Identity-Preserving Talking Head Generation on CelebV-HQ easy (same-video)
0.573Spk SimID-LoRA (Ours)
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| ID-LoRA (Ours)Video Backbone=LTX-22026.03 | 0.573 | 0.87 | 8.2 | 4.43 | 0.372 | 10.6 | |
| CosyVoice 3.0 + WAN2.2Video Backbone=WAN2.22026.03 | 0.51 | 0.886 | 11.27 | 1.53 | 0.301 | 27.3 | |
| Kling 2.6 ProCommercial Model=true2026.03 | 0.487 | 0.847 | 10.24 | 3.01 | 0.327 | 4.6 | |
| ElevenLabs + WAN2.2Video Backbone=WAN2.22026.03 | 0.413 | 0.879 | 11.98 | 1.51 | 0.266 | 18 | |
| VoiceCraft + WAN2.2Video Backbone=WAN2.22026.03 | 0.412 | 0.879 | 10.99 | 1.48 | 0.335 | 24.1 |