VASA-1: Lifelike Audio-Driven Talking Faces

Single portrait photo + speech audio = hyper-realistic talking face video with precise lip-audio sync, lifelike facial behavior, and naturalistic head movements, generated in real time.