Voice AI BenchmarksBy

FISH AUDIOTEXT-TO-SPEECH7,650 SAMPLES / 30 DAYSLAST RUN OCT 14, 2026, 16:30 UTC

official resource

S2.1 Pro Free text-to-speech benchmarks

S2.1 Pro Free, hosted by Fish Audio (FishAudio, fish.audio), measures mean 849 ms time to first audio (31st of 32) and 1.5% word error rate (2nd of 32) among TTS systems. Results cover the last 30 days. Last measured .

S2.1 Pro Free is Fish Audio's free service route for the S2.1 Pro family.

Time to First Audio#31 / 32
849ms
TTFA Network Roundtrip#31 / 32
833ms
TTFA Leading Silence#9 / 32
16ms
Word Error Rate#2 / 32
1.5%

Overview

Separate results show the latency and intelligibility of the free endpoint without combining it with the production route.

S2.1 Pro Free is tested every day on a fixed set of text prompts, measuring how quickly audible speech starts and how intelligible the result is.

Technical specifications

Made by
Fish Audio
Hosted by
Fish Audio
Source
Official API
Licensing
Proprietary
Deployment
Cloud
Region
US
Features
Emotion control, Multilingual, Voice cloning

How S2.1 Pro Free ranks

Full TTS dashboard
Time to First Audio — every measured TTS modelMilliseconds · lower is better · 30-day averageEvery TTS model ranked on Time to First Audio, with S2.1 Pro Free highlighted.
  1. #1vui59 ms
  2. #2Default63 ms
  3. #3Qwen3 TTS Fast68 ms
  4. #4TTS Flash 291 ms
  5. #5Qwen3 TTS 1.7b107 ms
  6. #6Simba 3.0111 ms
  7. #7Palabra TTS120 ms
  8. #8Simba 3.2126 ms
  9. #9drama3-preview167 ms
  10. #10TTS 2191 ms
  11. #11Flash v2.5192 ms
  12. #12Eleven v4 Turbo201 ms
  13. #13Flux TTS205 ms
  14. #14TTS RT v1255 ms
  15. #15TTS Rt v2256 ms
  16. #16Mist v3287 ms
  17. #17Phantom Z 3.4 conversational291 ms
  18. #18Sonic 3.5293 ms
  19. #19Aura 2297 ms
  20. #20Sonic 3.6299 ms
  21. #21Falcon 2299 ms
  22. #22Coda332 ms
  23. #23S2.1 Pro341 ms
  24. #24Grok TTS363 ms
  25. #25S1389 ms
  26. #26Lightning v3.1 Pro392 ms
  27. #27Daytona TTS486 ms
  28. #28Aura 2 En570 ms
  29. #29Chirp 3 HD597 ms
  30. #30Qwen3 TTS Flash Realtime718 ms
  31. #31S2.1 Pro Free849 ms
Show all 32 models
  1. #32GPT-4o mini TTS2585 ms
median of all models · 289 ms
TTS models over the last 30 days, ranked on Time to First Audio.
#ModelHostTTFAWERSamples
1vuiFluxions59 ms1,748
2Gradium TTS Beta 202609Gradium63 ms1,920
3Qwen3 TTS FastNari Labs68 ms1,920
4TTS Flash 2Inworld AI91 ms1,724
5Qwen3 TTS 1.7bBaseten107 ms1,248
6Simba 3.0SpeechifyAI111 ms1,919
7Palabra TTS v1Palabra120 ms1,580
8Simba 3.2SpeechifyAI126 ms1,916
9Drama 3 Preview Us EastFish Audio167 ms166
10TTS 2Inworld AI191 ms1,724
11Flash v2.5ElevenLabs192 ms1,910
12Eleven v4 TurboElevenLabs201 ms1,230
13Flux TTSDeepgram205 ms1,783
14TTS RT v1Soniox255 ms1,622
15TTS Rt v2Soniox256 ms1,619
16Mist v3Rime287 ms602
17Phantom Z 3.4 conversationalDeepdub291 ms1,655
18Sonic 3.5Cartesia293 ms1,920
19Aura 2Deepgram297 ms1,852
20Sonic 3.6Cartesia299 ms1,920
21Falcon 2Murf299 ms1,919
22CodaRime332 ms602
23S2.1 ProFish Audio341 ms1,920
24Grok TTSSpaceXAI363 ms1,915
25S1Fish Audio389 ms1,919
26Lightning v3.1 ProSmallest392 ms1,920
27Daytona TTSGuava486 ms1,152
28Aura 2 EnCloudflare570 ms1,919
29Chirp 3 HDGoogle597 ms1,920
30Qwen3 TTS Flash RealtimeAlibaba718 ms1,919
31S2.1 Pro FreeFish Audio849 ms1,913
32GPT-4o mini TTSOpenAI2585 ms1,846
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: 2nd of 32 on Word Error Rate.

