Generalized-Speech-Animation

USC CS621 Course Project

Implementation of "A Deep Learning Approach for Generalized Speech Animation" based on tensorflow
Use the Penn Phonetics Lab Forced Aligner (P2FA) to extract phonemetic labels
Use the flame-fitting method for 2d-to-3d mapping ("FLAME, a lightweight and expressive generic face model")
Use menpo to train the AAM model; this part is under the utilities folder

Requirements:

• tensorflow
• menpo

Below is a simple demo video: