Video-to-Audio Generation on AVSync15
701.21FVDLTX-2
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| LTX-2Setting=Unified (larger/speech)2025.10 | 701.21 | 10.92 | 30.37 | 32.13 | 28.78 | 25.8 | |
| BridgeDiTSetting=Default2025.10 | 765.74 | 5.34 | 28.52 | 35.95 | 34.59 | 27.5 | |
| OviSetting=Unified (larger/speech)2025.10 | 790.94 | 5.58 | 26.38 | 34.12 | 31.93 | 25.3 | |
| BridgeDiT (VGG-SS, OOD)Setting=OOD, Training dataset=VGGSound-SS2025.10 | 826.5 | 12.95 | 28.34 | 34.47 | 26.27 | 23.9 | |
| Wan + ThinkSoundSetting=Cascaded V2A2025.10 | 828.33 | 17.31 | 28.12 | 30.92 | 20.7 | 16.9 | |
| Wan + PrismAudioSetting=Cascaded V2A2025.10 | 828.33 | 9.79 | 28.12 | 33.53 | 25.25 | 23.3 | |
| SANA + AudioLDM (DCA)Setting=Cross-Backbone, DCA=Enabled2025.10 | 906.29 | 9.46 | 28.46 | 32.67 | 31.85 | 25.6 | |
| SANA + AudioLDM (separate)Setting=Cross-Backbone2025.10 | 1,033.39 | 17.09 | 27.95 | 29.66 | 23.39 | 19.3 |