Analysis of computer music using machine learning models

Final year dissertation project

This project created a framework for analysing the possible sounds that a software synthesizer can make. It then analysed two simple synthesizers using the framework created.

The framework itself is composed of three main components: sound generators, feature extractors, and learning algorithms.

For a detailed explanation of the diagram above, including design and implementation of it, refer to the Dissertation Full.pdf document.

Repository Structure

Code: all the code written in this project, including some extra libraries used (such as the ctcsound library).

Sounds: sounds referred to in the dissertation document.

Dissertation Full: the full dissertation, available to read.

Dependencies

• Csound 6.14
• Python 3.7.5
• Essentia_streaming_extractor_music for the correct CPU architecture (two are provided)
• scikit-learn 0.22.2
• pandas
• plotly
• matplotlib