Emotional Speech Synthesis on ESD English (test)
78.39Score (Neutral)EmoShift
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| EmoShiftnumber of parameters=10 M, alpha=1, variant=Default2026.01 | 78.39 | 88.06 | 65.48 | 65.48 | 73.87 | 74.26 | |
| EmoShiftnumber of parameters=10 M, alpha=3, variant=Best2026.01 | 77.1 | 87.1 | 61.61 | 72.26 | 81.61 | 75.94 | |
| CosyVoice-SFT-Shiftnumber of parameters=321 M, components=EmoSteer layer + fine-tuning2026.01 | 74.84 | 88.39 | 65.81 | 62.26 | 73.23 | 72.91 | |
| CosyVoicenumber of parameters=0 M, version=backbone2026.01 | 74.19 | 86.45 | 61.61 | 61.61 | 64.52 | 69.68 | |
| CosyVoice-SFTnumber of parameters=311 M, tuning=fully fine-tuned2026.01 | 70 | 89.68 | 60.32 | 54.19 | 74.52 | 69.74 |