Application of deep learning in music visualisation

Abstract

The aim of this report is to study the field of music visualisation techniques, and the applications of deep learning in the field. Current music visualisation techniques, both with and without deep learning, were reviewed in order to identify a potential research area that could be expanded upon. This report documents the research and development of a tool that leverages Deep Learning techniques in Music Visualisation. The script was implemented using Python, utilising external libraries NVIDIA's StyleGAN2 for training and testing, matplotlib for drawing, and external application splitter.ai for music processing. A user study was then conducted to verify the effectiveness of the tool.