Voice Conversion on ELHE HE portion (160 unmodified utterances)
2.9CERGT
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| GTDescription=Ground-truth healthy speech2026.01 | 2.9 | 4 | 3.48 | 4.11 | 3.2 | 0.89 | — | |
| Ours (adapted StreamVC)Guided loss=WavLM2026.01 | 40.9 | 3.26 | 3.32 | 3.93 | 2.94 | 0.84 | 0.34 | |
| Ours (adapted StreamVC)Guided loss=WavLM + HF2026.01 | 41.9 | 3.76 | 3.43 | 4 | 3.09 | 0.87 | 0.34 | |
| Ours (adapted StreamVC)Guided loss=WEO + HF2026.01 | 44.9 | 3.69 | 3.39 | 4.02 | 3.05 | 0.86 | 0.34 | |
| Ours (adapted StreamVC)Guided loss=WavLM + HF + F02026.01 | 46.7 | 3.7 | 3.42 | 3.97 | 3.06 | 0.87 | 0.35 | |
| Ours (adapted StreamVC)Guided loss=WEO + WavLM + HF2026.01 | 47.1 | 3.7 | 3.38 | 3.98 | 3.02 | 0.87 | 0.34 | |
| Ours (adapted StreamVC)Guided loss=none2026.01 | 53.7 | 3.17 | 3.29 | 3.88 | 2.9 | 0.86 | 0.35 | |
| Ours (adapted StreamVC)Guided loss=BNF + HF2026.01 | 55.41 | 3.82 | 3.42 | 3.98 | 3.07 | 0.88 | 0.35 | |
| XVCModel type=Offline SOTA VC2026.01 | 61.2 | 3.59 | 3.32 | 4.02 | 3 | 0.63 | 0.37 | |
| ELDescription=Electrolaryngeal speech source2026.01 | 88.2 | -0.28 | 3.14 | 3.12 | 2.41 | 0.55 | 0.62 | |
| FreeVCModel type=Offline SOTA VC2026.01 | 140.3 | 3.52 | 3.27 | 3.99 | 2.91 | 0.71 | 0.4 |