Asonitis; Luca A. Lanzendörfer; Frédéric Berdoz; Roger Wattenhofer; | arxiv-cs.CL | 2026-05-09 |
| 270 | WASIL: In-the-Wild Arabic Spoken Interactions with LLMs Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We provide low-cost gold transcripts via multi-ASR agreement-guided post-editing and annotate answerability (answerable, ambiguous/needs-clarification, unsupported, not-a-request/noise) to separate intrinsic unanswerability from ASR-induced degradation. |
ZIEN SHEIKH ALI et. al. | arxiv-cs.SD | 2026-05-09 |
| 271 | Real Time Translation and Emotional Intelligent Voice Model Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Current speech-to-speech translation systems often struggle to capture the original speaker’s vocal identity and emotional tone, resulting in robotic and unnatural conversations. To address this, we introduce the Real Time Translation and Emotional Intelligent Voice Model. |
Nooh C. H.; | International Journal for Research in Applied Science and … | 2026-05-08 |
| 272 | Beyond Single Ground Truth: Reference Monism As Epistemic Injustice in ASR Evaluation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose WER-Range: reporting performance across legitimate conventions rather than assuming a single correct answer. |
ANNA SEO GYEONG CHOI et. al. | arxiv-cs.CL | 2026-05-07 |
| 273 | Bangla-WhisperDiar: Fine-Tuning Whisper and PyAnnote for Bangla Long-Form Speech Recognition and Speaker Diarization Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We describe our complete pipeline, including data preprocessing, text normalization, audio augmentation, training strategies, inference optimization, and post-processing for both tasks. |
MOHAMMED AMAN BHUIYAN et. al. | arxiv-cs.SD | 2026-05-06 |
| 274 | Hearing Without Noticing? Attention-Aware Stealthy Black-box Adversarial Audio Attacks Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we introduce the first music carrier selection algorithm and an attention-aware stealthiness loss function to generate stealthy AEs. |
Cheng'an Wei; Yue Zhao; Kai Chen; | icml | 2026-05-05 |
| 275 | A Comprehensive Analysis of Tokenization and Self-Supervised Learning in End-to-End Automatic Speech Recognition Applied on French Language Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we conduct a qualitative study on the French language that investigates the impact of subword tokenization algorithms and self-supervised learning models from different linguistic and acoustic perspectives, using a comprehensive set of evaluation metrics. |
Thibault Bañeras-Roux; Mickael Rouvier; Jane Wottawa; Richard Dufour; | arxiv-cs.CL | 2026-05-05 |
| 276 | MoST: Mixing Speech and Text with Modality-Aware Mixture of Experts Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: MAMoE simultaneously enhances modality-specific learning and cross-modal understanding through two complementary components: modality-specific expert groups that capture domain-specific patterns and shared experts that facilitate information transfer between modalities. Building on this architecture, we develop an efficient transformation pipeline that adapts the pretrained MoE language model through strategic post-training on ASR and TTS datasets, followed by fine-tuning with a carefully curated speech-text instruction dataset. |
Yuxuan Lou; Kai Yang; Yang You; | icml | 2026-05-05 |
| 277 | A Study of Data Selection Strategies for Pre-Training Self-Supervised Speech Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We systematically examine how curated subsets of pre-training data influence Automatic Speech Recognition (ASR) performance. |
R. Whetten; T. Parcollet; M. Dinarelli; Y. Estève; | icassp | 2026-05-04 |
| 278 | Align2speak: Improving TTS for Low Resource Languages Via ASR-Guided Online Preference Optimization Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose a framework based on Group Relative Policy Optimization (GRPO) to adapt an autoregressive, multilingual TTS model to new languages. |
S. Hussain; | icassp | 2026-05-04 |
| 279 | Knowledge Distillation Via Generative Reconstruction Pathways for End-to-End Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we propose a generative reconstruction pathway knowledge distillation (GRP-KD) framework that leverages flow matching and diffusion processes to achieve richer knowledge transfer. |
S. -H. Jeong; D. -H. Kim; J. -H. Chang; | icassp | 2026-05-04 |
| 280 | Advancing LLM-Based Multi-Channel Multi-Speaker Speech Recognition with Global Cross-Channel Attention and Sentence-Ordered First-In First-Out Serialized Output Training Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: While multi-microphone input outperforms single-microphone setups in multi-speaker scenarios, multi-channel multi-speaker ASR remains challenging due to data scarcity, complex acoustic conditions, and cross-channel dependency modeling limitations. To address these challenges, we propose a novel framework integrating a Large Language Model (LLM) into multi-channel multi-speaker ASR for the first time. |
G. Wan; | icassp | 2026-05-04 |
| 281 | SS-JDSC: Single-Speaker Japanese Dysarthric Speech Corpus Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Existing dysarthric-speech corpora have facilitated research progress, but suffer from two major limitations: restricted language coverage and limited data per speaker. In this paper, the SS-JDSC, the first open-source corpus of Japanese dysarthric speech for automatic speech recognition (ASR), is presented to address these challenges. |
A. Ogasawara; S. Takamichi; J. Yang; G. Suenaga; Y. Tan; | icassp | 2026-05-04 |
| 282 | Proficiency-Aware Adaptation and Data Augmentation for Robust L2 ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Using the Common European Framework of Reference for Languages (CEFR)-graded Speak & Improve corpus, we show that naïve fine-tuning of Whisper reduces the average word error rate (WER) but simultaneously widens performance disparities and disproportionately harms lower-proficiency learners. To address this, we propose two strategies: (i) proficiency-aware multitask learning, jointly optimizing ASR with proficiency classification, and (ii) targeted augmentation, applying spectrogram masking to low-proficiency speech to counter imbalance. |
L. Sun; C. Zhu; S. Shi; | icassp | 2026-05-04 |
| 283 | Chunkwise Aligners for Streaming Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose the Chunkwise Aligner, a novel architecture for streaming automatic speech recognition (ASR). |
W. S. Teo; T. Moriya; M. Mimura; | icassp | 2026-05-04 |
| 284 | SED: Structural Entropy Based Speech Discretization for Discrete Token-Based ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we propose SED, a new Structural Entropy-based Speech Discretization method that models speech features as graph nodes and performs adaptive clustering by minimizing 2D Structural Entropy. |
L. Dong; W. Wang; Y. Xiang; Y. Xian; S. Gao; | icassp | 2026-05-04 |
| 285 | Tokenchain: A Discrete Speech Chain Via Semantic Token Modeling Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose TokenChain, a fully discrete speech chain coupling semantic-token ASR with a two-stage TTS: an autoregressive text-to-semantic model co-trained with ASR and a masked-generative semantic-to-acoustic model for synthesis only. |
M. Wang; S. Nakamura; | icassp | 2026-05-04 |
| 286 | Adapting Diarization-Conditioned Whisper for End-to-End Multi-Talker Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose a speaker-attributed (SA) Whisper-based model for multi-talker speech recognition that combines target-speaker modeling with serialized output training (SOT). |
M. KOCOUR et. al. | icassp | 2026-05-04 |
| 287 | BEST-RQ-based Self-Supervised Learning for Whisper Domain Adaptation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose BEARD (BEST-RQ Encoder Adaptation with Re-training and Distillation), a novel framework designed to adapt Whisper’s encoder with unlabeled data. |
R. Bagat; I. Illina; E. Vincent; | icassp | 2026-05-04 |
| 288 | Advanced Modeling of Interlanguage Speech Intelligibility Benefit with L1-L2 Multi-task Learning Using Differentiable K-means for Accent-robust Discrete Token-based ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this study, we propose a more advanced modeling of ISIB. |
K. Onda; S. Fukayama; D. Saito; N. Minematsu; | icassp | 2026-05-04 |
| 289 | Lattice-Guided Consistency Regularization of Dual-Mode Transducers for Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce a novel approach to train a dual-mode transducer model that supports both autoregressive (AR) and non-autoregressive (NAR) inference using lattice-based consistency regularization for Automatic Speech Recognition (ASR). |
W. Ding; H. Xu; J. Balam; J. Lai; | icassp | 2026-05-04 |
| 290 | Medical ASR Enhancement By Domain-Specific Reinforcement Fine-Tuning Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: State-of-the-art systems optimize all tokens equally, under-weighting rare, domain-critical terms. We propose a medical-aware reinforcement fine-tuning (RFT) framework that generates multiple hypotheses, extracts medical terms with LLMs, validates them via UMLS, and optimizes a composite reward combining WER, medical WER (MWER), and length regularization. |
C. Wang; | icassp | 2026-05-04 |
| 291 | MNV-17: A High-Quality Performative Mandarin Dataset for Nonverbal Vocalization Recognition in Speech IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Progress in NV-aware ASR has been hindered by the lack of high-quality, well-annotated datasets. To address this gap, we introduce MNV-17, a 7.55-hour performative Mandarin speech dataset. |
J. Mai; | icassp | 2026-05-04 |
| 292 | Synthetic Data Domain Adaptation for ASR Via LLM-Based Text and Phonetic Respelling Augmentation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose a synthetic-data-based domain adaptation framework with two contributions: (1) a large language model (LLM)-based text augmentation pipeline with a filtering strategy that balances lexical diversity, perplexity, and domain-term coverage, and (2) phonetic respelling augmentation (PRA), a novel method that introduces pronunciation variability through LLM-generated orthographic pseudo-spellings. |
N. Yamashita; K. Nagatsuka; H. Kokubo; K. Dohi; T. V. Ho; | icassp | 2026-05-04 |
| 293 | Streaming Speech Recognition with Decoder-Only Large Language Models and Latency Optimization Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we propose a novel streaming ASR approach that integrates a read/write policy network with monotonic chunkwise attention (MoChA) to dynamically segment speech embeddings. |
G. WAN et. al. | icassp | 2026-05-04 |
| 294 | Discrete Diffusion for Generative Modeling of Text-Aligned Speech Tokens IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper introduces a discrete diffusion model (DDM) framework for text-aligned speech tokenization and reconstruction. |
P. -J. KU et. al. | icassp | 2026-05-04 |
| 295 | CALM: Joint Contextual Acoustic-Linguistic Modeling for Personalization of Multi-Speaker ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present CALM, a joint Contextual Acoustic-Linguistic Modeling framework for multi-speaker automatic speech recognition (ASR). |
M. Shakeel; Y. Fukumoto; C. Maeda; C. -J. Lin; S. Watanabe; | icassp | 2026-05-04 |
| 296 | LOTUSDIS: A Thai Far-Field Meeting Corpus for Robust Conversational ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present LOTUSDIS, a publicly available Thai meeting corpus designed to advance far-field conversational ASR. |
P. Tipaksorn; S. Thatphithakkul; V. Chunwijitra; K. Thangthai; | icassp | 2026-05-04 |
| 297 | Leveraging Segment-Level Speech Representations for LLM-Based Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper proposes an LLM-ASR framework based on segment-level speech representations. |
S. Jiang; L. Dong; W. Wang; S. Gao; | icassp | 2026-05-04 |
| 298 | Phoneme-Level Visual Speech Recognition Via Point-Visual Fusion and Language Model Reconstruction Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To address this, our work focuses on phoneme-level modeling, which provides a finer linguistic granularity than visemes while reducing the reliance on large-scale pretraining. We propose a novel phoneme-based two-stage framework that fuses visual and landmark motion features, followed by a Large Language Model (LLM) for sentence reconstruction to address these challenges. |
M. K. K. Teng; H. Zhang; T. Saitoh; | icassp | 2026-05-04 |
| 299 | GLoRIA: Gated Low-Rank Interpretable Adaptation for Dialectal ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose GLoRIA, a parameter-efficient adaptation framework that leverages metadata (e.g., coordinates) to modulate low-rank updates in a pre-trained encoder. |
P. Mehralian; M. Farasyn; A. Breitbarth; A. -S. Ghyselen; H. Van Hamme; | icassp | 2026-05-04 |
| 300 | Low Resource Multimodal Translation of Nepali Spoken Words Into Emotion-Conditioned Sign Language Avatars Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Sign language communication systems, that integrate emotional expression remain underexplored, particularly for low-resource languages. This pilot study presents NEST-V1 (Nepali Emotion and Speech Transformer – Version 1), a proof-of-concept multimodal framework that demonstrates the feasibility of generating emotion-conditioned Nepali Sign Language avatars from spoken input. |
Jatin Bhusal; Salma Tamang; | arxiv-cs.CL | 2026-05-04 |
| 301 | Whisper-QF: Leveraging Dual Cross-Attention Q-Former for Speech Emotion Recognition With Multi-Task Learning Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we propose Whisper-QF, a multi-task framework that integrates Whisper with a lightweight query-based transformer module (Q-Former) to enhance emotional understanding. |
Z. Zhuang; T. Wei; Y. Shi; S. Wang; J. Xiao; | icassp | 2026-05-04 |
| 302 | Train Short, Infer Long: Speech-LLM Enables Zero-Shot Streamable Joint ASR and Diarization on Long Audio IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we present an end-to-end speech large language model (Speech-LLM) for Joint strEamable DIarization and aSr (JEDIS-LLM). |
M. Shi; X. Xiao; R. Fan; S. Ling; J. Li; | icassp | 2026-05-04 |
| 303 | OMNI-AVSR: Towards Unified Multimodal Speech Recognition With Large Language Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To this end, we present Omni-AVSR, a unified audio-visual LLM that combines efficient multi-granularity training with parameter-efficient adaptation. |
U. Cappellazzo; X. Liu; P. Ma; S. Petridis; M. Pantic; | icassp | 2026-05-04 |
| 304 | Speaking Clearly: A Simplified Whisper-Based Codec for Low-Bitrate Speech Coding Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Through empirical analysis, we discover that targeted architectural simplification can unlock the acoustic modeling potential of Whisper, a text-aligned Automatic Speech Recognition (ASR) model. Based on this finding, we propose SimWhisper-Codec, a novel codec that balances the semantic and acoustic preservation by leveraging a frozen, simplified Whisper encoder without requiring external supervision. |
X. Zhang; L. Li; X. Lu; J. Liu; K. A. Lee; | icassp | 2026-05-04 |
| 305 | Windowed SummaryMixing: An Efficient Fine-Tuning of Self-Supervised Learning Models for Low-Resource Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we introduce Windowed SummaryMixing (WSM), which enhances SM by integrating local neighborhood summaries alongside the global summary, maintaining efficiency while improving temporal dependencies. |
A. S. Menon; K. Tripathi; R. Gohil; P. Wasnik; | icassp | 2026-05-04 |
| 306 | BBPE16: UTF-16-Based Byte-Level Byte-Pair Encoding for Improved Multilingual Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose BBPE16, a UTF-16-based BBPE tokenizer that represents most modern scripts with a uniform 2-byte code unit. |
H. Kim; H. Kim; M. Lee; K. Lee; | icassp | 2026-05-04 |
| 307 | WAV2LEV: Predicting Levenshtein Edit Operation Sequences For Fine-Grained Estimation of Automatic Speech Recognition Error Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose WAV2LEV, a novel paradigm for WER estimation which predicts the underlying sequences of Levenshtein edit operations (substitutions, deletions, insertions and matches) from which the WER can be computed. |
H. Donnelly; K. Shi; G. Penn; | icassp | 2026-05-04 |
| 308 | Confidence-Guided Error Correction for Disordered Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In particular, we propose confidence-informed prompting, where word-level uncertainty estimates are embedded directly into LLM training to improve robustness and generalization across speakers and datasets. |
A. Hernandez; T. Arias-Vergara; A. Maier; P. A. Pérez-Toro; | icassp | 2026-05-04 |
| 309 | Exploring SSL Discrete Tokens for Multilingual Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: With the advancement of self-supervised learning (SSL) in speech-related tasks, there has been growing interest in using discrete tokens generated by SSL models for automatic … |
M. Cui; | icassp | 2026-05-04 |
| 310 | Decoder-Only Conformer with Modality-Aware Sparse Mixtures of Experts for ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present a decoder-only Conformer for automatic speech recognition (ASR) that processes speech and text in a single stack without external speech encoders or pretrained large language models (LLM). |
J. Lee; M. Mimura; | icassp | 2026-05-04 |
| 311 | Contextual Biasing for ASR in Speech LLM with Common Word Cues and Bias Word Position Prediction Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we explore contextual biasing in SLLM based on acoustic cues associated with a set of common words whose pronunciations are partially similar to those of the target bias words. |
S. Novitasari; T. Fukuda; G. Kurata; G. Saon; | icassp | 2026-05-04 |
| 312 | Scaling Multi-Talker ASR with Speaker-Agnostic Activity Streams Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we propose a method that decouples the inference cost of activity-conditioned ASR systems from the number of speakers by converting speaker-specific activity outputs into two speaker-agnostic streams. |
X. He; A. Polok; J. Villalba; T. Thebaud; M. Maciejewski; | icassp | 2026-05-04 |
| 313 | DOMA: Leveraging Diffusion Language Models with Adaptive Prior for Intent Classification and Slot Filling Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose DOMA, a model-agnostic framework that refines ASR transcripts for ICSF with diffusion language models (DLMs) and adaptive prior. |
S. Yang; | icassp | 2026-05-04 |
| 314 | Adversarial Fine-Tuning on Speech Foundation Model with Vulnerable Attention Consistency Regularization for Robust Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, for the first time, we study how to fine-tune Transformer-based SFMs for adversarially robust ASR, especially preserving utility while improving robustness. |
Y. Wang; B. Wu; L. Liu; | icassp | 2026-05-04 |
| 315 | Towards Robust Dysarthric Speech Recognition: LLM-Agent Post-ASR Correction Beyond WER Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This mismatch is particularly problematic for dysarthric speech, where articulatory imprecision and disfluencies can cause severe semantic distortions. To bridge this gap, we introduce a Large Language Model(LLM)-based agent for post-ASR correction: a Judge–Editor over the top-k ASR hypotheses that keeps high-confidence spans, rewrites uncertain segments, and operates in both zero-shot and fine-tuned modes. |
X. Zheng; S. Dong; B. Phukon; M. Hasegawa-Johnson; C. D. Yoo; | icassp | 2026-05-04 |
| 316 | Joint Autoregressive Modeling of Multi-Talker Overlapped Speech Recognition and Translation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose a novel end-to-end model for multi-talker speech recognition and translation that simultaneously generates multiple transcriptions and their corresponding translations from overlapped speech. |
T. Tanaka; | icassp | 2026-05-04 |
| 317 | Toward Conversational User Interface Via Voice Command Correction Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose a speech-command-based correction system that enables users to issue natural-language instructions to refine recognition outputs with minimal effort. |
S. -J. Ding; C. -H. Chang; | icassp | 2026-05-04 |
| 318 | TAGARELA – A Portuguese Speech Dataset from Podcasts Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Despite significant advances in speech processing, Portuguese remains under-resourced due to the scarcity of public, large-scale, and high-quality datasets. To address this gap, we present a new dataset, named TAGARELA, composed of over 8,972 hours of podcast audio, specifically curated for training automatic speech recognition (ASR) and text-to-speech (TTS) models. |
F. S. De Oliveira; | icassp | 2026-05-04 |
| 319 | Reducing Prompt Sensitivity in LLM-Based Speech Recognition Through Learnable Projection Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents a comprehensive analysis of commonly used prompts across diverse datasets, showing that prompt choice significantly affects ASR performance and introduces instability, with no single prompt performing best across all cases. |
S. Burdisso; | icassp | 2026-05-04 |
| 320 | Mitigating Attention Sinks and Massive Activations in Audio-Visual Speech Recognition with LLMs Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In natural language processing, recent work has revealed attention sinks, tokens that attract disproportionately high attention, and associated massive activations in which some features of sink tokens exhibit huge activation in LLMs. In this work, we are the first to study these phenomena in multimodal speech recognition. |
A. Anand; U. Cappellazzo; S. Petridis; M. Pantic; | icassp | 2026-05-04 |
| 321 | Do We Really Need Self-attention for Streaming Automatic Speech Recognition? Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This work questions the suitability of transformers for specific domains. |
Y. Dkhissi; V. Vielzeuf; E. Allesiardo; A. Larcher; | icassp | 2026-05-04 |
| 322 | Advancing Semi-Supervised Child Speech Recognition with Omni-Temporal Classification Under Label Noise Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we investigate alternative alignment modeling for training child speech ASR systems, relaxing the strict alignment between speech and transcripts. |
J. Xie; J. H. L. Hansen; | icassp | 2026-05-04 |
| 323 | Listen, But Don’t Leak: Sensitive Data Protection for Privacy Aware Automatic Speech Recognition with Acoustic Triggers Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this study, we introduce Protective Acoustic Triggering (PAT), a novel method for embedding privacy redaction directly into ASR models. |
T. Roy; N. T. Vu; | icassp | 2026-05-04 |
| 324 | Speaker Attributed Automatic Speech Recognition Using Speech Aware Llms Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To address limitations in training data, we propose a data augmentation method that uses artificially concatenated multi-speaker conversations. |
H. Aronowitz; Z. Kons; A. Dekel; G. Saon; R. Hoory; | icassp | 2026-05-04 |
| 325 | NGPT As A Scalable Architecture for Speech Recognition and Translation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: The model introduces a hyperspherical normalization method that enables stable large-scale training without convergence difficulties. |
N. Tadevosyan; | icassp | 2026-05-04 |
| 326 | Bridging The Front-End and Back-End for Robust ASR Via Cross-Attention-Based U-Net Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: However, the simple linear fusion mechanism in OA only yields ASR improvements under mild conditions, while its performance degrades substantially in complex acoustic environments. To overcome this limitation, we propose a cross-attention-based U-Net module designed to effectively achieve interactive feature fusion. |
T. NING et. al. | icassp | 2026-05-04 |
| 327 | Mixture To Beamformed Mixture: Leveraging Beamformed Mixture As Weak-Supervision for Speech Enhancement and Noise-Robust ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In multi-channel speech enhancement and robust automatic speech recognition (ASR), beamforming can typically improve the signal-to-noise ratio (SNR) of the target speaker and produce reliable enhancement with little distortion to target speech. With this observation, we propose to leverage beamformed mixture, which has a higher SNR of the target speaker than the input mixture, as a weak supervision to train deep neural networks (DNNs) to enhance the input mixture. |
Z. -Q. Wang; R. Pang; | icassp | 2026-05-04 |
| 328 | Reference Microphone Selection for Guided Source Separation Based on The Normalized L-P Norm Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we propose two reference microphone selection methods for GSS-based speech enhancement that are based on the normalized ℓp-norm, either using only the normalized ℓp-norm or combining the normalized ℓp-norm and the SNR to account for both differences in SNR and ELR across microphones. |
A. LOHMANN et. al. | icassp | 2026-05-04 |
| 329 | Noise-Robust AV-ASR Using Visual Features Both in The Whisper Encoder and Decoder Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Our proposed dual-use method shows consistent noise robustness improvement, e.g., a 35% relative improvement (WER: 4.41% vs. 6.83%) based on Whisper small, and a 57% relative improvement (WER: 4.07% vs. 9.53%) based on Whisper medium, compared to typical reference middle fusion in babble noise with a signal-to-noise ratio (SNR) of 0dB. |
Z. LI et. al. | icassp | 2026-05-04 |
| 330 | Audio-Conditioned Diffusion LLMs for ASR and Deliberation Processing Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we present an empirical study on using the diffusion-based large language model LLaDA for automatic speech recognition (ASR). |
M. WANG et. al. | icassp | 2026-05-04 |
| 331 | LLM-Based Post-ASR Error Correction for Disordered Speech Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present the first systematic study of large language model (LLM)-based post-ASR error correction for disordered speech. |
H. Wen; M. Assefa; A. Semsayan; E. Feo-Flushing; | icassp | 2026-05-04 |
| 332 | SSVD-O: Parameter-Efficient Fine-Tuning with Structured SVD for Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Building on our prior work, this paper introduces SSVD-Outer (SSVD-O), an extension of the structured SVD-guided (SSVD) fine-tuning method. |
P. Wang; S. Watanabe; H. Van Hamme; | icassp | 2026-05-04 |
| 333 | AccLID: Accent-aware Language Identification for Robust Multilingual Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose AccLid, a multi-modal ranking framework that improves accent robustness by integrating acoustic, temporal, linguistic, and phonetic evidence. |
R. Singh; | icassp | 2026-05-04 |
| 334 | A Multi-Task Approach Towards Robust Vietnamese Audio-Based Toxic Span Detection Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Detecting toxic spans in spoken content is crucial for content moderation, but conventional two-stage pipelines that run Automatic Speech Recognition (ASR) followed by a Toxic Span Detection (TSD) model suffer from high latency and system complexity. To address these limitations, we propose a multi-task framework for Vietnamese audio-based toxic span detection that unifies ASR and TSD modules within a single model. |
V. L. -P. Huynh; H. B. Do; L. T. Nguyen; | icassp | 2026-05-04 |
| 335 | Enhancing Multilingual LLM-Based ASR with Mixture of Experts and Dynamic Downsampling Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To this end, this work proposes a projector-based LLM-ASR framework targeting the key challenges of multilingual generalization and modality alignment. |
G. Lin; Z. Chen; Y. Fu; K. Li; W. -Q. Zhang; | icassp | 2026-05-04 |
| 336 | Group Relative Policy Optimization for Text-to-Speech with Large Language Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper proposes a GRPO-based approach to enhance the performance of large language model (LLM)-based text-to-speech (TTS) models by deriving rewards from an off-the-shelf automatic speech recognition (ASR) model. |
C. Liu; Y. -J. Hu; Y. -Y. Gao; S. -L. Zhang; Z. -H. Ling; | icassp | 2026-05-04 |
| 337 | TTA: Transcribe, Translate and Alignment for Cross-Lingual Speech Representation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To this end, we propose a lightweight TTA model specialized in speech semantics for more effective LLM integration. |
W. Liu; J. Li; Y. Shao; D. Yu; | icassp | 2026-05-04 |
| 338 | TextlessRAG: End-to-End Visual Document RAG By Speech Without Text Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose TextlessRAG, the first end-to-end framework for speech-based question answering over large-scale document images. |
P. XIE et. al. | icassp | 2026-05-04 |
| 339 | Synthesized Data Selection Via Score Distribution Matching for Te Reo Māori Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Fine-tuning pre-trained Automatic Speech Recognition (ASR) models with synthesized datasets leads to suboptimal performance in low-resource ASR, due to domain mismatches between true and synthesized speech data. To address this issue, we propose a novel score-distribution-matching method for data selection, which filters synthesized data by explicitly aligning its score distribution with the prior distribution of the true data score. |
Z. Wang; F. Hou; R. Wang; | icassp | 2026-05-04 |
| 340 | HD-PPT: Hierarchical Decoding of Content- and Prompt-Preference Tokens for Instruction-Based TTS Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Although instruction-based Text-to-Speech (Instruct-TTS) models are proposed, these models still lack fine-grained control due to the modality gap between single-level text instructions and multilevel speech tokens. To address this limitation, we propose HD-PPT, a framework that transforms speech synthesis into a structured, hierarchical task. |
S. Nie; X. Xing; J. Xing; B. Liu; X. Xu; | icassp | 2026-05-04 |
| 341 | Target-Speaker LLM-ASR with Speaker-Aware Speech Encoder Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we propose a TS-ASR system based on large language models (LLMs), featuring a speaker-aware speech encoder (SASE) that extends conventional LLM-based ASR to the target-speaker scenario while preserving the original structure. |
M. Kim; S. Kim; | icassp | 2026-05-04 |
| 342 | Probing The Hidden Talent of ASR Foundation Models for L2 English Oral Assessment Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we explore the untapped potential of Whisper [1], a well-established automatic speech recognition (ASR) foundation model, in the context of L2 spoken language assessment (SLA). |
F. -A. Chao; B. -C. Yan; B. Chen; | icassp | 2026-05-04 |
| 343 | Mind The Shift: Using Delta SSL Embeddings to Enhance Child ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Self-supervised learning (SSL) models have achieved impressive results across many speech tasks, yet child automatic speech recognition (ASR) remains challenging due to limited data and pretraining domain mismatch. |
Z. Wang; N. B. Shankar; K. Zhang; Z. Wang; A. Alwan; | icassp | 2026-05-04 |
| 344 | Improving Automatic Speech Recognition By Mitigating Distortions Introduced By Speech Enhancement Under Drone Noise Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper proposes a method to improve automatic speech recognition (ASR) performance against drone noise-contaminated audio recordings by mitigating distortions resulted from speech enhancement (SE). |
R. Miura; T. Osaki; B. Yen; T. Ashizawa; K. Nakadai; | icassp | 2026-05-04 |
| 345 | Lamer-SSL: Layer-Aware Mixture of Lora Experts for Continual Multilingual Expansion of Self-Supervised Models Without Forgetting Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Despite their impressive performance, self-supervised speech models often struggle to generalize to new languages and tend to forget previously acquired knowledge during continual training. To address this, we propose Lamer-SSL, a parameter-efficient framework that integrates a Layer-Aware MixturE of LoRA Experts (Lamer) module with a replay strategy. |
J. Xu; M. Wu; X. Chen; X. Wu; H. Meng; | icassp | 2026-05-04 |
| 346 | Bridging The Gap: A Comparative Exploration of Speech-LLM and End-to-End Architecture for Multilingual Conversational ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we present an enhanced LLM-based ASR framework that com-bines fine-tuned Whisper and mHuBERT encoders with an LLM to enrich speech representations. |
Y. Mei; D. Xu; J. Liang; Y. Long; | icassp | 2026-05-04 |
| 347 | Revisiting Direct Speech-to-Text Translation with Speech LLMS: Better Scaling Than Cot Prompting? Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we systematically compare CoT and Direct prompting under increasing amounts of S2TT data. |
O. Pareras; G. I. Gállego; F. Costa; C. España-Bonet; J. Hernando; | icassp | 2026-05-04 |
| 348 | Production-Scale Dynamic Vocabulary ASR Biasing with Word-Level FST and Robust Training Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Third, the general ASR accuracy degrades relative to a vanilla ASR system and usually deletes entities that are hard to recognize, which poses an issue for the no-bias condition, a scenario that can be quite common in production deployment. In this work, we present several enhancements to the original DynVoc model that mitigate these problems, scale the solution to production-level ASR and achieve substantial gains over the vanilla DynVoc method. |
J. E. García Lainez; T. Sun; S. Ling; Y. Gong; H. Wang; | icassp | 2026-05-04 |
| 349 | Understanding Frechet Speech Distance for Synthetic Speech Quality Evaluation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Fréchet Distance has emerged as a promising alternative, yet its reliability depends heavily on the choice of embeddings and experimental settings. In this work, we comprehensively evaluate Fréchet Speech Distance (FSD) and its variant Speech Maximum Mean Discrepancy (SMMD) under varied embeddings and conditions. |
J. -W. Kim; D. Agarwal; F. Cerina; | icassp | 2026-05-04 |
| 350 | PhoenixDSR: Phoneme-Guided and LLM-Enhanced Dysarthric Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present PhoenixDSR, a phoneme-mediated framework that decouples acoustic variability from linguistic decoding. |
Y. WU et. al. | icassp | 2026-05-04 |
| 351 | Scale: Semantic Chunking and Label-Delay Engine For Streaming Speech-LLM Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Existing approaches typically employ fixed-size chunking to maintain low latency, which often compromises recognition accuracy. We propose SCALE, a streaming ASR framework that addresses this challenge through three key techniques: (a) dynamic chunk boundary prediction leveraging semantic information, replacing rigid fixed-size chunking, (b) intra-chunk bidirectional attention mechanism for efficient acoustic context modeling, and (c) label delay training strategy enabling stable word predictions with smaller chunks. |
A. Jaiswal; | icassp | 2026-05-04 |
| 352 | PAC: Pronunciation-Aware Contextualized Large Language Model-Based Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents a Pronunciation-Aware Contextualized (PAC) framework to address two key challenges in Large Language Model (LLM)-based Automatic Speech Recognition (ASR) systems: effective pronunciation modeling and robust homophone discrimination. |
L. FU et. al. | icassp | 2026-05-04 |
| 353 | In-Sync: Adaptation of Speech Aware Large Language Models for ASR with Word Level Timestamp Predictions Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we extend an existing speech-aware language model to predict timestamps directly alongside transcripts. |
X. FAN et. al. | icassp | 2026-05-04 |
| 354 | Language-Infused Retrieval-Augmented CTC with Adaptive Soft-Hard Gating for Robust Code-Switching ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose Language-Infused Retrieval-Augmented CTC with Adaptive Soft-Hard Gating (LIRA-CTC), which explicitly incorporates language posterior information into the retrieval process. |
Z. Liang; S. Nakamura; | icassp | 2026-05-04 |
| 355 | MedSpeak: A Knowledge Graph-Aided ASR Error Correction Framework for Spoken Medical QA Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To this end, we propose MedSpeak, a novel knowledge graph-aided ASR error correction framework that refines noisy transcripts and improves downstream answer prediction by leveraging both semantic relationships and phonetic information encoded in a medical knowledge graph, together with the reasoning power of LLMs. |
Y. Song; | icassp | 2026-05-04 |
| 356 | Impact of Phonetics on Speaker Identity in Adversarial Voice Attack Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we analyze adversarial audio at the phonetic level and show that perturbations are associated with systematic phonetic tendencies, such as vowel centralization and consonant substitutions. |
D. K. Dar; Q. Yan; L. Xiao; A. Ross; | icassp | 2026-05-04 |
| 357 | From Hallucination to Articulation: Language Model-Driven Losses for Ultra Low-Bitrate Neural Speech Coding Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: It is the generative decoder’s attempt to synthesize plausible outputs from excessively compressed tokens missing some semantic information. In this work, we propose language model-driven losses (LM loss) and show they may alleviate PHs better than a semantic distillation (SD) objective in very-low-bitrate settings. |
J. Yi; M. Kim; | icassp | 2026-05-04 |
| 358 | OCR-Enhanced Multimodal ASR Can Read While Listening Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we propose Donut-Whisper, an audio-visual ASR model with dual encoder to leverage visual information to improve speech recognition performance in both English and Chinese. |
J. Chen; C. Tang; Y. Li; G. Sun; C. Zhang; | icassp | 2026-05-04 |
| 359 | Dual-Grained Routing Guided Multi-LoRA Experts for Multilingual Low-Resource Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we propose a novel dual-grained routing guided Multi-LoRA approach for multilingual low-resource ASR. |
Y. Wang; H. Zhang; H. Wang; L. Sun; M. Song; | icassp | 2026-05-04 |
| 360 | VOICE2MODE: Phonation Mode Classification in Singing Using Self-Supervised Speech Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present voice2mode, a method for classification of four singing phonation modes (breathy, neutral (modal), flow, and pressed) using embeddings extracted from large self-supervised speech models. |
A. A. Justus; R. Agrawal; S. R. Kadiri; S. Narayanan; | icassp | 2026-05-04 |
| 361 | Chain of Correction for Full-Text Speech Recognition with Large Language Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: However, challenges remain regarding stability, controllability, completeness, and fluency. To mitigate these issues, this paper proposes the Chain of Correction (CoC), which uses a multi-turn chat format to correct errors segment by segment, guided by pre-recognized text and full-text context for better semantic understanding. |
Z. TANG et. al. | icassp | 2026-05-04 |
| 362 | Inverse-Hessian Regularization for Continual Learning in ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we propose Inverse Hessian Regularization (IHR), a memory-free approach for CL in ASR that incorporates curvature information into the merging step. |
S. Vander Eeckt; H. Van Hamme; | icassp | 2026-05-04 |
| 363 | Frontend Token Enhancement for Token-Based Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we introduce a frontend system that estimates clean speech tokens from noisy speech and evaluate it on an ASR backend using semantic tokens. |
T. Ashihara; S. Horiguchi; K. Matsuura; T. Ochiai; M. Delcroix; | icassp | 2026-05-04 |
