Video-to-Audio Generation on VGGSound 10 (test)
5.32FADMMAudio
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| MMAudioTrain Data=180K, Configuration=V2A2026.03 | 5.32 | 1.64 | 17.18 | 0.33 | |
| CAFATrain Data=200K, Configuration=TV2A2026.03 | 12.6 | 2.02 | 13.45 | 0.21 | |
| ReWaSTrain Data=160K, Configuration=TV2A2026.03 | 14.71 | 2.69 | 8.45 | 0.15 | |
| AVControlTrain Data=7.8K, Configuration=joint audio and video via single LoRA2026.03 | 57.25 | 7.74 | 34.51 | 0.13 |