Designing and Building a 1st stage dataset for embodied music-making (musicking).

The aim of this research project was to design and develop a dataset that captured embodied music-making (hereafter musicking (Small 1989)) for use in human-computer interaction between AI and human musicians. The proposed solution to this challenge was to capture embodied musicking through data harvesting of physical and sonic elements from a musician embodied in the flow of musicking. This report describes the small-scale, first stage proof-of- concept design, development and deployment of such a system. The hard question here was to develop a dataset that could inform perception of music-AI so that it could co-create within the realtime flow of musicking with other machines and/or human musicians. The results highlight the importance that any modes of capture other than real-world musicking, would be a setup for failure.