Data Science · DevOps · Astrophysics


MLCompose is a Python library for algorithmic music composition. The goal of the project is to be able to ingest MIDI music files, use the data to train machine learning models, and generate new AI-composed music output as either new MIDI files or fully engraved sheet music.

The full ML models are still in internal development, but in the meantime, we have released the underlying Python modules for music representation and algorithmic composition. From note values and durations to obscure time signatures and measures with automatic rest-padding, music object classes have been designed to be easily incorporated into complex processing pipelines.

Find the source code on GitHub.