Cultural Economies of Adaptive and Affective Music AI
Eric Drott, Associate Professor Of Music Theory, University of Texas at Austin
Startups seeking to commercialize music AI have multiplied since 2010. Firms in this sector imagine several end-uses for automatically generated music (Collins 2018, Dredge 2018, Du Sautoy 2019): soundtracks for video and film, interactive scores for gaming etc. Other envisaged uses are more novel, extending the personalization characteristic of digital services to algorithmic music. For startups like LifeScore, AI.Music and Endel, the future is one where adaptive and affective music AI ‘shapeshifts’ recordings in response to users’ actions, surroundings and emotional states. In turn, commercial interest in adaptive and affective music AI follows from the recomposition of music distribution that streaming platforms have brought about.
Streaming’s transformation of music from a good that is purchased to one that is rented (Born and Durham 2022) encourages music’s reimagination as a service tailored to individuals’ shifting needs. Similarly, much work in adaptive and affective music AI is animated by this ‘music as a service’ paradigm. Combining ethnography and history, this study maps the cultural economies of adaptive and affective music AI. It places current efforts to commercialize music AI in historical context through archival research on corporate records, tracing prior efforts in the 1950s and 1960s to develop commercial systems for automatic music generation (Klein 1957, Olson and Belar 1961). The project thereby illuminates continuities and caesuras across commercial music AI’s history.