AnyTalk generates 3D speech animations for arbitrary characters

AnyTalk generates 3D speech animations for arbitrary characters without requiring any animation data, using a video generation model. It analyzes vocal data to drive synchronized articulation and infer emotional expression from speech tone.