Text-to-Speech on WordVoice zh 5A (test)
0.0142Dur-MAEGround-Truth
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Ground-Truth2026.07 | 0.0142 | — | 0.0116 | 0.0094 | 5.2 | 5.85 | |
| Mel-Recon2026.07 | 0.0257 | 1.19 | 0.0266 | 0.0552 | 13.98 | 13.9 | |
| WordVoice-ControlMode=Manual Control2026.07 | 0.0349 | 2.86 | 0.0486 | 0.11 | 12.72 | 20.25 | |
| WordVoice-FreeMode=Free2026.07 | 0.05 | 2.58 | 0.0963 | 0.1855 | 31.33 | 30.97 | |
| CosyVoice32026.07 | 0.0549 | 2.31 | 0.103 | 0.203 | 32.47 | 32.31 |