| 364 | CTC-DID: CTC-Based Arabic Dialect Identification for Streaming Applications Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper proposes a Dialect Identification (DID) approach inspired by the Connectionist Temporal Classification (CTC) loss function as used in Automatic Speech Recognition (ASR). |
M. U. Farooq; O. Saz; | icassp | 2026-05-04 |
| 365 | Whisper: Courtside Edition – Enhancing ASR Performance Through LLM-Driven Context Generation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce Whisper: Courtside Edition, a novel multi-agent large language model (LLM) pipeline tailored for the domain of basketball commentary, that enhances Whisper’s transcription accuracy without any model retraining. |
Y. Ron; S. Gilboa; T. Dubnov; | icassp | 2026-05-04 |
| 366 | Variational Low-Rank Adaptation for Personalized Impaired Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This work introduces a novel ASR personalization method based on Bayesian Low-rank Adaptation for data-efficient fine-tuning. |
N. Pokel; P. Moure; R. Boehringer; S. -C. Liu; Y. Gao; | icassp | 2026-05-04 |
| 367 | Dynamic Multi-Expert Projectors with Stabilized Routing for Multilingual Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: While effective in monolingual settings, a single projector struggles to capture the diverse acoustic-to-semantic mappings required for multilingual ASR. To address this, we propose SMEAR-MoE, a stabilized Mixture-of-Experts projector that ensures dense gradient flow to all experts, preventing expert collapse while enabling cross-lingual sharing. |
I. Pandey; A. Mittal; V. Bahuguna; G. Ramakrishnan; | icassp | 2026-05-04 |
| 368 | HarmoniFuse: A Component-Selective and Prompt-Adaptive Framework for Multi-Task Speech Language Modeling Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Such designs risk task interference and performance degradation, especially under limited data conditions. To address these limitations, we propose HarmoniFuse, a component-selective and prompt-adaptive framework for multi-task speech language modeling. |
Y. SI et. al. | icassp | 2026-05-04 |
| 369 | Evaluating ASR-LLM Setups for Japanese Speech Recognition with Multi-Pass Augmented Generative Error Correction Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This work explores how LLM-based GER can enhance and expand the capabilities of Japanese ASR, providing a systematic evaluation on Japanese data with 0.9-2.6k text utterances. |
Y. Ko; S. Li; C. -H. H. Yang; T. Kawahara; | icassp | 2026-05-04 |
| 370 | Sequence-Level Unsupervised Training in Speech Recognition: A Theoretical Study Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce two conditions under which unsupervised speech recognition is possible. |
Z. Yang; J. Barkoczi; R. Schlüter; H. Ney; | icassp | 2026-05-04 |
| 371 | A Dual-Channel ASR-LLM Architecture with A Progressive Training Strategy for Low-Resource Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To overcome this limitation and enable a more comprehensive flow of information, we propose a novel dual-channel architecture,termed DCPT, that provides the LLM with both information streams simultaneously, integrating acoustic evidence from the encoder as a foundational anchor with evolving linguistic hypotheses from the decoder as dynamic guidance. To train this complex system effectively, we introduce a progressive two-stage training strategy that first establishes a robust foundational alignment before learning the dynamic fusion, a decoupled approach that ensures stability and efficiency. |
Q. Ma; H. Jia; S. Huang; L. He; | icassp | 2026-05-04 |
| 372 | When Audio-Language Models Fail to Leverage Multimodal Context for Dysarthric Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce a benchmark built on the Speech Accessibility Project (SAP) dataset that tests whether diagnosis labels, clinician-derived speech ratings, and progressively richer clinical descriptions improve transcription accuracy for dysarthric speech. |
PEHUÉN MOURE et. al. | arxiv-cs.AI | 2026-05-04 |
| 373 | Learning to Align with Unbalanced Optimal Transport in Linguistic Knowledge Transfer for ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we propose an unbalanced optimal transport-based alignment model that explicitly handles distributional mismatch and structural asymmetries with soft and partial matching between acoustic and linguistic modalities. |
X. Lu; P. Shen; H. Kawai; | icassp | 2026-05-04 |
| 374 | Multi Stage Training with Dynamic Data Balancing for Multilingual Speech Recognition and Translation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Training large-scale multilingual speech models is often hindered by severe data imbalances across tasks, languages, and corpora. We introduce a systematic, multi-stage training framework to ad-dress this challenge. |
N. KOLUGURI et. al. | icassp | 2026-05-04 |
| 375 | TMS: Text-Prompted Multi-Channel Speech Separation on Smart Glasses Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: When conversational speech overlaps, the performance of on-device audio assistants on smart glasses degrades markedly. We address this with a text-prompted multichannel speech separation framework. |
Y. Liu; | icassp | 2026-05-04 |
| 376 | Whisper with Benefits: A Unified Approach to Speech and Speaker Attribute Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present a unified approach that extends Whisper, a state-of-the-art end-to-end ASR model, to simultaneously recognize linguistic content and speaker attributes, including gender, accent, age range, and roles such as doctor and patient, without additional computational overhead. |
R. N. Q. K. Duong; | icassp | 2026-05-04 |
| 377 | Mixtures of Lightweight Articulatory Experts for Multilingual Asr Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose mixtures of lightweight articulatory experts (MoLAE) trained with a novel multilabel articulatory CTC objective. |
M. Mimura; J. Lee; R. Magoshi; T. Kawahara; | icassp | 2026-05-04 |
| 378 | Multi-Task Learning For Speech Quality Assessment Using ASR-Derived Entropy Features Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This work introduces a novel approach that exploits entropy computed from automatic speech recognition (ASR) model predictions as a quality indicator for Non-reference speech quality assessment approach. |
T. D. Do; B. Thang Ta; V. H. Do; | icassp | 2026-05-04 |
| 379 | Federated Heterogeneous Language Model Optimization for Hybrid Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper proposes a heterogeneous LM optimization task and introduces a match-and-merge paradigm with two algorithms: the Genetic Match-and-Merge Algorithm (GMMA), using genetic operations to evolve and pair LMs, and the Reinforced Match-and-Merge Algorithm (RMMA), leveraging reinforcement learning for efficient convergence. |
M. Hong; | icassp | 2026-05-04 |
| 380 | Ara-BEST-RQ: Multi Dialectal Arabic SSL Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present Ara-BEST-RQ, a family of self-supervised learning (SSL) models specifically designed for multi-dialectal Arabic speech processing. |
H. Elleuch; R. Whetten; S. Mdhaffar; Y. Estève; F. Bougares; | icassp | 2026-05-04 |
| 381 | LESS: Large Language Model Enhanced Semi-Supervised Learning for Speech Foundational Models Using In-the-wild Data Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To address the challenges, we introduce LESS (Large Language Model Enhanced Semi-supervised Learning), a versatile framework that uses Large Language Models (LLMs) to correct pseudo-labels generated on in-the-wild data. |
W. Ding; F. Qian; | icassp | 2026-05-04 |
| 382 | Accent-Invariant Automatic Speech Recognition Via Saliency-Driven Spectrogram Masking Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Pre-trained transformer-based models have significantly advanced automatic speech recognition (ASR), yet they remain sensitive to accent and dialectal variations, resulting in elevated word error rates (WER) in linguistically diverse languages such as English and Persian. To address this challenge, we propose an accent-invariant ASR framework that integrates accent and dialect classification into the recognition pipeline. |
M. H. Sameti; S. H. Moridani; A. Zarean; H. Sameti; | icassp | 2026-05-04 |
| 383 | HATS: An Open Data Set Integrating Human Perception Applied to The Evaluation of Automatic Speech Recognition Metrics Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We investigated the relationship between human preferences and various ASR evaluation metrics, including lexical and embedding-based ones, the latter being those that correlate supposedly the most with human perception. |
Thibault Bañeras Roux; Jane Wottawa; Mickael Rouvier; Teva Merlin; Richard Dufour; | arxiv-cs.CL | 2026-04-30 |
| 384 | AppTek Call-Center Dialogues: A Multi-Accent Long-Form Benchmark for English ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This work presents the AppTek Call-Center Dialogues corpus, a collection of spontaneous, role-played agent-customer conversations spanning fourteen English accents covering sixteen service-oriented scenarios. |
EUGEN BECK et. al. | arxiv-cs.CL | 2026-04-30 |
| 385 | Qualitative Evaluation of Language Model Rescoring in Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we propose to study and understand the impact of rescoring using language models in ASR systems by means of several metrics often used in other natural language processing (NLP) tasks in addition to the WER. |
Thibault Bañeras-Roux; Mickaël Rouvier; Jane Wottawa; Richard Dufour; | arxiv-cs.CL | 2026-04-30 |
| 386 | Multimodal LLMs Are Not All You Need for Pediatric Speech Language Pathology Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose a cascading approach from binary classification to type, and symptom classification. |
Darren Fürst; Sebastian Steindl; Ulrich Schäfer; | arxiv-cs.CL | 2026-04-29 |
| 387 | Graph-Based Phonetic Error Correction of Noisy ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose a structured ASR correction framework, that we call G-SPIN, that combines phonetic graph modeling with contextual language understanding. |
Pratik Rakesh Singh; Mohammadi Zaki; Aneesh Mukkamala; Pankaj Wasnik; | arxiv-cs.CL | 2026-04-29 |
| 388 | StarDrinks: An English and Korean Test Set for SLU Evaluation in A Drink Ordering Scenario Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Drink ordering, for example, involves diverse named entities, drink types, sizes, customizations, and brand-specific terminology, as well as spontaneous speech phenomena such as hesitations and self-corrections. To address this gap, we introduce StarDrinks, a test set in English and Korean containing speech utterances features, transcriptions, and annotated slots. |
MARCELY ZANON BOITO et. al. | arxiv-cs.CL | 2026-04-29 |
| 389 | Few-Shot Accent Synthesis for ASR with LLM-Guided Phoneme Editing Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose a pipeline that adapts a text-to-speech (TTS) decoder to a target-accent speaker using fewer than ten reference utterances and employs large language model (LLM)-based phoneme editing to generate accent-conditioned pronunciations. |
Yurii Halychanskyi; Nimet Beyza Bozdag; Mark Hasegawa-Johnson; Dilek Hakkani-Tür; Volodymyr Kindratenko; | arxiv-cs.SD | 2026-04-29 |
| 390 | Assessing The Reliability, Accuracy, and Relevance of Artificial Intelligence Speech Recognition for Clinical Documentation: A Scoping Review Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Methods The scoping review employed the methodology developed by Arksey and O’Malley in 2005 and further expanded by Levac and Colquhoun 2010. |
Samuel Atiku; Kehinde Owolanke; Olufisayo Olakotan; | Journal of Evaluation in Clinical Practice | 2026-04-29 |
| 391 | WhisperPipe: A Resource-Efficient Streaming Architecture for Real-Time Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present WhisperPipe, a novel streaming architecture that achieves bounded memory consumption while maintaining transcription quality through three key innovations a hybrid Voice Activity Detection (VAD) pipeline combining Silero VAD with energy-based filtering to reduce false activations by 34%, a dynamic buffering mechanism with overlapping context windows that prevents information loss at segment boundaries, and an adaptive processing strategy that balances latency and accuracy based on speech characteristics. |
Erfan Ramezani; Mohammad Mahdi Giahi; Mohammad Erfan Zarabadipour; Amir Reza Yosefian; Hamid Ghadiri; | arxiv-cs.CL | 2026-04-28 |
| 392 | A Comprehensive Review of Speech Recognition Systems: Architectures, Techniques, and Future Directions Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents a comprehensive review of speech recognition systems, focusing on architectural developments, modeling techniques, datasets, evaluation metrics, and applications. |
Geddamuri Purna Narayanamma; Nandamuri Navya Sri Madhu Viharika; Tetala Shaharsha Devi; Miriyala Kavya; | INTERNATIONAL JOURNAL OF SCIENTIFIC RESEARCH IN ENGINEERING … | 2026-04-27 |
| 393 | AI-Driven Web-Based Video Conferencing System with Real-Time Meeting Intelligence and Analytics Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents an AI-Driven Web-Based Video Conferencing System that integrates real-time communication with an embedded artificial intelligence analytics pipeline. |
Asina Begam A; Ramya K; Anushiya V; Dharshani A; Dharshini T; | INTERNATIONAL JOURNAL OF SCIENTIFIC RESEARCH IN ENGINEERING … | 2026-04-27 |
| 394 | RAS: A Reliability Oriented Metric for Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce an abstention-aware transcription framework that enables ASR models to explicitly abstain from uncertain segments. |
WENBIN HUANG et. al. | arxiv-cs.SD | 2026-04-27 |
| 395 | Applying Deep Learning Algorithms for Speech Recognition in Speech-Impaired Children Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper investigates the application of deep learning algorithms for automatic speech recognition (ASR) specifically adapted to the speech patterns of children with speech-language disorders. |
Madagala Surya Satya Sai Anjali4 ; Dangeti Praveena1,; Guthi Lalitha Veera Bharathi2; Inti Visalakshi3 ; | INTERNATIONAL JOURNAL OF SCIENTIFIC RESEARCH IN ENGINEERING … | 2026-04-27 |
| 396 | 2nd of The 5th PVUW MeViS-Audio Track: ASR-SaSaSa2VA Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Despite their promise, these approaches are computationally intensive, struggle with aligning temporal audio cues to dynamic video content, and depend on large paired audio-video datasets. To address these challenges, we present ASR-SaSaSa2VA, a resource-efficient framework for audio-guided video segmentation. |
Zhiyu Wang; Xudong Kang; Shutao Li; | arxiv-cs.CV | 2026-04-26 |
| 397 | Identifying and Typifying Demographic Unfairness in Phoneme-level Embeddings of Self-supervised Speech Recognition Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper proposes a framework typifying two types of error that can occur in modeling phonemes in ASR systems: random error/high variance in phoneme embedding, vs systematic error/embedding bias. |
Felix Herron; Solange Rossato; Alexandre Allauzen; François Portet; | arxiv-cs.CL | 2026-04-24 |
| 398 | Evaluation of Automatic Speech Recognition Using Generative Large Language Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper evaluates their relevance through three approaches: (1) selecting the best hypothesis between two candidates, (2) computing semantic distance using generative embeddings, and (3) qualitative classification of errors. |
THIBAULT BAÑERAS-ROUX et. al. | arxiv-cs.CL | 2026-04-23 |
| 399 | UAF: A Unified Audio Front-end LLM for Full-Duplex Speech Interaction Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In our development of speech assistant, we observed that optimizing the speech front-end is equally crucial as advancing the back-end unified model for achieving seamless, responsive interactions. To bridge this gap, we propose the first unified audio front-end LLM (UAF) tailored for full-duplex speech systems. |
Yadong Li; Guoxin Wu; Haiping Hou; Biye Li; | arxiv-cs.AI | 2026-04-21 |
| 400 | Voice of India: A Large-Scale Benchmark for Real-World Speech Recognition in India Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In addition, strict single reference WER penalizes natural spelling variation in Indian languages, including non standardized spellings of code-mixed English origin words. To address these limitations, we introduce Voice of India, a closed source benchmark built from unscripted telephonic conversations covering 15 major Indian languages across 139 regional clusters. |
KAUSHAL BHOGALE et. al. | arxiv-cs.CL | 2026-04-21 |
| 401 | Hard to Be Heard: Phoneme-Level ASR Analysis of Phonologically Complex, Low-Resource Endangered Languages Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present a phoneme-level analysis of automatic speech recognition (ASR) for two low-resourced and phonologically complex East Caucasian languages, Archi and Rutul, based on curated and standardized speech-transcript resources totaling approximately 50 minutes and 1 hour 20 minutes of audio, respectively. |
V. S. D. S. Mahesh Akavarapu; Michael Daniel; Gerhard Jäger; | arxiv-cs.CL | 2026-04-20 |
| 402 | Where Do Self-Supervised Speech Models Become Unfair? Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To our knowledge, we present the first layerwise fairness analysis of pretrained self-supervised speech encoder models (S3Ms), probing each embedding layer for speaker identification (SID) automatic speech recognition (ASR). |
Felix Herron; Maja Hjuler; Solange Rossato; Alexandre Allauzen; François Portet; | arxiv-cs.CL | 2026-04-20 |
| 403 | Efficient Punctuation Restoration Via Weighted Lookahead Scoring Method for Streaming ASR Systems Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper proposes a non-autoregressive scoring method (no free-form generation) that preserves the input transcript and makes a decision at each word boundary. |
Sungmook Woo; Hyungu Kang; Chanwoo Kim; | arxiv-cs.CL | 2026-04-17 |
| 404 | MUSCAT: MUltilingual, SCientific ConversATion Benchmark Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: However, there is currently no dataset benchmarking this situation. We propose a new benchmark to evaluate current Automatic Speech Recognition (ASR) systems, whether they are able to handle these challenges. |
SUPRITI SINHAMAHAPATRA et. al. | arxiv-cs.CL | 2026-04-17 |
| 405 | Elderly-Contextual Data Augmentation Via Speech Synthesis for Elderly ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we address data scarcity in EASR through a data augmentation pipeline that combines large language model (LLM)-based transcript paraphrasing with text-to-speech (TTS) synthesis. |
MINSIK LEE et. al. | arxiv-cs.CL | 2026-04-15 |
| 406 | Pushing The Limits of On-Device Streaming ASR: A Compact, High-Accuracy English Model for Low-Latency Inference Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Through a comprehensive benchmark of over 50 configurations spanning OpenAI Whisper, NVIDIA Nemotron, Parakeet TDT, Canary, Conformer Transducer, and Qwen3-ASR, we identify NVIDIA’s Nemotron Speech Streaming as the strongest candidate for real-time English streaming on resource-constrained hardware. |
NENAD BANFIC et. al. | arxiv-cs.AI | 2026-04-15 |
| 407 | Diffusion Language Models for Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we explore variants for their use in speech recognition. |
Davyd Naveriani; Albert Zeyer; Ralf Schlüter; Hermann Ney; | arxiv-cs.CL | 2026-04-15 |
| 408 | Multilingual Conference Audio Transcription, Speaker Diarization And Translation System Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Conventional approaches based on manual transcription, speaker identification, and translation are inefficient, costly, and susceptible to errors caused by overlapping speech, accents, background noise, and domain-specific terminology. To overcome these limitations, this work presents an AI-driven framework that integrates Automatic Speech Recognition (ASR), speaker diarization, and Neural Machine Translation (NMT). |
Ms.S Aswini; | International Research Journal on Advanced Engineering Hub … | 2026-04-13 |
| 409 | BlasBench: An Open Benchmark for Irish Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: The best open model (omniASR LLM 7B) achieves 30.65% WER on Common Voice and 39.09% on FLEURS. |
Jyoutir Raj; John Conway; | arxiv-cs.CL | 2026-04-12 |
| 410 | From Speech to Profile: A Protocol-Driven LLM Agent for Psychological Profile Generation Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: The psychological profile that structurally documents the case of a depression patient is essential for psychotherapy. Large language models can be applied to summarize the … |
XINGJIAN YANG et. al. | arxiv-cs.SD | 2026-04-11 |
| 411 | Few-Shot Contrastive Adaptation for Audio Abuse Detection in Low-Resource Indic Languages Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We investigate whether Contrastive Language-Audio Pre-training (CLAP) can support abusive speech detection directly from audio. |
Aditya Narayan Sankaran; Reza Farahbakhsh; Noel Crespi; | arxiv-cs.SD | 2026-04-10 |
| 412 | Transcribing Children’s Speech: ASR Performance and Obtaining Reliable Orthographic Transcriptions Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Automatic speech recognition (ASR) has the potential to substantially reduce manual annotation effort in child speech research by generating automatic transcriptions. |
Gus Lathouwers; Lingyun Gao; Catia Cucchiarini; Helmer Strik; | arxiv-cs.CL | 2026-04-10 |
| 413 | Interactive ASR: Towards Human-Like Interaction and Semantic Coherence Evaluation for Agentic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Second, interactive correction-an essential component of human communication-has rarely been systematically studied in ASR research. In this paper, we integrate these two perspectives under an agentic framework for interactive ASR. |
PENG WANG et. al. | arxiv-cs.CL | 2026-04-10 |
| 414 | Closing The Speech-Text Gap with Limited Audio for Effective Domain Adaptation in LLM-Based ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: However, this introduces a modality gap, as the LLM is not exposed to the noisy representations produced by the speech projector. We investigate whether small amounts of speech can mitigate this mismatch. |
THIBAULT BAÑERAS-ROUX et. al. | arxiv-cs.CL | 2026-04-07 |
| 415 | Benchmarking Multilingual Speech Models on Pashto: Zero-Shot ASR, Script Failure, and Cross-Domain Evaluation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper reports the first reproducible multi-model evaluation on public Pashto data, covering zero-shot ASR, script-level failure, and cross-domain evaluation of fine-tuned models. |
Hanif Rahman; | arxiv-cs.CL | 2026-04-06 |
| 416 | AI-Based Inclusive Assistive And Learning Support Platform For Visually Impaired And Autism Students Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper discusses the design and development of an Artificial Intelligence (AI) enabled platform for assistive and learning support in an inclusive education context. |
Dr. A. Abdul Azeez Khan; Dr. K. Javubar Sathick; Mohamed Aaqil M.H; Pooja .K; | International Research Journal on Advanced Engineering Hub … | 2026-04-03 |
| 417 | Development and Multi-center Evaluation of Domain-adapted Speech Recognition for Human-AI Teaming in Real-world Gastrointestinal Endoscopy Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Here, we present EndoASR, a domain-adapted ASR system designed for real-time deployment in endoscopic workflows. |
RUIJIE YANG et. al. | arxiv-cs.CL | 2026-04-02 |
| 418 | CV-18 NER: Augmented Common Voice for Named Entity Recognition from Arabic Speech Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce CV-18 NER, the first publicly available dataset for NER from Arabic speech, created by augmenting the Arabic Common Voice 18 corpus with manual NER annotations following the fine-grained Wojood schema (21 entity types). |
Youssef Saidi; Haroun Elleuch; Fethi Bougares; | arxiv-cs.CL | 2026-04-02 |
| 419 | Speech LLMs Are Contextual Reasoning Transcribers Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To reduce the modality gap, this paper introduces a CTC-guided Modality Adapter, which uses CTC non-blank token probabilities to weight LLM embeddings, efficiently aligning speech encoder outputs with the LLM’s textual latent space. |
Keqi Deng; Ruchao Fan; Bo Ren; Yiming Wang; Jinyu Li; | arxiv-cs.CL | 2026-04-01 |
| 420 | Adapting Text LLMs to Speech Via Multimodal Depth Up-Scaling Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose Multimodal Depth Upscaling, an extension of an emerging strategy in continual LLM pre-training, where new transformer layers are inserted into a frozen text LLM and only the added layers are trained on speech data. |
Kazuki Yano; Jun Suzuki; Shinji Watanabe; | arxiv-cs.CL | 2026-04-01 |
| 421 | A Transformer-Based Method for Bidirectional French–Lingala Machine Translation in Speech and Text Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we propose a deep neural network pipeline for bidirectional French–Lingala automatic translation, covering both text-to-text and voice-to-text scenarios, by integrating Long Short-Term Memory (LSTM) and Transformer models on a specialized parallel corpus. |
REAGAN E. MANDIYA et. al. | Applied Sciences | 2026-03-31 |
| 422 | FLEURS-Kobani: Extending The FLEURS Dataset for Northern Kurdish Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present FLEURS-Kobani, a Northern Kurdish (ISO 639-3 KMR) spoken extension of the FLEURS benchmark. |
Daban Q. Jaff; Mohammad Mohammadamini; | arxiv-cs.CL | 2026-03-31 |
| 423 | Investigation on The Robustness of Acoustic Foundation Models on Post Exercise Speech Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we benchmark acoustic foundation models on post-exercise speech under a unified evaluation protocol. |
XIANGYUAN XUE et. al. | arxiv-cs.SD | 2026-03-29 |
| 424 | On The Role of Encoder Depth: Pruning Whisper and LoRA Fine-Tuning in SLAM-ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we analyze the effects of layer pruning in the Whisper encoder when used as the acoustic backbone of SLAM-ASR. |
Ganesh Pavan Kartikeya Bharadwaj Kolluri; Michael Kampouridis; Ravi Shekhar; | arxiv-cs.CL | 2026-03-29 |
| 425 | Two-Stage Acoustic Adaptation with Gated Cross-Attention Adapters for LLM-Based Multi-Talker Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: While acoustic-enriched prompts outperform the SOT-only baseline, prefix-only conditioning remains inadequate for three-talker mixtures. We therefore propose a lightweight gated residual cross-attention adapter and design a two-stage acoustic adaptation framework based on low-rank updates (LoRA). |
Hao Shi; Yuan Gao; Xugang Lu; Tatsuya Kawahara; | arxiv-cs.SD | 2026-03-28 |
| 426 | Distilling Conversations: Abstract Compression of Conversational Audio Context for LLM-based ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we study whether multimodal context from prior turns improves LLM-based ASR and how to represent that context efficiently. |
SHASHI KUMAR et. al. | arxiv-cs.CL | 2026-03-27 |
| 427 | Automatic Speech Recognition for Documenting Endangered Languages: Case Study of Ikema Miyakoan Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present an ongoing effort to develop an ASR system for Ikema based on field recordings. |
Chihiro Taguchi; Yukinori Takubo; David Chiang; | arxiv-cs.CL | 2026-03-27 |
| 428 | Back to Basics: Revisiting ASR in The Age of Voice Agents Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce WildASR, a multilingual (four-language) diagnostic benchmark sourced entirely from real human speech that factorizes ASR robustness along three axes: environmental degradation, demographic shift, and linguistic diversity. |
GEEYANG TAY et. al. | arxiv-cs.AI | 2026-03-26 |
| 429 | CLAR: CIF-Localized Alignment for Retrieval-Augmented Speech LLM-Based Contextual ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose CLAR, a dual-encoder speech-text retriever that uses Continuous Integrate-and-Fire (CIF) to learn monotonic token-level alignments without timestamps. |
Shangkun Huang; Huan Shen; Wei Zou; Yunzhang Chen; | arxiv-cs.SD | 2026-03-26 |
| 430 | Goodness-of-pronunciation Without Phoneme Time Alignment Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper proposes to overcome incompatibilities for feature extraction with weakly-supervised models, easing expansion of speech evaluation to low-resource languages. |
Jeremy H. M. Wong; Nancy F. Chen; | arxiv-cs.CL | 2026-03-26 |
| 431 | A Sociolinguistic Analysis of Automatic Speech Recognition Bias in Newcastle English Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This study investigates ASR bias through a sociolinguistic analysis of Newcastle English, a regional variety of North-East England that has been shown to challenge current speech recognition technologies. |
Dana Serditova; Kevin Tang; | arxiv-cs.CL | 2026-03-25 |
| 432 | Ethio-ASR: Joint Multilingual Speech Recognition and Language Identification for Ethiopian Languages Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present Ethio-ASR, a suite of multilingual CTC-based automatic speech recognition (ASR) models jointly trained on five Ethiopian languages: Amharic, Tigrinya, Oromo, Sidaama, and Wolaytta. |
BADR M. ABDULLAH et. al. | arxiv-cs.CL | 2026-03-24 |
| 433 | Precision-Varying Prediction (PVP): Robustifying ASR Systems Against Adversarial Attacks Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We observe that changing the precision of an ASR model during inference reduces the likelihood of adversarial attacks succeeding. |
Matías Pizarro; Raghavan Narasimhan; Asja Fischer; | arxiv-cs.LG | 2026-03-23 |
| 434 | Who Spoke What When? Evaluating Spoken Language Models for Conversational ASR with Semantic and Overlap-Aware Metrics Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To capture meaning-altering errors that conventional metrics miss, we introduce tcpSemER, which extends tcpWER by replacing Levenshtein distance with embedding-based semantic similarity. |
NAOHIRO TAWARA et. al. | arxiv-cs.CL | 2026-03-23 |
| 435 | SeaAlert: Critical Information Extraction From Maritime Distress Communications with Large Language Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents SeaAlert, an LLM-based framework for robust analysis of maritime distress communications. |
Tomer Atia; Yehudit Aperstein; Alexander Apartsin; | arxiv-cs.CL | 2026-03-23 |
| 436 | Ara-Best-RQ: Multi Dialectal Arabic SSL Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present Ara-BEST-RQ, a family of self-supervised learning (SSL) models specifically designed for multi-dialectal Arabic speech processing. |
Haroun Elleuch; Ryan Whetten; Salima Mdhaffar; Yannick Estève; Fethi Bougares; | arxiv-cs.CL | 2026-03-23 |
| 437 | LoASR-Bench: Evaluating Large Speech Language Models on Low-Resource Automatic Speech Recognition Across Language Families Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: As a result, it is essential to evaluate SpeechLMs on low-resource languages to ensure their generalizability across different language families. To address this problem, we propose \textbf{LoASR-Bench}, a comprehensive benchmark designed to evaluate \textbf{lo}w-resource \textbf{a}utomatic \textbf{s}peech \textbf{r}ecognition (\textbf{ASR}) of the latest SpeechLMs across diverse language families. |
Jianan Chen; Xiaoxue Gao; Tatsuya Kawahara; Nancy F. Chen; | arxiv-cs.CL | 2026-03-20 |
| 438 | Demonstration of Adapt4Me: An Uncertainty-Aware Authoring Environment for Personalizing Automatic Speech Recognition to Non-normative Speech Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Personalizing Automatic Speech Recognition (ASR) for non-normative speech remains challenging because data collection is labor-intensive and model training is technically complex. To address these limitations, we propose Adapt4Me, a web-based decentralized environment that operationalizes Bayesian active learning to enable end-to-end personalization without expert supervision. |
Niclas Pokel; Yiming Zhao; Pehuén Moure; Yingqiang Gao; Roman Böhringer; | arxiv-cs.HC | 2026-03-20 |
| 439 | Zipper-LoRA: Dynamic Parameter Decoupling for Speech-LLM Based Multilingual Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In such scenarios, a stability-plasticity dilemma often arises: fully shared Parameter-Efficient Fine-Tuning (PEFT) can cause negative inter-lingual interference for under-represented languages, while fully language-specific tuning limits the cross-lingual beneficial knowledge transfer needed for low-resource tasks. To address this, we propose Zipper-LoRA, a novel rank-level decoupling framework with three variants (Static, Hard, and Soft) that dynamically synthesizes LoRA updates from shared and language-specific subspaces. |
Yuxiang Mei; Delai Qiu; Shengping Liu; Jiaen Liang; Yanhua Long; | arxiv-cs.CL | 2026-03-18 |
| 440 | RECOVER: Robust Entity Correction Via Agentic Orchestration of Hypothesis Variants for Evidence-based Recovery Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: If the entities are entirely absent from the ASR output, post-ASR correction becomes difficult. To address this, we introduce RECOVER, an agentic correction framework that serves as a tool-using agent. |
Abhishek Kumar; Aashraya Sachdeva; | arxiv-cs.CL | 2026-03-17 |
| 441 | Is Semi-Automatic Transcription Useful in Corpus Creation? Preliminary Considerations on The KIParla Corpus Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Analyses combining alignment-based metrics, descriptive statistics and statistical modeling provide a systematic framework to monitor transcription behavior across annotators and workflows. |
Martina Simonotti; Ludovica Pannitto; Eleonora Zucchini; Silvia Ballarè; Caterina Mauri; | arxiv-cs.CL | 2026-03-17 |
| 442 | Polyglot-Lion: Efficient Multilingual ASR for Singapore Via Balanced Fine-Tuning of Qwen3-ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present Polyglot-Lion, a family of compact multilingual automatic speech recognition (ASR) models tailored for the linguistic landscape of Singapore, covering English, Mandarin, Tamil, and Malay. |
Quy-Anh Dang; Chris Ngo; | arxiv-cs.CL | 2026-03-17 |
| 443 | Lost in Transcription: Subtitle Errors in Automatic Speech Recognition Reduce Speaker and Content Evaluations Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we examined how subtitle errors affect evaluations of speakers and their content using a preregistered online experiment (N=207, U.S.-based crowdworkers). |
Kowe Kadoma; Priyal Shrivastava; Mor Naaman; | arxiv-cs.HC | 2026-03-16 |
| 444 | Tagarela – A Portuguese Speech Dataset from Podcasts Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Despite significant advances in speech processing, Portuguese remains under-resourced due to the scarcity of public, large-scale, and high-quality datasets. To address this gap, we present a new dataset, named TAGARELA, composed of over 8,972 hours of podcast audio, specifically curated for training automatic speech recognition (ASR) and text-to-speech (TTS) models. |
FREDERICO SANTOS DE OLIVEIRA et. al. | arxiv-cs.CL | 2026-03-16 |
| 445 | Two-Stage Adaptation for Non-Normative Speech Recognition: Revisiting Speaker-Independent Initialization for Personalization Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we propose a two-stage adaptation framework consisting of speaker-independent fine-tuning (SI-FT) on multi-speaker non-normative data followed by SS-FT, and evaluate it through a controlled comparison with direct SS-FT under identical per-speaker conditions. |
Shan Jiang; Jiawen Qi; Chuanbing Huo; Yingqiang Gao; Qinyu Chen; | arxiv-cs.SD | 2026-03-16 |
| 446 | Vietnamese Automatic Speech Recognition: A Revisit Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: For low-resource languages, existing open-source ASR datasets often suffer from insufficient quality and inconsistent annotation, hindering the development of robust models. To address these challenges, we propose a novel and generalizable data aggregation and preprocessing pipeline designed to construct high-quality ASR datasets from diverse, potentially noisy, open-source sources. |
Thi Vu; Linh The Nguyen; Dat Quoc Nguyen; | arxiv-cs.CL | 2026-03-15 |
| 447 | TASTE-Streaming: Towards Streamable Text-Aligned Speech Tokenization and Embedding for Spoken Language Modeling Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: However, the dependence on an external ASR system and the use of a non-causal decoder limits streaming use. To address this limitation, we propose TASTE-S, a streamable extension of TASTE suitable for real-time usage. |
Liang-Hsuan Tseng; Hung-yi Lee; | arxiv-cs.CL | 2026-03-12 |
| 448 | Distilling LLM Semantic Priors Into Encoder-Only Multi-Talker ASR with Talker-Count Routing Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Large language models (LLMs) provide strong semantic priors that can improve multi-talker automatic speech recognition (MT-ASR), but using an LLM as an autoregressive decoder is computationally expensive and remains fragile under heavy overlap. In this paper, we propose an encoder-only MT-ASR framework that adapts an LLM to multi-talker conditioning and distills its semantic guidance into the encoder during training, while retaining fast CTC-style decoding at inference. |
HAO SHI et. al. | arxiv-cs.SD | 2026-03-11 |
| 449 | Huntington Disease Automatic Speech Recognition with Biomarker Supervision Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present a systematic HD-ASR study using a high-fidelity clinical speech corpus not previously used for end-to-end ASR training. |
Charles L. Wang; Cady Chen; Ziwei Gong; Julia Hirschberg; | arxiv-cs.LG | 2026-03-11 |
| 450 | Continued Pretraining for Low-Resource Swahili ASR: Achieving State-of-the-Art Performance with Minimal Labeled Data Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We investigate continued pretraining (CPT) for adapting wav2vec2-bert-2.0 to Swahili automatic speech recognition (ASR). |
Hillary Mutisya; John Mugane; | arxiv-cs.SD | 2026-03-11 |
| 451 | Uni-ASR: Unified LLM-Based Architecture for Non-Streaming and Streaming Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we propose Uni-ASR, a unified framework based on LLMs that integrates both non-streaming and streaming speech recognition capabilities. |
Yinfeng Xia; Jian Tang; Junfeng Hou; Gaopeng Xu; Haitao Yao; | arxiv-cs.SD | 2026-03-11 |
| 452 | Duration Aware Scheduling for ASR Serving Under Workload Drift Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Scheduling policies in large-scale Automatic Speech Recognition (ASR) serving pipelines play a key role in determining end-to-end (E2E) latency. |
Darshan Makwana; Yash Jogi; Harsh Kotta; Aayush Kubba; | arxiv-cs.LG | 2026-03-11 |
| 453 | AlphaFlowTSE: One-Step Generative Target Speaker Extraction Via Conditional AlphaFlow Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present AlphaFlowTSE, a one-step conditional generative model trained with a Jacobian-vector product (JVP)-free AlphaFlow objective. |
DUOJIA LI et. al. | arxiv-cs.SD | 2026-03-11 |
| 454 | SCENEBench: An Audio Understanding Benchmark Grounded in Assistive and Industrial Use Cases Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We assess five state-of-the-art LALMs and find critical gaps: performance varies across tasks, with some tasks performing below random chance and others achieving high accuracy. |
Laya Iyer; Angelina Wang; Sanmi Koyejo; | arxiv-cs.SD | 2026-03-10 |
| 455 | Ramsa: A Large Sociolinguistically Rich Emirati Arabic Speech Corpus for ASR and TTS Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: The paper highlights the challenges encountered and provides directions for future work. |
