| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Video Reconstruction | VoxCeleb2 | SSIM0.996 | 27 | |
| Talking Face Generation | VoxCeleb2 (test) | SSIM1 | 14 | |
| Face Verification | VoxCeleb2 (test) | VR @ FAR=1e-296.1 | 7 | |
| Retrieval | VoxCeleb2 | Rec@1099.43 | 6 | |
| Audio-driven talking face generation | VoxCeleb2 | Sc8.841 | 6 | |
| Speaker Identification | VoxCeleb2 40 x 30 (held-out) | Top-1 Accuracy71.25 | 1 |