Voice AI BenchmarksBy

GOOGLESPEECH-TO-TEXT113,793 SAMPLES / 30 DAYSLAST RUN SEP 15, 2026, 07:30 UTC

official resource

Chirp 2 speech-to-text benchmarks

Chirp 2, hosted by Google, measures mean 873 ms time to final segment (28th of 28) and 4.9% word error rate (12th of 30) among STT systems. Results cover the last 30 days. Coval benchmarks hosted endpoints daily using a consistent evaluation methodology. Last measured .

Chirp 2 is a Google Cloud Speech-to-Text model in the Chirp family.

Time to Final Segment#28 / 28
873ms
Word Error Rate#12 / 30
4.9%
Time to First Token#26 / 26
6072ms

Overview

Chirp 2 is a multilingual recognizer served through Cloud Speech-to-Text V2; the tested endpoint includes voice-activity and phrase-biasing controls.

Chirp 2 is tested every day on fixed public audio — clean, accented, noisy, reverberant, far-field, clipped and phone-codec speech — for transcription accuracy and streaming latency.

Technical specifications

Made by
Google
Hosted by
Google
Source
Official API
Licensing
Proprietary
Deployment
On-prem
Region
US
Features
Keyterm biasing, Multilingual, VAD

How Chirp 2 ranks

Full STT dashboard
Time to Final Segment — every measured STT modelMilliseconds · lower is better · 30-day averageEvery STT model ranked on Time to Final Segment, with Chirp 2 highlighted.
  1. #1Qwen3 ASR 1.7b39 ms
  2. #2Qwen3 ASR Fast46 ms
  3. #3STT RT v557 ms
  4. #4STT 165 ms
  5. #5Parakeet TDT 0.6B v381 ms
  6. #6Nova 389 ms
  7. #7Nova 292 ms
  8. #8Flux Multilingual98 ms
  9. #9Flux99 ms
  10. #10Ink 2122 ms
  11. #11Whisper Large v3via Baseten125 ms
  12. #12Scribe v2 Realtime133 ms
  13. #13Universal 3.5 Pro173 ms
  14. #14Grok STT207 ms
  15. #15Defaultvia Speechmatics209 ms
  16. #16Pulse212 ms
  17. #17Defaultvia Gradium246 ms
  18. #18resonant-1264 ms
  19. #19Whisper Large v3via Together AI296 ms
  20. #20Enhanced299 ms
  21. #21Gemini 3.5 Transcribe Live305 ms
  22. #22Voxtral Mini Transcribe Realtime 2602400 ms
  23. #23GPT Realtime Whisper551 ms
  24. #24GPT-4o mini Transcribe690 ms
  25. #25GPT-4o Transcribe754 ms
  26. #26Chirp 3776 ms
  27. #27Solaria 1803 ms
  28. #28Chirp 2873 ms
median of all models · 208 ms
STT models over the last 30 days, ranked on Time to Final Segment.
#ModelHostTTFSWERTTFTSamples
1Qwen3 ASR 1.7bBaseten39 ms931 ms1,256
2Qwen3 ASR FastNari46 ms1748 ms4,391
3STT RT v5Soniox57 ms1529 ms22,468
4STT 1Inworld AI65 ms1400 ms22,776
5Parakeet TDT 0.6B v3Together AI81 ms1215 ms22,428
6Nova 3Deepgram89 ms1418 ms22,795
7Nova 2Deepgram92 ms1419 ms22,673
8Flux MultilingualDeepgram98 ms1163 ms12,191
9FluxDeepgram99 ms1076 ms12,183
10Ink 2Cartesia122 ms1827 ms22,802
11Whisper Large v3Baseten125 ms912 ms1,252
12Scribe v2 RealtimeElevenLabs133 ms2175 ms22,806
13Universal 3.5 ProAssemblyAI173 ms1040 ms20,193
14Grok STTxAI207 ms—22,793
15DefaultSpeechmatics209 ms1428 ms22,816
16PulseSmallest212 ms2011 ms22,796
17DefaultGradium246 ms1980 ms22,779
18resonant-1Reson8264 ms—22,800
19Whisper Large v3Together AI296 ms1338 ms22,685
20EnhancedSpeechmatics299 ms1492 ms22,816
21Gemini 3.5 Transcribe LiveGemini305 ms1660 ms16,343
22Voxtral Mini Transcribe Realtime 2602Mistral400 ms1850 ms22,526
23GPT Realtime WhisperOpenAI551 ms1814 ms22,752
24GPT-4o mini TranscribeOpenAI690 ms—22,770
25GPT-4o TranscribeOpenAI754 ms—22,771
26Chirp 3Google776 ms5974 ms22,815
27Solaria 1Gladia803 ms1801 ms22,379
28Chirp 2Google873 ms6072 ms22,729
—Nemotron 3.5 ASR StreamingTogether AI—1548 ms22,687
—Universal StreamingAssemblyAI—1513 ms20,175
Under-sampled models are excluded; tied models share a place. Dotted WER values split into substitutions, deletions and insertions on hover or tap.

Highest relative placement: 12th of 30 on Word Error Rate.

