Gesture-to-Speech Synthesis on PATS Different Text (test)
1.9434Gesture OffsetGesture2Speech
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Gesture2Speechvariant=multimodal MoE2026.03 | 1.9434 | 0.0475 | 18.97 | 12.15 | 3.54 | 3.39 | 2.69 | |
| Gesture2Speechvariant=XTTS V22026.03 | 2.0554 | 0.0433 | 19.22 | 12.8 | 3.25 | 3.18 | 2.65 | |
| Gesture2Speechvariant=unimodal MoE2026.03 | 2.5915 | 0.0411 | 19.89 | 12.69 | 3.4 | 3.39 | 2.65 | |
| Gesture2Speechvariant=H-MoE2026.03 | 3.2073 | 0.0265 | 20.56 | 13.53 | 3.55 | 3.32 | 2.61 | |
| Gesture2Speechvariant=GPT-SoVITS2026.03 | 4.9933 | 0.0047 | 34.29 | 24.17 | 3.42 | 2.75 | 2.33 |