Rania Al-Sabbagh; | arxiv-cs.CL | 2026-03-09 |
| 456 | Nwāchā Munā: A Devanagari Speech Corpus and Proximal Transfer Benchmark for Nepal Bhasha ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we introduce Nwāchā Munā, a newly curated 5.39-hour manually transcribed Devanagari speech corpus for Nepal Bhasha, and establish the first benchmark using script-preserving acoustic modeling. |
RISHIKESH KUMAR SHARMA et. al. | arxiv-cs.CL | 2026-03-08 |
| 457 | Seeing The Context: Rich Visual Context-Aware Speech Recognition Via Multimodal Reasoning Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Current AVSR approaches primarily focus on lip motion, largely overlooking rich context present in the video such as speaking scene and on-screen text. To tackle such CAVSR (AVSR including rich visual Context), we propose VASR designed to see and reason the visual context to improve speech recognition. |
WENJIE TIAN et. al. | arxiv-cs.SD | 2026-03-07 |
| 458 | The Talking Robot: Distortion-Robust Acoustic Models for Robot-Robot Communication Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present Artoo, a learned acoustic communication system for robots that replaces hand-designed signal processing with end-to-end co-trained neural networks. |
Hanlong Li; Karishma Kamalahasan; Jiahui Li; Kazuhiro Nakadai; Shreyas Kousik; | arxiv-cs.RO | 2026-03-07 |
| 459 | Recognition of Speech Using Deep Learning-based Optimization Technique Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, a new technique called Fractional Super Bilterling Optimization with Random Multimodal Deep Learning-Convolutional Neural Network (FrSBO_RMDL-CNN) is proposed for speech recognition from unclear speech signals. |
Raja Bhargava; N. Arivazhagan; Kunchala Suresh babu; | International Journal of Wavelets, Multiresolution and … | 2026-03-06 |
| 460 | Speak in Context: Multilingual ASR with Speech Context Alignment Via Contrastive Learning Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we introduce a context-aware multilingual ASR framework that supports diverse languages and accents while preserving the modularity of pretrained models. |
Yuchen Zhang; Haralambos Mouratidis; Ravi Shekhar; | arxiv-cs.CL | 2026-03-06 |
| 461 | Language-Aware Distillation for Multilingual Instruction-Following Speech LLMs with ASR-Only Supervision Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: While recent distillation-based approaches train performant English-only Speech LLMs using only annotated ASR data by aligning text and speech using only a lightweight projector, these models under-perform when scaled to multilingual settings due to language interference in the shared projector. We address this by introducing language-aware distillation using a query bank and a gating network that selects or mixes query tokens using a Q-Former projector. |
SHREYAS GOPAL et. al. | arxiv-cs.CL | 2026-03-06 |
| 462 | TW-Sound580K: A Regional Audio-Text Dataset with Verification-Guided Curation for Localized Audio-Language Modeling Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present TW-Sound580K, a Taiwanese audio-text instruction dataset developed through a Verify-Generate-Critique (VGC) protocol. |
HAO-HUI XIE et. al. | arxiv-cs.SD | 2026-03-05 |
| 463 | Beyond Word Error Rate: Auditing The Diversity Tax in Speech Recognition Through Dataset Cartography Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we explore the limitations of relying solely on lexical counts by systematically evaluating a broader class of non-linear and semantic metrics. |
Ting-Hui Cheng; Line H. Clemmensen; Sneha Das; | arxiv-cs.LG | 2026-03-05 |
| 464 | Which Data Matter? Embedding-Based Data Selection for Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: While such heterogeneous data benefit generalist models designed for broad deployment, they pose challenges for specialist models targeting specific domains: specialist models lack the capacity to learn from all available data, and one must pay closer attention to addressing the mismatch between training and test conditions. In this work, we study targeted data selection as a strategy to address these challenges, selecting relevant subsets from 100k hours of in-the-wild training data to optimize performance on target domains. |
ZAKARIA ALDENEH et. al. | arxiv-cs.SD | 2026-03-05 |
| 465 | Boosting ASR Robustness Via Test-Time Reinforcement Learning with Audio-Text Semantic Rewards Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: However, by treating model confidence as a learning signal, these methods may reinforce high-confidence errors, leading to confirmation bias that undermines adaptation. To overcome these limitations, we present ASR-TRA, a novel Test-time Reinforcement Adaptation framework inspired by causal intervention. |
Linghan Fang; Tianxin Xie; Li Liu; | arxiv-cs.SD | 2026-03-05 |
| 466 | Measuring The Redundancy of Decoder Layers in SpeechLLMs Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Speech Large Language Models route speech encoder representations into an LLM decoder that typically accounts for over 90% of total parameters. We study how much of this decoder capacity is actually needed for speech tasks. |
Adel Moumen; Guangzhi Sun; Philip C Woodland; | arxiv-cs.CL | 2026-03-05 |
| 467 | Federated Heterogeneous Language Model Optimization for Hybrid Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper proposes a heterogeneous LM optimization task and introduces a match-and-merge paradigm with two algorithms: the Genetic Match-and-Merge Algorithm (GMMA), using genetic operations to evolve and pair LMs, and the Reinforced Match-and-Merge Algorithm (RMMA), leveraging reinforcement learning for efficient convergence. |
MENGZE HONG et. al. | arxiv-cs.CL | 2026-03-05 |
| 468 | FLAMA: Frame-Level Alignment Margin Attack for Scene Text and Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: As a result, decisive frames are treated implicitly, and optimization can become unnecessarily diffuse over long input sequences, hindering convergence and perceptual quality. To address the above issues, we propose FLAMA, a unified Frame-Level Alignment Margin Attack, which could be used for both STR and ASR models. |
Yikun Xu; Zhiheng Xu; Pengwen Dai; | Electronics | 2026-03-04 |
| 469 | When Denoising Hinders: Revisiting Zero-Shot ASR with SAM-Audio and Whisper Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Recent advances in automatic speech recognition (ASR) and speech enhancement have led to a widespread assumption that improving perceptual audio quality should directly benefit recognition accuracy. In this work, we rigorously examine whether this assumption holds for modern zero-shot ASR systems. |
Akif Islam; Raufun Nahar; Md. Ekramul Hamid; | arxiv-cs.SD | 2026-03-04 |
| 470 | WhisperAlign: Word-Boundary-Aware ASR and WhisperX-Anchored Pyannote Diarization for Long-Form Bengali Speech Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents our solution for the DL Sprint 4.0, addressing the dual challenges of Bengali Long-Form Speech Recognition (Task 1) and Speaker Diarization (Task 2). |
Aurchi Chowdhury; Rubaiyat -E-Zaman; Sk. Ashrafuzzaman Nafees; | arxiv-cs.SD | 2026-03-04 |
| 471 | Speech Recognition Assisted By Large Language Models to Command Software Orally — Application to An Augmented and Virtual Reality Web App for Immersive Molecular Graphics Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This project successfully developed, evaluated and integrated a Voice User Interface (VUI) into a web application that we are developing for immersive molecular graphics. |
Fabio Cortes Rodriguez; Luciano Abriata; | arxiv-cs.HC | 2026-03-03 |
| 472 | An Investigation Into Various Approaches For Bengali Long-Form Speech Transcription and Bengali Speaker Diarization Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents a multistage approach developed for the DL Sprint 4.0 – Bengali Long-Form Speech Recognition and DL Sprint 4.0 – Bengali Speaker Diarization competitions on Kaggle, addressing the challenge of who spoke when/what in hour-long recordings. |
Epshita Jahan; Khandoker Md Tanjinul Islam; Pritom Biswas; Tafsir Al Nafin; | arxiv-cs.SD | 2026-03-03 |
| 473 | Sequence-Level Unsupervised Training in Speech Recognition: A Theoretical Study Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce two conditions under which unsupervised speech recognition is possible. |
Zijian Yang; Jörg Barkoczi; Ralf Schlüter; Hermann Ney; | arxiv-cs.SD | 2026-03-02 |
| 474 | From Black Box to Glass Box: Cross-Model ASR Disagreement to Prioto Review in Ambient AI Scribe Documentation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We investigate whether cross-model disagreement among heterogeneous ASR systems can act as a reference-free uncertainty signal to prioritize human verification in medical transcription workflows. |
Abdolamir Karbalaie; Fernando Seoane; Farhad Abtahi; | arxiv-cs.SD | 2026-03-02 |
| 475 | GLoRIA: Gated Low-Rank Interpretable Adaptation for Dialectal ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose GLoRIA, a parameter-efficient adaptation framework that leverages metadata (e.g., coordinates) to modulate low-rank updates in a pre-trained encoder. |
Pouya Mehralian; Melissa Farasyn; Anne Breitbarth; Anne-Sophie Ghyselen; Hugo Van hamme; | arxiv-cs.CL | 2026-03-02 |
| 476 | VietSuperSpeech: A Large-Scale Vietnamese Conversational Speech Dataset for ASR Fine-Tuning in Chatbot, Customer Support, and Call Center Applications Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce VietSuperSpeech, a large-scale Vietnamese automatic speech recognition (ASR) dataset of 52,023 audio-text pairs totaling 267.39 hours, with a distinctive focus on casual conversational speech. |
LOAN DO et. al. | arxiv-cs.SD | 2026-03-02 |
| 477 | RO-N3WS: Enhancing Generalization in Low-Resource ASR with Diverse Romanian Speech Benchmarks Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce RO-N3WS, a benchmark Romanian speech dataset designed to improve generalization in automatic speech recognition (ASR), particularly in low-resource and out-of-distribution (OOD) conditions. |
Alexandra Diaconu; Mădălina Vînaga; Bogdan Alexe; | arxiv-cs.CL | 2026-03-02 |
| 478 | End-to-End Simultaneous Dysarthric Speech Reconstruction with Frame-Level Adaptor and Multiple Wait-k Knowledge Distillation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this study, we propose an end-to-end simultaneous DSR system with two key innovations: 1) A frame-level adaptor module is introduced to bridge ASR and TTS. |
Minghui Wu; Haitao Tang; Jiahuan Fan; Ruizhi Liao; Yanyong Zhang; | arxiv-cs.SD | 2026-03-01 |
| 479 | DARS: Dysarthria-Aware Rhythm-Style Synthesis for ASR Enhancement Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: While text-to-speech (TTS)-based data augmentation has shown potential, existing methods often fail to accurately model the pathological rhythm and acoustic style of dysarthric speech. To address this, we propose DARS, a dysarthria-aware rhythm-style synthesis framework based on the Matcha-TTS architecture. |
MINGHUI WU et. al. | arxiv-cs.SD | 2026-03-01 |
| 480 | ACES: Accent Subspaces for Coupling, Explanations, and Stress-Testing in Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: ASR systems exhibit persistent performance disparities across accents, yet the internal mechanisms underlying these gaps remain poorly understood. We introduce ACES, a representation-centric audit that extracts accent-discriminative subspaces and uses them to probe model fragility and disparity. |
Swapnil Parekh; | arxiv-cs.SD | 2026-02-28 |
| 481 | Breeze Taigi: Benchmarks and Models for Taiwanese Hokkien Speech Recognition and Synthesis Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce Breeze Taigi, a comprehensive framework centered on standardized benchmarks for evaluating Taigi speech recognition and synthesis systems. |
YU-SIANG LAN et. al. | arxiv-cs.CL | 2026-02-26 |
| 482 | A Holistic Framework for Robust Bangla ASR and Speaker Diarization with Optimized VAD and CTC Alignment Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents a robust framework specifically engineered for extended Bangla content by leveraging preexisting models enhanced with novel optimization pipelines for the DL Sprint 4.0 contest. |
Zarif Ishmam; Zarif Mahir; Shafnan Wasif; Md. Ishtiak Moin; | arxiv-cs.SD | 2026-02-26 |
| 483 | Make It Hard to Hear, Easy to Learn: Long-Form Bengali ASR and Speaker Diarization Via Extreme Augmentation and Perfect Alignment Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, detailing our submission to the DL Sprint 4.0 competition, we systematically evaluate various architectures and approaches for long-form Bengali speech. |
Sanjid Hasan; Risalat Labib; A H M Fuad; Bayazid Hasan; | arxiv-cs.SD | 2026-02-26 |
| 484 | Challenges in Automatic Speech Recognition for Adults with Cognitive Impairment Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Results show that ASR errors are significantly higher for individuals with dementia, revealing a critical usability gap. To better understand these disparities, we conducted an acoustic analysis of speech features and found that a speaker’s intensity, voice quality, and pause ratio predicted ASR accuracy. |
MICHELLE COHN et. al. | arxiv-cs.HC | 2026-02-26 |
| 485 | ShobdoSetu: A Data-Centric Framework for Bengali Long-Form Speech Recognition and Speaker Diarization Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we present our system for the DL Sprint 4.0 Bengali Long-Form Speech Recognition (Task~1) and Bengali Speaker Diarization Challenge (Task~2). |
Md. Nazmus Sakib; Shafiul Tanvir; Mesbah Uddin Ahamed; H. M. Aktaruzzaman Mukdho; | arxiv-cs.CL | 2026-02-25 |
| 486 | Robust Long-Form Bangla Speech Processing: Automatic Speech Recognition and Speaker Diarization Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We describe our end-to-end system for Bengali long-form speech recognition (ASR) and speaker diarization submitted to the DL Sprint 4.0 competition on Kaggle. |
MD. Sagor Chowdhury; Adiba Fairooz Chowdhury; | arxiv-cs.CL | 2026-02-25 |
| 487 | Efficient Dialect-Aware Modeling and Conditioning for Low-Resource Taiwanese Hakka Speech Processing Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Traditional ASR models often encounter difficulties in this context, as they tend to conflate essential linguistic content with dialect-specific variations across both phonological and lexical dimensions. To address these challenges, we propose a unified framework grounded in the Recurrent Neural Network Transducers (RNN-T). |
AN-CI PENG et. al. | arxiv-cs.CL | 2026-02-25 |
| 488 | Mitigating Structural Noise in Low-Resource S2TT: An Optimized Cascaded Nepali-English Pipeline with Punctuation Restoration Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents and evaluates an optimized cascaded Nepali speech-to-English text translation (S2TT) system, focusing on mitigating structural noise introduced by Automatic Speech Recognition (ASR). |
Tangsang Chongbang; Pranesh Pyara Shrestha; Amrit Sarki; Anku Jaiswal; | arxiv-cs.CL | 2026-02-25 |
| 489 | 823-OLT @ BUET DL Sprint 4.0: Context-Aware Windowing for ASR and Fine-Tuned Speaker Diarization in Bengali Long Form Audio Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present frameworks for long form Bengali speech intelligence that address automatic speech recognition using a Whisper Medium based model and speaker diarization using a finetuned segmentation model. |
Ratnajit Dhar; Arpita Mallik; | arxiv-cs.SD | 2026-02-24 |
| 490 | ReHear: Iterative Pseudo-Label Refinement for Semi-Supervised Speech Recognition Via Audio Large Language Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Semi-supervised learning in automatic speech recognition (ASR) typically relies on pseudo-labeling, which often suffers from confirmation bias and error accumulation due to noisy supervision. To address this limitation, we propose ReHear, a framework for iterative pseudo-label refinement that integrates an instruction-tuned, audio-aware large language model (LLM) into the self-training loop. |
Zefang Liu; Chenyang Zhu; Sangwoo Cho; Shi-Xiong Zhang; | arxiv-cs.CL | 2026-02-21 |
| 491 | Whisper: Courtside Edition Enhancing ASR Performance Through LLM-Driven Context Generation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce Whisper: Courtside Edition, a novel multi-agent large language model (LLM) pipeline that enhances Whisper transcriptions without retraining. |
Yonathan Ron; Shiri Gilboa; Tammuz Dubnov; | arxiv-cs.CL | 2026-02-21 |
| 492 | The Cascade Equivalence Hypothesis: When Do Speech LLMs Behave Like ASR→LLM Pipelines? Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Speech LLMs are widely understood to be better than ASR$\rightarrow$LLM cascades since they have access to the audio directly, and not just the transcript. In this paper, we … |
Jayadev Billa; | ArXiv | 2026-02-19 |
| 493 | The Cascade Equivalence Hypothesis: When Do Speech LLMs Behave Like ASR$\rightarrow$LLM Pipelines? Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Current speech LLMs largely perform implicit ASR: on tasks solvable from a transcript, they are behaviorally and mechanistically equivalent to simple Whisper$\to$LLM cascades. We show this through matched-backbone testing across four speech LLMs and six tasks, controlling for the LLM backbone for the first time. |
Jayadev Billa; | arxiv-cs.CL | 2026-02-19 |
| 494 | AI-Driven Voice Translator for Cross-Language Communication Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Our solution uses CTC, GRU, and Parallel WaveGAN. |
GEENU RAMU; P.KALYAN CHAKRAVARTHI; | International Journal of Scientific Research in Engineering … | 2026-02-18 |
| 495 | SENS-ASR: Semantic Embedding Injection in Neural-transducer for Streaming Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we present SENS-ASR, an approach to enhance the transcription quality of Streaming-ASR by reinforcing the acoustic information with semantic information. |
Youness Dkhissi; Valentin Vielzeuf; Elys Allesiardo; Anthony Larcher; | arxiv-cs.CL | 2026-02-17 |
| 496 | Speech-to-LaTeX: New Models and Datasets for Converting Spoken Equations and Sentences Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Based on ASR post-correction, prior work requires 2 transcriptions, focuses only on isolated equations, has a limited test set, and provides neither training data nor multilingual coverage. To address these issues, we present the first fully open-source large-scale dataset, comprising over 66,000 human-annotated audio samples of mathematical equations and sentences in English and Russian, drawn from diverse scientific domains. |
DMITRII KORZH et. al. | iclr | 2026-02-17 |
| 497 | Confident and Adaptive Generative Speech Recognition Via Conformal Risk Control Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose an adaptive framework that dynamically determines the optimal number of hypotheses for each input using risk control. |
Amit Damri; Bracha Laufer-Goldshtein; | iclr | 2026-02-17 |
| 498 | CTC-DRO: Robust Optimization for Reducing Language Disparities in Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present CTC-DRO, which addresses the shortcomings of the group DRO objective by smoothing the group weight update to prevent overemphasis on consistently high-loss groups, while using input length-matched batching to mitigate CTC’s scaling issues. |
MARTIJN BARTELDS et. al. | iclr | 2026-02-17 |
| 499 | Bengali-Loop: Community Benchmarks for Long-Form Bangla ASR and Speaker Diarization Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Bengali (Bangla) remains under-resourced in long-form speech technology despite its wide use. We present Bengali-Loop, two community benchmarks to address this gap: (1) a long-form ASR corpus of 191 recordings (158.6 hours, 792k words) from 11 YouTube channels, collected via a reproducible subtitle-extraction pipeline and human-in-the-loop transcript verification; and (2) a speaker diarization corpus of 24 recordings (22 hours, 5,744 annotated segments) with fully manual speaker-turn labels in CSV format. |
H. M. SHADMAN TABIB et. al. | arxiv-cs.SD | 2026-02-15 |
| 500 | From Scarcity to Scale: A Release-Level Analysis of The Pashto Common Voice Dataset Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents a release-level analysis of the Pashto component of the Mozilla Common Voice corpus, focusing on version 24.0 (December 2025) and contextualizing trends across major releases. |
Jandad Jahani; Mursal Dawodi; Jawid Ahmad Baktash; | arxiv-cs.CL | 2026-02-15 |
| 501 | Voice2mode: Phonation Mode Classification in Singing Using Self-Supervised Speech Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present voice2mode, a method for classification of four singing phonation modes (breathy, neutral (modal), flow, and pressed) using embeddings extracted from large self-supervised speech models. |
Aju Ani Justus; Ruchit Agrawal; Sudarsana Reddy Kadiri; Shrikanth Narayanan; | arxiv-cs.SD | 2026-02-14 |
| 502 | ViMedCSS: A Vietnamese Medical Code-Switching Speech Dataset & Benchmark Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we construct a 34-hour \textbf{Vi}etnamese \textbf{Med}ical \textbf{C}ode-\textbf{S}witching \textbf{S}peech dataset (ViMedCSS) containing 16,576 utterances. |
TUNG X. NGUYEN et. al. | arxiv-cs.CL | 2026-02-13 |
| 503 | Lamer-SSL: Layer-aware Mixture of LoRA Experts for Continual Multilingual Expansion of Self-supervised Models Without Forgetting Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Despite their impressive performance, self-supervised speech models often struggle to generalize to new languages and tend to forget previously acquired knowledge during continual training. To address this, we propose Lamer-SSL, a parameter-efficient framework that integrates a Layer-Aware MixturE of LoRA Experts (Lamer) module with a replay strategy. |
Jing Xu; Minglin Wu; Xueyuan Chen; Xixin Wu; Helen Meng; | arxiv-cs.CL | 2026-02-13 |
| 504 | Towards Explainable Reference-free Speech Intelligibility Evaluation of People with Pathological Speech Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: While existing objective assessments, particularly reference-based approaches, can capture intelligibility changes, they are often hindered by lack of explainability and the need for labor-intensive manual transcriptions. To address these issues, this work proposes the reference-free, explainable ASR Inconsistency Score. |
Bence Mark Halpern; Thomas Tienkamp; Defne Abur; Thomas Tienkamp; | arxiv-cs.SD | 2026-02-13 |
| 505 | Moonshine V2: Ergodic Streaming Encoder ASR for Latency-Critical Speech Applications Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To better meet the needs of on-device, streaming ASR use cases we introduce Moonshine v2, an ergodic streaming-encoder ASR model that employs sliding-window self-attention to achieve bounded, low-latency inference while preserving strong local context. |
Manjunath Kudlur; Evan King; James Wang; Pete Warden; | arxiv-cs.CL | 2026-02-12 |
| 506 | ViSpeechFormer: A Phonemic Approach for Vietnamese Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Vietnamese has a phonetic orthography, where each grapheme corresponds to at most one phoneme and vice versa. Exploiting this high grapheme-phoneme transparency, we propose ViSpeechFormer (\textbf{Vi}etnamese \textbf{Speech} Trans\textbf{Former}), a phoneme-based approach for Vietnamese Automatic Speech Recognition (ASR). |
Khoa Anh Nguyen; Long Minh Hoang; Nghia Hieu Nguyen; Luan Thanh Nguyen; Ngan Luu-Thuy Nguyen; | arxiv-cs.CL | 2026-02-10 |
| 507 | Where Are We At with Automatic Speech Recognition for The Bambara Language? Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper introduces the first standardized benchmark for evaluating Automatic Speech Recognition (ASR) in the Bambara language, utilizing one hour of professionally recorded Malian constitutional text. |
SEYDOU DIALLO et. al. | arxiv-cs.CL | 2026-02-10 |
| 508 | Prototype-Based Disentanglement for Controllable Dysarthric Speech Synthesis Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we propose ProtoDisent-TTS, a prototype-based disentanglement TTS framework built on a pre-trained text-to-speech backbone that factorizes speaker timbre and dysarthric articulation within a unified latent space. |
HAOSHEN WANG et. al. | arxiv-cs.SD | 2026-02-09 |
| 509 | An In-Depth Review of Speech Enhancement Algorithms: Classifications, Underlying Principles, Challenges, and Emerging Trends Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Speech enhancement aims to improve speech quality and intelligibility in noisy environments and is important in applications such as hearing aids, mobile communications and … |
Nisreen Talib Abdulhusein; Basheera M. Mahmmod; | Algorithms | 2026-02-07 |
| 510 | Automated Legal Document Drafting in The Indian Judicial System: A Survey of AI-driven Approaches Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: The research states that we have done evaluation of 20 academic papers from 2017 to 2025 which aims to investigate Automatic Speech Recognition (ASR), Natural Language Processing (NLP) and transfer learning models for legal systems. |
Hemamalini S; | International Journal for Research in Applied Science and … | 2026-02-06 |
| 511 | Speech Emotion Recognition Leveraging OpenAI’s Whisper Representations and Attentive Pooling Methods Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This work explores the capabilities of Whisper, a pre-trained ASR system, in speech emotion recognition by proposing two attention-based pooling methods, Multi-head Attentive Average Pooling and QKV Pooling, designed to efficiently reduce the dimensionality of Whisper representations while preserving emotional features. |
Ali Shendabadi; Parnia Izadirad; Mostafa Salehi; Mahmoud Bijankhan; | arxiv-cs.AI | 2026-02-05 |
| 512 | Speaker-Aware Simulation Improves Conversational Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we adapt and implement the SASC framework for Hungarian conversational ASR. |
Máté Gedeon; Péter Mihajlik; | arxiv-cs.SD | 2026-02-04 |
| 513 | Frontend Token Enhancement for Token-Based Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we introduce a frontend system that estimates clean speech tokens from noisy speech and evaluate it on an ASR backend using semantic tokens. |
Takanori Ashihara; Shota Horiguchi; Kohei Matsuura; Tsubasa Ochiai; Marc Delcroix; | arxiv-cs.SD | 2026-02-04 |
| 514 | An ASR Transformer-Based Model for Kannada Speech-to-Text Transcription Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This work presents a dialect-aware and noise-robust Kannada automatic speech recognition (ASR) system that bridges the gap between low-resource linguistic contexts and state-of-the-art deep learning models. |
Chandrika Prasad; Veenga Gode Swamy Rao; Geetha J; R China Appala Naidu; | Journal of Artificial Intelligence and Technology | 2026-02-03 |
| 515 | Multimodal Consistency-Guided Reference-Free Data Selection for ASR Accent Adaptation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: However, common pseudo-label selection heuristics are largely text-centric (e.g., perplexity (PPL) filtering) and can prefer fluent yet acoustically mismatched hypotheses, leading to error amplification when fine-tuning. To address this, we introduce a multimodal consistency-guided, reference-free data selection pipeline for ASR accent adaptation under a transductive, label-free protocol. |
Ligong Lei; Wenwen Lu; Xudong Pang; Zaokere Kadeer; Aishan Wumaier; | arxiv-cs.CL | 2026-02-03 |
| 516 | BBPE16: UTF-16-based Byte-level Byte-pair Encoding for Improved Multilingual Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose BBPE16, a UTF-16-based BBPE tokenizer that represents most modern scripts with a uniform 2-byte code unit. |
Hyunsik Kim; Haeri Kim; Munhak Lee; Kyungmin Lee; | arxiv-cs.CL | 2026-02-02 |
| 517 | Mixture-of-Experts with Intermediate CTC Supervision for Accented Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce Moe-Ctc, a Mixture-of-Experts architecture with intermediate CTC supervision that jointly promotes expert specialization and generalization. |
Wonjun Lee; Hyounghun Kim; Gary Geunbae Lee; | arxiv-cs.CL | 2026-02-02 |
| 518 | A Mixed Reality Tool with Automatic Speech Recognition for 3D CAD Based Visualization and Automatic Dimension Generation in The Industry 5.0 Shipyard Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Industry 5.0 is composed of a variety of complex tasks and challenging processes requiring specialized labor and multidisciplinary coordination. Specifically, when it comes to … |
Aida Vidal-Balea; Antón Valladares-Poncela; Javier Vilar-Martínez; T. Fernández-Caramés; Paula Fraga-Lamas; | Multimodal Technol. Interact. | 2026-02-01 |
| 519 | MedSpeak: A Knowledge Graph-Aided ASR Error Correction Framework for Spoken Medical QA Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To this end, we propose MedSpeak, a novel knowledge graph-aided ASR error correction framework that refines noisy transcripts and improves downstream answer prediction by leveraging both semantic relationships and phonetic information encoded in a medical knowledge graph, together with the reasoning power of LLMs. |
YUTONG SONG et. al. | arxiv-cs.CL | 2026-01-31 |
| 520 | Qwen3-ASR Technical Report Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this report, we introduce Qwen3-ASR family, which includes two powerful all-in-one speech recognition models and a novel non-autoregressive speech forced alignment model. |
XIAN SHI et. al. | arxiv-cs.CL | 2026-01-29 |
| 521 | Understanding Frechet Speech Distance for Synthetic Speech Quality Evaluation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Fréchet Distance has emerged as a promising alternative, yet its reliability depends heavily on the choice of embeddings and experimental settings. In this work, we comprehensively evaluate Fréchet Speech Distance (FSD) and its variant Speech Maximum Mean Discrepancy (SMMD) under varied embeddings and conditions. |
June-Woo Kim; Dhruv Agarwal; Federica Cerina; | arxiv-cs.SD | 2026-01-29 |
| 522 | A Study of Data Selection Strategies for Pre-training Self-Supervised Speech Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We systematically examine how curated subsets of pre-training data influence Automatic Speech Recognition (ASR) performance. |
Ryan Whetten; Titouan Parcollet; Marco Dinarelli; Yannick Estève; | arxiv-cs.SD | 2026-01-28 |
| 523 | Text-only Adaptation in LLM-based ASR Through Text Denoising IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce a novel text-only adaptation method that emulates the audio projection task by treating it as a text denoising task. |
SERGIO BURDISSO et. al. | arxiv-cs.SD | 2026-01-28 |
| 524 | SW-ASR: A Context-Aware Hybrid ASR Pipeline for Robust Single Word Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper reviews recent deep learning approaches and proposes a modular framework for robust single-word detection. |
Manali Sharma; Riya Naik; Buvaneshwari G; | arxiv-cs.SD | 2026-01-27 |
| 525 | Advanced Modeling of Interlanguage Speech Intelligibility Benefit with L1-L2 Multi-Task Learning Using Differentiable K-Means for Accent-Robust Discrete Token-Based ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this study, we propose a more advanced modeling of ISIB. |
Kentaro Onda; Satoru Fukayama; Daisuke Saito; Nobuaki Minematsu; | arxiv-cs.SD | 2026-01-27 |
| 526 | Dynamic Multi-Expert Projectors with Stabilized Routing for Multilingual Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: While effective in monolingual settings, a single projector struggles to capture the diverse acoustic-to-semantic mappings required for multilingual ASR. To address this, we propose SMEAR-MoE, a stabilized Mixture-of-Experts projector that ensures dense gradient flow to all experts, preventing expert collapse while enabling cross-lingual sharing. |
Isha Pandey; Ashish Mittal; Vartul Bahuguna; Ganesh Ramakrishnan; | arxiv-cs.CL | 2026-01-27 |
| 527 | Mind The Shift: Using Delta SSL Embeddings to Enhance Child ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Self-supervised learning (SSL) models have achieved impressive results across many speech tasks, yet child automatic speech recognition (ASR) remains challenging due to limited data and pretraining domain mismatch. |
Zilai Wang; Natarajan Balaji Shankar; Kaiyuan Zhang; Zihan Wang; Abeer Alwan; | arxiv-cs.CL | 2026-01-27 |
| 528 | MetaSICL: Globalizing Auditory LLMs for Underserved Speakers and Languages Via Meta Speech In-Context Learning Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Generative AI for speech and audio is increasingly expected to serve users across languages, cultures, and communities, yet current auditory Large Language Models (LLMs) are still largely trained and evaluated on high-resource data. |
Haolong Zheng; Siyin Wang; Zengrui Jin; Mark Hasegawa-Johnson; | arxiv-cs.SD | 2026-01-26 |
| 529 | Unheard in The Digital Age: Rethinking AI Bias and Speech Diversity Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Given the current state of the debate, this article focuses on the structural biases that shape perceptions of atypical speech and are now being encoded into artificial intelligence. |
Onyedikachi Hope Amaechi-Okorie; Branislav Radeljic; | arxiv-cs.HC | 2026-01-26 |
| 530 | Language Family Matters: Evaluating LLM-Based ASR Across Linguistic Boundaries Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Prior work trains a separate connector per language, overlooking linguistic relatedness. We propose an efficient and novel connector-sharing strategy based on linguistic family membership, enabling one connector per family, and empirically validate its effectiveness across two multilingual LLMs and two real-world corpora spanning curated and crowd-sourced speech. |
Yuchen Zhang; Ravi Shekhar; Haralambos Mouratidis; | arxiv-cs.CL | 2026-01-26 |
| 531 | OCR-Enhanced Multimodal ASR Can Read While Listening Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we propose Donut-Whisper, an audio-visual ASR model with dual encoder to leverage visual information to improve speech recognition performance in both English and Chinese. |
Junli Chen; Changli Tang; Yixuan Li; Guangzhi Sun; Chao Zhang; | arxiv-cs.SD | 2026-01-26 |
| 532 | VIBEVOICE-ASR Technical Report IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This report presents VibeVoice-ASR, a general-purpose speech understanding framework built upon VibeVoice, designed to address the persistent challenges of context fragmentation and multi-speaker complexity in long-form audio (e.g., meetings, podcasts) that remain despite recent advancements in short-form speech recognition. |
ZHILIANG PENG et. al. | arxiv-cs.SD | 2026-01-26 |
| 533 | DLLM-ASR: A Faster Diffusion LLM-based Framework for Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose dLLM-ASR, an efficient dLLM-based ASR framework that formulates dLLM’s decoding as a prior-guided and adaptive denoising process. |
WENJIE TIAN et. al. | arxiv-cs.SD | 2026-01-25 |
| 534 | BanglaRobustNet: A Hybrid Denoising-Attention Architecture for Robust Bangla Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Bangla, one of the most widely spoken languages, remains underrepresented in state-of-the-art automatic speech recognition (ASR) research, particularly under noisy and speaker-diverse conditions. This paper presents BanglaRobustNet, a hybrid denoising-attention framework built on Wav2Vec-BERT, designed to address these challenges. |
Md Sazzadul Islam Ridoy; Mubaswira Ibnat Zidney; Sumi Akter; Md. Aminur Rahman; | arxiv-cs.SD | 2026-01-24 |
| 535 | Post-ASR Correction for Low-Resource Rajasthani Language Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose a multi-view 1 , character-level sequence-to-sequence (Seq2Seq) model that uses a gated fusion mechanism to dynamically weigh information from the two ASR outputs. |
Abhishek Bhandari; Gaurav Harit; | ACM Transactions on Asian and Low-Resource Language … | 2026-01-24 |
| 536 | Benchmarking Von ASR-Modellen Im Deutschen Medizinischen Kontext: Eine Leistungsanalyse Anhand Von Anamnesegesprächen Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this article, we present a curated dataset of simulated doctor-patient conversations and evaluate a total of 29 different ASR models. |
THOMAS SCHUSTER et. al. | arxiv-cs.CL | 2026-01-23 |
| 537 | Sink or SWIM: Tackling Real-Time ASR at Scale Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we present SWIM, a novel real-time ASR system built on top of OpenAI’s Whisper model that enables true model-level parallelization for scalable, multilingual transcription. |
Federico Bruzzone; Walter Cazzola; Matteo Brancaleoni; Dario Pellegrino; | arxiv-cs.SD | 2026-01-22 |
| 538 | Retrieval-Augmented Self-Taught Reasoning Model with Adaptive Chain-of-Thought for ASR Named Entity Correction Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: However, these approaches have yet to fully exploit the sophisticated reasoning capabilities inherent to LLMs. To bridge this gap, we propose a novel retrieval-augmented generation framework for correcting named entity errors in ASR. |
JUNJIE AN et. al. | arxiv-cs.CL | 2026-01-21 |
| 539 | SpeakerLM: End-to-End Versatile Speaker Diarization and Recognition with Multimodal Large Language Models IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: The cascaded systems suffer from several limitations, such as error propagation, difficulty in handling overlapping speech, and lack of joint optimization for exploring the synergy between SD and ASR tasks. To address these limitations, we introduce SpeakerLM, a unified multimodal large language model for SDR that jointly performs SD and ASR in an end-to-end manner. |
HAN YIN et. al. | aaai | 2026-01-20 |
| 540 | IO-RAE: Information-Obfuscation Reversible Adversarial Example for Audio Privacy Protection Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper introduces an Information-Obfuscation Reversible Adversarial Example (IO-RAE) framework, the pioneering method designed to safeguard audio privacy using reversible adversarial examples. |
JIAJIE ZHU et. al. | aaai | 2026-01-20 |
| 541 | Beyond Transcription: Mechanistic Interpretability in ASR IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we adapt and systematically apply established interpretability methods such as logit lens, linear probing, and activation patching, to examine how acoustic and semantic information evolves across layers in ASR systems. |
NETA GLAZER et. al. | aaai | 2026-01-20 |
| 542 | STEAMROLLER: A Multi-Agent System for Inclusive Automatic Speech Recognition for People Who Stutter Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present STEAMROLLER, a real time system that transforms stuttered speech into fluent output through a novel multi-stage, multi-agent AI pipeline. |
ZIQI XU et. al. | aaai | 2026-01-20 |
| 543 | BiCycle: Group-wise Recursive Transformer Based on ASR Mechanism Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: While RT has been successfully applied to large language models (LLMs), its effectiveness in automatic speech recognition (ASR) remains limited, despite the parallel trend of model scaling in the speech domain. In this paper, we reveal that conventional RT designs for LLMs are suboptimal for speech recognition, primarily because they do not fully consider the layer-wise specialization inherent in the ASR architecture, where lower layers focus on phonetic features and upper layers capture linguistic localization. |
Min Ho Jang; Eun Seo Seo; Jin Young Kim; Hyeongsoo Lim; Ji Won Yoon; | aaai | 2026-01-20 |
| 544 | Lost in Transcription: How Speech-to-Text Errors Derail Code Understanding Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Voice-based interfaces offer a more inclusive modality, but spoken queries involving code present unique challenges due to the presence of non-standard English usage, domain-specific vocabulary, and custom identifiers such as variable and function names, often combined with code-mixed expressions. In this work, we develop a multilingual speech-driven framework for code understanding that accepts spoken queries in a user native language, transcribes them using Automatic Speech Recognition (ASR), applies code-aware ASR output refinement using Large Language Models (LLMs), and interfaces with code models to perform tasks such as code question answering and code retrieval through benchmarks such as CodeSearchNet, CoRNStack, and CodeQA. |
Jayant Havare; Ashish Mittal; Srikanth Tamilselvam; Ganesh Ramakrishnan; | arxiv-cs.SE | 2026-01-20 |
| 545 | HPSU: A Benchmark for Human-Level Perception in Real-World Spoken Speech Understanding Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To this end, we introduce the Human-level Perception in Spoken Speech Understanding (HPSU), a new benchmark for fully evaluating the human-level perceptual and understanding capabilities of Speech LLMs. |
CHEN LI et. al. | aaai | 2026-01-20 |
| 546 | WenetSpeech-Yue: A Large-Scale Cantonese Speech Corpus with Multi-dimensional Annotation IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: However, for Cantonese (Yue Chinese), spoken by approximately 84.9 million native speakers worldwide, limited annotated resources have hindered progress and resulted in suboptimal ASR and TTS performance. To address this challenge, we propose WenetSpeech-Pipe, an integrated pipeline for building large-scale speech corpus with multi-dimensional annotation tailored for speech understanding and generation. |
LONGHAO LI et. al. | aaai | 2026-01-20 |
| 547 | Hearing More with Less: Multi-Modal Retrieval-and-Selection Augmented Conversational LLM-Based ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper proposes a multi-modal retrieval-and-selection method named MARS that augments conversational LLM-ASR by enabling it to retrieve and select the most relevant acoustic and textual historical context for the current utterance. |
Bingshen Mu; Hexin Liu; Hongfei Xue; Kun Wei; Lei Xie; | aaai | 2026-01-20 |
| 548 | Speech Recognition Model Improves Text-to-Speech Synthesis Using Fine-Grained Reward Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this context, we present an intriguing finding: encoder-decoder ASR models, such as Whisper, leverage their extensive pre-training to precisely capture word-level mismatches between speech and text within their cross-attention mechanisms, thereby providing a fine-grained reward signal. Building upon this insight, we propose a novel TTS optimization method, which we term Word-level TTS Alignment by ASR-driven Attentive Reward (W3AR). |
Guansu Wang; Peijie Sun; | aaai | 2026-01-20 |
| 549 | Speech-Aware Long Context Pruning and Integration for Contextualized Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This challenge arises primarily due to constrained model context windows and the sparsity of relevant information within extensive contextual noise. To solve this, we propose the SAP^2 method, a novel framework that dynamically prunes and integrates relevant contextual keywords in two stages. |
YIMING RONG et. al. | aaai | 2026-01-20 |
| 550 | ConfASR: A Conformer Block Accelerator for Speech Recognition Optimized for Edge Devices Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Attention-based neural networks, like transformers, have significantly improved automatic speech recognition (ASR). Adding convolution operations to transformers results in … |
MALTE WABNITZ et. al. | 2026 31st Asia and South Pacific Design Automation … | 2026-01-19 |
| 551 | Arab Voices: Mapping Standard and Dialectal Arabic Speech Technology Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To reduce fragmentation and support reproducible evaluation, we introduce Arab Voices, a standardized framework for DA ASR. |
Peter Sullivan; AbdelRahim Elmadany; Alcides Alcoba Inciarte; Muhammad Abdul-Mageed; | arxiv-cs.CL | 2026-01-19 |
| 552 | Typhoon ASR Real-time: FastConformer-Transducer for Thai Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present Typhoon ASR Real-time, a 115M-parameter FastConformer-Transducer model for low-latency Thai speech recognition. |
WARIT SIRICHOTEDUMRONG et. al. | arxiv-cs.CL | 2026-01-19 |
| 553 | DUAP: Dual-task Universal Adversarial Perturbations Against Voice Control Systems Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To fill this gap, we first utilize gradient analysis to reveal that ASR and SR exhibit no inherent conflicts. Building on this, we propose Dual-task Universal Adversarial Perturbation (DUAP). |
Suyang Sun; Weifei Jin; Yuxin Cao; Wei Song; Jie Hao; | arxiv-cs.CR | 2026-01-19 |
| 554 | SSVD-O: Parameter-Efficient Fine-Tuning with Structured SVD for Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Building on our prior work, this paper introduces SSVD-Outer (SSVD-O), an extension of the structured SVD-guided (SSVD) fine-tuning method. |
Pu Wang; Shinji Watanabe; Hugo Van hamme; | arxiv-cs.SD | 2026-01-18 |
| 555 | CTC-DID: CTC-Based Arabic Dialect Identification for Streaming Applications Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper proposes a Dialect Identification (DID) approach inspired by the Connectionist Temporal Classification (CTC) loss function as used in Automatic Speech Recognition (ASR). |
Muhammad Umar Farooq; Oscar Saz; | arxiv-cs.CL | 2026-01-17 |
| 556 | WenetSpeech-Wu: Datasets, Benchmarks, and Models for A Unified Chinese Wu Dialect Speech Processing Ecosystem Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we present WenetSpeech-Wu, the first large-scale, multi-dimensionally annotated open-source speech corpus for the Wu dialect, comprising approximately 8,000 hours of diverse speech data. |
CHENGYOU WANG et. al. | arxiv-cs.SD | 2026-01-16 |
| 557 | Unifying Speech Recognition, Synthesis and Conversion with Autoregressive Transformers Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we present General-Purpose Audio (GPA), a unified audio foundation model that integrates multiple core speech tasks within a single large language model (LLM) architecture. |
Runyuan Cai; Yu Lin; Yiming Wang; Chunlin Fu; Xiaodong Zeng; | arxiv-cs.SD | 2026-01-15 |
| 558 | Acoustic and Tonal Modeling of The Tpuri Language Through A Multi-Modular Hybrid Approach Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents a hybrid multimodular ASR architecture for tpuri, a Mboum-Day Niger-Congo language spoken in Cameroon and Chad that exhibits contrastive lexical tone, vowel length and nasalisation. |
Jules Paulin Bayang Souloukna; Patrick Nounamou Dabou; Paul Dabou Patrick; | EAI Endorsed Transactions on Intelligent Systems and … | 2026-01-14 |
| 559 | SLAM-LLM: A Modular, Open-Source Multimodal Large Language Model Framework and Best Practice for Speech, Language, Audio and Music Processing IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present SLAM-LLM, an open-source deep learning framework designed to train customized MLLMs, focused on speech, language, audio, and music processing. |
ZIYANG MA et. al. | arxiv-cs.SD | 2026-01-14 |
| 560 | MCGA: A Multi-task Classical Chinese Literary Genre Audio Corpus Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: While existing research has primarily focused on text and visual modalities, the audio corpus within this domain remains largely underexplored. To bridge this gap, we propose the Multi-task Classical Chinese Literary Genre Audio Corpus (MCGA). |
YEXING DU et. al. | arxiv-cs.CL | 2026-01-14 |
| 561 | Wav2Vec-based Audio Data Augmentation for Low-Resource Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This article focuses on performing ADA techniques namely the addition of noise, pitch shifting, increasing or decreasing of speed and adding reverberation to the audio signals. |
P. Haritha; P. Shanmugavadivu; | International Research Journal on Advanced Engineering Hub … | 2026-01-13 |
| 562 | Robust CAPTCHA Using Audio Illusions in The Era of Large Language Models: from Evaluation to Advances Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we introduce AI-CAPTCHA, a unified framework that offers (i) an evaluation framework, ACEval, which includes advanced LALM- and ASR-based solvers, and (ii) a novel audio CAPTCHA approach, IllusionAudio, leveraging audio illusions. |
ZIQI DING et. al. | arxiv-cs.SD | 2026-01-13 |
| 563 | SITA: Learning Speaker-Invariant and Tone-Aware Speech Representations for Low-Resource Tonal Languages Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: A key challenge is learning representations that are robust to nuisance variation such as gender while remaining tone-aware for different lexical meanings. To address this, we propose SITA, a lightweight adaptation recipe that enforces Speaker-Invariance and Tone-Awareness for pretrained wav2vec-style encoders. |
TIANYI XU et. al. | arxiv-cs.CL | 2026-01-13 |
| 564 | Towards Comprehensive Semantic Speech Embeddings for Chinese Dialects Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Building dialect-to-Mandarin speech-LLMs requires speech representations with cross-dialect semantic alignment between Chinese dialects and Mandarin. In this paper, we achieve such a cross-dialect semantic alignment by training a speech encoder with ASR (automatic speech recognition)-only data, as demonstrated by speech-to-speech retrieval on a new benchmark of spoken Chinese varieties that we contribute. |
Kalvin Chang; Yiwen Shao; Jiahong Li; Dong Yu; | arxiv-cs.CL | 2026-01-12 |
| 565 | Task Arithmetic with Support Languages for Low-Resource ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we consider training on a particular language to be a task, and we generate task vectors by fine-tuning variants of the Whisper ASR system. |
Emma Rafkin; Dan DeGenaro; Xiulin Yang; | arxiv-cs.CL | 2026-01-11 |
| 566 | Doing More with Less: Data Augmentation for Sudanese Dialect Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents a comprehensive study of data augmentation techniques for fine-tuning OpenAI Whisper models and establishes the first benchmark for the Sudanese dialect. |
Ayman Mansour; | arxiv-cs.CL | 2026-01-11 |
| 567 | Multimodal In-context Learning for ASR of Low-resource Languages Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In-context learning (ICL) with large language models (LLMs) addresses this problem, but prior work largely focuses on high-resource languages covered during training and text-only settings. This paper investigates whether speech LLMs can learn unseen languages with multimodal ICL (MICL), and how this learning can be used to improve ASR. |
Zhaolin Li; Jan Niehues; | arxiv-cs.CL | 2026-01-09 |
| 568 | Dialect Matters: Cross-Lingual ASR Transfer for Low-Resource Indic Language Varieties Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We conduct an empirical study of cross-lingual transfer using spontaneous, noisy, and code-mixed speech across a wide range of Indic dialects and language varieties. |
Akriti Dhasmana; Aarohi Srivastava; David Chiang; | arxiv-cs.CL | 2026-01-07 |
| 569 | WESR: Scaling and Evaluating Word-level Event-Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Based on the refined taxonomy, we introduce WESR-Bench, an expert-annotated evaluation set (900+ utterances) with a novel position-aware protocol that disentangles ASR errors from event detection, enabling precise localization measurement for both discrete and continuous events. |
CHENCHEN YANG et. al. | arxiv-cs.CL | 2026-01-07 |
| 570 | Dynamic Quantization Error Propagation in Encoder-Decoder ASR Quantization Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Existing solutions like Quantization Error Propagation (QEP) are suboptimal for ASR due to the model’s heterogeneity, processing acoustic features in the encoder while generating text in the decoder. To address this, we propose Fine-grained Alpha for Dynamic Quantization Error Propagation (FADE), which adaptively controls the trade-off between cross-layer error correction and local quantization. |
XINYU WANG et. al. | arxiv-cs.SD | 2026-01-05 |
| 571 | Multi-channel Multi-speaker Transformer for Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Based on these, we propose the multi-channel multi-speaker transformer (M2Former) for far-field multi-speaker ASR in this paper. |
Guo Yifan; Tian Yao; Suo Hongbin; Wan Yulong; | arxiv-cs.SD | 2026-01-05 |
| 572 | Bridging The Gap: A Comparative Exploration of Speech-LLM and End-to-end Architecture for Multilingual Conversational ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we present an enhanced LLM-based ASR framework that combines fine-tuned Whisper and mHuBERT encoders with an LLM to enrich speech representations. |
Yuxiang Mei; Dongxing Xu; Jiaen Liang; Yanhua Long; | arxiv-cs.CL | 2026-01-04 |
| 573 | The Human–machine Correlation in Automated Speech Evaluation: A Three-level Meta-analysis Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Moderator analysis revealed significant moderating effects on the overall human–machine correlation from 10 variables: publication year, publication type, unit of sample, age group, level of task constraints, rater expertise, inter-rater reliability, system developer, feature engineering , and algorithm type . |
Yanxin Wang; Shangchao Min; | Language Testing | 2026-01-03 |
| 574 | IO-RAE: Information-Obfuscation Reversible Adversarial Example for Audio Privacy Protection Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper introduces an Information-Obfuscation Reversible Adversarial Example (IO-RAE) framework, the pioneering method designed to safeguard audio privacy using reversible adversarial examples. |
JIAJIE ZHU et. al. | arxiv-cs.SD | 2026-01-03 |
| 575 | NeMo@IWSLT 2026: Cascaded System for Simultaneous Speech Translation Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: This paper describes the NVIDIA NeMo team’s submission to the IWSLT 2026 Simultaneous Speech Translation (SimulST) tracks. We use a cascaded architecture combining a dual-mode … |
LILIT GRIGORYAN et. al. | International Workshop on Spoken Language Translation | 2026-01-01 |
| 576 | Cross-Attention Reprogramming for ASR: Bridging Discrete Speech Units and Pretrained Language Models Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: In automatic speech recognition (ASR), an emerging trend involves converting continuous speech features into sequences of discrete speech units (DSUs) via quantization. A key … |
Pei-Jun Liao; Hung-Yi Lee; Hsin-Min Wang; | IEEE Access | 2026-01-01 |
| 577 | AfriVox: Probing Multilingual and Accent Robustness of Speech LLMs Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Recent advances in multimodal and speech-native large language models (LLMs) have delivered impressive speech recognition, translation, understanding, and question-answering … |
BUSAYO AWOBADE et. al. | Conference of the European Chapter of the Association for … | 2026-01-01 |
| 578 | Post-ASR Correction in Hindi: Comparing Language Models and Large Language Models in Low-Resource Scenarios Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Automatic Speech Recognition (ASR) systems for low-resource languages like Hindi often produce erroneous transcripts due to limited annotated data and linguistic complexity. … |
Rishabh Kumar; Amrith Krishna; Ganesh Ramakrishnan; P. Jyothi; | Conference of the European Chapter of the Association for … | 2026-01-01 |
| 579 | Building ASR Resources for The Hutsul Dialect of Ukrainian Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Dialectal speech remains largely underexplored in Automatic Speech Recognition (ASR) research, particularly for Slavic languages. While Ukrainian ASR systems have rapidly improved … |
R. Kyslyi; A. Orlovskyi; Pavlo Khomenko; Bohdan Onyshchenko; Zakhar Guzii; | VarDial@EAC | 2026-01-01 |
| 580 | Children’s Speech Recognition in Slovak Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: This paper addresses the challenge of Automatic Speech Recognition (ASR) for Slovak children’s speech, a low-resource scenario characterized by significant linguistic variability … |
Stanislav Ondáš; Ján Staš; Matus Pleva; | IEEE Access | 2026-01-01 |
| 581 | Multi-Model ASR Integration With Reliability Weighting for Automated Speech Disorder Screening Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Speech sound disorders affect communication development in children, requiring early detection for timely intervention. Traditional clinical assessment relies on manual phonetic … |
Selina S. Sung; Seunghee Ha; Tae-Jin Yoon; Ju-hyun So; | IEEE Access | 2026-01-01 |
| 582 | Index-ASR Technical Report Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Second, they provide limited support for flexible and fine-grained contextual customization. To address these challenges, we propose Index-ASR, a large-scale LLM-based ASR system designed to simultaneously enhance robustness and support customizable hotword recognition. |
ZHESHU SONG et. al. | arxiv-cs.SD | 2025-12-31 |
| 583 | PROFASR-BENCH: A Benchmark for Context-Conditioned ASR in High-Stakes Professional Speech Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present ProfASR-Bench, a professional-talk evaluation suite for high-stakes applications across finance, medicine, legal, and technology. |
Deepak Babu Piskala; | arxiv-cs.CL | 2025-12-29 |
| 584 | Fine-Tuning Whisper Model for Mandar Speech Recognition: Approach and Performance Evaluation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This study aims to enhance the performance of Automatic Speech Recognition (ASR) systems by fine-tuning the Whisper model using a Mandar-specific dataset. |
Jafar Jafar; Mar Athul Wazithah Tb; Firman Aziz; Rosary Iriany; Norma Nasir; | Journal of Applied Engineering and Technological Science … | 2025-12-29 |
| 585 | Kunnafonidilaw Ka Cadeau: An ASR Dataset of Present-day Bambara Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present Kunkado, a 160-hour Bambara ASR dataset compiled from Malian radio archives to capture present-day spontaneous speech across a wide range of topics. |
Yacouba Diarra; Panga Azazia Kamate; Nouhoum Souleymane Coulibaly; Michael Leventhal; | arxiv-cs.CL | 2025-12-22 |
| 586 | From Speech to Subtitles: Evaluating ASR Models in Subtitling Italian Television Programs Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents a case study on developing a professional subtitling system for an Italian media company. |
Alessandro Lucca; Francesco Pierri; | arxiv-cs.CL | 2025-12-22 |
| 587 | Automated Speech-fluency Explanations for Schizophrenia Diagnosis Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this study, we present a fully automated and explainable pipeline for detecting schizophrenia from audio recordings of verbal fluency tests, collected from 126 Slovene-speaking participants (68 healthy controls, 58 individuals diagnosed with schizophrenia), leveraging recent advancements in automatic speech recognition (ASR) and large language model (LLM) systems. |
Rok Rajher; Mila Marinković; Polona Rus Prelog; Jure Žabkar; | Scientific Reports | 2025-12-22 |
| 588 | X-Talk: On The Underestimated Potential of Modular Speech-to-Speech Dialogue System IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present X-Talk, an open-source framework that champions a decoupled, modular design for LLM-driven speech-to-speech (S2S) systems. |
ZHANXUN LIU et. al. | arxiv-cs.SD | 2025-12-21 |
| 589 | Incorporating Error Level Noise Embedding for Improving LLM-Assisted Robustness in Persian Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This study presents a robust noise-sensitive ASR error correction framework that combines multiple hypotheses and noise-aware modeling. |
Zahra Rahmani; Hossein Sameti; | arxiv-cs.CL | 2025-12-19 |
| 590 | When De-noising Hurts: A Systematic Study of Speech Enhancement Effects on Modern Medical ASR Systems Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present a systematic evaluation of MetricGAN-plus-voicebank denoising on four state-of-the-art ASR systems: OpenAI Whisper, NVIDIA Parakeet, Google Gemini Flash 2.0, Parrotlet-a using 500 medical speech recordings under nine noise conditions. |
SUJAL CHONDHEKAR et. al. | arxiv-cs.SD | 2025-12-19 |
| 591 | Voice Assistant for Desktop Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents the design and implementation of a Voice Assistant for Desktop, an intelligent system that enables users to interact with desktop computers using natural language voice commands. |
Geeta Patil; | International Journal for Research in Applied Science and … | 2025-12-18 |
| 592 | Bridging The Reality Gap: Efficient Adaptation of ASR Systems for Challenging Low-Resource Domains Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We quantify this gap by showing that a robust multilingual model (IndicWav2Vec) degrades to a stark 40.94% Word Error Rate (WER) when deployed on real-world clinical audio (Gram Vaani), rendering it unusable for practical applications. To address these challenges and bring ASR closer to deployment, we propose an efficient, privacy-preserving adaptation framework. |
DARSHIL CHAUHAN et. al. | arxiv-cs.CL | 2025-12-18 |
| 593 | Marco-ASR: A Principled and Metric-Driven Framework for Fine-Tuning Large-Scale ASR Models for Domain Adaptation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This challenge is amplified for modern Large Language Model (LLM)-based ASR systems, whose massive scale and complex training dynamics make effective fine-tuning non-trivial. To address this gap, this paper proposes a principled and metric-driven fine-tuning framework for adapting both traditional and LLM-based ASR models to specialized domains. |
XUANFAN NI et. al. | arxiv-cs.SD | 2025-12-17 |
| 594 | Adapting Speech Language Model to Singing Voice Synthesis Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we adapt a 1.7B parameter TTS pretrained SLM for singing voice synthesis (SVS), using only a 135-hour synthetic singing corpus, ACE-Opencpop. |
YIWEN ZHAO et. al. | arxiv-cs.SD | 2025-12-16 |
| 595 | Reproducing and Dissecting Denoising Language Models for Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To this end, we introduce DLM-sum, a novel method for decoding from multiple ASR hypotheses, which consistently outperforms the previously proposed DSR decoding method. |
Dorian Koch; Albert Zeyer; Nick Rossenbach; Ralf Schlüter; Hermann Ney; | arxiv-cs.NE | 2025-12-15 |
| 596 | Adaptive Edge-Cloud Inference for Speech-to-Action Systems Using ASR and Large Language Models Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Voice-based interaction has emerged as a natural and intuitive modality for controlling IoT devices. However, speech-driven edge devices face a fundamental trade-off between … |
Mohammad Jalili Torkamani; Israt Zarin; | ArXiv | 2025-12-14 |
| 597 | Adaptive Edge-Cloud Inference for Speech-to-Action Systems Using ASR and Large Language Models (ASTA) Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents ASTA, an adaptive speech-to-action solution that dynamically routes voice commands between edge and cloud inference to balance performance and system resource utilization. |
Mohammad Jalili Torkamani; Israt Zarin; | arxiv-cs.SD | 2025-12-14 |
| 598 | System X: A Mobile Voice-Based AI System for EMR Generation and Clinical Decision Support in Low-Resource Maternal Healthcare Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present the design, implementation, and in-situ deployment of a smartphone-based voice-enabled AI system for generating electronic medical records (EMRs) and clinical risk alerts in maternal healthcare settings. |
MARYAM MUSTAFA et. al. | arxiv-cs.HC | 2025-12-13 |
| 599 | Leveraging 5G Networks for Event Video Summarization Via Multimodal Analysis Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents an end-to-end framework that leverages 5G connectivity, Automatic Speech Recognition (ASR), Optical Character Recognition (OCR), and Large Language Models (LLMs) to generate concise, context-aware event summaries in the form of video highlights and textual overviews, eliminating the need to watch full videos. |
ALEXANDROS VROCHIDIS et. al. | Discover Artificial Intelligence | 2025-12-12 |
| 600 | TRIDENT: A Redundant Architecture for Caribbean-Accented Emergency Speech Triage Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present TRIDENT (Transcription and Routing Intelligence for Dispatcher-Empowered National Triage), a three-layer dispatcher-support architecture designed to structure emergency call inputs for human application of established triage protocols (the ESI for routine operations and START for mass casualty events), even when automatic speech recognition fails. |
Elroy Galbraith; Chadwick Sutherland; Donahue Morgan; | arxiv-cs.CL | 2025-12-11 |
| 601 | Efficient ASR for Low-Resource Languages: Leveraging Cross-Lingual Unlabeled Data Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present a systematic investigation into cross-lingual continuous pretraining for low-resource languages, using Perso-Arabic languages (Persian, Arabic, and Urdu) as our primary case study. |
Srihari Bandarupalli; Bhavana Akkiraju; Charan Devarakonda; Vamsiraghusimha Narsinga; Anil Kumar Vuppala; | arxiv-cs.CL | 2025-12-08 |
| 602 | MNSC: Advancing Singlish Speech Understanding with Carefully Curated Corpora Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Singlish, a Creole language rooted in English, is a key focus in linguistic research within multilingual and multicultural contexts. However, its spoken form remains … |
BIN WANG et. al. | 2025 IEEE Automatic Speech Recognition and Understanding … | 2025-12-06 |
| 603 | The Development and Experimental Evaluation of A Multilingual Speech Corpus for Low-Resource Turkic Languages Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This article presents the development and experimental evaluation of a speech corpus focused on Turkic languages, intended for use in speech synthesis and automatic translation tasks. |
AIDANA KARIBAYEVA et. al. | Applied Sciences | 2025-12-05 |
| 604 | Balancing Specialization and Generalization Trade-Off for Speech Recognition Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In some scenarios, we are interested in optimizing performance for the target domain (specialization)while preserving the general capabilities of the pretrained model. In this work, we study this effect for various finetuning strategies that aim to preserve pretrained model capabilities. |
Sebastian Cygert; Piotr Despot-Mładanowicz; Andrzej Czyżewski; | Electronics | 2025-12-05 |
| 605 | Enhancing Automatic Speech Recognition Through Integrated Noise Detection Architecture Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This research presents a novel approach to enhancing automatic speech recognition systems by integrating noise detection capabilities directly into the recognition architecture. |
Karamvir Singh; | arxiv-cs.SD | 2025-12-02 |
| 606 | Spoken Conversational Agents with Large Language Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This tutorial distills the path from cascaded ASR/NLU to end-to-end, retrieval-and vision-grounded systems. |
Chao-Han Huck Yang; Andreas Stolcke; Larry Heck; | arxiv-cs.CL | 2025-12-02 |
| 607 | Swivuriso: The South African Next Voices Multilingual Speech Dataset Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper introduces Swivuriso, a 3000-hour multilingual speech dataset developed as part of the African Next Voices project, to support the development and benchmarking of automatic speech recognition (ASR) technologies in seven South African languages. |
VUKOSI MARIVATEE et. al. | arxiv-cs.CL | 2025-12-01 |
| 608 | Benchmarking Automatic Speech Recognition Models for African Languages Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Large pre-trained systems such as Whisper, XLS-R, MMS, and W2v-BERT have expanded access to ASR technology, but their comparative behavior in African low-resource contexts has not been studied in a unified and systematic way. In this work, we benchmark four state-of-the-art ASR models across 13 African languages, fine-tuning them on progressively larger subsets of transcribed data ranging from 1 to 400 hours. |
ALVIN NAHABWE et. al. | arxiv-cs.CL | 2025-11-30 |
| 609 | ASR Under The Stethoscope: Evaluating Biases in Clinical Speech Recognition Across Indian Languages Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this study, we conduct the first systematic audit of ASR performance on real world clinical interview data spanning Kannada, Hindi, and Indian English, comparing leading models including Indic Whisper, Whisper, Sarvam, Google speech to text, Gemma3n, Omnilingual, Vaani, and Gemini. |
SUBHAM KUMAR et. al. | arxiv-cs.CL | 2025-11-30 |
| 610 | ZO-ASR: Zeroth-Order Fine-Tuning of Speech Foundation Models Without Back-Propagation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce ZO-ASR, a memory-efficient Zeroth-Order (ZO) method that avoids Back-Propagation (BP) and activation memory by estimating gradients via forward passes. |
YUEZHANG PENG et. al. | arxiv-cs.MM | 2025-11-30 |
| 611 | Deep Learning Techniques for Hindi Automatic Speech Recognition: A Comprehensive Survey Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This study examines multiple models on publicly available speech datasets to evaluate their performance for practical implementation. |
Hetal Gaudani; Narendra M Patel; | International Journal of Latest Technology in Engineering … | 2025-11-28 |
| 612 | Scaling HuBERT for African Languages: From Base to Large and XL Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: By conducting a carefully controlled experimental study focused exclusively on Sub-Saharan languages, covering automatic speech recognition (ASR) and language identification (LID) tasks, we demonstrate that larger architectures significantly improve performance by effectively leveraging large audio datasets. |
Antoine Caubrière; Elodie Gauthier; | arxiv-cs.CL | 2025-11-28 |
| 613 | Supplementary Resources and Analysis for Automatic Speech Recognition Systems Trained on The Loquacious Dataset Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: The main goal of the Loquacious dataset is to provide properly defined training and test partitions across many acoustic and language domains, with an open license suitable for both academia and industry. To further promote the benchmarking and usability of this new dataset, we present additional resources in the form of n-gram language models (LMs), a grapheme-to-phoneme (G2P) model and pronunciation lexica, with open and public access. |
NICK ROSSENBACH et. al. | arxiv-cs.CL | 2025-11-27 |
| 614 | ASR Error Correction in Low-Resource Burmese with Alignment-Enhanced Transformers Using Phonetic Features Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper investigates sequence-to-sequence Transformer models for automatic speech recognition (ASR) error correction in low-resource Burmese, focusing on different feature integration strategies including IPA and alignment information. |
Ye Bhone Lin; Thura Aung; Ye Kyaw Thu; Thazin Myint Oo; | arxiv-cs.CL | 2025-11-26 |
| 615 | SingingSDS: A Singing-Capable Spoken Dialogue System for Conversational Roleplay Applications Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present SingingSDS, a cascaded SDS that responds through singing rather than speaking, fostering more affective, memorable, and pleasurable interactions in character-based roleplay and interactive entertainment scenarios. |
JIONGHAO HAN et. al. | arxiv-cs.SD | 2025-11-25 |
| 616 | Zero-Shot Context-Aware ASR for Diverse Arabic Varieties Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Zero-shot ASR for Arabic remains challenging: while multilingual models perform well on Modern Standard Arabic (MSA), error rates rise sharply on dialectal and accented speech due … |
Bashar Talafha; Amin Abu Alhassan; M. Abdul-Mageed; | Annual Meeting of the Association for Computational … | 2025-11-24 |
| 617 | Context-Aware Whisper for Arabic ASR Under Linguistic Varieties Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce techniques such as prompt reordering, speaker-aware prefix synthesis, and modality-specific retrieval (lexical, semantic, acoustic) to improve transcription in real-world, zero-shot settings. |
Bashar Talafha; Amin Abu Alhassan; Muhammad Abdul-Mageed; | arxiv-cs.CL | 2025-11-24 |
| 618 | AIRHILT: A Human-in-the-Loop Testbed for Multimodal Conflict Detection in Aviation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce AIRHILT (Aviation Integrated Reasoning, Human-in-the-Loop Testbed), a modular and lightweight simulation environment designed to evaluate multimodal pilot and air traffic control (ATC) assistance systems for aviation conflict detection. |
Omar Garib; Jayaprakash D. Kambhampaty; Olivia J. Pinon Fischer; Dimitri N. Mavris; | arxiv-cs.RO | 2025-11-23 |
| 619 | A Multimodal Conversational Agent for Tabular Data Analysis Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this article, we present Talk2Data, a multimodal LLM-driven conversational agent for intuitive data exploration. |
Mohammad Nour Al Awad; Sergey Ivanov; Olga Tikhonova; Ivan Khodnenko; | arxiv-cs.AI | 2025-11-23 |
| 620 | WER Is Unaware: Assessing How ASR Errors Distort Clinical Understanding in Patient Facing Dialogue Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Our analysis reveals that WER and a comprehensive suite of existing metrics correlate poorly with the clinician-assigned risk labels (No, Minimal, or Significant Impact). To bridge this evaluation gap, we introduce an LLM-as-a-Judge, programmatically optimized using GEPA to replicate expert clinical assessment. |
ZACHARY ELLIS et. al. | arxiv-cs.CL | 2025-11-20 |
| 621 | Smart Voice Assistant Using Machine Learning and Deep Learning Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This project focuses on understanding the architecture, working principles, applications, and challenges of Smart Voice Assistants. |
Gorelal Verma; Deepesh Dewangan; | INTERNATIONAL JOURNAL OF SCIENTIFIC RESEARCH IN ENGINEERING … | 2025-11-19 |
| 622 | AfriSpeech-MultiBench: A Verticalized Multidomain Multicountry Benchmark Suite for African Accented English ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present AfriSpeech-MultiBench, the first domain-specific evaluation suite for over 100 African English accents across 10+ countries and seven application domains: Finance, Legal, Medical, General dialogue, Call Center, Named Entities and Hallucination Robustness. |
Gabrial Zencha Ashungafac; Mardhiyah Sanni; Busayo Awobade; Alex Gichamba; Tobi Olatunji; | arxiv-cs.CL | 2025-11-18 |
| 623 | Toward Conversational Hungarian Speech Recognition: Introducing The BEA-Large and BEA-Dialogue Datasets Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: The results highlight the persistent difficulty of conversational ASR, particularly due to disfluencies, overlaps, and informal speech patterns. By releasing these datasets and baselines, we aim to advance Hungarian speech technology and offer a methodological framework for developing spontaneous and conversational benchmarks in other languages. |
MÁTÉ GEDEON et. al. | arxiv-cs.CL | 2025-11-17 |
| 624 | Distinguishing Repetition Disfluency from Morphological Reduplication in Bangla ASR Transcripts: A Novel Corpus and Benchmarking Analysis Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Standard disfluency correction fails by erroneously deleting valid linguistic information. To solve this, we introduce the first publicly available, 20,000-row Bangla corpus, manually annotated to explicitly distinguish between these two phenomena in noisy ASR transcripts. |
Zaara Zabeen Arpa; Sadnam Sakib Apurbo; Nazia Karim Khan Oishee; Ajwad Abrar; | arxiv-cs.CL | 2025-11-17 |
| 625 | ELYADATA & LIA at NADI 2025: ASR and ADI Subtasks Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper describes Elyadata \& LIA’s joint submission to the NADI multi-dialectal Arabic Speech Processing 2025. |
Haroun Elleuch; Youssef Saidi; Salima Mdhaffar; Yannick Estève; Fethi Bougares; | arxiv-cs.CL | 2025-11-13 |
| 626 | Omnilingual ASR: Open-Source Multilingual Speech Recognition for 1600+ Languages Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Expanding ASR coverage has been costly and limited by architectures that restrict language support, making extension inaccessible to most–all while entangled with ethical concerns when pursued without community collaboration. To transcend these limitations, we introduce Omnilingual ASR, the first large-scale ASR system designed for extensibility. |
OMNILINGUAL ASR TEAM et. al. | arxiv-cs.CL | 2025-11-12 |
| 627 | A Prosodically Annotated Bengali and Assamese Audiobook Corpus for Sentence Boundary Detection Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Bengali and Assamese are two widely spoken Indo-Aryan languages, but speech resources for them are still very limited. In particular, there is no existing dataset that marks … |
Priyanjana Chowdhury; Sanghamitra Nath; Utpal Sharma; | 2025 28th Conference of the Oriental COCOSDA International … | 2025-11-12 |
| 628 | H-PRM: A Pluggable Hotword Pre-Retrieval Module for Various Speech Recognition Systems Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we introduce a novel hotword customization system that utilizes a hotword pre-retrieval module (H-PRM) to identify the most relevant hotword candidate by measuring the acoustic similarity between the hotwords and the speech segment. |
HUANGYU DAI et. al. | cikm | 2025-11-10 |
| 629 | E2E-VGuard: Adversarial Prevention for Production LLM-based End-To-End Speech Synthesis Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Therefore, this E2E speech synthesis also requiresnew security mechanisms. To tackle these challenges, we propose E2E-VGuard, aproactive defense framework for two emerging threats: (1) production LLM-basedspeech synthesis, and (2) the novel attack arising from ASR-driven E2Escenarios. |
ZHISHENG ZHANG et. al. | arxiv-cs.SD | 2025-11-10 |
| 630 | CLiFT-ASR: A Cross-Lingual Fine-Tuning Framework for Low-Resource Taiwanese Hokkien Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In addition, prior studies have rarely exploredstaged strategies that integrate both annotation types. To address this gap, wepresent CLiFT-ASR, a cross-lingual fine-tuning framework that builds onMandarin HuBERT models and progressively adapts them to Taiwanese Hokkien. |
HUNG-YANG SUNG et. al. | arxiv-cs.CL | 2025-11-10 |
| 631 | Towards Fine-Grained Code-Switch Speech Translation with Semantic Space Alignment Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Previous studies tend to rely on the model itself to implicitly learn semantic modeling during training, and resort to inefficient and costly manual annotations for these two challenges. To mitigate these limitations, we propose enhancing Large Language Models (LLMs) with a Mixture of Experts (MoE) speech projector, where each expert specializes in the semantic subspace of a specific language, enabling fine-grained modeling of speech features. |
YAN GAO et. al. | arxiv-cs.CL | 2025-11-09 |
| 632 | Enabling Differentially Private Federated Learning for Speech Recognition: Benchmarks, Adaptive Optimizers, and Gradient Clipping Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To address this gap, we establish **the first benchmark for FL with DP** in end-to-end ASR. Our approach centers on per-layer clipping and layer-wise gradient normalization: theoretical analysis reveals that these techniques together mitigate clipping bias and gradient heterogeneity across layers in deeper models. |
MARTIN PELIKAN et. al. | nips | 2025-11-07 |
