Human-Object Interaction Video Generation on GroundedInter Audio-Driven 1.0 (test)
0.2732VLM-QAInteractAvatar
Evaluation results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| InteractAvatarInput Signal=Audio-Driven, Inference Mode=TA2V2026.02 | 0.2732 | 0.931 | 0.133 | 0.765 | 0.803 | 0.671 | 0.857 | 0.286 | 0.899 | 0.945 | 6.04 | |
| Hunyuan-AvatarInput Signal=Audio-Driven2026.02 | 0.2688 | 0.745 | 0.112 | 0.321 | 0.666 | 0.661 | 0.841 | 0.285 | 0.873 | 0.934 | 5.98 | |
| OminiAvatarInput Signal=Audio-Driven2026.02 | 0.2676 | 0.732 | 0.118 | 0.305 | 0.732 | 0.639 | 0.847 | 0.282 | 0.885 | 0.94 | 5.95 | |
| Fantasy-TalkingInput Signal=Audio-Driven2026.02 | 0.2535 | 0.416 | 0.098 | 0.127 | 0.593 | 0.673 | 0.855 | 0.276 | 0.889 | 0.941 | 5.79 | |
| Wan-S2VInput Signal=Audio-Driven2026.02 | 0.2465 | 0.336 | 0.063 | 0.095 | 0.619 | 0.67 | 0.87 | 0.281 | 0.885 | 0.955 | 5.43 |