About this document

Speech-Driven Facial Animation Algorithm by ryl0903 is a document available to read on EtoBox.

The document presents a new algorithm for generating lip-synchronized facial animation from vocal audio using spectral-dimensional bidirectional LSTM and temporal attention. This allows efficient speech-driven facial animation without pre-trained speech recognition. Deformation gradients are also introduced as the motion representation, allowing better synthesis of nuanced local motions and easier generalization to different face topologies.

Author

ryl0903

Language

EN