| 633 | Objective Soups: Multilingual Multi-Task Modeling for Speech Processing Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This leads to an important question: Is it more effective to separate highly conflicting objectives into different optimization levels or to keep them in a single level? To address this question, this paper investigates three multi-objective MSP formulations, which we refer to as \textbf{objective soup recipes}. |
A F M SAIF et. al. | nips | 2025-11-07 |
| 634 | E2E-VGuard: Adversarial Prevention for Production LLM-based End-To-End Speech Synthesis Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Therefore, this E2E speech synthesis also requires new security mechanisms. To tackle these challenges, we propose E2E-VGuard, a proactive defense framework for two emerging threats: (1) production LLM-based speech synthesis, and (2) the novel attack arising from ASR-driven E2E scenarios. |
ZHISHENG ZHANG et. al. | nips | 2025-11-07 |
| 635 | MoME: Mixture of Matryoshka Experts for Audio-Visual Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: However, current MRL-based methods treat each scale independently during training, limiting cross-scale generalization, robustness at high compression, and interpretability. To overcome these limitations, we propose MoME (Mixture of Matryoshka Experts), a novel framework that integrates sparse Mixture-of-Experts (MoE) into MRL-based LLMs for AVSR. |
UMBERTO CAPPELLAZZO et. al. | nips | 2025-11-07 |
| 636 | BlockDecoder: Boosting ASR Decoders with Context and Merger Modules Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We observe a systematic pattern across the attention distributions of decoder layers in prior architectures: the initial layers direct most attention towards building textual context, while the later layers largely focus on merging acoustic and textual information for the final predictions. Leveraging this key insight, we propose **BlockDecoder**, a novel decoder architecture comprising two distinct components: a text encoder that is purely text-based, and a **Merger** that combines information from the audio encoder and text encoder to generate output tokens. |
Darshan Prabhu; Preethi Jyothi; | nips | 2025-11-07 |
| 637 | VITA-Audio: Fast Interleaved Audio-Text Token Generation for Efficient Large Speech-Language Model Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: However, the existing speech models still experience high latency when generating the first audio token during streaming, which poses a significant bottleneck for deployment. To address this issue, we propose VITA-Audio, an end-to-end large speech model with fast audio-text token generation. |
ZUWEI LONG et. al. | nips | 2025-11-07 |
| 638 | CantoASR: Prosody-Aware ASR-LALM Collaboration for Low-Resource Cantonese Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce CantoASR,a collaborative ASR-LALM error correction framework that integrates forcedalignment for acoustic feature extraction, a LoRA-finetuned Whisper forimproved tone discrimination, and an instruction-tuned Qwen-Audio forprosody-aware correction. |
DAZHONG CHEN et. al. | arxiv-cs.CL | 2025-11-06 |
| 639 | WST: Weakly Supervised Transducer for Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: The Recurrent Neural Network-Transducer (RNN-T) is widely adopted inend-to-end (E2E) automatic speech recognition (ASR) tasks but depends heavilyon large-scale, high-quality annotated data, which are often costly anddifficult to obtain. To mitigate this reliance, we propose a Weakly SupervisedTransducer (WST), which integrates a flexible training graph designed torobustly handle errors in the transcripts without requiring additionalconfidence estimation or auxiliary pre-trained models. |
DONGJI GAO et. al. | arxiv-cs.CL | 2025-11-05 |
| 640 | Intelligent Navigation Assistant for Campuses Using Speech Recognition, NLP And A* Algorithm Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Abstract Campus navigation often produce challenges for students, staff, and visitors, specifically in large complex. This research presents an intelligent navigation assistant that combines automatic speech recognition (ASR), Natural Language Processing (NLP), and the A* pathfinding algorithm to tackle these issues through voice interaction. |
Kirti A. Satpute; Mrunal Wakadkar; Shruti Nimbalkar; Sneha Shelar; Prof. Kalpana Sonval; | INTERNATIONAL JOURNAL OF SCIENTIFIC RESEARCH IN ENGINEERING … | 2025-11-04 |
| 641 | Energy-Efficient Hardware Acceleration of Whisper ASR on A CGLA Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: While ASICs offer highefficiency, they lack the programmability to adapt to evolving algorithms. Toaddress this trade-off, we implement and evaluate Whisper’s core computationalkernel on the IMAX, a general-purpose Coarse-Grained Linear Arrays (CGLAs)accelerator. |
Takuto Ando; Yu Eto; Ayumu Takeuchi; Yasuhiko Nakashima; | arxiv-cs.AR | 2025-11-04 |
| 642 | Transcription Accuracy of Automatic Speech Recognition for Orthodontic Clinical Records Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: The aim of this study was to investigate the transcriptional accuracy of ASR systems using orthodontic clinical records as the experimental model. |
R. O’KANE et. al. | Journal of Dental Research | 2025-11-03 |
| 643 | CoVoGER: A Multilingual Multitask Benchmark for Speech-to-text Generative Error Correction with Large Language Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce CoVoGER, a benchmark for GER that covers both ASR and speech-to-text translation (ST) across 15 languages and 28 language pairs. |
Zhengdong Yang; Zhen Wan; Sheng Li; Chao-Han Huck Yang; Chenhui Chu; | emnlp | 2025-11-02 |
| 644 | Visual-Aware Speech Recognition for Noisy Scenarios Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: However, current Automatic Speech Recognition (ASR) or Audio-Visual Speech Recognition (AVSR) models often struggle in noisy scenarios. To solve this task, we propose a model that improves transcription by correlating noise sources to visual cues. |
Balaji Darur; Karan Singla; | emnlp | 2025-11-02 |
| 645 | ASR-EC Benchmark: Evaluating Large Language Models on Chinese ASR Error Correction IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We apply LLMs to ASR error correction in three paradigms. |
Victor Junqiu Wei; Weicheng Wang; Di Jiang; Yuanfeng Song; Lu Wang; | emnlp | 2025-11-02 |
| 646 | Idiosyncratic Versus Normative Modeling of Atypical Speech Recognition: Dysarthric Case Studies Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To investigate this, we compare four strategies: (a) *normative* models trained on typical speech (no personalization), (b) *idiosyncratic* models completely personalized to individuals, (c) *dysarthric-normative* models trained on other dysarthric speakers, and (d) *dysarthric-idiosyncratic* models which combine strategies by first modeling normative patterns before adapting to individual speech. In this case study, we find the dysarthric-idiosyncratic model performs better than the idiosyncratic approach while requiring less than half as much personalized data (36. |
Vishnu Raja; Adithya V Ganesan; Anand Syamkumar; Ritwik Banerjee; H. Schwartz; | emnlp | 2025-11-02 |
| 647 | Enhancing Speech Large Language Models with Prompt-Aware Mixture of Audio Encoders Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we propose Prompt-aware Mixture (PaM) to enhance the Speech LLM that uses multiple audio encoders. |
WEIQIAO SHAN et. al. | emnlp | 2025-11-02 |
| 648 | In-Context Learning Boosts Speech Recognition Via Human-like Adaptation to Speakers and Language Varieties IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce a scalable framework that allows for in-context learning (ICL) in Phi-4 Multimodal (Phi-4-MM) using interleaved task prompts and audio-text pairs, and find that as few as 12 example utterances (~50 seconds) at inference time reduce word error rates by a relative 19. |
NATHAN ROLL et. al. | emnlp | 2025-11-02 |
| 649 | From Tens of Hours to Tens of Thousands: Scaling Back-Translation for Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts Related Code View Save Highlight: This paper introduces Speech Back-Translation, a a scalable pipeline that improves multilingual ASR models by converting large-scale text corpora into synthetic speech via off-the-shelf text-to-speech (TTS) models. |
Tianduo Wang; Lu Xu; Wei Lu; Shanbo Cheng; | emnlp | 2025-11-02 |
| 650 | Generative Annotation for ASR Named Entity Correction Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose a novel NEC method that utilizes speech sound features to retrieve candidate entities. |
YUANCHANG LUO et. al. | emnlp | 2025-11-02 |
| 651 | Dynamic Model-Bank Test-Time Adaptation for Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To alleviate the risk of performance collapse due to error accumulation, we propose Dynamic Model-bank Single-Utterance Test-time Adaptation (DMSUTA), a sustainable continual TTA framework based on adaptive ASR model ensembling. |
YANSHUO WANG et. al. | emnlp | 2025-11-02 |
| 652 | Layer-wise Minimal Pair Probing Reveals Contextual Grammatical-Conceptual Hierarchy in Speech Representations Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Transformer-based speech language models (SLMs) have significantly improved neural speech recognition and understanding. |
Linyang He; Qiaolin Wang; Xilin Jiang; Nima Mesgarani; | emnlp | 2025-11-02 |
| 653 | Beyond WER: Probing Whisper’s Sub‐token Decoder Across Diverse Language Resource Levels Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: While large multilingual automatic speech recognition (ASR) models achieve remarkable performance, the internal mechanisms of the end-to-end pipeline, particularly concerning fairness and efficacy across languages, remain underexplored. This paper introduces a fine-grained analysis of Whisper’s multilingual decoder, examining its sub-token hypotheses during transcription across languages with various resource levels. |
Siyu Liang; Nicolas Ballier; Gina-Anne Levow; Richard Wright; | emnlp | 2025-11-02 |
| 654 | Do Slides Help? Multi-modal Context for Automatic Transcription of Conference Talks Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In a first step, we create a benchmark for multi-modal presentation including an automatic analysis of transcribing domain-specific terminology. |
Supriti Sinhamahapatra; Jan Niehues; | emnlp | 2025-11-02 |
| 655 | LiteASR: Efficient Automatic Speech Recognition with Low-Rank Approximation IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts Related Code View Save Highlight: We introduce LiteASR, a low-rank compression scheme for ASR encoders that significantly reduces inference costs while maintaining transcription accuracy. |
Keisuke Kamahori; Jungo Kasai; Noriyuki Kojima; Baris Kasikci; | emnlp | 2025-11-02 |
| 656 | Towards Language-Agnostic STIPA: Universal Phonetic Transcription to Support Language Documentation at Scale Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce a new dataset for South Levantine Arabic and present the first large-scale evaluation of STIPA models across 51 language families. |
Jacob Lee Suchardt; Hana El-Shazli; Pierluigi Cassotti; | emnlp | 2025-11-02 |
| 657 | Do Modern Speech LLMs and Re-scoring Techniques Improve Bilingual ASR Performance for Basque and Spanish in Domain-specific Contexts? Related Papers Related Patents Related Grants Related Venues Related Experts View Save |
Ander González-Docasal; Juan Camilo Vásquez-Correa; Haritz Arzelus; Aitor Álvarez; S. Moreno-Acevedo; | Comput. Speech Lang. | 2025-11-01 |
| 658 | Enhancing XR Theatre with Remote Scent Delivery Using Audio Speech Recognition and Convolutional Neural Network-Based Scene Detection Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: This paper presents a multi-sensory solution designed to enhance the immersive experience of remote audiences for theatrical plays. Using the Macbettu play as a case study, the … |
A. SIMISCUKA et. al. | 2025 IEEE 6th International Symposium on the Internet of … | 2025-10-29 |
| 659 | Explainable Disentanglement on Discrete Speech Representations for Noise-Robust ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We show that our approachimproves alignment between clean/noisy speech and text, producing speech tokensthat display a high degree of noiseinvariance, and improves ASR performance.Keeping Whisper frozen, we show an 82% reduction in error rate compared toWhisper, and 35% improvement over baseline methods on the VBDemand test set.Further analyses show that the learned token space generalizes well to bothseen and unseen acoustic conditions. |
SHREYAS GOPAL et. al. | arxiv-cs.CL | 2025-10-29 |
| 660 | BEST-RQ-Based Self-Supervised Learning for Whisper Domain Adaptation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose BEARD (BEST-RQ Encoder Adaptation with Re-trainingand Distillation), a novel framework designed to adapt Whisper’s encoder usingunlabeled data. |
Raphaël Bagat; Irina Illina; Emmanuel Vincent; | arxiv-cs.CL | 2025-10-28 |
| 661 | RegSpeech12: A Regional Corpus of Bengali Spontaneous Speech Across Dialects Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: The Bengali language, spoken extensively across South Asia and amongdiasporic communities, exhibits considerable dialectal diversity shaped bygeography, culture, and history. … |
MD. REZUWAN HASSAN et. al. | arxiv-cs.CL | 2025-10-28 |
| 662 | POWSM: A Phonetic Open Whisper-Style Speech Foundation Model IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we introduce POWSM (Phonetic Open Whisper-style SpeechModel), the first unified framework capable of jointly performing multiplephone-related tasks. |
CHIN-JOU LI et. al. | arxiv-cs.CL | 2025-10-28 |
| 663 | Ming-Flash-Omni: A Sparse, Unified Architecture for Multimodal Perception and Generation IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose Ming-Flash-Omni, an upgraded version of Ming-Omni, built upon asparser Mixture-of-Experts (MoE) variant of Ling-Flash-2.0 with 100 billiontotal parameters, of which only 6.1 billion are active per token. |
INCLUSION AI et. al. | arxiv-cs.CV | 2025-10-28 |
| 664 | Arabic Little STT: Arabic Children Speech Recognition Dataset Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Moreover, the absence of child-specificspeech corpora is an essential gap that poses significant challenges. Toaddress this gap, we present our created dataset, Arabic Little STT, a datasetof Levantine Arabic child speech recorded in classrooms, containing 355utterances from 288 children (ages 6 – 13). |
Mouhand Alkadri; Dania Desouki; Khloud Al Jallad; | arxiv-cs.CL | 2025-10-27 |
| 665 | MoTAS: MoE-Guided Feature Selection from TTS-Augmented Speech for Enhanced Multimodal Alzheimer’s Early Screening Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: However, challenges such as limited data and the lack of fine-grained, adaptive feature selection often hinder performance. To address these issues, we propose MoTAS, a robust framework designed to enhance AD screening efficiency. |
Yongqi Shao; Bingxin Mei; Cong Tan; Hong Huo; Tao Fang; | mm | 2025-10-27 |
| 666 | Are ASR Foundation Models Generalized Enough to Capture Features of Regional Dialects for Low-resource Languages? Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To investigate theeffects of dialectal variations on ASR we develop a 78-hour annotated BengaliSpeech-to-Text (STT) corpus named Ben-10. |
TAWSIF TASHWAR DIPTO et. al. | arxiv-cs.CL | 2025-10-27 |
| 667 | LRW-Persian: Lip-reading in The Wild Dataset for Persian Language Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce LRW-Persian, the largest in-the-wild Persianword-level lipreading dataset, comprising $743$ target words and over$414{,}000$ video samples extracted from more than $1{,}900$ hours of footageacross $67$ television programs. |
Zahra Taghizadeh; Mohammad Shahverdikondori; Arian Noori; Alireza Dadgarnia; | arxiv-cs.CV | 2025-10-26 |
| 668 | A Sociophonetic Analysis of Racial Bias in Commercial ASR Systems Using The Pacific Northwest English Corpus Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents a systematic evaluation of racial bias in four majorcommercial automatic speech recognition (ASR) systems using the PacificNorthwest English (PNWE) corpus. |
Michael Scott; Siyu Liang; Alicia Wassink; Gina-Anne Levow; | arxiv-cs.CL | 2025-10-25 |
| 669 | Speaking Clearly: A Simplified Whisper-Based Codec for Low-Bitrate Speech Coding Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Throughempirical analysis, we discover that targeted architectural simplification canunlock the acoustic modeling potential of Whisper, a text-aligned AutomaticSpeech Recognition (ASR) model. |
Xin Zhang; Lin Li; Xiangni Lu; Jianquan Liu; Kong Aik Lee; | arxiv-cs.SD | 2025-10-23 |
| 670 | Multimodal Speech Analysis for Early Detection of Mild Cognitive Impairment: A Scalable Approach Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Early detection of dementia is critical for timely intervention, yet conventional diagnostic methods remain resource intensive and limited in scalability. Speech analysis offers a … |
Muhammad Bilal; Waleed Abdulla; Gary Cheung; L. Tippett; S. R. Shahamiri; | 2025 Asia Pacific Signal and Information Processing … | 2025-10-22 |
| 671 | HIPA-MoE: A Parameter-Efficient Fine-Tuning Architecture with Hierarchical Adapter-Based Mixture-Of-Experts for Multilingual ASR Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Multilingual automatic speech recognition (MASR) has advanced significantly with self-supervised pretraining (SSL). However, conventional fine-tuning remains constrained by data … |
Xun Lu; Xuyang Wang; Gaofeng Cheng; Lin Zheng; Pengyuan Zhang; | 2025 Asia Pacific Signal and Information Processing … | 2025-10-22 |
| 672 | Reducing Orthographic Dependency on Paired Data By Probabilistic Integration Via Syllabogram for Japanese Dialogue Speech Recognition Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Character end-to-end ASR (C-ASR) models fail to recognize the spelling of words that are absent from paired data (audio and text). This orthographic dependency on the paired data … |
Ryu Takeda; Kazunori Komatani; | 2025 Asia Pacific Signal and Information Processing … | 2025-10-22 |
| 673 | EvolveCaptions: Real-Time Collaborative ASR Adaptation for DHH Speakers Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Current ASR systems struggle to reliably recognize the speech of Deaf and Hard of Hearing (DHH) individuals, particularly in real-time communication. Existing personalization … |
Liang-Yuan Wu; Dhruv Jain; | Proceedings of the 27th International ACM SIGACCESS … | 2025-10-22 |
| 674 | Re-evaluating Minimum Bayes Risk Decoding for Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we evaluate MBRdecoding for ASR and ST tasks on English and Japanese using Whisper and itsderivative models. |
Yuu Jinnai; | arxiv-cs.CL | 2025-10-22 |
| 675 | Chain-of-Thought Distillation for ASR Error Correction with Multimodal Large Language Models Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Large language models (LLMs) have demonstrated strong performance in automatic speech recognition (ASR) error correction. However, current approaches rely heavily on annotated … |
SHAOMENG YANG et. al. | 2025 Asia Pacific Signal and Information Processing … | 2025-10-22 |
| 676 | CARTGPT: Real-Time Correction of CART Captions Using Large Language Models Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Communication Access Realtime Translation (CART) is a widely used captioning technology among deaf and hard of hearing (DHH) individuals, valued for its high accuracy and ability … |
Liang-Yuan Wu; Andrea Kleiver; Dhruv Jain; | Proceedings of the 27th International ACM SIGACCESS … | 2025-10-22 |
| 677 | MLMA: Towards Multilingual ASR With Mamba-based Architectures Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, weintroduce MLMA (Multilingual Language Modeling with Mamba for ASR), a newapproach that leverages the Mamba architecture — an efficient state-spacemodel optimized for long-context sequence processing — for multilingual ASR.Using Mamba, MLMA implicitly incorporates language-aware conditioning andshared representations to support robust recognition across diverse languages.Experiments on standard multilingual benchmarks show that MLMA achievescompetitive performance compared to Transformer-based architectures. |
Mohamed Nabih Ali; Daniele Falavigna; Alessio Brutti; | arxiv-cs.CL | 2025-10-21 |
| 678 | VALLR: Visual ASR Language Model for Lip Reading IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Current methods typically attempt to predict words or characters directly from these visual cues, but this approach frequently encounters high error rates due to coarticulation effects and viseme ambiguity. We propose a novel two-stage, phoneme-centric framework for Visual Automatic Speech Recognition (V-ASR) that addresses these longstanding challenges. |
Marshall Thomas; Edward Fish; Richard Bowden; | iccv | 2025-10-20 |
| 679 | Hallucination Benchmark for Speech Foundation Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Consequently, there is a critical need for new evaluationframeworks that can effectively identify and assess models with a heightenedpropensity for generating hallucinated content. |
Alkis Koudounas; Moreno La Quatra; Manuel Giollo; Sabato Marco Siniscalchi; Elena Baralis; | arxiv-cs.CL | 2025-10-18 |
| 680 | Probing The Hidden Talent of ASR Foundation Models for L2 English Oral Assessment Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we explore the untapped potential of Whisper, awell-established automatic speech recognition (ASR) foundation model, in thecontext of L2 spoken language assessment (SLA). |
Fu-An Chao; Bi-Cheng Yan; Berlin Chen; | arxiv-cs.CL | 2025-10-18 |
| 681 | RLAIF-SPA: Optimizing LLM-based Emotional Speech Synthesis Via RLAIF Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Existing methods often relyon costly emotion annotations or optimize indirect objectives that fail tocapture the emotional expressiveness and perceptual naturalness of speech,leading to generated speech that is accurate but emotionally flat. To addressthese challenges, we propose the RLAIF-SPA framework, incorporating aReinforcement Learning from AI Feedback (RLAIF) mechanism to employ AutomaticSpeech Recognition (ASR) and Large Language Model (LLM) techniques torespectively judge semantic accuracy and prosodic-emotional label alignment asa direct reward for emotional expressiveness and intelligibility optimization.Specifically, it leverages Prosodic Label Alignment to enhance expressivequality by jointly considering semantic accuracy and prosodic-emotionalalignment along four fine-grained dimensions: Structure, Emotion, Speed, andTone. |
QING YANG et. al. | arxiv-cs.CL | 2025-10-16 |
| 682 | Toward Responsible ASR for African American English Speakers: A Scoping Review of Bias and Equity in Speech Technology Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose a governance-centered ASR lifecycle as an emergent interdisciplinary framework for responsible ASR development and offer implications for researchers, practitioners, and policymakers seeking to address language marginalization in speech AI systems. |
JAY L. CUNNINGHAM et. al. | Proceedings of the AAAI/ACM Conference on AI, Ethics, and … | 2025-10-15 |
| 683 | End-to-end Speech Recognition with Similar Length Speech and Text Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In thispaper, we focus on speech recognition in those cases where the length of speechaligns closely with that of the corresponding text. |
Peng Fan; Wenping Wang; Fei Deng; | arxiv-cs.CL | 2025-10-12 |
| 684 | Proficiency-Aware Adaptation and Data Augmentation for Robust L2 ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Using theCEFR-graded Speak and Improve corpus, we show that naive fine-tuning of Whisperreduces average WER but simultaneously widens disparities anddisproportionately harms lower-level learners. To address this, we propose twostrategies: (i) proficiency-aware multitask learning, jointly optimizing ASRwith proficiency classification, and (ii) targeted augmentation, applyingspectrogram masking to low-proficiency speech to counter imbalance. |
Ling Sun; Charlotte Zhu; Shuju Shi; | arxiv-cs.SD | 2025-10-12 |
| 685 | End-to-end Automatic Speech Recognition and Speech Translation: Integration of Speech Foundational Models and LLMs Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper explores acombined end-to-end architecture of pre-trained speech encoders and LargeLanguage Models (LLMs) for performing both Automatic Speech Recognition (ASR)and ST simultaneously. |
Nam Luu; Ondřej Bojar; | arxiv-cs.CL | 2025-10-11 |
| 686 | Voice-Enabled Local Language Translator Using Generative AI Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents a voice-enabled local language translator powered by Generative AI, integrating Automatic Speech Recognition (ASR), Natural Language Understanding (NLU), generative translation, and Text-to-Speech (TTS) synthesis for accurate, context-aware output. |
Dr. V. Shanmugapriya; J N Pravanthika; | International Scientific Journal of Engineering and … | 2025-10-10 |
| 687 | Accent-Invariant Automatic Speech Recognition Via Saliency-Driven Spectrogram Masking Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: For Persian, we introduce a newly collected dataset spanning multipleregional accents, establishing the first systematic benchmark for accentvariation in Persian ASR that fills a critical gap in multilingual speechresearch and provides a foundation for future studies on low-resource,linguistically diverse languages. |
Mohammad Hossein Sameti; Sepehr Harfi Moridani; Ali Zarean; Hossein Sameti; | arxiv-cs.CL | 2025-10-10 |
| 688 | Speech Recognition and Synthesis Models and Platforms for The Kazakh Language Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This study provides a comprehensive analysis of existing speech recognition and synthesis models, emphasizing their applicability and adaptation to the Kazakh language. |
Aidana Karibayeva; Vladislav Karyukin; Balzhan Abduali; Dina Amirova; | Information | 2025-10-10 |
| 689 | Serial-Parallel Dual-Path Architecture for Speaking Style Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we propose a novel serial-paralleldual-path architecture for SSR that leverages acoustic-linguistic bimodalinformation. |
GUOJIAN LI et. al. | arxiv-cs.SD | 2025-10-09 |
| 690 | How Much Speech Data Is Necessary for ASR in African Languages? An Evaluation of Data Scaling in Kinyarwanda and Kikuyu Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We evaluate Whisper’sperformance through comprehensive experiments on two Bantu languages:systematic data scaling analysis on Kinyarwanda using training sets from 1 to1,400 hours, and detailed error characterization on Kikuyu using 270 hours oftraining data. |
BENJAMIN AKERA et. al. | arxiv-cs.CL | 2025-10-08 |
| 691 | Open ASR Leaderboard: Towards Reproducible and Transparent Multilingual and Long-Form Speech Recognition Evaluation IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present the Open ASRLeaderboard, a fully reproducible benchmark and interactive leaderboardcomparing 60+ open-source and proprietary systems across 11 datasets, includingdedicated multilingual and long-form tracks. |
VAIBHAV SRIVASTAV et. al. | arxiv-cs.CL | 2025-10-08 |
| 692 | Linguistically Informed Tokenization Improves ASR for Underresourced Languages Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We find that alinguistically informed phonemic tokenization system substantially improves WERand CER compared to a baseline orthographic tokenization scheme. |
Massimo Daul; Alessio Tosolini; Claire Bowern; | arxiv-cs.CL | 2025-10-07 |
| 693 | BanglaTalk: Towards Real-Time Speech Assistance for Bengali Regional Dialects Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we present BanglaTalk, thefirst real-time speech assistance system for Bengali regional dialects.BanglaTalk follows the client-server architecture and uses the Real-timeTransport Protocol (RTP) to ensure low-latency communication. |
Jakir Hasan; Shubhashis Roy Dipta; | arxiv-cs.CL | 2025-10-07 |
| 694 | How I Built ASR for Endangered Languages with A Spoken Dictionary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Inthis paper, we explore how little data, and in what form, is needed to buildASR for critically endangered languages. |
Christopher Bartley; Anton Ragni; | arxiv-cs.CL | 2025-10-06 |
| 695 | Phonetic Analysis of Real and Synthetic Speech Using HuBERT Embeddings: Perspectives for Deepfake Detection Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: The growing sophistication of speech generated by Artificial Intelligence (AI) has introduced new challenges in audio deepfake detection. Text-to-speech (TTS) and voice conversion … |
DIA ELHAK TEMMAR et. al. | 2025 IEEE International Conference on Systems, Man, and … | 2025-10-05 |
| 696 | Integrating Speech Recognition Into Intelligent Information Systems: From Statistical Models to Deep Learning Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Automatic speech recognition (ASR) has advanced rapidly, evolving from early template-matching systems to modern deep learning frameworks. This review systematically traces ASR’s … |
Chao-Ji Wu; Yi Pan; Haipang Wu; Lei Ning; | Informatics | 2025-10-04 |
| 697 | Revisiting Direct Speech-to-Text Translation with Speech LLMs: Better Scaling Than CoT Prompting? Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we systematicallycompare CoT and Direct prompting under increasing amounts of S2TT data. |
Oriol Pareras; Gerard I. Gállego; Federico Costa; Cristina España-Bonet; Javier Hernando; | arxiv-cs.CL | 2025-10-03 |
| 698 | Listening or Reading? Evaluating Speech Awareness in Chain-of-Thought Speech-to-Text Translation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Analyzing CoT through attribution methods, robustnessevaluations with corrupted transcripts, and prosody-awareness, we find that itlargely mirrors cascaded behavior, relying mainly on transcripts while barelyleveraging speech. |
JACOBO ROMERO-DÍAZ et. al. | arxiv-cs.CL | 2025-10-03 |
| 699 | EvolveCaptions: Empowering DHH Users Through Real-Time Collaborative Captioning Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present EvolveCaptions, a real-time, collaborative ASRadaptation system that supports in-situ personalization with minimal effort.Hearing participants correct ASR errors during live conversations. |
Liang-Yuan Wu; Dhruv Jain; | arxiv-cs.HC | 2025-10-02 |
| 700 | Automatic Speech Recognition (ASR) for African Low-Resource Languages: A Systematic Literature Review Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: ASR has achieved remarkable global progress, yet African low-resourcelanguages remain rigorously underrepresented, producing barriers to digitalinclusion across the continent with more than +2000 languages. This systematicliterature review (SLR) explores research on ASR for African languages with afocus on datasets, models and training methods, evaluation techniques,challenges, and recommends future directions. |
SUKAIRAJ HAFIZ IMAM et. al. | arxiv-cs.CL | 2025-10-01 |
| 701 | Backdoor Attacks Against Speech Language Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we present the firstsystematic study of audio backdoor attacks against speech language models. |
Alexandrine Fortier; Thomas Thebaud; Jesús Villalba; Najim Dehak; Patrick Cardinal; | arxiv-cs.CL | 2025-10-01 |
| 702 | Automatic Spoken-To-Sign Language Translation: A Review of Key Technologies, Limitations, and Interdisciplinary Opportunities Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: The communication barrier between deaf and hearing individuals remains a global challenge in academic, professional, and social contexts. Although numerous countries officially … |
W. FERREIRA et. al. | 2025 27th Symposium on Virtual and Augmented Reality (SVR) | 2025-09-30 |
| 703 | ASR Under Noise: Exploring Robustness for Sundanese and Javanese Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We investigate the robustness of Whisper-based automatic speech recognition(ASR) models for two major Indonesian regional languages: Javanese andSundanese. |
Salsabila Zahirah Pranida; Muhammad Cendekia Airlangga; Rifo Ahmad Genadi; Shady Shehata; | arxiv-cs.CL | 2025-09-30 |
| 704 | Beyond WER: Probing Whisper’s Sub-token Decoder Across Diverse Language Resource Levels Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: While large multilingual automatic speech recognition (ASR) models achieveremarkable performance, the internal mechanisms of the end-to-end pipeline,particularly concerning fairness and efficacy across languages, remainunderexplored. This paper introduces a fine-grained analysis of Whisper’smultilingual decoder, examining its sub-token hypotheses during transcriptionacross languages with various resource levels. |
Siyu Liang; Nicolas Ballier; Gina-Anne Levow; Richard Wright; | arxiv-cs.CL | 2025-09-29 |
| 705 | Confidence-Guided Error Correction for Disordered Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We investigate the use of large language models (LLMs) as post-processingmodules for automatic speech recognition (ASR), focusing on their ability toperform error correction for disordered speech. |
Abner Hernandez; Tomás Arias Vergara; Andreas Maier; Paula Andrea Pérez-Toro; | arxiv-cs.CL | 2025-09-29 |
| 706 | HiKE: Hierarchical Evaluation Framework for Korean-English Code-Switching Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, weintroduce HiKE: the Hierarchical Korean-English code-switching benchmark, thefirst globally accessible evaluation framework for Korean-English CS, aiming toprovide a means for the precise evaluation of multilingual ASR models and tofoster research in the field. |
Gio Paik; Yongbeom Kim; Soungmin Lee; Sangmin Ahn; Chanwoo Kim; | arxiv-cs.CL | 2025-09-29 |
| 707 | Game-Oriented ASR Error Correction Via RAG-Enhanced LLM Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: However, general ASR systems struggle withgaming-specific challenges like short phrases, rapid speech, jargon, and noise,leading to frequent errors. To address this, we propose the GO-AEC framework,which integrates large language models, Retrieval-Augmented Generation (RAG),and a data augmentation strategy using LLMs and TTS. |
Yan Jiang; Yongle Luo; Qixian Zhou; Elvis S. Liu; | arxiv-cs.AI | 2025-09-28 |
| 708 | MeanFlowSE: One-Step Generative Speech Enhancement Via MeanFlow Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Speech enhancement (SE) recovers clean speech from noisy signals and is vitalfor applications such as telecommunications and automatic speech recognition(ASR). While generative … |
YIKE ZHU et. al. | arxiv-cs.SD | 2025-09-27 |
| 709 | Align2Speak: Improving TTS for Low Resource Languages Via ASR-Guided Online Preference Optimization Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Developing high-quality text-to-speech (TTS) systems for low-resourcelanguages is challenging due to the scarcity of paired text and speech data. Incontrast, automatic speech … |
SHEHZEEN HUSSAIN et. al. | arxiv-cs.AI | 2025-09-25 |
| 710 | Lightweight Front-end Enhancement for Robust ASR Via Frame Resampling and Sub-Band Pruning Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper proposes optimizations to reduce SE computational costswithout compromising ASR performance. |
Siyi Zhao; Wei Wang; Yanmin Qian; | arxiv-cs.SD | 2025-09-25 |
| 711 | SwissGPC V1.0 – The Swiss German Podcasts Corpus Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: We present SwissGPC v1.0, the first mid-to-large-scale corpus of spontaneous Swiss German speech, developed to support research in ASR, TTS, dialect identification, and related … |
Samuel Stucki; Mark Cieliebak; Jan Deriu; | ArXiv | 2025-09-24 |
| 712 | SloPalSpeech: A 2,8000-Hour Slovak Speech Corpus from Parliamentary Data Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We developed a robust processingpipeline to align and segment long-form recordings into clean, 30-secondaudio-transcript pairs suitable for model training. We use this dataset tofine-tune several OpenAI Whisper models (small, medium, large-v3, andlarge-v3-turbo), achieving significant Word Error Rate (WER) reductions onstandard Slovak benchmarks like Common Voice and FLEURS. |
Erik Božík; Marek Šuppa; | arxiv-cs.CL | 2025-09-23 |
| 713 | M4SER: Multimodal, Multirepresentation, Multitask, and Multistrategy Learning for Speech Emotion Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To address thischallenge, in our previous work, we introduced two auxiliary tasks, namely, ASRerror detection and ASR error correction, and we proposed a novel multimodalfusion (MF) method for learning modality-specific and modality-invariantrepresentations across different modalities. |
JIAJUN HE et. al. | arxiv-cs.HC | 2025-09-23 |
| 714 | LOTUSDIS: A Thai Far-field Meeting Corpus for Robust Conversational ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present LOTUSDIS, a publicly available Thai meeting corpus designed toadvance far-field conversational ASR. |
Pattara Tipaksorn; Sumonmas Thatphithakkul; Vataya Chunwijitra; Kwanchiva Thangthai; | arxiv-cs.CL | 2025-09-23 |
| 715 | Explore The Reinforcement Learning for The LLM Based ASR and TTS System Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this study, we propose alightweight RL framework tailored for audio-based LLMs that can process audioinputs and generate audio outputs. |
CHANGFENG GAO et. al. | arxiv-cs.SD | 2025-09-22 |
| 716 | Conversational Orientation Reasoning: Egocentric-to-Allocentric Navigation with Multimodal Chain-of-Thought Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce ConversationalOrientation Reasoning (COR), a new benchmark designed for Traditional Chineseconversational navigation projected from real-world environments, addressingegocentric-to-allocentric reasoning in non-English and ASR-transcribedscenarios. We propose a multimodal chain-of-thought (MCoT) framework, whichintegrates ASR-transcribed speech with landmark coordinates through astructured three-step reasoning process: (1) extracting spatial relations, (2)mapping coordinates to absolute directions, and (3) inferring user orientation.A curriculum learning strategy progressively builds these capabilities onTaiwan-LLM-13B-v2.0-Chat, a mid-sized model representative ofresource-constrained settings. |
Yu Ti Huang; | arxiv-cs.LG | 2025-09-20 |
| 717 | A Systematic Literature Review on Bias Evaluation and Mitigation in Automatic Speech Recognition Models for Low-Resource African Languages IF:3 Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: With recent advancements in speech recognition, it is crucial to ensure that automatic speech recognition (ASR) systems do not exhibit systematic biases, such as those related to … |
Joyce Nakatumba‐Nabende; Sulaiman Kagumire; Caroline Kantono; Peter Nabende; | ACM Computing Surveys | 2025-09-20 |
| 718 | Idiosyncratic Versus Normative Modeling of Atypical Speech Recognition: Dysarthric Case Studies Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: State-of-the-art automatic speech recognition (ASR) models like Whisper,perform poorly on atypical speech, such as that produced by individuals withdysarthria. |
Vishnu Raja; Adithya V Ganesan; Anand Syamkumar; Ritwik Banerjee; H Andrew Schwartz; | arxiv-cs.SD | 2025-09-20 |
| 719 | MNV-17: A High-Quality Performative Mandarin Dataset for Nonverbal Vocalization Recognition in Speech IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Progress in NV-aware ASR hasbeen hindered by the lack of high-quality, well-annotated datasets. To addressthis gap, we introduce MNV-17, a 7.55-hour performative Mandarin speechdataset. |
JIALONG MAI et. al. | arxiv-cs.SD | 2025-09-19 |
| 720 | Thinking in Cocktail Party: Chain-of-Thought and Reinforcement Learning for Target Speaker Automatic Speech Recognition Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Target Speaker Automatic Speech Recognition (TS-ASR) aims to transcribe thespeech of a specified target speaker from multi-speaker mixtures in cocktailparty scenarios. Recent … |
