Exchange Evaluation Protocol

Researchers have developed STREAM, a new AI model that uses a diffusion transformer architecture to generate dance motions from both text and music cues, overcoming the modality collapse problem where…