Latency vs accuracy

TTFA vs WEREach point is one measured model · S2.1 Pro Free highlighted · 30-day averagesTime to First Audio against Word Error Rate for every measured TTS model, with S2.1 Pro Free highlighted.
vui — 59 ms, 2.1%Default — 63 ms, 1.9%Qwen3 TTS Fast — 68 ms, 1.8%TTS Flash 2 — 91 ms, 1.7%Qwen3 TTS 1.7b — 107 ms, 1.8%Simba 3.0 — 111 ms, 1.5%Palabra TTS — 120 ms, 2.6%Simba 3.2 — 126 ms, 1.9%drama3-preview — 167 ms, 2.0%TTS 2 — 191 ms, 1.6%Flash v2.5 — 192 ms, 1.8%Eleven v4 Turbo — 201 ms, 1.7%Flux TTS — 205 ms, 1.7%TTS RT v1 — 255 ms, 2.4%TTS Rt v2 — 256 ms, 2.4%Mist v3 — 287 ms, 2.1%Phantom Z 3.4 conversational — 291 ms, 1.8%Sonic 3.5 — 293 ms, 1.5%Aura 2 — 297 ms, 1.8%Sonic 3.6 — 299 ms, 1.5%Falcon 2 — 299 ms, 1.7%Coda — 332 ms, 9.5%S2.1 Pro — 341 ms, 1.5%Grok TTS — 363 ms, 1.6%S1 — 389 ms, 1.8%Lightning v3.1 Pro — 392 ms, 1.5%Daytona TTS — 486 ms, 1.6%Aura 2 En — 570 ms, 2.1%Chirp 3 HD — 597 ms, 1.5%Qwen3 TTS Flash Realtime — 718 ms, 3.0%GPT-4o mini TTS — 2585 ms, 1.8%S2.1 Pro Free — 849 ms, 1.5%

Where the errors come from

WER compositionS2.1 Pro Free's Word Error Rate split by error type · 30-day averageS2.1 Pro Free's WER split into substitutions, deletions and insertions.
  • S2.1 Pro Free1.5%
SubstitutionsDeletionsInsertions

Averages and tail latency

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

S2.1 Pro Free latency distributionMilliseconds · shared axis across metrics · last 30 daysS2.1 Pro Free's latency percentiles per metric: p25–p75 band, p50 tick, whisker to max.
  • Time to First Audiop50 844 ms · Max 6744 ms
  • TTFA Network Roundtripp50 833 ms · Max 6744 ms
  • TTFA Leading Silencep50 5 ms · Max 141 ms
band p25–p75 · tick p50 · whisker to max with p90 and p95 stops
Average and percentile values per metric, with the number of samples behind each row.
MetricAveragep25p50p75p90p95MaxSamples
Time to First Audio849 ms759 ms844 ms935 ms1041 ms1160 ms6744 ms1,913
TTFA Network Roundtrip833 ms748 ms833 ms908 ms1009 ms1138 ms6744 ms1,913
TTFA Leading Silence16 ms1 ms5 ms22 ms50 ms74 ms141 ms1,913
Word Error Rate1.5%0.0%0.0%1.8%6.9%12.1%43.8%1,911

Last 30 days

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

Time to First Audio — daily p50Line p50 · band p25–p75 · UTC daysS2.1 Pro Free's daily median Time to First Audio over the last 30 days.
S2.1 Pro Free · Sep 23: 963 msS2.1 Pro Free · Sep 24: 884 msS2.1 Pro Free · Sep 25: 879 msS2.1 Pro Free · Sep 26: 897 msS2.1 Pro Free · Sep 27: 900 msS2.1 Pro Free · Sep 28: 1,002 msS2.1 Pro Free · Sep 29: 931 msS2.1 Pro Free · Sep 30: 849 msS2.1 Pro Free · Oct 1: 987 msS2.1 Pro Free · Oct 2: 884 msS2.1 Pro Free · Oct 3: 906 msS2.1 Pro Free · Oct 4: 878 msS2.1 Pro Free · Oct 5: 830 msS2.1 Pro Free · Oct 6: 881 msS2.1 Pro Free · Oct 7: 896 msS2.1 Pro Free · Oct 8: 629 msS2.1 Pro Free · Oct 9: 411 msS2.1 Pro Free · Oct 10: 771 msS2.1 Pro Free · Oct 11: 758 msS2.1 Pro Free · Oct 12: 750 msS2.1 Pro Free · Oct 13: 1,002 msS2.1 Pro Free · Oct 14: 802 ms
Word Error Rate — daily averageDaily average · UTC daysS2.1 Pro Free's daily Word Error Rate over the last 30 days.
S2.1 Pro Free · Sep 23: 4.7%S2.1 Pro Free · Sep 24: 3.6%S2.1 Pro Free · Sep 25: 3.5%S2.1 Pro Free · Sep 26: 3.3%S2.1 Pro Free · Sep 27: 4.5%S2.1 Pro Free · Sep 28: 5.1%S2.1 Pro Free · Sep 29: 3.7%S2.1 Pro Free · Sep 30: 1.8%S2.1 Pro Free · Oct 1: 4.9%S2.1 Pro Free · Oct 2: 4.2%S2.1 Pro Free · Oct 3: 3.8%S2.1 Pro Free · Oct 4: 7.3%S2.1 Pro Free · Oct 5: 2.3%S2.1 Pro Free · Oct 6: 5.2%S2.1 Pro Free · Oct 7: 5.7%S2.1 Pro Free · Oct 8: 6.0%S2.1 Pro Free · Oct 9: 2.9%S2.1 Pro Free · Oct 10: 1.4%S2.1 Pro Free · Oct 11: 2.5%S2.1 Pro Free · Oct 12: 4.5%S2.1 Pro Free · Oct 13: 4.8%S2.1 Pro Free · Oct 14: 5.8%

How fast is S2.1 Pro Free?

On Fish Audio, S2.1 Pro Free measures mean 849 ms time to first audio (31st of 32). Last measured 2026-10-14.

How accurate is S2.1 Pro Free?

On Fish Audio, S2.1 Pro Free measures 1.5% word error rate (2nd of 32). Last measured 2026-10-14.

Who hosts S2.1 Pro Free?

S2.1 Pro Free is created by Fish Audio and served by Fish Audio. Coval measures each hosted endpoint separately.

Limits of this comparison

The free route is measured separately from the production route using the same prompts.

  • Free-tier capacity and policy can change, so readers should use the live window rather than treating one snapshot as permanent.

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