Voice AI BenchmarksBy

DATASETSTTWILDASR / FLEURSACTIVE

manifest

WildASR accents speech-to-text benchmark dataset

845 English clips spanning demographic accents.

Items
845
fixed public inputs
Models measured
31
last 30 days
Current leader · WER#1 / 31
2%
Universal 3.6 Provia AssemblyAI

How models rank on WildASR accents

Full STT dashboard
Word Error Rate on WildASR accentsPercent · lower is better · 30-day average on this datasetEvery STT model measured on the WildASR accents dataset, ranked on Word Error Rate.
  1. #1Universal 3.6 Pro2.0%
  2. #2GPT-4o Transcribe2.3%
  3. #3resonant-12.7%
  4. #4Universal 3.5 Pro2.8%
  5. #5Qwen3 ASR Fast3.0%
  6. #6GPT-4o mini Transcribe3.1%
  7. #7Qwen3 ASR 1.7b3.1%
  8. #8Grok STT3.4%
  9. #9Gemini 3.5 Transcribe Live3.9%
  10. #10Scribe v2 Realtime4.0%
  11. #11Chirp 34.3%
  12. #12Linden 14.4%
Show all 31 models
  1. #13Ink 24.6%
  2. #14Solaria 14.6%
  3. #15Whisper Large v3via Baseten4.7%
  4. #16Daytona STT4.7%
  5. #17Voxtral Mini Transcribe Realtime 26024.8%
  6. #18Pulse5.1%
  7. #19GPT Realtime Whisper5.6%
  8. #20Whisper Large v3via Together AI5.8%
  9. #21Chirp 25.9%
  10. #22Nova 36.0%
  11. #23Parakeet TDT 0.6B v36.3%
  12. #24Universal Streaming6.5%
  13. #25STT RT v56.7%
  14. #26Flux6.9%
  15. #27Nova 27.5%
  16. #28Default8.6%
  17. #29STT 19.3%
  18. #30Flux Multilingual11.2%
  19. #31Nemotron 3.5 ASR Streaming20.2%
median of all models · 4.7%
STT models on the WildASR accents dataset over the last 30 days, ranked on Word Error Rate.
#ModelHostTTFSWERTTFTSamples
1Universal 3.6 ProAssemblyAI124 ms443 ms1,270
2GPT-4o TranscribeOpenAI674 ms—1,402
3resonant-1Reson8241 ms—1,403
4Universal 3.5 ProAssemblyAI209 ms440 ms1,403
5Qwen3 ASR FastNari Labs44 ms1757 ms1,403
6GPT-4o mini TranscribeOpenAI649 ms—1,402
7Qwen3 ASR 1.7bBaseten24 ms280 ms699
8Grok STTSpaceXAI208 ms—1,399
9Gemini 3.5 Transcribe LiveGemini292 ms841 ms1,401
10Scribe v2 RealtimeElevenLabs119 ms2068 ms1,402
11Chirp 3Google662 ms4434 ms1,403
12Linden 1Speechmatics312 ms833 ms1,400
13Ink 2Cartesia151 ms1191 ms1,403
14Solaria 1Gladia709 ms1165 ms1,374
15Whisper Large v3Baseten90 ms415 ms699
16Daytona STTGuava437 ms3546 ms247
17Voxtral Mini Transcribe Realtime 2602Mistral400 ms1070 ms1,387
18PulseSmallest210 ms1199 ms1,403
19GPT Realtime WhisperOpenAI636 ms1052 ms1,401
20Whisper Large v3Together AI216 ms491 ms1,391
21Chirp 2Google793 ms4570 ms1,403
22Nova 3Deepgram82 ms969 ms1,369
23Parakeet TDT 0.6B v3Together AI99 ms507 ms1,396
24Universal StreamingAssemblyAI—847 ms1,403
25STT RT v5Soniox46 ms805 ms1,239
26FluxDeepgram102 ms688 ms1,369
27Nova 2Deepgram84 ms1002 ms1,369
28DefaultGradium355 ms1245 ms1,402
29STT 1Inworld AI58 ms1048 ms1,305
30Flux MultilingualDeepgram103 ms458 ms1,369
31Nemotron 3.5 ASR StreamingTogether AI—951 ms1,391
Under-sampled models are excluded; tied models share a place. Dotted WER values split into substitutions, deletions and insertions on hover or tap.

Inside the dataset

845 fixed public inputs — every model is tested on exactly these.

  • CLIP 1 · 8.0s

    “if the problem is a variable or function with multiple misrecognized words, add the whole phrase to your vocabulary.”

  • CLIP 2 · 7.5s

    “palm oil is cheap, but workers on illegal plantations get exploited and huge areas of jungle get destroyed for its production.”

  • CLIP 3 · 6.1s

    “the man looked up from his book and, noticing nothing newsworthy, returned his gaze to the page and continued reading.”

3 of 845 clips, streamed from the public benchmark storage bucket.

Source: bosonai/WildASR demographic_shift__en__demographic_accent_en · License: Apache-2.0 (WildASR)

What it tests

This set measures recognition across a wider range of English accents than the clean baseline. Its 845 clips make it one of the largest STT datasets in the active benchmark.

This condition is derived from the WildASR environment family and FLEURS read speech. Each paired condition uses the same utterance and transcript as the WildASR clean baseline, so the WER change isolates the effect of the test condition.

Metrics reported for this dataset

Same datasets, prompts and metric definitions for every model, measured by Coval’s open-source runner. Full methodology on the overview.

Evaluate your own voice agent

Use Coval to test your production configuration, prompts and calls—not only the public benchmark endpoints.

Book a Demo