Zero-shot Voice Cloning on CV3-Eval (Multilingual)
4.08CV3-Eval Score (zh)CosyVoice 2
Evaluation Results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| CosyVoice 2mode=zero-shot cloning2026.06 | 4.08 | 6.32 | 12.58 | 11.96 | 9.13 | 19.7 | — | — | — | — | — | |
| CosyVoice 3-1.5Bmode=zero-shot cloning, parameters=1.5B2026.06 | 3.91 | 4.99 | 9.77 | 10.55 | 7.57 | 5.69 | 6.43 | 4.47 | 11.8 | 10.5 | 6.64 | |
| VoxCPM2mode=zero-shot cloning, inference recipe=reference + continuation2026.06 | 3.65 | 5 | 8.55 | 8.48 | 5.96 | 5.69 | 4.77 | 3.8 | 9.85 | 4.25 | 5.21 | |
| Fish Audio S2mode=zero-shot cloning2026.06 | 2.65 | 2.43 | 9.1 | 4.4 | 3.96 | 2.76 | 2.22 | 2 | 6.26 | 2.04 | 2.78 |