PPIG 2022 - 33rd Annual Workshop
Sound-based music style modelling, for a free improvisation musical agent
Andrea Bolzoni
Abstract: This paper presents the first stages of development of an improvising musical agent capable of interacting with a human musician in a free improvised music context. This research aims to explore the creative potential of a co-creative system that draws upon two approaches of music and sound generation: the well-established practice of modelling musical styles with Markov-based models and recent developments in neural-network-based audio synthesis. At this preliminary stage, the focus is on the definition of style in a sound-based music context, and the outline of a formal evaluation framework for style imitation systems.