An Intelligent System for Chinese Dance Creation Using Generative Artificial Intelligence

We present an innovative and practical system for creating Chinese dance using generative artificial intelligence techniques. A full-attention cross-modal transformer is utilized to generate 3D dance motions that match the music and are rendered with realistic visual effects. Furthermore, we fine-tuned a pre-training model on a dedicated dataset to generate easily understandable narrations that enhance users' understanding of the generated dance motion sequences. The experiment was conducted to evaluate the quality of the generated content, and its outcomes suggest the prospective applicability of the proposed system.