Text-to-Speech on VCTK
1.7WERGroundtruth
Evaluation results
| Method | Links | ||
|---|---|---|---|
| GroundtruthEvaluation Protocol=zero-shot2024.04 | 1.7 | — | |
| Ground Truthzero-shot=true2024.02 | 1.9 | 0.93 | |
| CSLM-SFT2026.04 | 2.7 | — | |
| CSLM-base2026.04 | 2.8 | — | |
| Ground Truthcalculation_method=whisper-large-v32026.04 | 3.5 | — | |
| CosyVoicemodel_version=CosyVoice-300M-SFT2026.04 | 3.7 | — | |
| SpeechGPT2026.04 | 5.7 | — | |
| USLMzero-shot=true2024.02 | 6.5 | 0.84 | |
| SpeechAlign-BoNEvaluation Protocol=zero-shot2024.04 | 7.5 | 0.79 | |
| VALL-Ezero-shot=true2024.02 | 7.9 | 0.75 | |
| SpeechAlign-DPO-Iter3Evaluation Protocol=zero-shot2024.04 | 7.9 | 0.83 | |
| SpeechAlign-DPO-Iter2Evaluation Protocol=zero-shot2024.04 | 8 | 0.83 | |
| AnyGPTzero-shot=true2024.02 | 8.5 | 0.77 | |
| SpeechAlign-RLHF-PPOEvaluation Protocol=zero-shot2024.04 | 8.5 | 0.8 | |
| SpeechAlign-DPO-Iter1Evaluation Protocol=zero-shot2024.04 | 8.5 | 0.82 | |
| AnyGPT2026.04 | 8.5 | — | |
| SpeechAlign-sftEvaluation Protocol=zero-shot2024.04 | 8.8 | 0.79 | |
| Continue SFTEvaluation Protocol=zero-shot2024.04 | 9.8 | 0.8 | |
| SpeechAlign-CoHEvaluation Protocol=zero-shot2024.04 | 10.2 | 0.81 |