Audio-Driven 3D Talking Head

Oct 1, 2025

Based on 3DGS, 3DMM, and generative models, this project develops a 3DMM-guided diffusion framework for audio-driven talking head generation, enabling controllable facial motion and high-fidelity video synthesis.

Key contributions:

  • Designed a structured parameter encoding and conditional injection strategy
  • Improved temporal consistency and expression realism
  • Enables voice-driven speaker head modeling with photorealistic quality
Deep Learning 3DGS
ZIYUE ZENG
ZIYUE ZENG
Master’s Student → Incoming PhD Student at Waseda University

Pursuing socially meaningful, grounded research with enduring passion and diligence.