The repository of the paper SelfTalk: A Self-Supervised Commutative Training Diagram to Comprehend 3D Talking Faces ACM MM 2023
The proposed SelfTalk utilizes a cross-modal network system to generate coherent and visually comprehensible 3D talking faces by reducing the domain gap between different modalities.
We released our code in the Psyche AI's official repository, please visit SelfTalk_release.