Emotion Voice Conversion on Emotion Speech Dataset (ESD) (test)
3.61UT-MOSv2 ScoreVEVO
Evaluation Results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| VEVOControl Input=Speech Reference, Training Data(h)=101k2026.07 | 3.61 | — | — | — | — | 8.48 | — | 32.42 | 0.52 | 0.64 | 3.87 | |
| TRACE-EVCControl Input=Natural Language Prompt, Training Data(h)=11.92026.07 | 3.57 | — | — | — | — | 9.96 | — | 93 | 0.82 | 0.68 | 3.98 | |
| StepAudioEditXiteration=22026.03 | 3.395 | 0.425 | 6.5 | 3.308 | 3.005 | 4.43 | 0.634 | 20.9 | — | — | — | |
| TRACE-EVC w/o Emo-CompassControl Input=Natural Language Prompt, Training Data(h)=11.92026.07 | 3.38 | — | — | — | — | 10.9 | — | 43.08 | 0.65 | 0.66 | — | |
| SGEVCControl Input=Emotion Label, Training Data(h)=11.92026.07 | 3.23 | — | — | — | — | 9.81 | — | 80.44 | 0.8 | 0.6 | 3.84 | |
| DurFlex-EVCControl Input=Emotion Label, Training Data(h)=11.92026.07 | 3.17 | — | — | — | — | 10.05 | — | 89.5 | 0.81 | 0.56 | 3.67 | |
| ZESTControl Input=Speech Reference, Training Data(h)=11.92026.07 | 3.16 | — | — | — | — | 8.09 | — | 78.58 | 0.79 | 0.55 | 3.36 | |
| VEVO2026.03 | 3.06 | 0.476 | 5.4 | 3.409 | 3.093 | 10.08 | 0.696 | 42.4 | — | — | — | |
| ProposedDeep-Prefix=true2026.03 | 2.96 | 0.5 | 4.5 | 3.431 | 3.152 | 6.28 | 0.85 | 85.5 | — | — | — | |
| ProposedDeep-Prefix=false2026.03 | 2.946 | 0.503 | 4.42 | 3.416 | 3.133 | 5.74 | 0.842 | 83.5 | — | — | — | |
| StarGANv2-VC-EVC2026.03 | 2.318 | 0.565 | 3.24 | 3.33 | 3.037 | 5.87 | 0.686 | 36 | — | — | — | |
| GenVC2026.03 | 2.157 | 0.238 | 20.87 | 3.328 | 3.021 | 8.06 | 0.681 | 32.48 | — | — | — |