whatmodel.sh / tts / Qwen3-TTS 12Hz 0.6B Base
Qwen3-TTS 12Hz 0.6B Base
↗ share
+ contribute
Qwen3-TTS 12Hz 0.6B Base is an open-weights text-to-speech model from Alibaba (Qwen), released January 2026. Its published size is 0.6B parameters. Licence: apache-2.0.
Catalogue identity notes
Qwen/Qwen3-TTS-12Hz-0.6B-Base, the 0.6B reference-audio cloning checkpoint. Distinct from the 1.7B Base (qwen3-tts-12hz-1-7b-base) and the CustomVoice/VoiceDesign checkpoints.
Score within the Vapi Humanness · Bradley–Terry rating, 2026-09-25 snapshot comparison set only.
$ how it compares
The text-to-speech models ranked either side of Qwen3-TTS 12Hz 0.6B Base, out of 148 with a score, then more from Alibaba (Qwen).
| # | model | score | where | ||
|---|---|---|---|---|---|
| 42 |  ZipVoice | 62±3 | local | ||
| 43 | Ghana TTS 72k | 62±7 | comparison only | ||
| 44 | Qwen3-TTS 12Hz 0.6B Base | 62±2 | local | ||
| 45 |  Eleven Flash v2 | 62±3 | api $50.00/M chars | ||
| 46 |  DMOSpeech 2 | 62±4 | local | ||
| more from Alibaba (Qwen) | |||||
| 25 | Qwen3-TTS 12Hz 1.7B Base | 65±1 | local | ||
| 98 | Qwen3-TTS 12Hz 0.6B CustomVoice | 46±6 | comparison only | ||
see all text-to-speech models · find the ones that fit your machine
$ what it's good at
voice cloning▁·▁▁▁▁▁▁▇█▃▃ #16 of 37 · 43% of its score · 6 tests
Beats Gemini 3.1 Flash TTS Preview, Fun-CosyVoice3-0.5B-2512, IndexTTS-2 here, though they rank higher overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| Seed-TTS test-en · speaker similarity (WavLM-large SV) | 70.5% ✎ | █·█·██·▄·▄·▄ beats 6/10 | 14% |
| Seed-TTS test-zh · speaker similarity (WavLM-large SV) | 76.5% ✎ | ▃··▃█▅▅··▃·▃ beats 5/10 | 13% |
| Seed-TTS test-en · Whisper large-v3 WER +1 | 1.7% ✎ | ▃··▃▅··▅·█▃▅ beats 9/11 | 5.6% |
| Seed-TTS test-en WER (transcriber not stated) lab | 1.3% ✎ | ▄········█▄█ beats 4/5 | 5.4% |
| Seed-TTS test-zh · Paraformer CER | 1.0% ✎ | ▂··▄··▂▂█·▄▂ beats 8/11 | 4.5% |
| Seed-TTS test-zh error rate (transcriber not stated) lab | 0.9% ✎ | █··········█ beats 0/1 | — |
other tests5 not in a task · 56% of its score
| test | result | vs the others in its run | counts |
|---|---|---|---|
| CV3-Eval hard-en · WER | 10.7% ✎ | █··██·██████ beats 2/8 | 17% |
| CV3-Eval ZH (zero-shot Chinese) · CER | 3.4% ✎ | █▄···█▄·█··▄ beats 6/8 | 14% |
| CV3-Eval hard-zh · CER | 10.7% ✎ | ▃▃·▃··▃▃█··▃ beats 1/8 | 11% |
| CV3-Eval hard-en · speaker similarity (ERes2Net SV) | 67.0% ✎ | ▄█▄·▄··▄··▄█ beats 3/8 | 11% |
| CV3-Eval EN (zero-shot English) · WER | 33.9% ✎ | ▃·······▃▅█▅ beats 0/8 | 4.4% |
$ where to run it
Other ways to run it
version
- Run it as published · needs ~2.52 GB · from huggingface.co/Qwen (official)
uv run --python 3.12 --with qwen-tts python -c "import torch, soundfile as sf; from qwen_tts import Qwen3TTSModel; model = Qwen3TTSModel.from_pretrained('Qwen/Qwen3-TTS-12Hz-0.6B-Base', device_map='cpu', dtype=torch.bfloat16); wavs, sr = model.generate_voice_clone(text='Hello, world.', language='English', ref_audio='https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen3-TTS-Repo/clone.wav', ref_text='Okay. Yeah. I resent you. I love you. I respect you. But you know what? You blew it! And thanks to you.'); sf.write('output_voice_clone.wav', wavs[0], sr)"
Official weights and documentation: huggingface.co/Qwen ↗
$ how the score was worked out · UltraEval-Audio (OpenBMB) · T… 93%, Qwen/Qwen3-TTS-12Hz-1.7B-Base… 5%, SGLang-Omni serving cookbooks… 1%
Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works
| source | results | share of the score |
|---|---|---|
 UltraEval-Audio (OpenBMB) · TTS replication results | 9 | 93% |
 Qwen/Qwen3-TTS-12Hz-1.7B-Base model card · lab-reported | 1 | 5% |
| SGLang-Omni serving cookbooks (SGLang-Omni team measurements) | 1 | 1% |
| prior · pull toward the middle | 1% | |
| weighted average | = 61.7 |
not counted (1)
| Qwen/Qwen3-TTS-12Hz-1.7B-Base model card | Seed-TTS test-zh error rate (transcriber not stated) [correct] | 0.9% | fewer than three models have this test yet, or this source ran it for this model alone |
$ improve this page
Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.
data updated · sources and licences · json