1 paper
Aaron Mir, Eduardo Alonso, Esther Mondragón
We propose a novel talking head synthesis pipeline called "DiT-Head", which is based on diffusion transformers and uses audio as a condition to drive the denoising process of a dif…