Yiru Zhang; Hang Su; Lichun Fan; Zhenbo Luo; Jian Luan; | arxiv-cs.SD | 2025-09-19 |
| 721 | Impact of Phonetics on Speaker Identity in Adversarial Voice Attack Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we analyze adversarialaudio at the phonetic level and show that perturbations exploit systematicconfusions such as vowel centralization and consonant substitutions. |
Daniyal Kabir Dar; Qiben Yan; Li Xiao; Arun Ross; | arxiv-cs.SD | 2025-09-18 |
| 722 | HARNESS: Lightweight Distilled Arabic Speech Foundation Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, weintroduce HArnESS, the first Arabic-centric self-supervised speech modelfamily, designed to capture Arabic speech nuances. |
Vrunda N. sukhadia; Shammur Absar Chowdhury; | arxiv-cs.CL | 2025-09-18 |
| 723 | Speech Language Models for Under-Represented Languages: Insights from Wolof Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present our journey in training a speech language model for Wolof, anunderrepresented language spoken in West Africa, and share key insights. |
Yaya Sy; Dioula Doucouré; Christophe Cerisara; Irina Illina; | arxiv-cs.CL | 2025-09-18 |
| 724 | Frustratingly Easy Data Augmentation for Low-Resource ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper introduces three self-contained data augmentation methods forlow-resource Automatic Speech Recognition (ASR). |
Katsumi Ibaraki; David Chiang; | arxiv-cs.CL | 2025-09-18 |
| 725 | Canary-1B-v2 & Parakeet-TDT-0.6B-v3: Efficient and High-Performance Models for Multilingual ASR and AST IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This report introduces Canary-1B-v2, a fast, robust multilingual model forAutomatic Speech Recognition (ASR) and Speech-to-Text Translation (AST). |
MONICA SEKOYAN et. al. | arxiv-cs.CL | 2025-09-17 |
| 726 | Conducting Mission-Critical Voice Experiments with Automated Speech Recognition and Crowdsourcing Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper describes our efforts to develop the methodology and tools forhuman-subject experiments with MCV. |
JAN JANAK et. al. | arxiv-cs.NI | 2025-09-17 |
| 727 | GLAD: Global-Local Aware Dynamic Mixture-of-Experts for Multi-Talker ASR Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: End-to-end multi-talker automatic speech recognition (MTASR) facessignificant challenges in accurately transcribing overlapping speech,especially under high-overlap conditions. To … |
Yujie Guo; Jiaming Zhou; Yuhang Jia; Shiwan Zhao; Yong Qin; | arxiv-cs.SD | 2025-09-16 |
| 728 | Assessing Whisper for Infant Research: Benchmarking ASR Accuracy and Failure Analysis on Caregiver-Infant Interactions Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Transcribing naturalistic caregiver-child interactions is a labor-intensive task in developmental research. While automatic speech recognition (ASR) models offer potential … |
Yueyan Tang; Victoria Hennessy; Tairan Liu; G. Deák; | 2025 IEEE International Conference on Development and … | 2025-09-16 |
| 729 | PAC: Pronunciation-Aware Contextualized Large Language Model-based Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents a Pronunciation-Aware Contextualized (PAC) framework toaddress two key challenges in Large Language Model (LLM)-based Automatic SpeechRecognition (ASR) systems: effective pronunciation modeling and robusthomophone discrimination. |
LI FU et. al. | arxiv-cs.CL | 2025-09-16 |
| 730 | Fun-ASR Technical Report IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, wepresent Fun-ASR, a large-scale, LLM-based ASR system that synergisticallycombines massive data, large model capacity, LLM integration, and reinforcementlearning to achieve state-of-the-art performance across diverse and complexspeech recognition scenarios. |
KEYU AN et. al. | arxiv-cs.CL | 2025-09-15 |
| 731 | In-domain SSL Pre-training and Streaming ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this study, we investigate the benefits of domain-specific self-supervisedpre-training for both offline and streaming ASR in Air Traffic Control (ATC)environments. |
JAROD DURET et. al. | arxiv-cs.CL | 2025-09-15 |
| 732 | WhisTLE: Deeply Supervised, Text-Only Domain Adaptation for Pretrained Speech Recognition Transformers Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose WhisTLE, a deeply supervised,text-only adaptation method for pretrained encoder-decoder ASR models. |
Akshat Pandey; Karun Kumar; Raphael Tang; | arxiv-cs.CL | 2025-09-12 |
| 733 | Prominence-aware Automatic Speech Recognition for Conversational Speech Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper investigates prominence-aware automatic speech recognition (ASR)by combining prominence detection and speech recognition for conversationalAustrian German. |
Julian Linke; Barbara Schuppler; | arxiv-cs.CL | 2025-09-12 |
| 734 | Data-independent Beamforming for End-to-end Multichannel Multi-speaker ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we propose a beamforming approach that processesspecific angular sectors based on their spherical polar coordinates beforeapplying an end-to-end multichannel, multi-speaker ASR system. |
Can Cui; Paul Magron; Mostafa Sadeghi; Emmanuel Vincent; | arxiv-cs.SD | 2025-09-12 |
| 735 | TextlessRAG: End-to-End Visual Document RAG By Speech Without Text Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose TextlessRAG, thefirst end-to-end framework for speech-based question answering over large-scaledocument images. |
PEIJIN XIE et. al. | arxiv-cs.CV | 2025-09-09 |
| 736 | Phoneme-Aware Augmentation for Robust Cantonese ASR Under Low-Resource Conditions Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Cantonese automatic speech recognition (ASR) faces persistent challenges due to its nine lexical tones, extensive phonological variation, and the scarcity of professionally … |
Lusheng Zhang; Shie Wu; Zhongxun Wang; | Symmetry | 2025-09-08 |
| 737 | The ML-SUPERB 2.0 Challenge: Towards Inclusive ASR Benchmarking for All Language Varieties Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To advance state-of-the-art (SOTA) ASRmodels, we present the Interspeech 2025 ML-SUPERB 2.0 Challenge. |
WILLIAM CHEN et. al. | arxiv-cs.CL | 2025-09-08 |
| 738 | Phoneme-Level Speech Intelligibility Reduction Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: As noise masking is central to reducing fatigue in open-plan offices, identifying which elements of speech contribute to intelligibility is essential. This paper employs … |
Aine Drelingyte; Romain Serizel; Mathieu Lagrange; | 2025 33rd European Signal Processing Conference (EUSIPCO) | 2025-09-08 |
| 739 | TSPC: A Two-Stage Phoneme-Centric Architecture for Code-switching Vietnamese-English Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we propose a novel architecture forVietnamese-English CS ASR, a Two-Stage Phoneme-Centric model (TSPC). |
Minh N. H. Nguyen; Anh Nguyen Tran; Dung Truong Dinh; Nam Van Vo; | arxiv-cs.SD | 2025-09-07 |
| 740 | Enhancing The Robustness of Contextual ASR to Varying Biasing Information Volumes Through Purified Semantic Correlation Joint Modeling Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To this end, we propose a purified semanticcorrelation joint modeling (PSC-Joint) approach. |
YUE GU et. al. | arxiv-cs.CL | 2025-09-06 |
| 741 | Contextualized Token Discrimination for Speech Search Query Correction Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: With thegrowing popularity of speech search driven by Automated Speech Recognition(ASR) systems, this paper introduces a novel method named Contextualized TokenDiscrimination (CTD) to conduct effective speech query correction. |
JUNYU LU et. al. | arxiv-cs.SD | 2025-09-04 |
| 742 | WenetSpeech-Yue: A Large-scale Cantonese Speech Corpus with Multi-dimensional Annotation IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To address thischallenge, we propose WenetSpeech-Pipe, an integrated pipeline for buildinglarge-scale speech corpus with multi-dimensional annotation tailored for speechunderstanding and generation. |
LONGHAO LI et. al. | arxiv-cs.SD | 2025-09-04 |
| 743 | PARCO: Phoneme-Augmented Robust Contextual ASR Via Contrastive Entity Disambiguation Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Automatic speech recognition (ASR) systems struggle with domain-specificnamed entities, especially homophones. Contextual ASR improves recognition butoften fails to capture … |
Jiajun He; Naoki Sawada; Koichi Miyazaki; Tomoki Toda; | arxiv-cs.CL | 2025-09-04 |
| 744 | LatPhon: Lightweight Multilingual G2P for Romance Languages and English Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Grapheme-to-phoneme (G2P) conversion is a key front-end for text-to-speech(TTS), automatic speech recognition (ASR), speech-to-speech translation (S2ST)and alignment systems, … |
Luis Felipe Chary; Miguel Arjona Ramirez; | arxiv-cs.CL | 2025-09-03 |
| 745 | SSVD: Structured SVD for Parameter-Efficient Fine-Tuning and Benchmarking Under Domain Shift in ASR IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: While low-rank adaptation (LoRA) is widelyused in speech applications, its state-of-the-art variants, e.g., VeRA, DoRA,PiSSA, and SVFT, are developed mainly for language and vision tasks, withlimited validation in speech. This work presents the first comprehensiveintegration and benchmarking of these PEFT methods within ESPnet. |
Pu Wang; Shinji Watanabe; Hugo Van hamme; | arxiv-cs.CL | 2025-09-02 |
| 746 | CuckooAttack: Towards Practical Backdoor Attack Against Automatic Speech Recognition Systems Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Deep learning-based automatic speech recognition (ASR) systems are capable of transcribing input audio of arbitrary duration into character sequences, which are widely used in … |
BOWEN LI et. al. | IEEE Transactions on Dependable and Secure Computing | 2025-09-01 |
| 747 | CabinSep: IR-Augmented Mask-Based MVDR for Real-Time In-Car Speech Separation with Distributed Heterogeneous Arrays Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper proposes CabinSep, a lightweight neuralmask-based minimum variance distortionless response (MVDR) speech separationapproach, to reduce speech recognition errors in back-end automatic speechrecognition (ASR) models. |
RUNDUO HAN et. al. | arxiv-cs.SD | 2025-09-01 |
| 748 | Refining Transcripts With TV Subtitles By Prompt-Based Weakly Supervised Training of ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This study proposes a novel approach to using TV subtitles within a weaklysupervised (WS) Automatic Speech Recognition (ASR) framework. |
Xinnian Zhao; Hugo Van Hamme; | arxiv-cs.CL | 2025-09-01 |
| 749 | A Unified Denoising and Adaptation Framework for Self-Supervised Bengali Dialectal ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This challenge is compoundedby two persistent and intertwined factors: the language’s vast dialectaldiversity and the prevalence of acoustic noise in real-world environments.While state-of-the-art self-supervised learning (SSL) models have advanced ASRfor low-resource languages, they often lack explicit mechanisms to handleenvironmental noise during pre-training or specialized adaptation strategiesfor the complex phonetic and lexical variations across Bengali dialects. Thispaper introduces a novel, unified framework designed to address these dualchallenges simultaneously. |
Swadhin Biswas; Tuhin Sheikh; | arxiv-cs.SD | 2025-08-31 |
| 750 | Child Speech Assessment Through Large Language Model Speech Synthesis: Preliminary Results Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Mispronunciation detection and/or diagnosis (MDD) plays an important role in computer-aided language learning (CALL) systems. It remains a research challenge mainly because of the … |
Xinwei Cao; Zijian Fan; T. Svendsen; G. Salvi; | 2025 IEEE 35th International Workshop on Machine Learning … | 2025-08-31 |
| 751 | Serialized Output Prompting for Large Language Model-based Multi-Talker Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we proposeextracting serialized output prompts (SOP) and explicitly guiding the LLM usingstructured prompts to improve system performance (SOP-MT-ASR). |
HAO SHI et. al. | arxiv-cs.CL | 2025-08-31 |
| 752 | SpeechLLM: Unified Speech and Language Model for Enhanced Multi-Task Understanding in Low Resource Settings Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: While integrating speech encoder with LLM requires substantial data and resources, use cases face limitations due to insufficient availability. To address this, we propose a … |
Jaekwon Yoo; Kunal Chandiramani; Divya Tadimeti; Abenezer Girma; Chandra Dhir; | ArXiv | 2025-08-29 |
| 753 | Towards Inclusive Communication: A Unified Framework for Generating Spoken Language from Sign, Lip, and Audio Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, wepropose the first unified framework capable of handling diverse combinations ofsign language, lip movements, and audio for spoken-language text generation. |
Jeong Hun Yeo; Hyeongseop Rha; Sungjune Park; Junil Won; Yong Man Ro; | arxiv-cs.CV | 2025-08-28 |
| 754 | MoTAS: MoE-Guided Feature Selection from TTS-Augmented Speech for Enhanced Multimodal Alzheimer’s Early Screening Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Early screening for Alzheimer’s Disease (AD) through speech presents apromising non-invasive approach. However, challenges such as limited data andthe lack of fine-grained, … |
Yongqi Shao; Binxin Mei; Cong Tan; Hong Huo; Tao Fang; | arxiv-cs.SD | 2025-08-28 |
| 755 | Cross-Learning Fine-Tuning Strategy for Dysarthric Speech Recognition Via CDSD Database Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Dysarthric speech recognition faces challenges from severity variations anddisparities relative to normal speech. Conventional approaches individuallyfine-tune ASR models … |
Qing Xiao; Yingshan Peng; PeiPei Zhang; | arxiv-cs.SD | 2025-08-26 |
| 756 | Zero-shot Context Biasing with Trie-based Decoding Using Synthetic Multi-Pronunciation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we propose a synthesis-drivenmulti-pronunciation contextual biasing method that performs zero-shotcontextual ASR on a pretrained Whisper model. |
Changsong Liu; Yizhou Peng; Eng Siong Chng; | arxiv-cs.CL | 2025-08-25 |
| 757 | Talking to Robots: A Practical Examination of Speech Foundation Models for HRI Applications Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We evaluate four state-of-the-art ASR systems on eight publiclyavailable datasets that capture six dimensions of difficulty: domain-specific,accented, noisy, age-variant, impaired, and spontaneous speech. |
Theresa Pekarek Rosin; Julia Gachot; Henri-Leon Kordt; Matthias Kerzel; Stefan Wermter; | arxiv-cs.RO | 2025-08-25 |
| 758 | Whisper Based Cross-Lingual Phoneme Recognition Between Vietnamese and English Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Unlike many languages, Vietnamese relies on tonal variations todistinguish word meanings, whereas English features stress patterns andnon-standard pronunciations that hinder phoneme alignment between the twolanguages. To address this challenge, we propose a novel bilingual speechrecognition approach with two primary contributions: (1) constructing arepresentative bilingual phoneme set that bridges the differences betweenVietnamese and English phonetic systems; (2) designing an end-to-end systemthat leverages the PhoWhisper pre-trained encoder for deep high-levelrepresentations to improve phoneme recognition. |
Nguyen Huu Nhat Minh; Tran Nguyen Anh; Truong Dinh Dung; Vo Van Nam; Le Pham Tuyen; | arxiv-cs.CL | 2025-08-22 |
| 759 | Benchmarking Training Paradigms, Dataset Composition, and Model Scaling for Child ASR in ESPnet IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We compare flat-start training across multiple datasets, SSLrepresentations (WavLM, XEUS), and decoder architectures. |
ANYU YING et. al. | arxiv-cs.LG | 2025-08-22 |
| 760 | UniCoM: A Universal Code-Switching Speech Generator Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: However, systemscapable of handling this phenomenon remain underexplored, primarily due to thescarcity of suitable datasets. To resolve this issue, we propose UniversalCode-Mixer (UniCoM), a novel pipeline for generating high-quality, natural CSsamples without altering sentence semantics. |
Sangmin Lee; Woojin Chung; Seyun Um; Hong-Goo Kang; | arxiv-cs.CL | 2025-08-21 |
| 761 | Evaluating ASR Robustness to Spontaneous Speech Errors: A Study of WhisperX Using A Speech Error Database Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This analysis demonstrates the atabase’s effectivenessas a diagnostic tool for ASR system performance. |
John Alderete; Macarious Kin Fung Hui; Aanchan Mohan; | arxiv-cs.CL | 2025-08-18 |
| 762 | Overcoming Latency Bottlenecks in On-Device Speech Translation: A Cascaded Approach with Alignment-Based Streaming MT Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Although state-of-the-art ASR systems based onRecurrent Neural Network Transducers (RNN-T) can perform real-timetranscription, achieving streaming translation in real-time remains asignificant challenge. To address this issue, we propose a simultaneoustranslation approach that effectively balances translation quality and latency.We also investigate efficient integration of ASR and MT, leveraging linguisticcues generated by the ASR system to manage context and utilizing efficientbeam-search pruning techniques such as time-out and forced finalization tomaintain system’s real-time factor. |
ZEESHAN AHMED et. al. | arxiv-cs.CL | 2025-08-18 |
| 763 | Whispering Context: Distilling Syntax and Semantics for Long Speech Transcripts Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose a novel approach that enhances ASRby distilling contextual knowledge from LLaMA models into Whisper. |
Duygu Altinok; | arxiv-cs.CL | 2025-08-18 |
| 764 | R2S: Real-to-Synthetic Representation Learning for Training Speech Recognition Models on Synthetic Data Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: We investigate the use of synthetic speech to enhance the performance of Automatic Speech Recognition (ASR) systems. While pre-trained ASR models have demonstrated impressive … |
M. TRAN et. al. | Interspeech 2025 | 2025-08-17 |
| 765 | Towards Inclusive and Fair ASR: Insights from The SAPC Challenge for Optimizing Disordered Speech Recognition Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: ASR has advanced significantly, yet remains limited for impaired speakers due to data scarcity. In response to this gap, the Speech Accessibility Project (SAP) represents a … |
Nada Gohider; O. Basir; | Interspeech 2025 | 2025-08-17 |
| 766 | Unlearning LLM-Based Speech Recognition Models Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Recent advances in Large Language Models (LLMs) have been extended to the field of Automatic Speech Recognition (ASR), leveraging the exceptional generative capabilities of LLMs … |
Zhe Liu; | Interspeech 2025 | 2025-08-17 |
| 767 | Enhancing Target-speaker Automatic Speech Recognition Using Multiple Speaker Embedding Extractors with Virtual Speaker Embedding Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Target-speaker automatic speech recognition (TS-ASR) utilizes speaker embeddings to identify a target speaker in multi-talker environments. While high-performance speaker … |
Ju-Seok Seong; Jeonghwan Choi; Ye-Rin Jeoung; Ilseok Kim; Joon-Hyuk Chang; | Interspeech 2025 | 2025-08-17 |
| 768 | PPGs-BERT: Leveraging Phoneme Sequence and BERT for Alzheimer’s Disease Detection from Spontaneous Speech Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Alzheimer’s Disease (AD) is a neurodegenerative condition characterized by linguistic impairments. While ASR and LLMs show promise in AD detection, ASR often normalizes key … |
QI SUN et. al. | Interspeech 2025 | 2025-08-17 |
| 769 | Word-Level Error Analysis in Decoding Systems: From Speech Recognition to Brain-Computer Interfaces Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Brain-to-text (BTT) systems that decode attempted speech from neural activity have achieved 4 . 2% word error rate (WER). These systems demonstrate potential for daily use similar … |
Jingya Huang; Aashish N. Patel; Sowmya Manojna Narasimha; Gal Mishne; Vikash Gilja; | Interspeech 2025 | 2025-08-17 |
| 770 | Exploring Shared-Weight Mechanisms in Transformer and Conformer Architectures for Automatic Speech Recognition Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: In recent years, the increasing demand for parameter-efficient automatic speech recognition (ASR) systems has driven researchers to explore innovative architectures and techniques … |
Thomas Rolland; Alberto Abad; | Interspeech 2025 | 2025-08-17 |
| 771 | Assessing The Performance and Efficiency of Mamba ASR in Low-Resource Scenarios Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Mamba, a state space model-based architecture, is emerging as a strong alternative to Transformer models, showing equal or superior performance in sequence generation, including … |
RODOLFO ZEVALLOS et. al. | Interspeech 2025 | 2025-08-17 |
| 772 | Synthetic Dysarthric Speech: A Supplement, Not A Substitute for Authentic Data in Dysarthric Speech Recognition Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Dysarthric speech recognition (DSR) is an emerging field that can enhance social interactions and mental health for individuals with dysarthria. However, the lack of sufficient … |
Jingting Li; Keyi Feng; Xinran Zhao; Yan Wang; Su-Jing Wang; | Interspeech 2025 | 2025-08-17 |
| 773 | Visually-Adaptive Guided Robust Speech Recognition with Parameter-Efficient Adaptation Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Recent developments in large-scale speech foundation models have further pushed the boundaries of automatic speech recognition (ASR) capabilities, making them excellent candidates … |
ZHAO YANG et. al. | Interspeech 2025 | 2025-08-17 |
| 774 | Towards Adaptable and Intelligible Speech Synthesis in Noisy Environments Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: We present an investigation into adaptable speech synthesis for noisy environments. Leveraging a zero-shot TTS we synthesized a corpus of 1,200 speech samples from 100 sentences … |
Lubos Marcinek; J. Beskow; Joakim Gustafson; | Interspeech 2025 | 2025-08-17 |
| 775 | ASR Confidence Estimation Using True Class Lexical Similarity Score Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Deep Neural Networks (DNN) often exhibit overconfi-dence, leading to poor confidence calibration in Automatic Speech Recognition (ASR) models. State-Of-The-Art (SOTA) approaches … |
Nagarathna Ravi; Thishyan Raj T; Ravi Teja Chaganti; Vipul Arora; | Interspeech 2025 | 2025-08-17 |
| 776 | Switch Conformer with Universal Phonetic Experts for Multilingual ASR Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Multilingual end-to-end ASR presents significant challenges due to the need to accommodate diverse writing systems, lexi-cons, and grammatical structures. Existing methods often … |
M. Mimura; Jaeyoung Lee; Tatsuya Kawahara; | Interspeech 2025 | 2025-08-17 |
| 777 | Scaling Pseudo-labeling Data for End-to-end Low-resource Speech Translation (the Case of Kurdish Language) Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: In this paper we propose a pseudo-labeling pipeline to generate End-to-End Speech to Text Translation (E2E S2TT) data for low-resource languages. This pipeline allows us to … |
Mohammad MohammadAmini; Aghilas Sini; Marie Tahon; Antoine Laurent; | Interspeech 2025 | 2025-08-17 |
| 778 | CBA-Whisper: Curriculum Learning-Based AdaLoRA Fine-Tuning on Whisper for Low-Resource Dysarthric Speech Recognition Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Whisper is a powerful automatic speech recognition (ASR) model. However, its zero-shot performance on low-resource speech requires further improvement, especially in dysarthric … |
TIANYI TAN et. al. | Interspeech 2025 | 2025-08-17 |
| 779 | GLCLAP: A Novel Contrastive Learning Pre-trained Model for Contextual Biasing in ASR Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Recently, Automatic Speech Recognition (ASR) that supports prompts has shown remarkable versatility. For contextual biasing with these systems, a pivotal factor lies in obtaining … |
YUXIANG KONG et. al. | Interspeech 2025 | 2025-08-17 |
| 780 | Extended High-frequency Cues to Phoneme Recognition: Insights from ASR Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: There is emerging evidence that extended high frequencies (EHFs; >8 kHz) improve speech perception in noise. Yet, the mechanisms underlying this benefit remain unclear. We … |
Zhe-chen Guo; B. Chandrasekaran; | Interspeech 2025 | 2025-08-17 |
| 781 | End-to-End Speech Translation Guided By Robust Translation Capability of Large Language Model Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: We present an end-to-end speech translation (ST) model that uses a large language model (LLM) to guide the translation process. Recent advances in LLMs have shown strong … |
Yosuke Higuchi; Tetsuji Ogawa; Tetsunori Kobayashi; | Interspeech 2025 | 2025-08-17 |
| 782 | J-j-j-just Stutter: Benchmarking Whisper’s Performance Disparities on Different Stuttering Patterns Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Despite their prevalence in everyday technologies, automated speech recognition (ASR) systems often struggle with disfluent speech. To diagnose and address these technical … |
Charan Sridhar; Shaomei Wu; | Interspeech 2025 | 2025-08-17 |
| 783 | Adapting Whisper for Low-resource Hindi-English Code-Mix Speech with On-the-fly Augmentation & LLM-Synthesised Data Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Code-switching (CS) automatic speech recognition (ASR) faces challenges due to language confusion from accents, acoustic similarities, and seamless transitions. In multilingual … |
A. Biswas; Oleg Shevelev; Amine Abdaoui; V. Tyagi; Abdelmoumene Boumadane; | Interspeech 2025 | 2025-08-17 |
| 784 | Building An Accurate Open-Source Hebrew ASR System Through Crowdsourcing Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Automatic Speech Recognition (ASR) for Hebrew faces significant challenges due to limited resources and rich morphology. While recent advances have improved high-resource … |
Y. Marmor; Y. Lifshitz; Y. Snapir; Kinneret Misgav; | Interspeech 2025 | 2025-08-17 |
| 785 | Hybrid Data Sampling for ASR: Integrating Acoustic Diversity and Transcription Uncertainty Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Efficiently selecting training data is crucial for improving automatic speech recognition (ASR) models while minimizing annotation costs. This research extends TypiClust, a data … |
Komei Hiruta; Yosuke Yamano; Hideaki Tamori; | Interspeech 2025 | 2025-08-17 |
| 786 | Bidirectional Spoken-Written Text Conversion with Large Language Models Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Traditional ASR systems normalize transcriptions into spoken form for training, leading to spoken form outputs. In contrast, modern Transformer-based models directly map … |
MUYEOL CHOI et. al. | Interspeech 2025 | 2025-08-17 |
| 787 | Semi-Supervised Learning for Automatic Speech Recognition with Word Error Rate Estimation and Targeted Domain Data Selection Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: There is a growing demand for leveraging untranscribed multi-domain data in semi-supervised learning (SSL) for automatic speech recognition (ASR) to broaden its applications. … |
Chanho Park; Thomas Hain; | Interspeech 2025 | 2025-08-17 |
| 788 | Is Synthetic Data Truly Effective for Training Speech Language Models? Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: The development of Large Language Models (LLMs) has expanded beyond text-based tasks to speech applications such as Automatic Speech Recognition (ASR) and Automated Speech … |
Tomoya Mizumoto; Atsushi Kojima; Yusuke Fujita; Lianbo Liu; Yui Sudo; | Interspeech 2025 | 2025-08-17 |
| 789 | Improving Generalization of End-to-End ASR Through Diversity and Independence Regularization Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Automatic speech recognition (ASR) has been driven by representative end-to-end model architectures, including connectionist temporal classification (CTC), attention-based … |
Ye-Eun Ko; Mun-Hak Lee; Donghyun Kim; Joon-Hyuk Chang; | Interspeech 2025 | 2025-08-17 |
| 790 | Why Is Children’s ASR So Difficult? Analyzing Children’s Phonological Error Patterns Using SSL-based Phoneme Recognizers Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Children’s automatic speech recognition (child ASR) is generally more challenging than adult ASR, as children’s speech differs from that of adults and continuously evolves with a … |
Koharu Horii; Naohiro Tawara; Atsunori Ogawa; Shoko Araki; | Interspeech 2025 | 2025-08-17 |
| 791 | Pathology-Aware Speech Encoding and Data Augmentation for Dysarthric Speech Recognition Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Automatic speech recognition (ASR) for pathologic speech remains a major challenge due to high variability in articulation, phonation, and prosody distortions. In this work, we … |
Ilja Baumann; Dominik Wagner; K. Riedhammer; Tobias Bocklet; | Interspeech 2025 | 2025-08-17 |
| 792 | Beyond Traditional Speech Modifications : Utilizing Self Supervised Features for Enhanced Zero-Shot Children ASR Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Zero-shot automatic speech recognition (ASR) for children is challenging due to pronounced acoustic and linguistic mis-matches, speaker variability and limited annotated data. … |
Abhijit Sinha; H. Kathania; Mikko Kurimo; | Interspeech 2025 | 2025-08-17 |
| 793 | Simultaneous Masked and Unmasked Decoding with Speculative Decoding Masking for Fast ASR Without Accuracy Loss Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: In this paper, we introduce two methods, Simultaneous Masked and Unmasked Decoding (SMUD) and speculative decoding masking, into Partially autoregressive (PAR) decoding. These … |
K. Okabe; Hitoshi Yamamoto; | Interspeech 2025 | 2025-08-17 |
| 794 | Fine-tuning Strategies for Automatic Speech Recognition of Low-Resource Speech with Autism Spectrum Disorder Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Individuals with autism spectrum disorder (ASD) exhibit unique speech patterns that challenge conventional automatic speech recognition (ASR) systems. However, research on … |
Yeseul Park; Bowon Lee; | Interspeech 2025 | 2025-08-17 |
| 795 | Attention-Free Dual-Mode ASR with Latency-Controlled Selective State Spaces Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: This paper introduces a novel encoder architecture designed to enhance transducer-based dual-mode automatic speech recognition (ASR). Our approach leverages the selective … |
Takafumi Moriya; Masato Mimura; Kiyoaki Matsui; Hiroshi Sato; Kohei Matsuura; | Interspeech 2025 | 2025-08-17 |
| 796 | Exploring The Limits of Conformer CTC-Encoder for Speech Emotion Recognition Using Large Language Models Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Conformer CTC-Encoders have consistently delivered state-of-the-art results in the field of Automatic Speech Recognition (ASR); however, their merits for tasks that demand more … |
E. MORAIS et. al. | Interspeech 2025 | 2025-08-17 |
| 797 | Efficient Noise-Robust Hybrid Audiovisual Encoder with Joint Distillation and Pruning for Audiovisual Speech Recognition Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Powered by self-supervised learning (SSL) on vast amounts of unlabeled data, a computationally intensive audiovisual encoder—a hybrid architecture combining ResNet and … |
Zhengyang Li; Pascal Reichert; Thomas Graave; Patrick Blumenberg; Tim Fingscheidt; | Interspeech 2025 | 2025-08-17 |
| 798 | Fine-tuning Parakeet-TDT for Dysarthric Speech Recognition in The Speech Accessibility Project Challenge Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: We present our dysarthric speech recognition system submitted to the Interspeech 2025 Speech Accessibility Project Challenge. This challenge is a competition aimed at improving … |
Kaito Takahashi; Keigo Hojo; Toshimitsu Sakai; Yukoh Wakabayashi; N. Kitaoka; | Interspeech 2025 | 2025-08-17 |
| 799 | Towards Atypical Speech Transcription Using LLM-based ASR Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Training a linear transformation between speech encoders and LLMs enable LLMs to transcribe speech. SLAM-ASR is one such recently proposed architecture. This paper examines its … |
Jinda Zhang; Aanchan Mohan; | Interspeech 2025 | 2025-08-17 |
| 800 | CEREALES : A New Dataset of Quebec French Accented Speech with Applications to Speech Recognition Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Automatic speech recognition (ASR) systems have become widespread and have broad categories of usage and users. It is therefore crucial for ASR models to be robust to speaker … |
L. Maison; Thomas Soulas; Marie-Jean Meurs; | Interspeech 2025 | 2025-08-17 |
| 801 | Leveraging Geographic Metadata for Dialect-Aware Speech Recognition Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Automatic Speech Recognition (ASR) systems struggle with di-alectal variations due to significant phonetic, morphosyntactic, and lexical differences across regions. Traditional … |
Pouya Mehralian; Hugo Van hamme; | Interspeech 2025 | 2025-08-17 |
| 802 | Character Error Rate Estimation for Semi-Supervised Training of Speech Recognition for Arabic Dialects Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: The use of semi-supervised data for Automatic Speech Recognition (ASR) is nowadays commonplace and is the basis of the most advanced ASR models. For low-resourced languages, where … |
Chanho Park; Oscar Saz-Torralba; | Interspeech 2025 | 2025-08-17 |
| 803 | SepALM: Audio Language Models Are Error Correctors for Robust Speech Separation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: While contemporary speech separation technologies adeptly process lengthy mixed audio waveforms, they are frequently challenged by the intricacies of real-world environments, including noisy and reverberant settings, which can result in artifacts or distortions in the separated speech. To overcome these limitations, we introduce SepALM, a pioneering approach that employs audio language models (ALMs) to rectify and re-synthesize speech within the text domain following preliminary separation. |
Zhaoxi Mu; Xinyu Yang; Gang Wang; | ijcai | 2025-08-16 |
| 804 | Investigating Transcription Normalization in The Faetar ASR Benchmark Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We examine the role of transcription inconsistencies in the Faetar AutomaticSpeech Recognition benchmark, a challenging low-resource ASR benchmark. |
Leo Peckham; Michael Ong; Naomi Nagy; Ewan Dunbar; | arxiv-cs.CL | 2025-08-15 |
| 805 | Analysis of Domain Shift Across ASR Architectures Via TTS-Enabled Separation of Target Domain and Acoustic Conditions Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We analyze automatic speech recognition (ASR) modeling choices under domainmismatch, comparing classic modular and novel sequence-to-sequence (seq2seq)architectures. |
Tina Raissi; Nick Rossenbach; Ralf Schlüter; | arxiv-cs.SD | 2025-08-13 |
| 806 | Assessing The Feasibility of Lightweight Whisper Models for Low-Resource Urdu Transcription Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This study evaluates the feasibility of lightweight Whisper models (Tiny,Base, Small) for Urdu speech recognition in low-resource settings. |
Abdul Rehman Antall; Naveed Akhtar; | arxiv-cs.CL | 2025-08-13 |
| 807 | A Comparative Analysis on ASR System Combination for Attention, CTC, Factored Hybrid, and Transducer Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we compare model combination acrosspopular ASR architectures. |
NOURELDIN BAYOUMI et. al. | arxiv-cs.SD | 2025-08-13 |
| 808 | Revealing The Role of Audio Channels in ASR Performance Degradation Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Pre-trained automatic speech recognition (ASR) models have demonstratedstrong performance on a variety of tasks. However, their performance candegrade substantially when the input … |
Kuan-Tang Huang; Li-Wei Chen; Hung-Shin Lee; Berlin Chen; Hsin-Min Wang; | arxiv-cs.SD | 2025-08-12 |
| 809 | Munsit at NADI 2025 Shared Task 2: Pushing The Boundaries of Multidialectal Arabic ASR with Weakly Supervised Pretraining and Continual Supervised Fine-tuning Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we present a scalabletraining pipeline that combines weakly supervised learning with supervisedfine-tuning to develop a robust Arabic ASR model. |
Mahmoud Salhab; Shameed Sait; Mohammad Abusheikh; Hasan Abusheikh; | arxiv-cs.CL | 2025-08-12 |
| 810 | Bridging ASR and LLMs for Dysarthric Speech Recognition: Benchmarking Self-Supervised and Generative Approaches Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Speech Recognition (ASR) due to phoneme distortions and high variability.While self-supervised ASR models like Wav2Vec, HuBERT, and Whisper have shownpromise, their effectiveness in dysarthric speech remains unclear. This studysystematically benchmarks these models with different decoding strategies,including CTC, seq2seq, and LLM-enhanced decoding (BART,GPT-2, Vicuna). |
Ahmed Aboeitta; Ahmed Sharshar; Youssef Nafea; Shady Shehata; | arxiv-cs.SD | 2025-08-11 |
| 811 | A Small-footprint Acoustic Echo Cancellation Solution for Mobile Full-Duplex Speech Interactions Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To further optimize AEC’sdownstream applications, we introduce a novel post-processing strategyemploying tailored parameters designed specifically for tasks such as VoiceActivity Detection (VAD) and Automatic Speech Recognition (ASR), thus enhancingtheir overall efficacy. |
Yiheng Jiang; Tian Biao; | arxiv-cs.SD | 2025-08-10 |
| 812 | Fairness of Automatic Speech Recognition: Looking Through A Philosophical Lens Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper examines ASR bias through aphilosophical lens, arguing that systematic misrecognition of certain speechvarieties constitutes more than a technical limitation — it represents a formof disrespect that compounds historical injustices against marginalizedlinguistic communities. |
Anna Seo Gyeong Choi; Hoon Choi; | arxiv-cs.CL | 2025-08-09 |
| 813 | Whisfusion: Parallel ASR Decoding Via A Diffusion Transformer Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose Whisfusion, the first framework to fuse a pre-trainedWhisper encoder with a text diffusion decoder. |
TAEYOUN KWON et. al. | arxiv-cs.SD | 2025-08-09 |
| 814 | A Study on Regularization-Based Continual Learning Methods for Indic ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We employ a Conformer-based hybrid RNN-T/CTC model,initially pretrained on Hindi, which is then incrementally trained on eightadditional Indian languages, for a total sequence of nine languages. |
Gokul Adethya T; S. Jaya Nirmala; | arxiv-cs.LG | 2025-08-08 |
| 815 | Improved Dysarthric Speech to Text Conversion Via TTS Personalization Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present a case study on developing a customized speech-to-text system fora Hungarian speaker with severe dysarthria. |
PÉTER MIHAJLIK et. al. | arxiv-cs.SD | 2025-08-08 |
| 816 | SPGISpeech 2.0: Transcribed Multi-speaker Financial Audio for Speaker-tagged Transcription Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce SPGISpeech 2.0, a dataset suitable for speaker-taggedtranscription in the financial domain. |
RAYMOND GROSSMAN et. al. | arxiv-cs.SD | 2025-08-07 |
| 817 | Efficient Scaling for LLM-based ASR IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Throughcomprehensive and controlled experiments, we find that pretraining the speechencoder before integrating it with the LLM leads to significantly betterscaling efficiency than the standard practice of joint post-training ofLLM-ASR. Based on this insight, we propose a new multi-stage LLM-ASR trainingstrategy, EFIN: Encoder First Integration. |
