DiffSHEG: A Diffusion-Based Approach for Real-Time Speech-Driven Holistic 3D Expression and Gesture Generation | IEEE Conference Publication | IEEE Xplore