This is a system to analyze and compose music using Answer Set Programming

The idea is to find reoccurring patterns inside similar but different musical pieces and exploit these similarities to guide a composition process. Instead of using genre-dependent constraints we use the result of our analysis to guide the search for a musical piece. Our analysis uses pattern mining to find reoccurring patterns inside the input files. These patterns get translated into rules used by the composer. We use MIDI files as input and output format as they are widely available and used. We divided our system into three parts:

1. MIDI -> ASP
2. pattern mining
3. composing

For a usage example see 'usage_example.py'

Dependencies

We use the clingo and Mido package for our system. They can be installed using Anaconda and pip respectively.

conda install -c potassco clingo
pip install mido

You can find more information about clingo on the Potassco website https://potassco.org/ and about Mido on their GitHub https://github.com/mido/mido