Bingshen Mu; Yiwen Shao; Kun Wei; Dong Yu; Lei Xie; | arxiv-cs.SD | 2025-08-06 |
| 818 | What Do Humans Hear When Interacting? Experiments on Selective Listening for Evaluating ASR of Spoken Dialogue Systems Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Spoken dialogue systems (SDSs) utilize automatic speech recognition (ASR) atthe front end of their pipeline. The role of ASR in SDSs is to recognizeinformation in user speech related to response generation appropriately.Examining selective listening of humans, which refers to the ability to focuson and listen to important parts of a conversation during the speech, willenable us to identify the ASR capabilities required for SDSs and evaluate them.In this study, we experimentally confirmed selective listening when humansgenerate dialogue responses by comparing human transcriptions for generatingdialogue responses and reference transcriptions. |
KIYOTADA MORI et. al. | arxiv-cs.CL | 2025-08-06 |
| 819 | NVSpeech: An Integrated and Scalable Pipeline for Human-Like Speech Modeling with Paralinguistic Vocalizations IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present NVSpeech, an integrated and scalable pipeline that bridgesthe recognition and synthesis of paralinguistic vocalizations, encompassingdataset construction, ASR modeling, and controllable TTS. |
HUAN LIAO et. al. | arxiv-cs.SD | 2025-08-06 |
| 820 | Pitch Accent Detection Improves Pretrained Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We show the performance of Automatic Speech Recognition (ASR) systems thatuse semi-supervised speech representations can be boosted by a complimentarypitch accent detection module, by introducing a joint ASR and pitch accentdetection model. |
David Sasu; Natalie Schluter; | arxiv-cs.CL | 2025-08-06 |
| 821 | Beyond Hard Sharing: Efficient Multi-Task Speech-to-Text Modeling with Supervised Mixture of Experts Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To address the issue, we propose a simple yeteffective Supervised Mixture of Experts (S-MoE). |
HOJUN JIN et. al. | arxiv-cs.CL | 2025-08-05 |
| 822 | RAG-Boost: Retrieval-Augmented Generation Enhanced LLM-based Speech Recognition Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: In this paper, we propose RAG-Boost (ST-ShinozakiLab Task I system), which enhances the baseline LLM-based ASR system of the MLC-SLM Challenge (task I) with a retrieval-augmented … |
Pengcheng Wang; Sheng Li; Takahiro Shinozaki; | ArXiv | 2025-08-05 |
| 823 | Speech-to-LaTeX: New Models and Datasets for Converting Spoken Equations and Sentences Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Based on ASRpost-correction, prior work requires 2 transcriptions, focuses only on isolatedequations, has a limited test set, and provides neither training data normultilingual coverage. To address these issues, we present the first fullyopen-source large-scale dataset, comprising over 66,000 human-annotated audiosamples of mathematical equations and sentences in both English and Russian,drawn from diverse scientific domains. |
DMITRII KORZH et. al. | arxiv-cs.CV | 2025-08-05 |
| 824 | Toward Low-Latency End-to-End Voice Agents for Telecommunications Using Streaming ASR, Quantized LLMs, and Real-Time TTS IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce a low-latency telecom AI voice agent pipeline for real-time,interactive telecommunications use, enabling advanced voice AI for call centerautomation, intelligent IVR (Interactive Voice Response), and AI-drivencustomer support. |
Vignesh Ethiraj; Ashwath David; Sidhanth Menon; Divya Vijay; | arxiv-cs.SD | 2025-08-05 |
| 825 | LLM-Enhanced Spoken Named Entity Recognition Leveraging ASR N-Best Hypotheses Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Identifying Personally Identifiable Information (PII) from spoken documents is crucial for privacy preservation in speech processing. Unlike written text, spoken language exhibits … |
Farha Azmi; Rong Tong; | 2025 International Conference on Asian Language Processing … | 2025-08-03 |
| 826 | Model-free Speculative Decoding for Transformer-based ASR with Token Map Drafting Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: However, this approach is impractical for devices lacking hardwareaccelerators like GPUs. To address this, we propose \emph{Token Map Drafting},a model-free SD technique that eliminates the need for a separate draft model.Instead, we leverage a precomputed n-gram token map derived fromdomain-specific training data, enabling efficient speculative decoding withminimal overhead. |
Tuan Vu Ho; Hiroaki Kokubo; Masaaki Yamamoto; Yohei Kawaguchi; | arxiv-cs.CL | 2025-07-29 |
| 827 | The Interspeech 2025 Speech Accessibility Project Challenge IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Hosted on EvalAI and leveraging the remote evaluation pipeline,the SAP Challenge evaluates submissions based on Word Error Rate and SemanticScore. |
XIUWEN ZHENG et. al. | arxiv-cs.AI | 2025-07-29 |
| 828 | Whisper Smarter, Not Harder: Adversarial Attack on Partial Suppression Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: However, recent studies have demonstrated thepossibility of adversarial attack on these models which could potentiallysuppress or disrupt model output. We investigate and verify the robustness ofthese attacks and explore if it is possible to increase their imperceptibility.We additionally find that by relaxing the optimisation objective from completesuppression to partial suppression, we can further decrease theimperceptibility of the attack. |
Zheng Jie Wong; Bingquan Shen; | arxiv-cs.SD | 2025-07-29 |
| 829 | A Deep Learning Automatic Speech Recognition Model for Shona Language Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This study presented the development of a deep learning-based AutomaticSpeech Recognition system for Shona, a low-resource language characterized byunique tonal and grammatical complexities. |
Leslie Wellington Sirora; Mainford Mutandavari; | arxiv-cs.CL | 2025-07-28 |
| 830 | Self-Improvement for Audio Large Language Model Using Unlabeled Speech IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose a self-improvement methodcalled SI-SDA, leveraging the information embedded in large-model decoding toevaluate the quality of generated pseudo labels and then perform domainadaptation based on reinforcement learning optimization. |
Shaowen Wang; Xinyuan Chen; Yao Xu; | arxiv-cs.SD | 2025-07-27 |
| 831 | MLLM-based Speech Recognition: When and How Is Multimodality Beneficial? Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Building on our prior work, this paper examines the conditions andmodel architectures under which multiple input modalities can improve automaticspeech recognition (ASR) accuracy in noisy environments. |
Yiwen Guan; Viet Anh Trinh; Vivek Voleti; Jacob Whitehill; | arxiv-cs.SD | 2025-07-25 |
| 832 | HITSZ’s End-To-End Speech Translation Systems Combining Sequence-to-Sequence Auto Speech Recognition Model and Indic Large Language Model for IWSLT 2025 in Indic Track Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents HITSZ’s submission for the IWSLT 2025 Indic track,focusing on speech-to-text translation (ST) for English-to-Indic andIndic-to-English language pairs. |
XUCHEN WEI et. al. | arxiv-cs.CL | 2025-07-25 |
| 833 | The Eloquence Team Submission for Task 1 of MLC-SLM Challenge Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we present our studies and experiments carried out for thetask 1 of the Challenge and Workshop on Multilingual Conversational SpeechLanguage Model (MLC-SLM), which focuses on advancing multilingualconversational speech recognition through the development of speech languagemodels architectures. |
Lorenzo Concina; Jordi Luque; Alessio Brutti; Marco Matassoni; Yuchen Zhang; | arxiv-cs.SD | 2025-07-25 |
| 834 | Enhancing Speech Emotion Recognition Leveraging Aligning Timestamps of ASR Transcripts and Speaker Diarization Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we investigate the impact of incorporating timestamp-basedalignment between Automatic Speech Recognition (ASR) transcripts and SpeakerDiarization (SD) outputs on Speech Emotion Recognition (SER) accuracy.Misalignment between these two modalities often reduces the reliability ofmultimodal emotion recognition systems, particularly in conversationalcontexts. |
Hsuan-Yu Wang; Pei-Ying Lee; Berlin Chen; | arxiv-cs.CL | 2025-07-25 |
| 835 | Phoneme-Level Visual Speech Recognition Via Point-Visual Fusion and Language Model Reconstruction Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Existingmethods often aim to predict words or characters directly from visual cues, butthey commonly suffer from high error rates due to viseme ambiguity and requirelarge amounts of pre-training data. We propose a novel phoneme-based two-stageframework that fuses visual and landmark motion features, followed by an LLMmodel for word reconstruction to address these challenges. |
Matthew Kit Khinn Teng; Haibo Zhang; Takeshi Saitoh; | arxiv-cs.CV | 2025-07-24 |
| 836 | Restoring Rhythm: Punctuation Restoration Using Transformer Models for Bangla, A Low-Resource Language Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this study, we explore the applicationof transformer-based models, specifically XLM-RoBERTa-large, to automaticallyrestore punctuation in unpunctuated Bangla text. |
Md Obyedullahil Mamun; Md Adyelullahil Mamun; Arif Ahmad; Md. Imran Hossain Emu; | arxiv-cs.CL | 2025-07-24 |
| 837 | Synthetic Voice Data for Automatic Speech Recognition in African Languages Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present the first systematic assessment of large-scale syntheticvoice corpora for African ASR. |
Brian DeRenzi; Anna Dixon; Mohamed Aymane Farhi; Christian Resch; | arxiv-cs.CL | 2025-07-23 |
| 838 | Application of Whisper in Clinical Practice: The Post-Stroke Speech Assessment During A Naming Task Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this study, we evaluate whether Whisper, a state-of-the-art ASRfoundation model, can be applied to transcribe and analyze speech from patientswith stroke during a commonly used picture-naming task. |
MILENA DAVUDOVA et. al. | arxiv-cs.SD | 2025-07-23 |
| 839 | BoSS: Beyond-Semantic Speech Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present aformalized framework for BoSS, leveraging cognitive relevance theories andmachine learning models to analyze temporal and contextual speech dynamics. |
QING WANG et. al. | arxiv-cs.SD | 2025-07-23 |
| 840 | The TEA-ASLP System for Multilingual Conversational Speech Recognition and Speech Diarization in MLC-SLM 2025 Challenge Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents the TEA-ASLP’s system submitted to the MLC-SLM 2025Challenge, addressing multilingual conversational automatic speech recognition(ASR) in Task I and speech diarization ASR in Task II. |
Hongfei Xue; Kaixun Huang; Zhikai Zhou; Shen Huang; Shidong Shang; | arxiv-cs.SD | 2025-07-23 |
| 841 | That Doesn’t Sound Right: Evaluating Speech Transcription Quality in Field Linguistics Corpora Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we explore methods for identifying speech transcriptions in fieldwork data that may be unsuitable for training ASR models. |
Eric Le Ferrand; Bo Jiang; Joshua Hartshorne; Emily Prud’hommeaux; | acl | 2025-07-21 |
| 842 | DNCASR: End-to-End Training for Speaker-Attributed ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper introduces DNCASR, a novel end-to-end trainable system designed for joint neural speaker clustering and automatic speech recognition (ASR), enabling speaker-attributed transcription of long multi-party meetings. |
Xianrui Zheng; Chao Zhang; Phil Woodland; | acl | 2025-07-21 |
| 843 | MultiMed: Multilingual Medical Speech Recognition Via Attention Encoder Decoder IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts Related Code View Save Highlight: In this work, we introduce MultiMed, the first multilingual medical ASR dataset, along with the first collection of small-to-large end-to-end medical ASR models, spanning five languages: Vietnamese, English, German, French, and Mandarin Chinese. |
KHAI LE-DUC et. al. | acl | 2025-07-21 |
| 844 | Data Quality Issues in Multilingual Speech Datasets: The Need for Sociolinguistic Awareness and Proactive Language Planning IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We provide a case analysis of Taiwanese Southern Min (nan_tw) that highlights the need for proactive language planning (e. g. orthography prescriptions, dialect boundary definition) and enhanced data quality control in the process of Automatic Speech Recognition (ASR) dataset creation. |
MINGFEI LAU et. al. | acl | 2025-07-21 |
| 845 | MISP-Meeting: A Real-World Dataset with Multimodal Cues for Long-form Meeting Transcription and Summarization Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce MISP-Meeting, a new real-world, multimodal dataset that covers subject-oriented long-form content. |
HangChen HangChen; Chao-Han Huck Yang; Jia-Chen Gu; Sabato Marco Siniscalchi; Jun Du; | acl | 2025-07-21 |
| 846 | GigaSpeech 2: An Evolving, Large-Scale and Multi-domain ASR Corpus for Low-Resource Languages with Automated Crawling, Transcription and Refinement IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts Related Code View Save Highlight: This paper presents GigaSpeech 2, a large-scale, multi-domain, multilingual speech recognition corpus. |
YIFAN YANG et. al. | acl | 2025-07-21 |
| 847 | Unifying Streaming and Non-streaming Zipformer-based ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present a unified framework that trains a single end-to-end ASR model for both streaming and non-streaming applications, leveraging future context information. |
BIDISHA SHARMA et. al. | acl | 2025-07-21 |
| 848 | Improving Language and Modality Transfer in Translation By Character-level Modeling Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To this end, we propose a character-based approach to improve adaptability to new languages and modalities. |
Ioannis Tsiamas; David Dale; Marta R. Costa-jussà; | acl | 2025-07-21 |
| 849 | NeKo: Cross-Modality Post-Recognition Error Correction with Tasks-Guided Mixture-of-Experts Language Model IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we present Mixture-of-Experts as a solution, highlighting that MoEs are much more than a scalability tool. |
YEN-TING LIN et. al. | acl | 2025-07-21 |
| 850 | ChildMandarin: A Comprehensive Mandarin Speech Dataset for Young Children Aged 3-5 IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts Related Code View Save Highlight: However, developing robust ASR models for young children’s speech remains challenging due to differences in pronunciation, tone, and pace compared to adult speech. In this paper, we introduce a new Mandarin speech dataset focused on children aged 3 to 5, addressing the scarcity of resources in this area. |
JIAMING ZHOU et. al. | acl | 2025-07-21 |
| 851 | Dialectal Coverage And Generalization in Arabic Speech Recognition IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts Related Code View Save Highlight: In this work, we introduce a suite of ASR models optimized to effectively recognize multiple variants of spoken Arabic, including MSA, various dialects, and code-switching. |
Amirbek Djanibekov; Hawau Olamide Toyin; Raghad Alshalan; Abdullah Alatir; Hanan Aldarmaki; | acl | 2025-07-21 |
| 852 | The Esethu Framework: Reimagining Sustainable Dataset Governance and Curation for Low-Resource Languages Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents the Esethu Framework, a sustainable data curation framework specifically designed to empower local communities and ensure equitable benefit-sharing from their linguistic resource. |
JENALEA RAJAB et. al. | acl | 2025-07-21 |
| 853 | WavRAG: Audio-Integrated Retrieval Augmented Generation for Spoken Dialogue Models IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: However, existing RAG frameworks are primarily designed for text-based LLMs and rely on Automatic Speech Recognition to process speech input, which discards crucial audio information, risks transcription errors, and increases computational overhead. Therefore, we introduce WavRAG, the first retrieval augmented generation framework with native, end-to-end audio support. |
YIFU CHEN et. al. | acl | 2025-07-21 |
| 854 | EchoVoices: Preserving Generational Voices and Memories for Seniors and Children Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Thesedemographics possess distinct vocal characteristics, linguistic styles, andinteraction patterns that challenge conventional ASR, TTS, and LLM systems. Toaddress this, we introduce EchoVoices, an end-to-end digital human pipelinededicated to creating persistent digital personas for seniors and children,ensuring their voices and memories are preserved for future generations. |
HAIYING XU et. al. | arxiv-cs.SD | 2025-07-20 |
| 855 | Weak Supervision Techniques Towards Enhanced ASR Models in Industry-level CRM Systems Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: However, this process faces thechallenge of discerning customer voices and intentions, and general pre-trainedautomatic speech recognition (ASR) models make it difficult to effectivelyaddress industry-specific speech recognition tasks. To address this issue, weinnovatively proposed a solution for fine-tuning industry-specific ASR models,which significantly improved the performance of the fine-tuned ASR models inindustry applications. |
ZHONGSHENG WANG et. al. | arxiv-cs.SD | 2025-07-19 |
| 856 | Open Automatic Speech Recognition Models for Classical and Modern Standard Arabic Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper introduces a universal methodologyfor Arabic speech and text processing designed to address unique challenges ofthe language. |
Lilit Grigoryan; Nikolay Karpov; Enas Albasiri; Vitaly Lavrukhin; Boris Ginsburg; | arxiv-cs.CL | 2025-07-18 |
| 857 | Optimizing ASR for Catalan-Spanish Code-Switching: A Comparative Analysis of Methodologies Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we improve ASRfor Catalan-Spanish CS by exploring three strategies: (1) generating syntheticCS data, (2) concatenating monolingual audio, and (3) leveraging real CS datawith language tokens. |
CARLOS MENA et. al. | arxiv-cs.CL | 2025-07-18 |
| 858 | Reading Between The Lines: Combining Pause Dynamics and Semantic Coherence for Automated Assessment of Thought Disorder Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This study integrates pause features with semanticcoherence metrics across three datasets: naturalistic self-recorded diaries(AVH, n = 140), structured picture descriptions (TOPSY, n = 72), and dreamnarratives (PsyCL, n = 43). |
FENG CHEN et. al. | arxiv-cs.CL | 2025-07-17 |
| 859 | NonverbalTTS: A Public English Corpus of Text-Aligned Nonverbal Vocalizations with Emotion Annotations for Text-to-Speech IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we introduce NonverbalTTS (NVTTS), a 17-hour open-accessdataset annotated with 10 types of NVs (e.g., laughter, coughs) and 8 emotionalcategories. |
Maksim Borisov; Egor Spirin; Daria Diatlova; | arxiv-cs.LG | 2025-07-17 |
| 860 | Improving Contextual ASR Via Multi-grained Fusion with Large Language Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we propose a novel multi-grained fusion approach that jointlyleverages the strengths of both token-level and phrase-level fusion with LargeLanguage Models (LLMs). |
Shilin Zhou; Zhenghua Li; | arxiv-cs.CL | 2025-07-16 |
| 861 | WhisperKit: On-device Real-time ASR with Billion-Scale Transformers Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Real-time Automatic Speech Recognition (ASR) is a fundamental building blockfor many commercial applications of ML, including live captioning, dictation,meeting transcriptions, … |
Atila Orhon; Arda Okan; Berkin Durmus; Zach Nagengast; Eduardo Pacheco; | arxiv-cs.AI | 2025-07-14 |
| 862 | DQLoRA: A Lightweight Domain-Aware Denoising ASR Via Adapter-guided Distillation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present a demo of DQLoRA, an Adapter-Guided Distillation framework forrobust speech recognition under low-resource and noisy conditions. |
Yiru Yang; | arxiv-cs.SD | 2025-07-14 |
| 863 | Supporting SENCOTEN Language Documentation Efforts with Automatic Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To supportthese on-the-ground efforts, the community is turning to digital technology.Automatic Speech Recognition (ASR) technology holds great promise foraccelerating language documentation and the creation of educational resources.However, developing ASR systems for SENCOTEN is challenging due to limited dataand significant vocabulary variation from its polysynthetic structure andstress-driven metathesis. To address these challenges, we propose an ASR-drivendocumentation pipeline that leverages augmented speech data from atext-to-speech (TTS) system and cross-lingual transfer learning with SpeechFoundation Models (SFMs). |
Mengzhe Geng; Patrick Littell; Aidan Pine; Marc Tessier; Roland Kuhn; | arxiv-cs.SD | 2025-07-14 |
| 864 | ClaritySpeech: Dementia Obfuscation in Speech Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Dementia, a neurodegenerative disease, alters speech patterns, creatingcommunication barriers and raising privacy concerns. Current speechtechnologies, such as automatic speech … |
Dominika Woszczyk; Ranya Aloufi; Soteris Demetriou; | arxiv-cs.CL | 2025-07-12 |
| 865 | ILT-Iterative LoRA Training Through Focus-Feedback-Fix for Multilingual Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Based on Whisper-large-v3 andQwen2-Audio, we conduct systematic experiments using a three-stage trainingprocess: Focus Training, Feed Back Training, and Fix Training. |
Qingliang Meng; Hao Wu; Wei Liang; Wei Xu; Qing Zhao; | arxiv-cs.CL | 2025-07-11 |
| 866 | The Impact of Automatic Speech Transcription on Speaker Attribution Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we conduct what is, to our knowledge, the firstcomprehensive study of the impact of automatic transcription on speakerattribution performance. |
Cristina Aggazzotti; Matthew Wiesner; Elizabeth Allyn Smith; Nicholas Andrews; | arxiv-cs.CL | 2025-07-11 |
| 867 | Edge-ASR: Towards Low-Bit Quantization of Automatic Speech Recognition Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Built on an extension of the LLM compressiontoolkit, our framework integrates edge-ASR models, diverse advancedquantization algorithms, a unified calibration and evaluation data pipeline,with detailed analysis tools. |
CHEN FENG et. al. | arxiv-cs.SD | 2025-07-10 |
| 868 | Code-Switching in End-to-End Automatic Speech Recognition: A Systematic Literature Review Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Motivated by a growing research interest into automatic speech recognition(ASR), and the growing body of work for languages in which code-switching (CS)often occurs, we present a systematic literature review of code-switching inend-to-end ASR models. |
Maha Tufail Agro; Atharva Kulkarni; Karima Kadaoui; Zeerak Talat; Hanan Aldarmaki; | arxiv-cs.CL | 2025-07-10 |
| 869 | Omni-Router: Sharing Routing Decisions in Sparse Mixture-of-Experts for Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To increase thecooperation between experts in different layers and encourage greaterspecialization, we use a shared router across different MoE layers. |
Zijin Gu; Tatiana Likhomanenko; Navdeep Jaitly; | arxiv-cs.CL | 2025-07-08 |
| 870 | How to Evaluate Automatic Speech Recognition: Comparing Different Performance and Bias Measures Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this study, we compare different performance and bias measures,from literature and proposed, to evaluate state-of-the-art end-to-end ASRsystems for Dutch. |
Tanvina Patel; Wiebke Hutiri; Aaron Yi Ding; Odette Scharenborg; | arxiv-cs.CL | 2025-07-08 |
| 871 | Benchmarking Akan ASR Models Across Domain-Specific Datasets: A Comparative Evaluation of Performance, Scalability, and Adaptability Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Most existing automatic speech recognition (ASR) research evaluate modelsusing in-domain datasets. |
MARK ATTA MENSAH et. al. | arxiv-cs.CL | 2025-07-03 |
| 872 | A Cookbook for Community-driven Data Collection of Impaired Speech in LowResource Languages Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This study presents an approach for collecting speech samples to buildAutomatic Speech Recognition (ASR) models for impaired speech, particularly,low-resource languages. |
SUMAYA AHMED SALIHS et. al. | arxiv-cs.CL | 2025-07-03 |
| 873 | Adaptability of ASR Models on Low-Resource Language: A Comparative Study of Whisper and Wav2Vec-BERT on Bangla Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Through systematic fine-tuning and hyperparameteroptimization, including learning rate, epochs, and model checkpoint selection,we have compared the models based on Word Error Rate (WER), Character ErrorRate (CER), Training Time, and Computational Efficiency. |
Md Sazzadul Islam Ridoy; Sumi Akter; Md. Aminur Rahman; | arxiv-cs.CL | 2025-07-02 |
| 874 | Dysarthric Speech Classification: A Comparative Analysis of Decision-Support Methods Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: This study compares data-driven and non-data-driven methods for classifying dysarthric speech employing the UA-Speech dataset. Data-driven deep-learning architectures, including … |
Davide Lillini; Carlo Aironi; Lucia Migliorelli; L. Gabrielli; Stefano Squartini; | 2025 International Joint Conference on Neural Networks … | 2025-06-30 |
| 875 | Bypassing Audio ReCAPTCHA with Automatic Speech Recognition Models Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: CAPTCHAs are challenges designed to distinguish humans from automated bots. However, with the growing capabilities of Automatic Speech Recognition (ASR) models, these challenges … |
PAUL AUBRY et. al. | 2025 IEEE European Symposium on Security and Privacy … | 2025-06-30 |
| 876 | Mind The Gap: Entity-Preserved Context-Aware ASR Structured Transcriptions Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose a novel training approach thatextends the semantic context of ASR models by adding overlapping contextwindows during training. |
Duygu Altinok; | arxiv-cs.CL | 2025-06-28 |
| 877 | AURA: Agent for Understanding, Reasoning, and Automated Tool Use in Voice-Driven Tasks Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce AURA (Agent for Understanding, Reasoning, andAutomated Tool Use), the first open-source, speech-native assistant capable ofcompleting complex, goal-driven tasks through dynamic tool invocation andmulti-turn conversation. |
Leander Melroy Maben; Gayathri Ganesh Lakshmy; Srijith Radhakrishnan; Siddhant Arora; Shinji Watanabe; | arxiv-cs.AI | 2025-06-28 |
| 878 | Boosting CTC-Based ASR Using LLM-Based Intermediate Loss Regularization Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Using the Conformer architecture and various LLaMA models, wedemonstrate significant improvements in Word Error Rate (WER) on theLibriSpeech, TEDLIUM2, and WSJ corpora, achieving state-of-the-art performancefor CTC-based ASR with minimal computational overhead. |
Duygu Altinok; | arxiv-cs.CL | 2025-06-28 |
| 879 | Mizo Automatic Speech Recognition: Leveraging Wav2vec 2.0 and XLS-R for Enhanced Accuracy in Low-Resource Language Processing Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: This study introduces a Mizo Automatic Speech Recognition (ASR) approach by fine-tuning Wav2vec 2.0 and XLS-R models. The research presents the newly developed Mizo speech … |
Andrew Bawitlung; S. Dash; Radha Mohan Pattanayak; | ACM Transactions on Asian and Low-Resource Language … | 2025-06-28 |
| 880 | Analyzing and Fine-Tuning Whisper Models for Multilingual Pilot Speech Transcription in The Cockpit Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper investigates and improves the transcription accuracyof cockpit conversations with Whisper models. |
Kartheek Kumar Reddy Nareddy; Sarah Ternus; Julia Niebling; | arxiv-cs.CL | 2025-06-27 |
| 881 | OWLS: Scaling Laws for Multilingual Speech Recognition and Translation Models IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we introduce OWLS, an open-access, reproducible suite of multilingual speech recognition and translation models spanning 0.25B to 18B parameters, with the 18B version being the largest speech model, to the best of our knowledge. |
WILLIAM CHEN et. al. | icml | 2025-06-25 |
| 882 | SING: Spatial Context in Large Language Model for Next-Gen Wearables Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we present a novel system architecture that incorporates spatial speech understanding into LLMs, enabling contextually aware and adaptive applications for wearable technologies.To address the lack of existing dataset for microstructure-assisted speech recordings, we synthetically create a dataset by using the LibriSpeech dataset. |
Ayushi Mishra; Yang Bai; Priyadarshan Narayanasamy; Nakul Garg; Nirupam Roy; | icml | 2025-06-25 |
| 883 | A Unified Speech LLM for Diarization and Speech Recognition in Multilingual Conversations Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Speech Large Language Models (Speech LLMs) have emerged as a crucial paradigm in recent years, extending the capabilities of traditional LLMs to speech tasks such as automatic … |
Phurich Saengthong; Boonnithi Jiaramaneepinit; Sheng Li; Manabu Okumura; Takahiro Shinozaki; | ArXiv | 2025-06-25 |
| 884 | Accurate, Fast, Cheap: Choose Three. Replacing Multi-Head-Attention with Bidirectional Recurrent Attention for Long-Form ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present astrong limited-context attention (LCA) baseline, and show that RA layers arejust as accurate while being more efficient. |
Martin Ratajczak; Jean-Philippe Robichaud; Jennifer Drexler Fox; | arxiv-cs.CL | 2025-06-24 |
| 885 | End-to-End Spoken Grammatical Error Correction Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: SGEC systems typicallyfollow a cascaded pipeline consisting of Automatic Speech Recognition (ASR),disfluency detection, and GEC, making them vulnerable to error propagationacross modules. This work examines an End-to-End (E2E) framework for SGEC andfeedback generation, highlighting challenges and possible solutions whendeveloping these systems. |
Mengjie Qian; Rao Ma; Stefano Bannò; Mark J. F. Gales; Kate M. Knill; | arxiv-cs.CL | 2025-06-23 |
| 886 | Our Collective Voices: The Social and Technical Values of A Grassroots Chinese Stuttered Speech Dataset Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: The lack of authentic stuttered speech data has significantly limited the development of stuttering friendly automatic speech recognition (ASR) models. In previous work, we … |
Jingjin Li; Qisheng Li; Rong Gong; Lezhi Wang; Shaomei Wu; | Proceedings of the 2025 ACM Conference on Fairness, … | 2025-06-23 |
| 887 | Adapting Foundation Speech Recognition Models to Impaired Speech: A Semantic Re-chaining Approach for Personalization of German Speech Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Speech impairments caused by conditions such as cerebral palsy or genetic disorders pose significant challenges for automatic speech recognition (ASR) systems. Despite recent … |
Niclas Pokel; Pehuen Moure; Roman Böhringer; Yingqiang Gao; | ArXiv | 2025-06-23 |
| 888 | Breaking The Transcription Bottleneck: Fine-tuning ASR Models for Extremely Low-Resource Fieldwork Languages IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we benchmark theperformance of two fine-tuned multilingual ASR models, MMS and XLS-R, on fivetypologically diverse low-resource languages with control of training dataduration. |
Siyu Liang; Gina-Anne Levow; | arxiv-cs.CL | 2025-06-20 |
| 889 | LM-SPT: LM-Aligned Semantic Distillation for Speech Tokenization Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Reducing the frame rate is anatural solution, but standard techniques, such as rigid average pooling acrossframes, can distort or dilute the semantic structure required for effective LMalignment. To address this, we propose LM-SPT, a speech tokenization methodthat introduces a novel semantic distillation. |
Daejin Jo; Jeeyoung Yun; Byungseok Roh; Sungwoong Kim; | arxiv-cs.CL | 2025-06-20 |
| 890 | Aligning ASR Evaluation with Human and LLM Judgments: Intelligibility Metrics Using Phonetic, Semantic, and NLI Approaches Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We identify twokey challenges: (1) Existing metrics do not adequately reflect intelligibility,and (2) while LLMs can refine ASR output, their effectiveness in correcting ASRtranscripts of dysarthric speech remains underexplored. |
Bornali Phukon; Xiuwen Zheng; Mark Hasegawa-Johnson; | arxiv-cs.LG | 2025-06-19 |
| 891 | Automatic Speech Recognition Biases in Newcastle English: An Error Analysis Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This study investigates ASR performance on NewcastleEnglish, a well-documented regional dialect known to be challenging for ASR. |
Dana Serditova; Kevin Tang; Jochen Steffens; | arxiv-cs.CL | 2025-06-19 |
| 892 | Exploiting Music Source Separation for Automatic Lyrics Transcription with Whisper Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: For long-form, we propose an algorithm using source separation as a vocal activity detector to derive segment boundaries, which results in a consistent reduction in WER relative to Whisper’s native long-form algorithm. |
Jaza Syed; Ivan Meresman Higgs; Ondřej Cífka; Mark Sandler; | arxiv-cs.SD | 2025-06-18 |
| 893 | Unifying Streaming and Non-streaming Zipformer-based ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present a unified framework that trains a single end-to-end ASR model for both streaming and non-streaming applications, leveraging future context information. |
BIDISHA SHARMA et. al. | arxiv-cs.SD | 2025-06-17 |
| 894 | AsyncSwitch: Asynchronous Text-Speech Adaptation for Code-Switched ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce AsyncSwitch, a novel asynchronous adaptation framework that leverages large-scale, text-rich web data to pre-expose ASR models to diverse code-switched domains before fine-tuning on paired speech-text corpora. |
Tuan Nguyen; Huy-Dat Tran; | arxiv-cs.CL | 2025-06-17 |
| 895 | Bi-directional Context-Enhanced Speech Large Language Models for Multilingual Conversational ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper introduces the integration of language-specific bi-directionalcontext into a speech large language model (SLLM) to improve multilingualcontinuous conversational automatic speech recognition (ASR). |
Yizhou Peng; Hexin Liu; Eng Siong Chng; | arxiv-cs.CL | 2025-06-16 |
| 896 | Adapting Whisper for Parameter-efficient Code-Switching Speech Recognition Via Soft Prompt Tuning Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Large-scale multilingual ASR models like Whisper excel in high-resource settings but face challenges in low-resource scenarios, such as rare languages and code-switching (CS), due … |
Hongli Yang; Yizhou Peng; Hao Huang; Sheng Li; | ArXiv | 2025-06-16 |
| 897 | Language-Aware Prompt Tuning for Parameter-Efficient Seamless Language Expansion in Multilingual ASR Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Recent advancements in multilingual automatic speech recognition (ASR) have been driven by large-scale end-to-end models like Whisper. However, challenges such as language … |
Hongli Yang; Sheng Li; Hao Huang; Ayiduosi Tuohan; Yizhou Peng; | ArXiv | 2025-06-16 |
| 898 | NTU Speechlab LLM-Based Multilingual ASR System for Interspeech MLC-SLM Challenge 2025 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This report details the NTU Speechlab system developed for the Interspeech2025 Multilingual Conversational Speech and Language Model (MLC-SLM) Challenge(Task I), where we achieved 5th place. |
YIZHOU PENG et. al. | arxiv-cs.CL | 2025-06-16 |
| 899 | Seewo’s Submission to MLC-SLM: Lessons Learned from Speech Reasoning Language Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce a multi-stage training pipeline that explicitly enhances reasoning and self-correction in speech language models for ASR. |
Bo Li; Chengben Xu; Wufeng Zhang; | arxiv-cs.CL | 2025-06-16 |
| 900 | SC-SOT: Conditioning The Decoder on Diarized Speaker Information for End-to-End Overlapped Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose Speaker-Conditioned Serialized Output Training (SC-SOT), an enhanced SOT-based training for E2E multi-talker ASR. |
Yuta Hirano; Sakriani Sakti; | arxiv-cs.SD | 2025-06-14 |
| 901 | An Exploration of Mamba for Speech Self-Supervised Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: While Mamba has demonstrated strong performance in language modeling, its potential as a speech self-supervised (SSL) model remains underexplored, with prior studies limited to isolated tasks. To address this, we explore Mamba-based HuBERT models as alternatives to Transformer-based SSL architectures. |
TZU-QUAN LIN et. al. | arxiv-cs.CL | 2025-06-14 |
| 902 | Adapting Whisper for Streaming Speech Recognition Via Two-Pass Decoding Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we fine-tune Whisper for streaming ASR using the WeNet toolkit by adopting a Unified Two-pass (U2) structure. |
HAORAN ZHOU et. al. | arxiv-cs.SD | 2025-06-13 |
| 903 | Enabling Automatic Transcription of Child-centered Audio Recordings from Real-world Environments Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we present an approach to automatically detect those utterances in longform audio that can be reliably transcribed with modern ASR systems, allowing automatic and relatively accurate transcription of a notable proportion of all speech in typical longform data. |
Daniil Kocharov; Okko Räsänen; | arxiv-cs.SD | 2025-06-13 |
| 904 | (SimPhon Speech Test): A Data-Driven Method for In Silico Design and Validation of A Phonetically Balanced Speech Test Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce the Simulated Phoneme Speech Test (SimPhon Speech Test) methodology, a novel, multi-stage computational pipeline for the in silico design and validation of a phonetically balanced minimal-pair speech test. |
Stefan Bleeck; | arxiv-cs.SD | 2025-06-13 |
| 905 | Task-specific, Personalized Automatic Speech Recognition Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Voice User Interfaces (VUI) are particularly useful if the operator has to work hands-free or if her cognitive load is very high. This is the case, e.g., when the operator can be … |
Fahrettin Gökgöz; Hussein Hasso; | Proceedings of the 33rd ACM Conference on User Modeling, … | 2025-06-13 |
| 906 | Improving Named Entity Transcription with Contextual LLM-based Revision Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we introduce a large language model (LLM) revision mechanism to revise incorrect named entities in ASR predictions by leveraging the LLM’s reasoning ability as well as local context (e.g., lecture notes) containing a set of correct named entities. |
Viet Anh Trinh; Xinlu He; Jacob Whitehill; | arxiv-cs.CL | 2025-06-12 |
| 907 | Efficient Multilingual ASR Finetuning Via LoRA Language Experts IF:3 Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Recent advancements in deep learning have significantly enhanced multilingual automatic speech recognition (ASR) due to the development of advanced model architectures and … |
JIAHONG LI et. al. | ArXiv | 2025-06-11 |
| 908 | Addressing Auditing Pitfalls in Automatic Speech Recognition Technologies: A Case Study of People with Aphasia Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Automatic Speech Recognition (ASR) systems’ growing use warrants robust auditing approaches to ensure equitable transcription quality, especially for people with speech disorders … |
