Text and Reference Speech Synthesis on First evaluation set ZH (Hard)
0.89WERCosyVoice2
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| CosyVoice22026.01 | 0.89 | 3.58 | 0.36 | |
| Ground-truth2026.01 | 4.55 | 3.92 | 0 | |
| FlexiVoice-Base2026.01 | 5.63 | 3.66 | -0.88 | |
| FlexiVoice2026.01 | 7.02 | 3.76 | 0.57 | |
| VoxInstruct2026.01 | 9.4 | 3.04 | -1.62 |