Benchmark Almanac

Open TTS Leaderboard 1.2.0

sahilmahendrakar/Paradee-8M-v1.0 NEW 0.96% tops its leaderboard Qwen3-TTS-12Hz-1.7B-Base 3.38% leads the models we track
Not in model cards
Version 1.2.0

Speech models read multilingual text aloud. The board compares transcriptions with the reference text and measures voice similarity.

The board adds UTMOS scores.

Made by
Hugging Face
Released
October 2026
Kind
Audio understanding and voice
Level
Everyday
Scored
Checked automatically
Answers
Open-ended
Tasks
All public
Measures
Capability
Language
Many languages
In and out
Text, Audio → Audio
Interaction
Single turn
Items from
Another benchmark
Run by
Anyone
Last model added
January 2026
Scores refreshed
November 20th, 2026
37
Entries on its own leaderboard, 3 models we track

sahilmahendrakar/Paradee-8M-v1.0 NEW tops the leaderboard with 0.96% Qwen3-TTS-12Hz-1.7B-Base leads with 3.38% among models released by January 2026

Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. The best model we track is Qwen3-TTS-12Hz-1.7B-Base, at 3.38%. This view shows only models we track, each at its best.

The newest model with a score here came out in January 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 sahilmahendrakar/Paradee-8M-v1.0 NEW Without voice cloning 0.96%
  2. 2 gradium/phonon-31m-en NEW ¹ Without voice cloning 1.08%
  3. 3 gradium/phonon-40m-en NEW ² Without voice cloning 1.09%
  4. 4 ResembleAI/chatterbox (ChatterboxTTS) Without voice cloning 1.20%
  5. 5 BreezeBlue/Breeze-TTS-2 Without voice cloning 1.22%
  6. 6 LiquidAI/LFM2.5-Audio-1.5B NEW Without voice cloning 1.27%
  7. 7 tencent/AuK Voice cloning 1.39%
  8. 8 tencent/AuK-Flash Voice cloning 1.40%
  9. 9 owensong/Inflect-Micro-v2 Without voice cloning 1.69%
  10. 10 microsoft/speecht5_tts Without voice cloning 1.79%
  11. 11 hexgrad/Kokoro-82M Without voice cloning 2.49%
  12. 12 k2-fsa/OmniVoice Without voice cloning 2.72%
  13. 13 vibevoice/VibeVoice-7B-hf Voice cloning 2.84%
  14. 14 IndexTeam/IndexTTS-2.5 Without voice cloning 2.94%
  15. 15 neuphonic/neutts-nano Without voice cloning 3.04%
  16. 16 neuphonic/neutts-nano-q8-gguf Without voice cloning 3.04%
  17. 17 Supertone/supertonic-3 Without voice cloning 3.06%
  18. 18 fishaudio/s2-pro Without voice cloning 3.16%
  19. 19 bosonai/higgs-tts-3-4b Voice cloning 3.23%
  20. 20 Qwen3-TTS-12Hz-1.7B-Base Voice cloning 3.38%
  21. 21 vibevoice/VibeVoice-1.5B-hf Voice cloning 3.40%
  22. 22 sesame/csm-1b Without voice cloning 3.40%
  23. 23 fishaudio/drama-3 NEW Without voice cloning 3.41%
  24. 24 Qwen3-TTS-12Hz-1.7B-CustomVoice Without voice cloning 3.58%
  25. 25 OpenMOSS-Team/MOSS-TTS Without voice cloning 3.77%
  26. 26 nvidia/magpie_tts_multilingual_357m Without voice cloning 3.80%
  27. 27 openbmb/VoxCPM2 Voice cloning 4.14%
  28. 28 owensong/Inflect-Nano-v1 Without voice cloning 4.18%
  29. 29 bosonai/higgs-tts-2-3b-base Without voice cloning 4.67%
  30. 30 kyutai/pocket-tts Without voice cloning 6.63%
  31. 31 ResembleAI/chatterbox (ChatterboxMultilingualTTS) Voice cloning 6.66%
  32. 32 facebook/seamless-m4t-v2-large Without voice cloning 6.90%
  33. 33 FunAudioLLM/Fun-CosyVoice3-0.5B-2512 (RL) Without voice cloning 7.48%
  34. 34 mistralai/Voxtral-4B-TTS-2603 Without voice cloning 8.20%
  35. 35 Qwen3-Omni-30B-A3B Without voice cloning 9.84%
  36. 36 microsoft/VibeVoice-Realtime-0.5B Without voice cloning 12.9%
  37. 37 FunAudioLLM/Fun-CosyVoice3-0.5B-2512 Without voice cloning 17.4%

