Video-to-Audio (V2A) on VGGSound
1.4KL DivergenceVATT
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| VATTTask=V2A2026.06 | 1.4 | 10.02 | 11.71 | 2.55 | 3.64 | 5.85 | 0.25 | |
| MMAudioTask=V2A2026.06 | 1.97 | 14.95 | 6.18 | 2.04 | 3.38 | 5.91 | 0.35 | |
| AudioX-BaseTask=V2A2026.06 | 1.98 | 12.35 | 6.94 | 2 | 3.75 | 6.31 | 0.28 | |
| AudioX-TurboTask=V2A2026.06 | 1.99 | 12.39 | 7.88 | 1.34 | 3.78 | 6.23 | 0.29 | |
| VABTask=V2A2026.06 | 2.3 | 8.15 | 20.21 | 3.05 | 3.52 | 5.93 | 0.24 | |
| FoleyCrafterTask=V2A2026.06 | 2.39 | 8.7 | 17.68 | 2.23 | 3.31 | 5.99 | 0.27 | |
| Seeing&HearingTask=V2A2026.06 | 2.58 | 5.15 | 27.21 | 5.23 | 3.42 | 5.33 | 0.36 | |
| FRIERENTask=V2A2026.06 | 2.58 | 6.91 | 50.88 | 3.13 | 2.98 | 6.06 | 0.2 | |
| Diff-FoleyTask=V2A2026.06 | 3.01 | 8.35 | 56.54 | 5.89 | 2.57 | 5.85 | 0.2 |