Perceptually Accurate 3D Talking Head Generation: New Definitions, Speech-Mesh Representation, and Evaluation Metrics
Lee Chae-Yeon, Oh Hyun-Bin, Han EunGi, Kim Sung-Bin, Suekyeong Nam, Tae-Hyun Oh
2025-03-31
Summary
This paper is about making AI-generated 3D talking heads that move their lips more realistically in sync with speech.
What's the problem?
Current AI talking heads often have lip movements that don't quite match the nuances of the spoken words.
What's the solution?
The researchers developed a new way to represent the relationship between speech and lip movements, and used it to train AI models that produce more realistic and expressive lip sync.
Why it matters?
This work matters because it can make virtual characters and avatars more believable and engaging.
Abstract
Recent advancements in speech-driven 3D talking head generation have made significant progress in lip synchronization. However, existing models still struggle to capture the perceptual alignment between varying speech characteristics and corresponding lip movements. In this work, we claim that three criteria -- Temporal Synchronization, Lip Readability, and Expressiveness -- are crucial for achieving perceptually accurate lip movements. Motivated by our hypothesis that a desirable representation space exists to meet these three criteria, we introduce a speech-mesh synchronized representation that captures intricate correspondences between speech signals and 3D face meshes. We found that our learned representation exhibits desirable characteristics, and we plug it into existing models as a perceptual loss to better align lip movements to the given speech. In addition, we utilize this representation as a perceptual metric and introduce two other physically grounded lip synchronization metrics to assess how well the generated 3D talking heads align with these three criteria. Experiments show that training 3D talking head generation models with our perceptual loss significantly improve all three aspects of perceptually accurate lip synchronization. Codes and datasets are available at https://perceptual-3d-talking-head.github.io/.