Identity-Preserving Talking Face Generation With Landmark and Appearance Priors | Read Paper on Bytez