K. Mei; Anna Seo Gyeong Choi; Hilke Schellmann; Mona Sloane; Allison Koenecke; | The 2026 ACM Conference on Fairness, Accountability, and … | 2025-06-10 |
| 909 | A Self-Refining Framework for Enhancing ASR Using TTS-Synthesized Data Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose a self-refining framework that enhances ASR performance with only unlabeled datasets. |
CHENG-KANG CHOU et. al. | arxiv-cs.CL | 2025-06-10 |
| 910 | Addressing Pitfalls in Auditing Practices of Automatic Speech Recognition Technologies: A Case Study of People with Aphasia Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we identify three pitfalls inexisting standard ASR auditing procedures, and demonstrate how addressing themimpacts audit results via a case study of six popular ASR systems’ performancefor aphasia speakers. |
Katelyn Xiaoying Mei; Anna Seo Gyeong Choi; Hilke Schellmann; Mona Sloane; Allison Koenecke; | arxiv-cs.CY | 2025-06-10 |
| 911 | Transcript-Prompted Whisper with Dictionary-Enhanced Decoding for Japanese Speech Annotation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we propose a method for annotating phonemic and prosodic labels on a given audio-transcript pair, aimed at constructing Japanese text-to-speech (TTS) datasets. |
Rui Hu; Xiaolong Lin; Jiawang Liu; Shixi Huang; Zhenpeng Zhan; | arxiv-cs.CL | 2025-06-09 |
| 912 | DeRAGEC: Denoising Named Entity Candidates with Synthetic Rationale for ASR Error Correction Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We present DeRAGEC, a method for improving Named Entity (NE) correction in Automatic Speech Recognition (ASR) systems. |
SOLEE IM et. al. | arxiv-cs.CL | 2025-06-09 |
| 913 | Technical Report: A Practical Guide to Kaldi ASR Optimization Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This technical report introduces innovative optimizations for Kaldi-based Automatic Speech Recognition (ASR) systems, focusing on acoustic model enhancement, hyperparameter tuning, and language model efficiency. |
Mengze Hong; Di Jiang; | arxiv-cs.SD | 2025-06-08 |
| 914 | Speech Recognition on TV Series with Video-guided Post-ASR Correction Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Existing approaches failto explicitly leverage the rich temporal and contextual information availablein the video. To address this limitation, we propose a Video-Guided Post-ASRCorrection (VPC) framework that uses a Video-Large Multimodal Model (VLMM) tocapture video context and refine ASR outputs. |
Haoyuan Yang; Yue Zhang; Liqiang Jing; John H. L. Hansen; | arxiv-cs.SD | 2025-06-08 |
| 915 | LiveCC: Learning Video LLM with Streaming Speech Transcription at Scale Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we explore large-scale training for Video LLM with cheap automatic speech recognition (ASR) transcripts. |
JOYA CHEN et. al. | cvpr | 2025-06-07 |
| 916 | Automatic Speech Recognition of African American English: Lexical and Contextual Effects Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This study focuses on two key AAE variables: Consonant Cluster Reduction(CCR) and ING-reduction. |
Hamid Mojarad; Kevin Tang; | arxiv-cs.CL | 2025-06-07 |
| 917 | Phonetically-Augmented Discriminative Rescoring for Voice Search Error Correction Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we propose a phonetic correction system that consists of (a) a phonetic search based on the ASR model’s output that generates phonetic alternatives that may not be considered by the E2E system, and (b) a rescorer component that combines the ASR model recognition and the phonetic alternatives, and select a final system output. |
CHRISTOPHE VAN GYSEL et. al. | arxiv-cs.CL | 2025-06-06 |
| 918 | ViCocktail: Automated Multi-Modal Data Collection for Vietnamese Audio-Visual Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This work presents a practical approach to generate AVSR datasets from raw video, refining existing techniques for improved efficiency and accessibility. |
Thai-Binh Nguyen; Thi Van Nguyen; Quoc Truong Do; Chi Mai Luong; | arxiv-cs.CL | 2025-06-05 |
| 919 | LLM-based Phoneme-to-grapheme for Phoneme-based Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: However, Weighted Finite State Transducer (WFST) based decoding is limited by its complex pipeline and inability to leverage large language models (LLMs). Therefore, we propose LLM-based phoneme-to-grapheme (LLM-P2G) decoding for phoneme-based ASR, consisting of speech-to-phoneme (S2P) and phoneme-to-grapheme (P2G). |
Te Ma; Min Bi; Saierdaer Yusuyin; Hao Huang; Zhijian Ou; | arxiv-cs.SD | 2025-06-05 |
| 920 | LESS: Large Language Model Enhanced Semi-Supervised Learning for Speech Foundational Models Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Although state-of-the-art Speech Foundational Models can produce high-quality text pseudo-labels, applying Semi-Supervised Learning (SSL) for in-the-wild real-world data remains … |
Wen Ding; Fan Qian; | ArXiv | 2025-06-05 |
| 921 | Customizing Speech Recognition Model with Large Language Model Feedback Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we propose areinforcement learning based approach for unsupervised domain adaptation,leveraging unlabeled data to enhance transcription quality, particularly thenamed entities affected by domain mismatch, through feedback from a LLM. |
Shaoshi Ling; Guoli Ye; | arxiv-cs.CL | 2025-06-05 |
| 922 | Efficient Data Selection for Domain Adaptation of ASR Using Pseudo-Labels and Multi-Stage Filtering Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Here, we explore different data selection pipelines and propose a robust approach that improves ASR adaptation by filtering pseudo-labels generated using Whisper (encoder-decoder) and Zipformer (transducer) models. |
PRADEEP RANGAPPA et. al. | arxiv-cs.CL | 2025-06-04 |
| 923 | Effects of Speaker Count, Duration, and Accent Diversity on Zero-Shot Accent Robustness in Low-Resource ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We systematically study how three different variables in training data — the number of speakers, the audio duration per each individual speaker, and the diversity of accents — affect ASR robustness towards unseen accents in a low-resource training regime. |
Zheng-Xin Yong; Vineel Pratap; Michael Auli; Jean Maillard; | arxiv-cs.CL | 2025-06-04 |
| 924 | LESS: Large Language Model Enhanced Semi-Supervised Learning for Speech Foundational Models Using In-the-wild Data Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To address the challenges, we introduceLESS (Large Language Model Enhanced Semi-supervised Learning), a versatileframework that uses Large Language Models (LLMs) to correct pseudo-labelsgenerated on in-the-wild data. |
Wen Ding; Fan Qian; | arxiv-cs.CL | 2025-06-04 |
| 925 | Overcoming Data Scarcity in Multi-Dialectal Arabic ASR Via Whisper Fine-Tuning Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We evaluate MSA training size effects, benefits of pre-training on MSA data, and dialect-specific versus dialect-pooled models. |
Ömer Tarik Özyilmaz; Matt Coler; Matias Valdenegro-Toro; | arxiv-cs.CL | 2025-06-03 |
| 926 | A Multi-Dialectal Dataset for German Dialect ASR and Dialect-to-Standard Speech Translation IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To enable studies of how robust models are towards dialectal variation, we present Betthupferl, an evaluation dataset containing four hours of read speech in three dialect groups spoken in Southeast Germany (Franconian, Bavarian, Alemannic), and half an hour of Standard German speech. |
Verena Blaschke; Miriam Winkler; Constantin Förster; Gabriele Wenger-Glemser; Barbara Plank; | arxiv-cs.CL | 2025-06-03 |
| 927 | Echoes of Phonetics: Unveiling Relevant Acoustic Cues for ASR Via Feature Attribution Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we apply a feature attribution technique to identify the relevant acoustic cues for a modern Conformer-based ASR system. |
Dennis Fucci; Marco Gaido; Matteo Negri; Mauro Cettolo; Luisa Bentivogli; | arxiv-cs.CL | 2025-06-02 |
| 928 | Whale: Large-Scale Multilingual ASR Model with W2v-BERT and E-Branchformer with Large Speech Data Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper reports on the development of a large-scale speech recognition model, Whale. |
Yosuke Kashiwagi; Hayato Futami; Emiru Tsunoo; Satoshi Asakawa; | arxiv-cs.CL | 2025-06-02 |
| 929 | HENT-SRT: Hierarchical Efficient Neural Transducer with Self-Distillation for Joint Speech Recognition and Translation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Existing NT-based ST approaches also suffer from high computational training costs. To address these issues, we propose HENT-SRT (Hierarchical Efficient Neural Transducer for Speech Recognition and Translation), a novel framework that factorizes ASR and translation tasks to better handle reordering. |
AMIR HUSSEIN et. al. | arxiv-cs.CL | 2025-06-02 |
| 930 | WCTC-Biasing: Retraining-free Contextual Biasing ASR with Wildcard CTC-based Keyword Spotting and Inter-layer Biasing Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Despite recent advances in end-to-end speech recognition methods, the output tends to be biased to the training data’s vocabulary, resulting in inaccurate recognition of proper nouns and other unknown terms. To address this issue, we propose a method to improve recognition accuracy of such rare words in CTC-based models without additional training or text-to-speech systems. |
Yu Nakagome; Michael Hentschel; | arxiv-cs.CL | 2025-06-01 |
| 931 | Fine-Tuning ASR for Stuttered Speech: Personalized Vs. Generalized Approaches Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, weinvestigate fine-tuning ASRs for stuttered speech, comparing generalized models(trained across multiple speakers) to personalized models tailored toindividual speech characteristics. |
Dena Mujtaba; Nihar Mahapatra; | arxiv-cs.SD | 2025-06-01 |
| 932 | PMF-CEC: Phoneme-Augmented Multimodal Fusion for Context-Aware ASR Error Correction With Error-Specific Selective Decoding Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: End-to-end automatic speech recognition (ASR) models often struggle to accurately recognize rare words. Previously, we introduced an ASR postprocessing method called error … |
Jiajun He; T. Toda; | IEEE Transactions on Audio, Speech and Language Processing | 2025-05-31 |
| 933 | Causal Structure Discovery for Error Diagnostics of Children’s ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Existing analysis methods examine the impact of these factors in isolation, neglecting interdependencies-such as age affecting ASR accuracy both directly and indirectly via pronunciation skills. In this paper, we introduce a causal structure discovery to unravel these interdependent relationships among physiology, cognition, extrinsic factors, and ASR errors. |
Vishwanath Pratap Singh; Md. Sahidullah; Tomi Kinnunen; | arxiv-cs.CL | 2025-05-31 |
| 934 | Chain-of-Thought Training for Open E2E Spoken Dialogue Systems Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose a simple yet effective strategy leveraging a chain-of-thought (CoT) formulation, ensuring that training on conversational data remains closely aligned with the multimodal language model (LM)’s pre-training on speech recognition~(ASR), text-to-speech synthesis (TTS), and text LM tasks. |
SIDDHANT ARORA et. al. | arxiv-cs.CL | 2025-05-31 |
| 935 | LID Models Are Actually Accent Classifiers: Implications and Solutions for LID on Accented Speech Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Our analysis reveals a simple method to enhance model robustness to accents through input chunking. (iii) We present an approach that integrates sequence-level information into our model without relying on monolingual ASR systems; this reduces accent-language confusion and significantly enhances performance on accented speech while maintaining comparable results on standard LID. |
Niyati Bafna; Matthew Wiesner; | arxiv-cs.CL | 2025-05-31 |
| 936 | ViToSA: Audio-Based Toxic Spans Detection on Vietnamese Speech Utterances Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose a pipeline that combines ASR and toxic spans detection for fine-grained identification of toxic content. |
Huy Ba Do; Vy Le-Phuong Huynh; Luan Thanh Nguyen; | arxiv-cs.CL | 2025-05-31 |
| 937 | Improving Multilingual Speech Models on ML-SUPERB 2.0: Fine-tuning with Data Augmentation and LID-Aware CTC Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper enhances multilingual LID and ASR on ML-SUPERB 2.0 by exploring multiple strategies for adapting SFMs, including frozen upstream training, partial fine-tuning, and low-rank adaptation. |
Qingzheng Wang; Jiancheng Sun; Yifan Peng; Shinji Watanabe; | arxiv-cs.SD | 2025-05-30 |
| 938 | SwitchLingua: The First Large-Scale Multilingual and Multi-Ethnic Code-Switching Dataset IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Although some code-switching datasets exist, most are limited to bilingual mixing within homogeneous ethnic groups, leaving a critical need for a large-scale, diverse benchmark akin to ImageNet in computer vision. To bridge this gap, we introduce \textbf{LinguaMaster}, a multi-agent collaboration framework specifically designed for efficient and scalable multilingual data synthesis. |
PENG XIE et. al. | arxiv-cs.CL | 2025-05-30 |
| 939 | Dynamic Context-Aware Streaming Pretrained Language Model For Inverse Text Normalization Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we introduce a streaming pretrained language model for ITN, leveraging pretrained linguistic representations for improved robustness. |
LUONG HO et. al. | arxiv-cs.CL | 2025-05-30 |
| 940 | Pseudo Labels-based Neural Speech Enhancement for The AVSR Task in The MISP-Meeting Challenge Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents our system for the MISP-Meeting Challenge Track 2. |
Longjie Luo; Shenghui Lu; Lin Li; Qingyang Hong; | arxiv-cs.SD | 2025-05-30 |
| 941 | MSDA: Combining Pseudo-labeling and Self-Supervision for Unsupervised Domain Adaptation in ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we investigate the Meta PL unsupervised domain adaptation framework for Automatic Speech Recognition (ASR). |
Dimitrios Damianos; Georgios Paraskevopoulos; Alexandros Potamianos; | arxiv-cs.CL | 2025-05-30 |
| 942 | Fewer Hallucinations, More Verification: A Three-Stage LLM-Based Framework for ASR Error Correction IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: However, directly using LLMs will encounter hallucinations problem, which may lead to the modification of the correct text. To address this problem, we propose the Reliable LLM Correction Framework (RLLM-CF), which consists of three stages: (1) error pre-detection, (2) chain-of-thought sub-tasks iterative correction, and (3) reasoning process verification. |
YANGUI FANG et. al. | arxiv-cs.CL | 2025-05-30 |
| 943 | Vedavani: A Benchmark Corpus for ASR on Vedic Sanskrit Poetry Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this study, we introduce Vedavani, the first comprehensive ASR study focused on Sanskrit Vedic poetry. |
SUJEET KUMAR et. al. | arxiv-cs.CL | 2025-05-30 |
| 944 | Contextualized Automatic Speech Recognition with Dynamic Vocabulary Prediction and Activation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we propose an encoder-based phrase-level contextualized ASR method that leverages dynamic vocabulary prediction and activation. |
Zhennan Lin; Kaixun Huang; Wei Ren; Linju Yang; Lei Xie; | arxiv-cs.SD | 2025-05-29 |
| 945 | Prompting Whisper for Improved Verbatim Transcription and End-to-end Miscue Detection Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To improve on current methods for reading error annotation, we propose a novel end-to-end architecture that incorporates the target reading text via prompting and is trained for both improved verbatim transcription and direct miscue detection. |
Griffin Dietz Smith; Dianna Yee; Jennifer King Chen; Leah Findlater; | arxiv-cs.LG | 2025-05-29 |
| 946 | Towards LLM-Empowered Fine-Grained Speech Descriptors for Explainable Emotion Recognition Related Papers Related Patents Related Grants Related Venues Related Experts Related Code View Save Highlight: This paper presents a novel end-to-end LLM-empowered explainable speech emotion recognition (SER) approach. |
YOUJUN CHEN et. al. | arxiv-cs.SD | 2025-05-29 |
| 947 | BeaverTalk: Oregon State University’s IWSLT 2025 Simultaneous Speech Translation System Related Papers Related Patents Related Grants Related Venues Related Experts Related Code View Save Highlight: This paper discusses the construction, fine-tuning, and deployment of BeaverTalk, a cascaded system for speech-to-text translation as part of the IWSLT 2025 simultaneous translation task. |
Matthew Raffel; Victor Agostinelli; Lizhong Chen; | arxiv-cs.CL | 2025-05-29 |
| 948 | Advancing Hearing Assessment: An ASR-Based Frequency-Specific Speech Test for Diagnosing Presbycusis Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents the development and simulated evaluation of a novel Automatic Speech Recognition (ASR)-based frequency-specific speech test designed to provide granular diagnostic insights. |
Stefan Bleeck; | arxiv-cs.SD | 2025-05-28 |
| 949 | Weakly Supervised Data Refinement and Flexible Sequence Compression for Efficient Thai LLM-based ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper proposes EThai-ASR, the first to apply large language models (LLMs) to Thai ASR and create an efficient LLM-based ASR system. |
MINGCHEN SHAO et. al. | arxiv-cs.SD | 2025-05-28 |
| 950 | AISHELL-5: The First Open-Source In-Car Multi-Channel Multi-Speaker Speech Dataset for Automatic Speech Diarization and Recognition Related Papers Related Patents Related Grants Related Venues Related Experts Related Code View Save Highlight: We firmly believe the AISHELL-5 dataset will significantly advance the research on ASR systems under complex driving scenarios by establishing the first publicly available in-car ASR benchmark. |
YUHANG DAI et. al. | arxiv-cs.SD | 2025-05-28 |
| 951 | Leveraging LLM for Stuttering Speech: A Unified Architecture Bridging Recognition and Event Detection Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper proposed an LLM-driven ASR-SED multi-task learning framework that jointly optimized the ASR and Stuttering Event Detection (SED) tasks. |
Shangkun Huang; Jing Deng; Jintao Kang; Rong Zheng; | arxiv-cs.SD | 2025-05-28 |
| 952 | Overlap-Adaptive Hybrid Speaker Diarization and ASR-Aware Observation Addition for MISP 2025 Challenge Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: For the automatic speech recognition (ASR) system, we proposed an ASR-aware observation addition method that compensates for the performance limitations of Guided Source Separation (GSS) under low signal-to-noise ratio conditions. |
SHANGKUN HUANG et. al. | arxiv-cs.SD | 2025-05-28 |
| 953 | GMU Systems for The IWSLT 2025 Low-Resource Speech Translation Shared Task Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper describes the GMU systems for the IWSLT 2025 low-resource speech translation shared task. |
Chutong Meng; Antonios Anastasopoulos; | arxiv-cs.CL | 2025-05-27 |
| 954 | Loquacious Set: 25,000 Hours of Transcribed and Diverse English Speech Recognition Data for Research and Commercial Use IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This work presents the Loquacious Set, a 25,000-hour curated collection of commercially usable English speech. |
Titouan Parcollet; Yuan Tseng; Shucong Zhang; Rogier van Dalen; | arxiv-cs.CL | 2025-05-27 |
| 955 | Leveraging LLM and Self-Supervised Training Models for Speech Recognition in Chinese Dialects: A Comparative Analysis Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Recent advancements in self-supervised learning have shown that self-supervised pre-training, combined with large language models (LLM), can effectively enhance ASR performance in low-resource scenarios. We aim to investigate the effectiveness of this paradigm for Chinese dialects. |
TIANYI XU et. al. | arxiv-cs.CL | 2025-05-27 |
| 956 | Unfolding A Few Structures for The Many: Memory-Efficient Compression of Conformer and Speech Foundation Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents a novel memory-efficient model compression approach for Conformer ASR and speech foundation systems. |
ZHAOQING LI et. al. | arxiv-cs.SD | 2025-05-27 |
| 957 | Towards One-bit ASR: Extremely Low-bit Conformer Quantization Using Co-training and Stochastic Precision Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we study model weight quantization, which directly reduces the memory footprint to accommodate computationally resource-constrained applications. |
ZHAOQING LI et. al. | arxiv-cs.SD | 2025-05-27 |
| 958 | Mixture of LoRA Experts for Low-Resourced Multi-Accent Automatic Speech Recognition IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We aim to improve the robustness of Automatic Speech Recognition (ASR) systems against non-native speech, particularly in low-resourced multi-accent settings. |
Raphaël Bagat; Irina Illina; Emmanuel Vincent; | arxiv-cs.CL | 2025-05-26 |
| 959 | Exploring Generative Error Correction for Dysarthric Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts Related Code View Save Highlight: In this work, we proposed a two-stage framework for the Speech Accessibility Project Challenge at INTERSPEECH 2025, which combines cutting-edge speech recognition models with LLM-based generative error correction (GER). |
Moreno La Quatra; Alkis Koudounas; Valerio Mario Salerno; Sabato Marco Siniscalchi; | arxiv-cs.CL | 2025-05-26 |
| 960 | In-context Language Learning for Endangered Languages in Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Prior research indicates LLMs can learn new languages for certain tasks without supervised data. We extend this investigation to speech recognition, investigating whether LLMs can learn unseen, low-resource languages through in-context learning (ICL). |
Zhaolin Li; Jan Niehues; | arxiv-cs.CL | 2025-05-26 |
| 961 | KIT’s Low-resource Speech Translation Systems for IWSLT2025: System Enhancement with Synthetic Data and Model Regularization Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper presents KIT’s submissions to the IWSLT 2025 low-resource track. |
ZHAOLIN LI et. al. | arxiv-cs.CL | 2025-05-26 |
| 962 | Automated Evaluation of Children’s Speech Fluency for Low-resource Languages Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper proposes a system to automatically assess fluency by combining a fine-tuned multilingual ASR model, an objective metrics extraction stage, and a generative pre-trained transformer (GPT) network. |
BOWEN ZHANG et. al. | arxiv-cs.SD | 2025-05-26 |
| 963 | Towards Pretraining Robust ASR Foundation Model with Acoustic-Aware Data Augmentation Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we examine how linguistic and acoustic diversity in training data affect the robustness of the ASR model and reveal that transcription generalization is primarily driven by acoustic variation rather than linguistic richness. |
Dancheng Liu; Amir Nassereldine; Chenhui Xu; Jinjun Xiong; | arxiv-cs.CL | 2025-05-26 |
| 964 | BR-ASR: Efficient and Scalable Bias Retrieval Framework for Contextual Biasing ASR in Speech LLM IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: While speech large language models (SpeechLLMs) have advanced standard automatic speech recognition (ASR), contextual biasing for named entities and rare words remains challenging, especially at scale. To address this, we propose BR-ASR: a Bias Retrieval framework for large-scale contextual biasing (up to 200k entries) via two innovations: (1) speech-and-bias contrastive learning to retrieve semantically relevant candidates; (2) dynamic curriculum learning that mitigates homophone confusion which negatively impacts the final performance. |
Xun Gong; Anqi Lv; Zhiming Wang; Huijia Zhu; Yanmin Qian; | arxiv-cs.SD | 2025-05-25 |
| 965 | An Effective Training Framework for Light-Weight Automatic Speech Recognition Models Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Existing approaches (pruning, distillation, layer skip etc.) transform the large models into smaller ones at the cost of significant performance degradation or require prolonged training of smaller models for better performance. To address these issues, we introduce an efficacious two-step representation learning based approach capable of producing several small sized models from a single large model ensuring considerably better performance in limited number of epochs. |
Abdul Hannan; Alessio Brutti; Shah Nawaz; Mubashir Noman; | arxiv-cs.CV | 2025-05-22 |
| 966 | Large Language Models Based ASR Error Correction for Child Conversations Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this study, we explore the use of LLMs in correcting ASR errors for conversational child speech. |
ANFENG XU et. al. | arxiv-cs.CL | 2025-05-22 |
| 967 | Differentiable K-means for Fully-optimized Discrete Token-based ASR Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper proposes the use of differentiable k-means, enabling the joint optimization of tokenization and downstream tasks. |
Kentaro Onda; Yosuke Kashiwagi; Emiru Tsunoo; Hayato Futami; Shinji Watanabe; | arxiv-cs.SD | 2025-05-22 |
| 968 | Prosodically Enhanced Foreign Accent Simulation By Discrete Token-based Resynthesis Only with Native Speech Corpora Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we integrate duration modification to the previous method to simulate foreign accents more accurately. |
Kentaro Onda; Keisuke Imoto; Satoru Fukayama; Daisuke Saito; Nobuaki Minematsu; | arxiv-cs.SD | 2025-05-21 |
| 969 | Discrete Tokens Exhibit Interlanguage Speech Intelligibility Benefit: An Analytical Study Towards Accent-robust ASR Only with Native Speech Data Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this study, we gained insight that contributes to achieving accent-robust ASR using only native speech data. |
Kentaro Onda; Keisuke Imoto; Satoru Fukayama; Daisuke Saito; Nobuaki Minematsu; | arxiv-cs.SD | 2025-05-21 |
| 970 | Selective Invocation for Multilingual ASR: A Cost-effective Approach Adapting to Speech Recognition Difficulty Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To overcome these, we propose SIMA, a selective invocation for multilingual ASR that adapts to the difficulty level of the input speech. |
Hongfei Xue; Yufeng Tang; Jun Zhang; Xuelong Geng; Lei Xie; | arxiv-cs.SD | 2025-05-21 |
| 971 | Vox-Profile: A Speech Foundation Model Benchmark for Characterizing Diverse Speaker and Speech Traits IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts Related Code View Save Highlight: We introduce Vox-Profile, a comprehensive benchmark to characterize rich speaker and speech traits using speech foundation models. |
TIANTIAN FENG et. al. | arxiv-cs.SD | 2025-05-20 |
| 972 | PersonaTAB: Predicting Personality Traits Using Textual, Acoustic, and Behavioral Cues in Fully-Duplex Speech Dialogs Related Papers Related Patents Related Grants Related Venues Related Experts Related Code View Save Highlight: We propose a pipeline that preprocesses raw audio recordings to create a dialogue dataset annotated with timestamps, response types, and emotion/sentiment labels. |
Sho Inoue; Shai Wang; Haizhou Li; | arxiv-cs.SD | 2025-05-20 |
| 973 | Towards Inclusive ASR: Investigating Voice Conversion for Dysarthric Speech Recognition in Low-Resource Languages Summary Related Papers Related Patents Related Grants Related Venues Related Experts Related Code View Save Abstract: Automatic speech recognition (ASR) for dysarthric speech remains challengingdue to data scarcity, particularly in non-English languages. To address this,we fine-tune a voice … |
CHIN-JOU LI et. al. | arxiv-cs.CL | 2025-05-20 |
| 974 | ClapFM-EVC: High-Fidelity and Flexible Emotional Voice Conversion with Dual Control from Natural Language and Speech Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper introduces ClapFM-EVC, a novel EVC framework capable of generating high-quality converted speech driven by natural language prompts or reference speech with adjustable emotion intensity. |
YU PAN et. al. | arxiv-cs.SD | 2025-05-19 |
| 975 | Improving Endpoint Detection in End-to-end Streaming ASR for Conversational Speech Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We propose methods to improve EP by addressing delayed emission along with EP mistakes. |
ANANDH C et. al. | arxiv-cs.CL | 2025-05-19 |
| 976 | Personalized Fine-Tuning with Controllable Synthetic Speech from LLM-Generated Transcripts for Dysarthric Speech Recognition IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we present our submission to the Speech Accessibility Project challenge for dysarthric speech recognition. |
DOMINIK WAGNER et. al. | arxiv-cs.SD | 2025-05-19 |
| 977 | ASR-FAIRBENCH: Measuring and Benchmarking Equity Across Speech Recognition Systems Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we introduce the ASR-FAIRBENCH leaderboard which is designed to assess both the accuracy and equity of ASR models in real-time. |
Anand Rai; Satyam Rahangdale; Utkarsh Anand; Animesh Mukherjee; | arxiv-cs.SD | 2025-05-16 |
| 978 | Survey of End-to-End Multi-Speaker Automatic Speech Recognition for Monaural Audio IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This survey provides a systematic taxonomy of E2E neural approaches for multi-speaker ASR, highlighting recent advances and comparative analysis. |
Xinlu He; Jacob Whitehill; | arxiv-cs.CL | 2025-05-16 |
| 979 | Multi-Stage Speaker Diarization for Noisy Classrooms Related Papers Related Patents Related Grants Related Venues Related Experts Related Code View Save Highlight: This study investigates the effectiveness of multi-stage diarization models using Nvidia’s NeMo diarization pipeline. |
Ali Sartaz Khan; Tolulope Ogunremi; Ahmed Adel Attia; Dorottya Demszky; | arxiv-cs.SD | 2025-05-16 |
| 980 | LegoSLM: Connecting LLM with Speech Encoder Using CTC Posteriors Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we propose a new paradigm, LegoSLM, that bridges speech encoders and LLMs using the ASR posterior matrices. |
Rao Ma; Tongzhou Chen; Kartik Audhkhasi; Bhuvana Ramabhadran; | arxiv-cs.CL | 2025-05-16 |
| 981 | Automatic Speech Recognition for African Low-Resource Languages: Challenges and Future Directions IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Automatic Speech Recognition (ASR) technologies have transformed human-computer interaction; however, low-resource languages in Africa remain significantly underrepresented in both research and practical applications. This study investigates the major challenges hindering the development of ASR systems for these languages, which include data scarcity, linguistic complexity, limited computational resources, acoustic variability, and ethical concerns surrounding bias and privacy. |
SUKAIRAJ HAFIZ IMAM et. al. | arxiv-cs.CL | 2025-05-16 |
| 982 | Lost in The Noise: Evaluating ASR Performance in Industrial and Environment Noise Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Automatic Speech Recognition (ASR) systems are increasingly use in real-world applications, yet their robustness in noisy environments remains a challenge. This study examines the … |
Sara M. Pearsell; Oliver Niebuhr; | 2025 IEEE 8th International Conference on Industrial … | 2025-05-12 |
| 983 | EvilHarmony: Stealthy Adversarial Attacks Against Black-Box Speech Recognition Systems Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Automatic Speech Recognition (ASR) systems are vulnerable to adversarial examples (AEs), where small, carefully designed perturbations are added to original audio to mislead the … |
XUEJING YUAN et. al. | 2025 IEEE Symposium on Security and Privacy (SP) | 2025-05-12 |
| 984 | Efficient Integration of ASR with Large Language Models to Enhance Video Search at Scale Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Currently, video is the most consumed content on the internet, with an increasing amount of information and knowledge shared as videos. This has led to a significant rise in video … |
QIANG ZHANG et. al. | Companion Proceedings of the ACM on Web Conference 2025 | 2025-05-08 |
| 985 | Teochew-Wild: The First In-the-wild Teochew Dataset with Orthographic Annotations Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This paper reports the construction of the Teochew-Wild, a speech corpus of the Teochew dialect. |
Linrong Pan; Chenglong Jiang; Gaoze Hou; Ying Gao; | arxiv-cs.CL | 2025-05-08 |
| 986 | VITA-Audio: Fast Interleaved Cross-Modal Token Generation for Efficient Large Speech-Language Model IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts Related Code View Save Highlight: Specifically, we introduce a lightweight Multiple Cross-modal TokenPrediction (MCTP) module that efficiently generates multiple audio tokenswithin a single model forward pass, which not only accelerates the inferencebut also significantly reduces the latency for generating the first audio instreaming scenarios. |
ZUWEI LONG et. al. | arxiv-cs.CL | 2025-05-06 |
| 987 | Breaking Down Power Barriers in On-Device Streaming ASR: Insights and Solutions Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We found that the influence of these parameters on power consumption varies depending on factors such as invocation frequency and memory allocation. Leveraging these insights, we propose design principles that enhance on-device speech recognition models by reducing power consumption with minimal impact on accuracy. |
YANG LI et. al. | naacl | 2025-05-04 |
| 988 | Wav2Prompt: End-to-End Speech Prompt Learning and Task-based Fine-tuning for Text-based LLMs IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: Wav2Prompt uses a straightforward training process with only the same data used to train an automatic speech recognition (ASR) model. |
Keqi Deng; Guangzhi Sun; Phil Woodland; | naacl | 2025-05-04 |
| 989 | AMPS: ASR with Multimodal Paraphrase Supervision Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we present a new technique AMPS, that augments a multilingual multimodal ASR system with paraphrase-based supervision for improved conversational ASR in multiple languages, including Hindi, Marathi, Malayalam, Kannada, and Nyanja. |
Abhishek Gupta; Amruta Parulekar; Sameep Chattopadhyay; Preethi Jyothi; | naacl | 2025-05-04 |
| 990 | Afrispeech-Dialog: A Benchmark Dataset for Spontaneous English Conversations in Healthcare and Beyond IF:3 Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: We introduce Afrispeech-Dialog, a benchmark dataset of 50 simulated medical and non-medical African-accented English conversations, designed to evaluate automatic speech recognition (ASR) and related technologies. |
MARDHIYAH SANNI et. al. | naacl | 2025-05-04 |
| 991 | Prepending or Cross-Attention for Speech-to-Text? An Empirical Comparison Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: To perform a controlled architectural comparison, we train all models from scratch rather than using large pretrained models and use comparable data and parameter settings, testing speech-to-text recognition (ASR) and translation (ST) on MuST-C v1. |
Tsz Kin Lam; Marco Gaido; Sara Papi; Luisa Bentivogli; Barry Haddow; | naacl | 2025-05-04 |
| 992 | Medical Spoken Named Entity Recognition Related Papers Related Patents Related Grants Related Venues Related Experts Related Code View Save Highlight: In this work, we present *VietMed-NER* – the first spoken NER dataset in the medical domain. |
KHAI LE-DUC et. al. | naacl | 2025-05-04 |
| 993 | BERSting at The Screams: A Benchmark for Distanced, Emotional and Shouted Speech Recognition Related Papers Related Patents Related Grants Related Venues Related Experts Related Code View Save Highlight: Here we present theB(asic) E(motion) R(andom phrase) S(hou)t(s) (BERSt) dataset. |
PAIGE TUTTÖSÍ et. al. | arxiv-cs.CL | 2025-04-30 |
| 994 | Retrieval-Enhanced Few-Shot Prompting for Speech Event Extraction Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we present a modular, pipeline-based SpeechEE framework that integrates high-performance ASR with semantic search-enhanced prompting of Large Language Models (LLMs). |
Máté Gedeon; | arxiv-cs.CL | 2025-04-30 |
| 995 | “It Feels Like We’re Not Meeting The Criteria”: Examining and Mitigating The Cascading Effects of Bias in Automatic Speech Recognition in Spoken Language Interfaces Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Researchers have demonstrated that Automatic Speech Recognition (ASR) systems perform differently across demographic groups (i.e. show bias), yet their downstream impact on spoken … |
Kelechi Ezema; Chelsea Chandler; Rosy Southwell; Niranjan Cholendiran; Sidney D’Mello; | Proceedings of the 2025 CHI Conference on Human Factors in … | 2025-04-25 |
| 996 | Live: Learning Video LLM with Streaming Speech Transcription at Scale Summary Related Papers Related Patents Related Grants Related Venues Related Experts View Save Abstract: Recent video large language models (Video LLMs) often depend on costly human annotations or proprietary APIs (e.g., GPT-4o) to produce training data, which limits their training … |
JOYA CHEN et. al. | 2025 IEEE/CVF Conference on Computer Vision and Pattern … | 2025-04-22 |
| 997 | Chinese-LiPS: A Chinese Audio-visual Speech Recognition Dataset with Lip-reading and Presentation Slides Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this paper, we release a multimodal Chinese AVSR dataset, Chinese-LiPS, comprising 100 hours of speech, video, and corresponding manual transcription, with the visual modality encompassing both lip-reading information and the presentation slides used by the speaker. |
JINGHUA ZHAO et. al. | arxiv-cs.MM | 2025-04-21 |
| 998 | Acoustic to Articulatory Inversion of Speech; Data Driven Approaches, Challenges, Applications, and Future Scope Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: This review paper considered the relevant works published in the last ten years (2011-2021). The selection criteria includes (a) type of AAI – Speaker Dependent and Speaker Independent AAI, (b) objectives of the work – Articulatory approximation, Articulatory Feature space selection and Automatic Speech Recognition (ASR), explore the correlation between acoustic and articulatory features, and framework for Computer-assisted language training, (c) Corpus – Simultaneously recorded speech (wav) and medical imaging models such as ElectroMagnetic Articulography (EMA), Electropalatography (EPG), Laryngography, Electroglottography (EGG), X-ray Cineradiography, Ultrasound, and real-time Magnetic Resonance Imaging (rtMRI), (d) Methods or models – recent works are considered, and therefore all the works are based on machine learning, (e) Evaluation – as AAI is a non-linear regression problem, the performance evaluation is mostly done by Correlation Coefficient (CC), Root Mean Square Error (RMSE), and also considered Mean Square Error (MSE), and Mean Format Error (MFE). |
Leena G Pillai; D. Muhammad Noorul Mubarak; | arxiv-cs.SD | 2025-04-17 |
| 999 | Dysarthria Normalization Via Local Lie Group Transformations for Robust ASR Related Papers Related Patents Related Grants Related Venues Related Experts Related Code View Save Highlight: We present a geometry-driven method for normalizing dysarthric speech by modeling time, frequency, and amplitude distortions as smooth, local Lie group transformations of spectrograms. |
Mikhail Osipov; | arxiv-cs.SD | 2025-04-16 |
| 1000 | Advancing Arabic Speech Recognition Through Large-Scale Weakly Supervised Learning Related Papers Related Patents Related Grants Related Venues Related Experts View Save Highlight: In this work, we employ weakly supervised learning to train an Arabic ASR model using the Conformer architecture. |
MAHMOUD SALHAB et. al. | arxiv-cs.AI | 2025-04-16 |