Source: Official Open TTS Leaderboard 1.2.0 leaderboard

Source: Official Open TTS Leaderboard 1.2.0 leaderboard

#EntryScoreSettingDateLinkSource
1 sahilmahendrakar/Paradee-8M-v1.0 NEW 0B 0.96% Without voice cloning huggingface.co Official leaderboard
2 gradium/phonon-31m-en NEW ¹ 0B 1.08% Without voice cloning huggingface.co Official leaderboard
3 gradium/phonon-40m-en NEW ² 0.1B 1.09% Without voice cloning huggingface.co Official leaderboard
4 ResembleAI/chatterbox (ChatterboxTTS) 0.8B 1.20% Without voice cloning huggingface.co Official leaderboard
5 BreezeBlue/Breeze-TTS-2 3.5B 1.22% Without voice cloning huggingface.co Official leaderboard
6 LiquidAI/LFM2.5-Audio-1.5B NEW 1.5B 1.27% Without voice cloning huggingface.co Official leaderboard
7 tencent/AuK 1.5B 1.39% Voice cloning huggingface.co Official leaderboard
8 tencent/AuK-Flash 1.5B 1.40% Voice cloning huggingface.co Official leaderboard
9 owensong/Inflect-Micro-v2 0B 1.69% Without voice cloning huggingface.co Official leaderboard
10 microsoft/speecht5_tts 0.1B 1.79% Without voice cloning huggingface.co Official leaderboard
11 hexgrad/Kokoro-82M 0.1B 2.49% Without voice cloning huggingface.co Official leaderboard
12 k2-fsa/OmniVoice 0.8B 2.72% Without voice cloning huggingface.co Official leaderboard
13 vibevoice/VibeVoice-7B-hf 9.3B 2.84% Voice cloning huggingface.co Official leaderboard
14 IndexTeam/IndexTTS-2.5 1.7B 2.94% Without voice cloning huggingface.co Official leaderboard
15 neuphonic/neutts-nano 1.1B 3.04% Without voice cloning huggingface.co Official leaderboard
16 neuphonic/neutts-nano-q8-gguf 0.8B 3.04% Without voice cloning huggingface.co Official leaderboard
17 Supertone/supertonic-3 0.1B 3.06% Without voice cloning huggingface.co Official leaderboard
18 fishaudio/s2-pro 5B 3.16% Without voice cloning huggingface.co Official leaderboard
19 bosonai/higgs-tts-3-4b 4.7B 3.23% Voice cloning huggingface.co Official leaderboard
20 Qwen3-TTS-12Hz-1.7B-Base 3.38% Voice cloning Jan 2026 huggingface.co Official leaderboard
21 vibevoice/VibeVoice-1.5B-hf 2.7B 3.40% Voice cloning huggingface.co Official leaderboard
22 sesame/csm-1b 1.8B 3.40% Without voice cloning huggingface.co Official leaderboard
23 fishaudio/drama-3 NEW 35B 3.41% Without voice cloning huggingface.co Official leaderboard
24 Qwen3-TTS-12Hz-1.7B-CustomVoice 3.58% Without voice cloning Jan 2026 huggingface.co Official leaderboard
25 OpenMOSS-Team/MOSS-TTS 8.5B 3.77% Without voice cloning huggingface.co Official leaderboard
26 nvidia/magpie_tts_multilingual_357m 0.3B 3.80% Without voice cloning huggingface.co Official leaderboard
27 openbmb/VoxCPM2 2.4B 4.14% Voice cloning huggingface.co Official leaderboard
28 owensong/Inflect-Nano-v1 0B 4.18% Without voice cloning huggingface.co Official leaderboard
29 bosonai/higgs-tts-2-3b-base 5.4B 4.67% Without voice cloning huggingface.co Official leaderboard
30 kyutai/pocket-tts 0.1B 6.63% Without voice cloning huggingface.co Official leaderboard
31 ResembleAI/chatterbox (ChatterboxMultilingualTTS) 0.8B 6.66% Voice cloning huggingface.co Official leaderboard
32 facebook/seamless-m4t-v2-large 1.7B 6.90% Without voice cloning huggingface.co Official leaderboard
33 FunAudioLLM/Fun-CosyVoice3-0.5B-2512 (RL) 0.1B 7.48% Without voice cloning huggingface.co Official leaderboard
34 mistralai/Voxtral-4B-TTS-2603 4B 8.20% Without voice cloning huggingface.co Official leaderboard
35 Qwen3-Omni-30B-A3B 9.84% Without voice cloning Sep 2025 huggingface.co Official leaderboard
36 microsoft/VibeVoice-Realtime-0.5B 1.4B 12.9% Without voice cloning huggingface.co Official leaderboard
37 FunAudioLLM/Fun-CosyVoice3-0.5B-2512 0.1B 17.4% Without voice cloning huggingface.co Official leaderboard
#ModelScoreSettingDateLinkSource
1 Qwen3-TTS-12Hz-1.7B-Base 3.38% Voice cloning Jan 2026 huggingface.co Official leaderboard
2 Qwen3-TTS-12Hz-1.7B-CustomVoice 3.58% Without voice cloning Jan 2026 huggingface.co Official leaderboard
3 Qwen3-Omni-30B-A3B 9.84% Without voice cloning Sep 2025 huggingface.co Official leaderboard

