Getting Motif-ated: Controllable AI Compositions from Injected Motif Prompts

Chao Péter Yang, Cynthia Rudin, Simon Mak, Yue Jiang, Stephen Ni-Hahn
May, 2026

Abstract

Getting Motif-ated extends structured symbolic music generation with controllable composition from injected motif prompts. Given a short user-specified motif, the model produces harmonically and melodically coherent compositions that develop the motif across a phrase while preserving the structural guarantees of hierarchical music analysis. The approach combines discrete diffusion with knowledge-based probabilistic modeling, enabling user-steerable generation without sacrificing interpretability or musical cohesion.

Type
Preprint
Publication
Submitted to the 40th Conference on Neural Information Processing Systems (NeurIPS 2026), Creative AI Track
Symbolic Music Generation Controllable Generation Diffusion Models Interpretability
Chao Péter Yang
Chao Péter Yang
Machine Learning Researcher

My research focuses on LLM alignment, agentic systems, and structure-aware generative modeling. I build preference-aligned language models, reliable multi-agent systems, and symbolic music generation models, and aim to bridge theory and practice to create both scientific and real-world impact.