Zero-shot Text-to-Speech on YoBind (test)
0.764SECS (s)Fish-Speech
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Fish-SpeechTraining Data=Real 10% + Synth 90%, DC=false, OS=false2026.03 | 0.764 | 4.081 | 9.358 | |
| Fish-SpeechTraining Data=Real 10% + Synth 90%, DC=false, OS=true2026.03 | 0.77 | 4.085 | 9.542 | |
| CosyVoice 2Training Data=Real 10% + Synth 90%, DC=false, OS=false2026.03 | 0.774 | 4.045 | 9.065 | |
| CosyVoice 2Training Data=Real 10% + Synth 90%, DC=false, OS=true2026.03 | 0.782 | 4.044 | 9.377 | |
| Fish-SpeechTraining Data=Real 10% + Synth 90%, DC=true, OS=false2026.03 | 0.792 | 4.471 | 9.965 | |
| Fish-SpeechTraining Data=Real 10% + Synth 90% (+Extra Synth), DC=true, OS=true2026.03 | 0.794 | 3.912 | 8.951 | |
| CosyVoice 2Training Data=Real 10% + Synth 90%, DC=true, OS=false2026.03 | 0.795 | 4.279 | 9.531 | |
| Fish-SpeechTraining Data=Real 10% + Synth 90%, DC=true, OS=true2026.03 | 0.799 | 4.174 | 9.553 | |
| CosyVoice 2Training Data=Real 10% + Synth 90%, DC=true, OS=true2026.03 | 0.804 | 4.023 | 9.006 | |
| CosyVoice 2Training Data=Real 10% + Synth 90% (+Extra Synth), DC=true, OS=true2026.03 | 0.805 | 3.684 | 8.364 | |
| Real-Only TrainingTraining Data=Real 10%, DC=false, OS=false2026.03 | 0.813 | 5.063 | 11.035 | |
| Real-Only TrainingTraining Data=Real 100%, DC=false, OS=false2026.03 | 0.84 | 4.743 | 10.424 |