Voice AI BenchmarksBy

INWORLD AISPEECH-TO-TEXT114,228 SAMPLES / 30 DAYSLAST RUN SEP 15, 2026, 08:00 UTC

official resource

STT 1 speech-to-text benchmarks

STT 1, hosted by Inworld AI, measures mean 65 ms time to final segment (4th of 28) and 4.4% word error rate (8th of 30) among STT systems. Results cover the last 30 days. Coval benchmarks hosted endpoints daily using a consistent evaluation methodology. Last measured .

Inworld STT 1 is Inworld's streaming transcription model for interactive systems.

Time to Final Segment#4 / 28
65ms
Word Error Rate#8 / 30
4.4%
Time to First Token#8 / 26
1400ms

Overview

The tested endpoint is multilingual and includes voice activity plus keyterm biasing.

STT 1 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
Inworld AI
Hosted by
Inworld AI
Source
Official API
Licensing
Proprietary
Deployment
Cloud
Region
US
Features
Keyterm biasing, Multilingual, VAD

How STT 1 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 STT 1 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
Show all 28 models
  1. #13Universal 3.5 Pro173 ms
  2. #14Grok STT207 ms
  3. #15Defaultvia Speechmatics209 ms
  4. #16Pulse212 ms
  5. #17Defaultvia Gradium246 ms
  6. #18resonant-1264 ms
  7. #19Whisper Large v3via Together AI296 ms
  8. #20Enhanced299 ms
  9. #21Gemini 3.5 Transcribe Live305 ms
  10. #22Voxtral Mini Transcribe Realtime 2602400 ms
  11. #23GPT Realtime Whisper551 ms
  12. #24GPT-4o mini Transcribe690 ms
  13. #25GPT-4o Transcribe754 ms
  14. #26Chirp 3776 ms
  15. #27Solaria 1805 ms
  16. #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 ms4,431
3STT RT v5Soniox57 ms1529 ms22,468
4STT 1Inworld AI65 ms1400 ms22,816
5Parakeet TDT 0.6B v3Together AI81 ms1215 ms22,468
6Nova 3Deepgram89 ms1418 ms22,835
7Nova 2Deepgram92 ms1419 ms22,713
8Flux MultilingualDeepgram98 ms1163 ms12,231
9FluxDeepgram99 ms1076 ms12,223
10Ink 2Cartesia122 ms1827 ms22,842
11Whisper Large v3Baseten125 ms912 ms1,252
12Scribe v2 RealtimeElevenLabs133 ms2175 ms22,846
13Universal 3.5 ProAssemblyAI173 ms1040 ms20,233
14Grok STTxAI207 ms—22,833
15DefaultSpeechmatics209 ms1428 ms22,856
16PulseSmallest212 ms2011 ms22,836
17DefaultGradium246 ms1980 ms22,819
18resonant-1Reson8264 ms—22,840
19Whisper Large v3Together AI296 ms1338 ms22,724
20EnhancedSpeechmatics299 ms1491 ms22,856
21Gemini 3.5 Transcribe LiveGemini305 ms1671 ms16,383
22Voxtral Mini Transcribe Realtime 2602Mistral400 ms1850 ms22,564
23GPT Realtime WhisperOpenAI551 ms1814 ms22,792
24GPT-4o mini TranscribeOpenAI690 ms—22,810
25GPT-4o TranscribeOpenAI754 ms—22,811
26Chirp 3Google776 ms5974 ms22,855
27Solaria 1Gladia805 ms1803 ms22,419
28Chirp 2Google873 ms6071 ms22,769
—Nemotron 3.5 ASR StreamingTogether AI—1548 ms22,727
—Universal StreamingAssemblyAI—1513 ms20,215
Under-sampled models are excluded; tied models share a place. Dotted WER values split into substitutions, deletions and insertions on hover or tap.