Latency vs accuracy

TTFS vs WEREach point is one measured model · Chirp 2 highlighted · 30-day averagesTime to Final Segment against Word Error Rate for every measured STT model, with Chirp 2 highlighted.
Qwen3 ASR 1.7b — 39 ms, 4.1%Qwen3 ASR Fast — 46 ms, 3.2%STT RT v5 — 57 ms, 5.7%STT 1 — 65 ms, 4.4%Parakeet TDT 0.6B v3 — 81 ms, 11.2%Nova 3 — 89 ms, 6.2%Nova 2 — 92 ms, 7.9%Flux Multilingual — 98 ms, 8.8%Flux — 99 ms, 6.6%Ink 2 — 122 ms, 4.8%Whisper Large v3 via Baseten — 125 ms, 5.6%Scribe v2 Realtime — 133 ms, 5.5%Universal 3.5 Pro — 173 ms, 3.1%Grok STT — 207 ms, 4.6%Default via Speechmatics — 209 ms, 5.4%Pulse — 212 ms, 5.3%Default via Gradium — 246 ms, 10.0%resonant-1 — 264 ms, 3.4%Whisper Large v3 via Together AI — 296 ms, 8.5%Enhanced — 299 ms, 4.3%Gemini 3.5 Transcribe Live — 305 ms, 3.9%Voxtral Mini Transcribe Realtime 2602 — 400 ms, 6.2%GPT Realtime Whisper — 551 ms, 5.1%GPT-4o mini Transcribe — 690 ms, 4.9%GPT-4o Transcribe — 754 ms, 4.6%Chirp 3 — 776 ms, 4.1%Solaria 1 — 803 ms, 7.1%Chirp 2 — 873 ms, 4.9%

Where the errors come from

WER compositionChirp 2's Word Error Rate split by error type · 30-day averageChirp 2's WER split into substitutions, deletions and insertions.
  • Chirp 24.9%
SubstitutionsDeletionsInsertions

Averages and tail latency

Averages hide slow outliers — these are the distributions behind each figure.

Chirp 2 latency distributionMilliseconds · shared axis across metrics · last 30 daysChirp 2's latency percentiles per metric: p25–p75 band, p50 tick, whisker to p99.
  • Time to Final Segmentp50 791 ms · p99 2174 ms
  • Time to First Tokenp50 6418 ms · p99 9102 ms
band p25–p75 · tick p50 · whisker to p99 with p90 and p95 stops
Average and percentile values per metric, with the number of samples behind each row.
MetricAveragep25p50p75p90p95p99Samples
Time to Final Segment873 ms697 ms791 ms924 ms1214 ms1440 ms2174 ms22,729
Word Error Rate4.9%0.0%0.0%6.7%14.3%20.0%43.6%22,766
Time to First Token6072 ms5733 ms6418 ms6666 ms7113 ms7497 ms9102 ms22,766

Last 30 days

Daily medians from the same measurement runs · gaps are days without qualifying runs.

Time to Final Segment — daily p50Line p50 · band p25–p75 · UTC daysChirp 2's daily median Time to Final Segment over the last 30 days.
Chirp 2 · Sep 1: 711 msChirp 2 · Sep 2: 786 msChirp 2 · Sep 3: 821 msChirp 2 · Sep 4: 781 msChirp 2 · Sep 5: 749 msChirp 2 · Sep 6: 789 msChirp 2 · Sep 7: 750 msChirp 2 · Sep 8: 828 msChirp 2 · Sep 9: 803 msChirp 2 · Sep 10: 819 msChirp 2 · Sep 11: 820 msChirp 2 · Sep 12: 793 msChirp 2 · Sep 13: 790 msChirp 2 · Sep 14: 750 msChirp 2 · Sep 15: 769 ms
Word Error Rate — daily averageDaily average · UTC daysChirp 2's daily Word Error Rate over the last 30 days.
Chirp 2 · Sep 1: 3.9%Chirp 2 · Sep 2: 5.3%Chirp 2 · Sep 3: 5.8%Chirp 2 · Sep 4: 5.9%Chirp 2 · Sep 5: 5.8%Chirp 2 · Sep 6: 5.5%Chirp 2 · Sep 7: 6.4%Chirp 2 · Sep 8: 6.2%Chirp 2 · Sep 9: 6.0%Chirp 2 · Sep 10: 5.3%Chirp 2 · Sep 11: 5.2%Chirp 2 · Sep 12: 5.8%Chirp 2 · Sep 13: 5.2%Chirp 2 · Sep 14: 5.6%Chirp 2 · Sep 15: 6.4%

Time to Final Segment by dataset

Chirp 2 by test conditionMilliseconds · lower is better · best condition firstChirp 2's Time to Final Segment on each benchmark dataset.
  1. WildASR accents770 ms
  2. WildASR clean866 ms
  3. WildASR clipping866 ms
  4. WildASR reverb869 ms
  5. WildASR noise gaps874 ms
  6. PipeCat (production)881 ms
  7. WildASR far-field901 ms
  8. WildASR phone codec902 ms
Time to Final Segment per dataset, with the number of samples behind each figure.
DatasetTTFSSamples
LibriSpeech930 ms1
ProductionPipeCat881 ms11,441
AccentsWildASR770 ms1,111
CleanWildASR866 ms4,547
ClippingWildASR866 ms1,133
Far-fieldWildASR901 ms1,139
Noise gapsWildASR874 ms1,138
Phone codecWildASR902 ms1,131
ReverbWildASR869 ms1,088

Strongest condition: WildASR accents at 770 ms · weakest: LibriSpeech at 930 ms.

How fast is Chirp 2?

On Google, Chirp 2 measures mean 873 ms time to final segment (28th of 28) and mean 6072 ms time to first token (26th of 26). Last measured 2026-09-15.

How accurate is Chirp 2?

On Google, Chirp 2 measures 4.9% word error rate (12th of 30). Last measured 2026-09-15.

Who hosts Chirp 2?

Chirp 2 is created by Google and served by Google. Coval measures each hosted endpoint separately.

Limits of this comparison

Coval keeps it distinct from Chirp 3 so a generation change can be evaluated on the same audio and scoring rules.

  • The results describe Coval's configured US endpoint; Cloud region, recognizer settings and adaptation choices can change production behavior.

Official sources

Results are re-measured daily using fixed datasets and reported over a rolling 30-day window. 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