Rankings by part

We read 9 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as 🇬🇧 WER, 🇨🇳 CER and 🇫🇷 WER. Qwen3-TTS-12Hz-1.7B-Base leads 6 of them, and Qwen3-TTS-12Hz-1.7B-CustomVoice leads the other 3. Arrows show how far a model moved from its overall rank among the models we track.

🇬🇧 WER

3 models we track have a score here.

  1. 1 Qwen3-TTS-12Hz-1.7B-CustomVoice ↑1 1.17%
  2. 2 Qwen3-TTS-12Hz-1.7B-Base ↓1 1.29%
  3. 3 Qwen3-Omni-30B-A3B 1.98%

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

Qwen
Own leaderboard Replaced by 1.3.0, Oct 2026

Source: Official Open TTS Leaderboard 1.2.0 leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% 2026 Qwen3-TTS-12Hz-1.7B-Base Qwen3-Omni-30B-A3B 0% 20% 40% 60% 80% 100% 2026 Qwen3-TTS-12Hz-1.7B-Base Qwen3-Omni-30B-A3B

Source: Official Open TTS Leaderboard 1.2.0 leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

Versions

More in Audio and video

All 695 in Audio and video
  • Video-MMMU In use 300 expert-level lecture videos with questions testing perception, comprehension and adaptation of knowledge. Accuracy. Qwen 3.8 Max 88.7%
  • LVBench In use Hours-long video understanding with multiple-choice questions on reasoning, grounding and summarization. Gemini 4 Argon 91.7%
  • MMVU In use 3,000 expert-annotated questions on videos across 27 disciplines. Multiple-choice and open-ended accuracy. Qwen 3.8 Max 82.4%
  • MLVU In use Multi-task long video understanding (minutes to hours) with multiple-choice and generation tasks. Accuracy. Qwen 3.8 Max 90.8%
  • FLEURS In use Models transcribe read speech in 102 languages built from FLoRes-101 sentences. Model cards usually report word error rate, so lower scores are better. Phi-4-multimodal 4.00%
  • VoiceBench In use Spoken instructions with speaker, environment and content variation. Voice assistant responses scored by GPT judge or rules. Qwen3.5-Omni-Plus 93.1%

Data audit

All 8 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 2 pass 3 notes
  • Pass Independent runs Scores from the official leaderboard (37 entries).
  • Pass Models scored 3 models we track, plus 34 other entries on its leaderboard.
  • Note Newest score The newest came 10 months ago, in January 2026, for Qwen3-TTS-12Hz-1.7B-Base. That’s expected for a test replaced by Open TTS Leaderboard 1.3.0.
  • Note Frontier models None of its models are on the capabilities index.
  • Note Headroom Lower is better, so there’s no ceiling to reach.
Leaderboard import 3 pass
  • Pass Importer We read the board on huggingface.co: 3 rows for models we track and 34 other entries at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check