cs.CLNov 12, 2026

Beyond Word Error Rate: A Failure Taxonomy of Streaming Dictation ASR

Authors: Harras Mansoor, Muhammad Wisal, Hassan Bhatti, Masood Shakir

Abstract

Speech recognition (ASR) systems are usually compared by word error rate (WER) on recorded files, after punctuation, capital letters and number format are removed. In dictation, the audio is streamed live and the text goes into the user's message as written, so what matters is which mistakes the user has to fix. The test set, streamed to eleven ASR systems, comprises 106 scripted English dictations from one speaker (read in a quiet room, with a fan on, and whispered), 20 clips in each of eleven other language and dialect sets, and 400 recordings that join clips in two languages. Every error was sorted by a fixed rule into four levels: recognition, output formatting, language handling and, for the self-hosted models, audio dropped before the model. Systems with similar WER failed in different ways. Zoom Scribe and OpenAI's GPT-Live-Transcribe have almost the same WER on quiet English (7.4% and 7.3%), but 34.7% and 8.8% on whispered English. Zoom and both AssemblyAI models left out 26-28% of whispered words, and still 21-26% with their end-of-turn detection relaxed or off; seven other systems left out at most 3%. Qwen3-ASR's 32% came from the speech detection in our own deployment (6% when relaxed). Across 107 EARS speakers, the same four systems lost 7-31% of whispered words, while six others lost under 2%. With only English selected, Gemini, Deepgram and our self-hosted Whisper lost the non-English part of a two-language recording in 86-99% of recordings, against 33-46% with both languages selected, while GPT-Live-Transcribe, OpenAI's realtime Whisper and Meta lost it in at most 20%. Running Whisper behind a Silero voice activity detector that drops audio judged to be silence raised whispered WER from 10.1% to 51.4%. Most English results come from one read speaker. The code, the recordings, both references and every transcript are released.

Figures & tables

Appendix figures & tables4 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

Oct 17, 2026

Ask, Don't Guess: Targeted Clarification for Speech Recognition Errors

Speech recognition in real time in far-field conversational settings still incurs significant errors. Identifying potential error spans during streaming ASR permits a recourse mechanism where the speaker can be asked to repeat the affected content under cleaner acoustic conditions. We propose PARDON, a new framework for online error localization, that identifies spans likely to benefit from this intervention. Our supervision uses clean audio to estimate how the recognizer's output might change under improved acoustic conditions. We probabilistically align noisy and clean decodes using a pair-HMM to obtain soft supervision. These soft targets train a counterfactual gain scorer by combining a suite of acoustic and language features to produce token-level scores, that are further grouped into spans for targeted recourse. An LLM uses the ASR transcript of the repetition to revise only the selected span. At a precision of 70%, PARDON achieves the highest recall by identifying up to 52% more erroneous tokens compared to the strongest baseline. Targeted repetition yields relative WER reductions of up to 24% under a budget that accounts for both interruption frequency and the lengths of re-uttered spans.
Mar 5, 2026

Beyond Word Error Rate: Auditing the Diversity Tax in Speech Recognition through Dataset Cartography

Automatic speech recognition (ASR) systems are predominantly evaluated using the Word Error Rate (WER). However, raw token-level metrics fail to capture semantic fidelity and routinely obscures the `diversity tax', the disproportionate burden on marginalized and atypical speaker due to systematic recognition failures. 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. To enable rigorous model auditing, we introduce the sample difficulty index (SDI), a novel metric that quantifies how intrinsic demographic and acoustic factors drive model failure. By mapping SDI on data cartography, we demonstrate that metrics EmbER and SemDist expose hidden systemic biases and inter-model disagreements that WER ignores. Finally, our findings are the first steps towards a robust audit framework for prospective safety analysis, empowering developers to audit and mitigate ASR disparities prior to deployment.
Oct 21, 2026

Text-Only LLM Correctors Break Correct Transcripts: A Repair-Damage Analysis of Audio Grounding in ASR Error Correction

LLM correctors of ASR output are judged by aggregate WER, which hides what they fix and what they break. We decompose every WER change exactly into repaired tokens, damaged tokens and insertions, and use this to ask what the source audio adds to a corrector. We compare correction with and without the audio, for the model's own draft and for the output of an independent recogniser, on three corpora. With Gemini 2.5 Flash, text-only correction of its own draft breaks 13-32% of already-correct transcripts; with the audio, 1-3%. Damage falls in both settings. Repair falls against the own draft and rises against the recogniser. On medical dictation, kept insertions cancel the gain in net WER. The open-weight Voxtral, which rarely edits, does not show this pattern. Prompts differ between the two arms, so the contrasts bound the effect of the audio and do not isolate it.