Evaluation of Automatic Speech Recognition Using Generative Large Language Models

Automatic Speech Recognition (ASR) is traditionally evaluated using Word Error Rate (WER), a metric that is insensitive to meaning. Embedding-based semantic metrics are better correlated with human perception, but decoder-based Large Language Models (LLMs) remain underexplored for this task. 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. On the HATS dataset, the best LLMs achieve 92--94\% agreement with human annotators for hypothesis selection, compared to 63\% for WER, also outperforming semantic metrics. Embeddings from decoder-based LLMs show performance comparable to encoder models. Finally, LLMs offer a promising direction for interpretable and semantic ASR evaluation.

Do LLM Decoders Listen Fairly? Benchmarking How Language Model Priors Shape Bias in Speech Recognition

As pretrained large language models replace task-specific decoders in speech recognition, a critical question arises: do their text-derived priors make recognition fairer or more biased across demographic groups? We evaluate nine models spanning three architectural generations (CTC with no language model, encoder-decoder with an implicit LM, and LLM-based with an explicit pretrained decoder) on about 43,000 utterances across five demographic axes (ethnicity, accent, gender, age, first language) using Common Voice 24 and Meta's Fair-Speech, a controlled-prompt dataset that eliminates vocabulary confounds. On clean audio, three findings challenge assumptions: LLM decoders do not amplify racial bias (Granite-8B has the best ethnicity fairness, max/min WER = 2.28); Whisper exhibits pathological hallucination on Indian-accented speech with a non-monotonic insertion-rate spike to 9.62% at large-v3; and audio compression predicts accent fairness more than LLM scale. We then stress-test these findings under 12 acoustic degradation conditions (noise, reverberation, silence injection, chunk masking) across both datasets, totaling 216 inference runs. Severe degradation paradoxically compresses fairness gaps as all groups converge to high WER, but silence injection amplifies Whisper's accent bias up to 4.64x by triggering demographic-selective hallucination. Under masking, Whisper enters catastrophic repetition loops (86% of 51,797 insertions) while explicit-LLM decoders produce 38x fewer insertions with near-zero repetition; high-compression audio encoding (Q-former) reintroduces repetition pathology even in LLM decoders. These results suggest that audio encoder design, not LLM scaling, is the primary lever for equitable and robust speech recognition.

Aligning Stuttered-Speech Research with End-User Needs: Scoping Review, Survey, and Guidelines

Atypical speech is receiving greater attention in speech technology research, but much of this work unfolds with limited interdisciplinary dialogue. For stuttered speech in particular, it is widely recognised that current speech recognition systems fall short in practice, and current evaluation methods and research priorities are not systematically grounded in end-user experiences and needs. In this work, we analyse these gaps through 1) a scoping review of papers that deal with stuttered speech and 2) a survey of 70 stakeholders, including adults who stutter and speech-language pathologists. By analysing these two perspectives, we propose a taxonomy of stuttered-speech research, identify where current research directions diverge from the needs articulated by stakeholders, and conclude by outlining concrete guidelines and directions towards addressing the real needs of the stuttering community.

ATIR: Towards Audio-Text Interleaved Contextual Retrieval

Audio carries richer information than text, including emotion, speaker traits, and environmental context, while also enabling lower-latency processing compared to speech-to-text pipelines. However, recent multimodal information retrieval research has predominantly focused on images, largely overlooking audio, especially in the setting of interleaved audio-text contextual retrieval. In this work, we introduce the Audio-Text Interleaved contextual Retrieval (ATIR) task, where queries can alternate between audio and text modalities. We construct an ATIR benchmark by integrating several Automatic Speech Recognition (ASR), QA, and retrieval datasets, ultimately unifying four types of contextual retrieval tasks. This benchmark substantially addresses the limitations of existing audio retrieval datasets in semantic retrieval. To study this task, we evaluate several off-the-shelf retrievers and train our ATIR model based on a Multimodal Large Language Model (MLLM). We further introduce a novel token compression mechanism that is orthogonal to existing compression methods, thereby alleviating the issue of excessive audio tokens in MLLM-based ATIR models. Experimental results demonstrate that our ATIR model achieves substantial improvements over strong baselines.

"This Wasn't Made for Me": Recentering User Experience and Emotional Impact in the Evaluation of ASR Bias

Studies on bias in Automatic Speech Recognition (ASR) tend to focus on reporting error rates for speakers of underrepresented dialects, yet less research examines the human side of system bias: how do system failures shape users' lived experiences, how do users feel about and react to them, and what emotional toll do these repeated failures exact? We conducted user experience studies across four U.S. locations (Atlanta, Gulf Coast, Miami Beach, and Tucson) representing distinct English dialect communities. Our findings reveal that most participants report technologies fail to consider their cultural backgrounds and require constant adjustment to achieve basic functionality. Despite these experiences, participants maintain high expectations for ASR performance and express strong willingness to contribute to model improvement. Qualitative analysis of open-ended narratives exposes the deeper costs of these failures. Participants report frustration, annoyance, and feelings of inadequacy, yet the emotional impact extends beyond momentary reactions. Participants recognize that systems were not designed for them, yet often internalize failures as personal inadequacy despite this critical awareness. They perform extensive invisible labor, including code-switching, hyper-articulation, and emotional management, to make failing systems functional. Meanwhile, their linguistic and cultural knowledge remains unrecognized by technologies that encode particular varieties as standard while rendering others marginal. These findings demonstrate that algorithmic fairness assessments based on accuracy metrics alone miss critical dimensions of harm: the emotional labor of managing repeated technological rejection, the cognitive burden of constant self-monitoring, and the psychological toll of feeling inadequate in one's native language variety.

