Audio Driven Talking Head Generation on HDTF 51 (test)
1SSIMGround Truth
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Ground Truth2025.04 | 1 | — | 0 | 0 | 8.25 | 6.87 | |
| Wav2Lipconditional_input=ground truth upper half of the face2025.04 | 0.937 | 34.31 | 18.05 | 3.39 | 8.84 | 6.48 | |
| IP-LAPconditional_input=ground truth upper half of the face2025.04 | 0.874 | 31.67 | 26.62 | 3.3 | 7.07 | 7.89 | |
| MGGTalk2025.04 | 0.731 | 30.92 | 18.73 | 3.75 | 7.68 | 6.91 | |
| Real3DPortrait2025.04 | 0.689 | 30.66 | 39.1 | 3.61 | 6.57 | 8.01 | |
| NeRFFaceSpeech2025.04 | 0.676 | 28.52 | 37.89 | 3.73 | 6.35 | 8.57 | |
| EchoMimic2025.04 | 0.655 | 29.65 | 22.33 | 3.79 | 6.26 | 8.77 | |
| AniTalker2025.04 | 0.62 | 30.42 | 25.74 | 3.87 | 6.12 | 8.23 | |
| SadTalker2025.04 | 0.567 | 28.6 | 32.77 | 4.07 | 7.63 | 7.42 |