Text-to-Speech on WordVoice 5A-en (test)
0.0188Duration MAEGround-Truth
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Ground-Truth2026.07 | 0.0188 | — | 0.0091 | 0.0081 | 11.58 | 7.01 | |
| Mel-Recon2026.07 | 0.0393 | 0.34 | 0.0247 | 0.0365 | 21.39 | 16.2 | |
| WordVoice-ControlMode=Manual Control2026.07 | 0.045 | 1.57 | 0.0475 | 0.0782 | 23.23 | 26.58 | |
| WordVoice-FreeMode=Free2026.07 | 0.0696 | 1.31 | 0.085 | 0.1568 | 42.37 | 39.98 | |
| CosyVoice32026.07 | 0.0806 | 1.06 | 0.0899 | 0.1765 | 43.81 | 40.92 |