Chain of Generation: Multi-Modal Gesture Synthesis via Cascaded Conditional Control

AAAI 2024

Zunnan Xu,  Yachao Zhang,  Sicheng Yang,  Ronghui Li,  Xiu Li 
SIGS, Tsinghua University
empty


Chain of Generation introduces a gesture synthesis method that enhances the generation of 3D gestures by utilizing multimodal information from human speech. It employs a cascaded conditional control approach to sequentially generate movements of different parts of the body, enhancing the quality of the gestures while reducing the need for extensive setup during inference.

Demo

BibTeX

@inproceedings{xu2024chain,
    title={Chain of generation: Multi-modal gesture synthesis via cascaded conditional control},
    author={Xu, Zunnan and Zhang, Yachao and Yang, Sicheng and Li, Ronghui and Li, Xiu},
    booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
    volume={38},
    number={6},
    pages={6387--6395},
    year={2024}
}