Fetching the paper…

DiffTalker: Co-driven audio-image diffusion for talking faces via intermediate landmarks · Around