Music can speak for you: making music with a deep net partner

This project aims to develop and evaluate a novel computational partner to aid composers and non-musicians to make personal music. One computational component learns to output musical structures that another component moulds towards user-desired features while encouraging innovation and exploration. Listeners’ evaluation of the musical outputs in terms of affect will be analysed, potentially allowing us to extend current music generation software considerably. The expected outcomes will be a tool for musicians, but also for untrained people, young and older, allowing such untrained people to make personalized music. The tool can thus provide benefits to the creative arts, and to the educational and wellbeing support sectors.