Music has an important, but often subsidiary, role in most games and interactive experiences. Efforts to make generative music more accessible can potentially help numerous indie developers and designers. In this paper we explore the extension and application of the Wave Function Collapse (WFC) algorithm in the realm of music. Recognizing music’s inherent complex, layered structure, we investigate the feasibility of adapting WFC, a method broadly used in procedural content generation, to handle musical semantics in a mixed-initiative setting. Our approach introduces a novel generative method utilizing multiple canvases and constraints, each defined with specific musical significance. At each level, new constraints are specified that influence the generator at the next levels. This approach is particularly fitting to capture the layered structure and complex patterns inherent to music, and it encourages a mixed-initiative, iterative composition process. We also provide a prototype implementation of the proposed algorithm in ProceduraLiszt, a declarative music editor designed to offer composers a platform for exploring highly-controllable generative composition. The effectiveness and creativity-support facilitated by this experimental setup are under evaluation and will be detailed in a future publication.