Voice of India: A Large-Scale Benchmark for Real-World Speech Recognition in India

Existing Indic ASR benchmarks often use scripted, clean speech and leaderboard driven evaluation that encourages dataset specific overfitting. 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. The dataset contains 306230 utterances, totaling 536 hours of speech from 36691 speakers with transcripts accounting for spelling variations. We also analyze performance geographically at the district level, revealing disparities. Finally, we provide detailed analysis across factors such as audio quality, speaking rate, gender, and device type, highlighting where current ASR systems struggle and offering insights for improving real world Indic ASR systems.

Reducing the Offline-Streaming Gap for Unified ASR Transducer with Consistency Regularization

Unification of automatic speech recognition (ASR) systems reduces development and maintenance costs, but training a single model to perform well in both offline and low-latency streaming settings remains challenging. We present a Unified ASR framework for Transducer (RNNT) training that supports both offline and streaming decoding within a single model, using chunk-limited attention with right context and dynamic chunked convolutions. To further close the gap between offline and streaming performance, we introduce an efficient Triton implementation of mode-consistency regularization for RNNT (MCR-RNNT), which encourages agreement across training modes. Experiments show that the proposed approach improves streaming accuracy at low latency while preserving offline performance and scaling to larger model sizes and training datasets. The proposed Unified ASR framework and the English model checkpoint are open-sourced.

Detecting Hallucinations in SpeechLLMs at Inference Time Using Attention Maps

Hallucinations in Speech Large Language Models (SpeechLLMs) pose significant risks, yet existing detection methods typically rely on gold-standard outputs that are costly or impractical to obtain. Moreover, hallucination detection methods developed for text-based LLMs do not directly capture audio-specific signals. We investigate four attention-derived metrics: AUDIORATIO, AUDIOCONSISTENCY, AUDIOENTROPY, and TEXTENTROPY, designed to capture pathological attention patterns associated with hallucination, and train lightweight logistic regression classifiers on these features for efficient inference-time detection. Across automatic speech recognition and speech-to-text translation tasks, evaluations on Qwen-2-Audio and Voxtral-3B show that our approach outperforms uncertainty-based and prior attention-based baselines on in-domain data, achieving improvements of up to +0.23 PR-AUC, and generalises to out-of-domain ASR settings. We further find that strong performance can be achieved with approximately 100 attention heads, improving out-of-domain generalisation compared to using all heads. While effectiveness is model-dependent and task-specific training is required, our results demonstrate that attention patterns provide a valuable tool for hallucination detection in SpeechLLMs.

UAF: A Unified Audio Front-end LLM for Full-Duplex Speech Interaction

Full-duplex speech interaction, as the most natural and intuitive mode of human communication, is driving artificial intelligence toward more human-like conversational systems. Traditional cascaded speech processing pipelines suffer from critical limitations, including accumulated latency, information loss, and error propagation across modules. To address these issues, recent efforts focus on the end-to-end audio large language models (LLMs) like GPT-4o, which primarily unify speech understanding and generation task. However, most of these models are inherently half-duplex, and rely on a suite of separate, task-specific front-end components, such as voice activity detection (VAD) and turn-taking detection (TD). 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. Our model reformulates diverse audio front-end tasks into a single auto-regressive sequence prediction problem, including VAD, TD, speaker recognition (SR), automatic speech recognition (ASR) and question answer (QA). It takes streaming fixed-duration audio chunk (e.g., 600 ms) as input, leverages a reference audio prompt to anchor the target speaker at the beginning, and regressively generates discrete tokens encoding both semantic content and system-level state controls (e.g., interruption signals). Experiments demonstrate that our model achieves leading performance across multiple audio front-end tasks and significantly enhances response latency and interruption accuracy in real-world interaction scenarios.

Where Do Self-Supervised Speech Models Become Unfair?

Speech encoder models are known to model members of some speaker groups (SGs) better than others. However, there has been little work in establishing why this occurs on a technological level. 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). We find S3Ms produce embeddings biased against certain SGs for both tasks, starting at the very first latent layers. Furthermore, we find opposite patterns of layerwise bias for SID vs ASR for all models in our study: SID bias is minimized in layers that minimize overall SID error; on the other hand, ASR bias is maximized in layers that minimize overall ASR error. The inverse bias/error relationship for ASR is unaffected when probing S3Ms that are finetuned for ASR, suggesting SG-level bias is established during pretraining and is difficult to remove.