Diffusion and graph networks improve music-conditioned dance generation and style control
A study using conditional diffusion, ST-GCN, cross-modal attention and AdaIN reported improved dance-generation quality, music synchronization and style control on ...
