# Decoupled facial representation

JoyVASA is an audio-driven facial animation generation method based on diffusion models, capable of generating facial dynamics and head movements with support for multilingual input.