Text-to-Speech on CV3-Eval Cross-lingual zh → en
4.32WERCosyVoice 3
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CosyVoice 3Number of Parameters=1.5B2026.06 | 4.32 | 66.4 | |
| dots.ttsTraining Stage=Pretrain2026.06 | 4.97 | 71.9 | |
| dots.ttsTraining Stage=MF, NFE (Number of Function Evaluations)=42026.06 | 5.24 | 70.9 | |
| dots.ttsTraining Stage=SOAR2026.06 | 5.66 | 72.8 | |
| CosyVoice 22026.06 | 6.47 | 64.3 |