Audio-driven face reenactment on YouTube face videos (test)
7.653SyncNet ScoreGround Truth
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Ground Truth2022.11 | 7.653 | — | — | — | |
| Personalized Generative Prior (Ours Audio)training duration=20 seconds video clip, input=audio features2022.11 | 4.866 | 32.57 | 0.957 | 0.035 | |
| DFRFtraining duration=20 seconds video clip, mode=few-shot2022.11 | 4.142 | 30.23 | 0.939 | 0.042 | |
| AD-NeRFtraining duration=20 seconds video clip, input=audio features2022.11 | 1.238 | 29.69 | 0.934 | 0.057 |