Preprint2026
Helping Music Co-Creation Agents 'Listen' Well: Hierarchical Self-Supervised World Models for Understanding and Generation
Scott H. Hawley
A hierarchical self-supervised world model for symbolic music uses JEPA-style objectives on MIDI piano-roll images, enabling understanding and generation without labels, with a flow-matching decoder for interactive co-creation.
0Aug 5, 2026PromptingComputer Vision
arXiv