Voice Conversion on LibriSpeech (test-clean)
2.87WERGT (Vocoder)
Evaluation Results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| GT (Vocoder)type=ground truth mel synthesis2025.06 | 2.87 | 1.36 | — | — | — | — | 4.14 | 4.19 | — | — | |
| R-VCdiffusion scheme=CFM, NFE=102025.06 | 3.47 | 1.39 | — | — | — | 0.931 | 4.1 | 4.05 | 4.12 | 0.34 | |
| R-VCNFE=22025.06 | 3.51 | 1.4 | — | — | — | 0.93 | 4.1 | 4.03 | 4.11 | 0.12 | |
| HierSpeech++2025.06 | 3.64 | 1.46 | — | — | — | 0.907 | 4.09 | 3.98 | 3.93 | 0.25 | |
| Diff-HierVC2025.06 | 4.34 | 2.14 | — | — | — | 0.894 | 3.75 | 3.64 | 3.8 | 0.47 | |
| FACodec-VC2025.06 | 4.68 | 1.96 | — | — | — | 0.902 | 2.88 | 3.21 | 3.91 | 0.1 | |
| SEF-VC2025.06 | 4.82 | 2.04 | — | — | — | 0.897 | 3.96 | 3.8 | 3.85 | 0.18 | |
| CosyVoice-VC2025.06 | 5.95 | 2.7 | — | — | — | 0.933 | 4.09 | 3.95 | 4.11 | 1.74 | |
| Testset topline2023.05 | 5.96 | 2.38 | — | — | — | — | — | — | — | — | |
| kNN-VCvocoder=HiFi-GAN (prematched), target_data=8 minutes2023.05 | 7.36 | 2.96 | 37.15 | — | — | — | — | — | — | — | |
| FreeVCtarget_data=8 minutes2023.05 | 7.61 | 3.17 | 8.97 | — | — | — | — | — | — | — | |
| YourTTSmode=text-free, target_data=8 minutes2023.05 | 11.93 | 5.51 | 25.32 | — | — | — | — | — | — | — | |
| VQMIVCtarget_data=8 minutes2023.05 | 59.46 | 37.55 | 2.22 | — | — | — | — | — | — | — |