![]() ![]() Obtained from a diffusion autoencoder (DAE). Method called DAE-Talker that leverages data-driven latent representations To address these limitations, we propose a novel Additionally, these methods require an external pretrained model forĮxtracting these representations, whose performance sets an upper bound on ![]() Representations like facial landmarks and 3DMM coefficients, which are designedīased on human knowledge and are insufficient to precisely describe facial One reason for this is the use of handcrafted intermediate Download a PDF of the paper titled DAE-Talker: High Fidelity Speech-Driven Talking Face Generation with Diffusion Autoencoder, by Chenpng Du and 7 other authors Download PDF Abstract: While recent research has made significant progress in speech-driven talkingįace generation, the quality of the generated video still lags behind that of
0 Comments
Leave a Reply. |