Zero-shot Voice Cloning on Seed-TTS-Eval zh-hard (test)
5.83WERCosyVoice 3
Evaluation results
| Method | Links | ||
|---|---|---|---|
| CosyVoice 3Params=1.5B2026.06 | 5.83 | 75.8 | |
| dots.tts (Pretrain)Params=2B, Variant=Pretrain2026.06 | 6.46 | 79.2 | |
| dots.tts (SOAR)Params=2B, Variant=SOAR2026.06 | 6.6 | 79.5 | |
| dots.tts (MF, NFE=4)Params=2B, Distillation Method=MeanFlow, NFE=42026.06 | 6.6 | 78.5 | |
| Qwen3-TTSParams=1.7B2026.06 | 6.76 | 74.8 | |
| IndexTTS 2Params=1.5B2026.06 | 7.12 | 75.5 | |
| dots.tts (MF, NFE=3)Params=2B, Distillation Method=MeanFlow, NFE=32026.06 | 7.19 | 78.6 | |
| Seed-TTS2026.06 | 7.59 | 77.6 | |
| dots.tts (MF, NFE=2)Params=2B, Distillation Method=MeanFlow, NFE=22026.06 | 7.74 | 76.7 | |
| VoxCPM 2Params=2B2026.06 | 8.13 | 75.3 | |
| F5-TTSParams=0.3B2026.06 | 8.67 | 71.3 | |
| FireRedTTS 2Params=1.5B2026.06 | 8.98 | 70.3 |