GeneFace: Generalized and High-Fidelity Audio-Driven 3D Talking Face Synthesis

TL;DR

GeneFace is a novel NeRF-based method that produces highly realistic 3D talking face videos from diverse audio inputs, overcoming previous generalization limitations.

Contribution

It introduces a variational motion generator trained on large lip-reading data and a domain-adaptive post-net for improved out-of-domain audio synthesis.

Findings

Outperforms previous methods in realism and generalization

Achieves high-fidelity results with diverse audio inputs

Effectively handles head-torso separation in 3D face synthesis

Abstract

Generating photo-realistic video portrait with arbitrary speech audio is a crucial problem in film-making and virtual reality. Recently, several works explore the usage of neural radiance field in this task to improve 3D realness and image fidelity. However, the generalizability of previous NeRF-based methods to out-of-domain audio is limited by the small scale of training data. In this work, we propose GeneFace, a generalized and high-fidelity NeRF-based talking face generation method, which can generate natural results corresponding to various out-of-domain audio. Specifically, we learn a variaitional motion generator on a large lip-reading corpus, and introduce a domain adaptative post-net to calibrate the result. Moreover, we learn a NeRF-based renderer conditioned on the predicted facial motion. A head-aware torso-NeRF is proposed to eliminate the head-torso separation problem. Extensive experiments show that our method achieves more generalized and high-fidelity talking face generation compared to previous methods.