Audio-controlled video-to-audio generation on Greatest Hits
69.71OnsetSync APFoleyGenEx + AA
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| FoleyGenEx + AAAdverb Augmentation=true2026.06 | 69.71 | 0.54 | 91.28 | 72.5 | |
| FoleyGenEx2026.06 | 69.38 | 0.54 | 90.85 | 72.16 | |
| MMAudio-S3Surrogate video configuration=The actual reference video corresponding to the reference audio is prepended2026.06 | 68.92 | 2.16 | 85.68 | 51.95 | |
| MMAudio-S2Surrogate video configuration=The first two seconds of the target video are duplicated and prepended2026.06 | 68.72 | 3.05 | 83.6 | 51.32 | |
| MMAudio-S1Surrogate video configuration=A two-second blank (zero-filled) video is prepended2026.06 | 65.53 | 4.77 | 81.85 | 50.63 | |
| CondFoleyGen2026.06 | 60 | 0.65 | 89.99 | 73.85 |