Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Composer Vector: Style-steering Symbolic Music Generation in a Latent Space

Xunyi Jiang, Mingyang Yao, Jingyue Huang, Julian McAuley

Abstract

Symbolic music generation has made significant progress, yet achieving fine-grained and flexible control over composer style remains challenging. Existing training-based methods for composer style conditioning depend on large labeled datasets. Besides, these methods typically support only single-composer generation at a time, limiting their applicability to more creative or blended scenarios. In this work, we propose Composer Vector, an inference-time steering method that operates directly in the model's latent space to control composer style without retraining. Through experiments on multiple symbolic music generation models, we show that Composer Vector effectively guides generations toward target composer styles, enabling smooth and interpretable control through a continuous steering coefficient. It also enables seamless fusion of multiple styles within a unified latent space framework. Overall, our work demonstrates that simple latent space steering provides a practical and general mechanism for controllable symbolic music generation, enabling more flexible and interactive creative workflows. Code and Demo are available here: https://github.com/JiangXunyi/Composer-Vector and https://jiangxunyi.github.io/composervector.github.io/

Composer Vector: Style-steering Symbolic Music Generation in a Latent Space

Abstract

Symbolic music generation has made significant progress, yet achieving fine-grained and flexible control over composer style remains challenging. Existing training-based methods for composer style conditioning depend on large labeled datasets. Besides, these methods typically support only single-composer generation at a time, limiting their applicability to more creative or blended scenarios. In this work, we propose Composer Vector, an inference-time steering method that operates directly in the model's latent space to control composer style without retraining. Through experiments on multiple symbolic music generation models, we show that Composer Vector effectively guides generations toward target composer styles, enabling smooth and interpretable control through a continuous steering coefficient. It also enables seamless fusion of multiple styles within a unified latent space framework. Overall, our work demonstrates that simple latent space steering provides a practical and general mechanism for controllable symbolic music generation, enabling more flexible and interactive creative workflows. Code and Demo are available here: https://github.com/JiangXunyi/Composer-Vector and https://jiangxunyi.github.io/composervector.github.io/

Paper Structure

This paper contains 34 sections, 1 equation, 8 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Composer Vector
  3. Style-steering Symbolic Music Generation
  4. Method
  5. Composer Vector Construction.
  6. Composer Style Steering.
  7. Format Preservation.
  8. Results
  9. Controlling Symbolic Music Generation with Composer Vector.
  10. Composer Vector can Provide Continuous Style Control.
  11. Multi-style Fusion through Composer Vector.
  12. Conclusion
  13. Related work
  14. Symbolic music generation.
  15. Inference time intervention.
  16. ...and 19 more sections

Figures (8)

  • Figure 1: Inference-time control pipeline and style control diagrams
  • Figure 2: NotaGen: CLAP and CLaMP similarity before and after steering.
  • Figure 3: t-SNE visualization of piece-level embeddings across composers.
  • Figure 4: Comparison of layer-wise composer style localization across different models.
  • Figure 5: ChatMusician: CLAP and CLaMP similarity before and after steering.
  • ...and 3 more figures