From Tokens to Faces: Investigating Discrete Speech Representations for 3D Facial Animation
Researchers have explored how different speech representations impact the quality of 3D facial animation. The study compared four families of speech representations, evaluating their effectiveness with two facial decoders using both objective and perceptual measures. Findings indicate that encoding phonetic classes within speech representations leads to more accurate facial animation predictions. AI
IMPACT This research could lead to more realistic and accurate AI-driven facial animation systems by optimizing the use of speech data.