Chord progression generation: evolving musical agents with the SAMUEL algorithm

This repository contains the work of my master thesis in Data Science and Scientific Computing. It is a custom implementation of the SAMUEL algorithm, mixing reinforcement learning with evolutionary algorithms, adapted to evolve a population of agents capable of writing chords progressions.

The (messy) files present in this repo are the following:

• Evolution.ipynb: containing the needed steps to evolve the agents
• preprocessing.ipynb: preprocessing of a corpus of songs nedded to compute the target metrics for the fitness function
• processed_verse.pkl: pickle file containing the results of the preprocessing
• thesis.py: contains all (or most of) the implemented functions and classes
• Thesis and slides: this folder contains the Master thesis document in PDF and the slides used during the discussion of my work.