Latency vs accuracy

TTFS vs WEREach point is one measured model · STT 1 highlighted · 30-day averagesTime to Final Segment against Word Error Rate for every measured STT model, with STT 1 highlighted.
Qwen3 ASR 1.7b — 39 ms, 4.1%Qwen3 ASR Fast — 46 ms, 3.2%STT RT v5 — 57 ms, 5.7%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, 6.7%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, 3.2%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 — 805 ms, 7.1%Chirp 2 — 873 ms, 4.9%STT 1 — 65 ms, 4.4%

Where the errors come from

WER compositionSTT 1's Word Error Rate split by error type · 30-day averageSTT 1's WER split into substitutions, deletions and insertions.
  • STT 14.4%
SubstitutionsDeletionsInsertions

Averages and tail latency

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

STT 1 latency distributionMilliseconds · shared axis across metrics · last 30 daysSTT 1's latency percentiles per metric: p25–p75 band, p50 tick, whisker to p99.
  • Time to Final Segmentp50 63 ms · p99 119 ms
  • Time to First Tokenp50 1435 ms · p99 2538 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 Segment65 ms54 ms63 ms70 ms78 ms84 ms119 ms22,816
Word Error Rate4.4%0.0%0.0%6.1%12.0%18.2%50.0%22,853
Time to First Token1400 ms1043 ms1435 ms1641 ms1768 ms2034 ms2538 ms22,853

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 daysSTT 1's daily median Time to Final Segment over the last 30 days.
STT 1 · Sep 1: 66 msSTT 1 · Sep 2: 64 msSTT 1 · Sep 3: 64 msSTT 1 · Sep 4: 64 msSTT 1 · Sep 5: 62 msSTT 1 · Sep 6: 63 msSTT 1 · Sep 7: 61 msSTT 1 · Sep 8: 63 msSTT 1 · Sep 9: 64 msSTT 1 · Sep 10: 62 msSTT 1 · Sep 11: 64 msSTT 1 · Sep 12: 63 msSTT 1 · Sep 13: 64 msSTT 1 · Sep 14: 63 msSTT 1 · Sep 15: 64 ms
Word Error Rate — daily averageDaily average · UTC daysSTT 1's daily Word Error Rate over the last 30 days.
STT 1 · Sep 1: 7.4%STT 1 · Sep 2: 4.6%STT 1 · Sep 3: 4.9%STT 1 · Sep 4: 4.8%STT 1 · Sep 5: 4.1%STT 1 · Sep 6: 5.0%STT 1 · Sep 7: 4.9%STT 1 · Sep 8: 4.9%STT 1 · Sep 9: 4.4%STT 1 · Sep 10: 4.8%STT 1 · Sep 11: 4.7%STT 1 · Sep 12: 4.0%STT 1 · Sep 13: 5.4%STT 1 · Sep 14: 4.3%STT 1 · Sep 15: 4.0%

Time to Final Segment by dataset

STT 1 by test conditionMilliseconds · lower is better · best condition firstSTT 1's Time to Final Segment on each benchmark dataset.
  1. WildASR phone codec57 ms
  2. WildASR accents62 ms
  3. PipeCat (production)63 ms
  4. WildASR reverb65 ms
  5. WildASR clean68 ms
  6. WildASR clipping71 ms
  7. WildASR noise gaps71 ms
  8. WildASR far-field73 ms
Time to Final Segment per dataset, with the number of samples behind each figure.
DatasetTTFSSamples
LibriSpeech81 ms1
ProductionPipeCat63 ms11,493
AccentsWildASR62 ms1,115
CleanWildASR68 ms4,568
ClippingWildASR71 ms1,138
Far-fieldWildASR73 ms1,139
Noise gapsWildASR71 ms1,140
Phone codecWildASR57 ms1,136
ReverbWildASR65 ms1,086

Strongest condition: WildASR phone codec at 57 ms · weakest: LibriSpeech at 81 ms.

How fast is STT 1?

On Inworld AI, STT 1 measures mean 65 ms time to final segment (4th of 28) and mean 1400 ms time to first token (8th of 26). Last measured 2026-09-15.

How accurate is STT 1?

On Inworld AI, STT 1 measures 4.4% word error rate (8th of 30). Last measured 2026-09-15.

Who hosts STT 1?

STT 1 is created by Inworld AI and served by Inworld AI. Coval measures each hosted endpoint separately.

Limits of this comparison

Coval measures it independently from Inworld's TTS 2 family so recognition and synthesis remain attributable to the correct stage.

  • Coval does not evaluate game-character behavior or other application features in Inworld's broader platform.

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