Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Exploring Multi-Modal Control in Music-Driven Dance Generation

Ronghui Li, Yuqin Dai, Yachao Zhang, Jun Li, Jian Yang, Jie Guo, Xiu Li

TL;DR

A unified framework capable of generating high-quality dance movements and supporting multi-modal control, including genre control, semantic control, and spatial control is proposed, thereby avoiding the degradation in dance quality when adding additional control information.

Abstract

Existing music-driven 3D dance generation methods mainly concentrate on high-quality dance generation, but lack sufficient control during the generation process. To address these issues, we propose a unified framework capable of generating high-quality dance movements and supporting multi-modal control, including genre control, semantic control, and spatial control. First, we decouple the dance generation network from the dance control network, thereby avoiding the degradation in dance quality when adding additional control information. Second, we design specific control strategies for different control information and integrate them into a unified framework. Experimental results show that the proposed dance generation framework outperforms state-of-the-art methods in terms of motion quality and controllability.

Exploring Multi-Modal Control in Music-Driven Dance Generation

TL;DR

A unified framework capable of generating high-quality dance movements and supporting multi-modal control, including genre control, semantic control, and spatial control is proposed, thereby avoiding the degradation in dance quality when adding additional control information.

Abstract

Existing music-driven 3D dance generation methods mainly concentrate on high-quality dance generation, but lack sufficient control during the generation process. To address these issues, we propose a unified framework capable of generating high-quality dance movements and supporting multi-modal control, including genre control, semantic control, and spatial control. First, we decouple the dance generation network from the dance control network, thereby avoiding the degradation in dance quality when adding additional control information. Second, we design specific control strategies for different control information and integrate them into a unified framework. Experimental results show that the proposed dance generation framework outperforms state-of-the-art methods in terms of motion quality and controllability.
Paper Structure (10 sections, 4 equations, 7 figures, 2 tables)

This paper contains 10 sections, 4 equations, 7 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Method
  3. Pre-training: Motion VQ-VAE
  4. Training basic Cross-modal GPT
  5. Multi modal control
  6. Inference: Unified Framework
  7. Experiment
  8. Setups
  9. Comparative Results
  10. Conclusion

Figures (7)

  • Figure 1: Generated dance from various control input and music.
  • Figure 2: Overview of our method. GEN means Genre Embedding Network.
  • Figure 3: Text control module.
  • Figure 4: Generated dance for the same music in different genres, showcasing the effective control of the given genre on the generated sequence and the diversity achieved.
  • Figure 5: Generated dances for the same music using different text controls.
  • ...and 2 more figures