BREAKING
NVIDIA Open-Sources Audio2Face-3D
From Audio to Facial Motion
1
Audio input
↓
2
Infer emotion
↓
3
Lip-sync
↓
4
3D face output
0
fps
Real-time speed
0
Model versions
Two Models, Open Tooling
Models
Hugging Face
●
Diffusion v3.0
●
Regression v2.3
●
Audio2Emotion
Tooling
MIT / Apache
●
Maya & Unreal 5
●
Training framework
●
NIM containers
Part of NVIDIA ACE
A Building Block for Digital Humans
AI NEWS BLITZ
NVIDIA has open-sourced Audio2Face-3D, turning audio into real-time 3D facial animation.