Composing the Assemblage: Probing Aesthetic and Technical Dimensions of Artistic Creation with Machine Learning

Abstract

Abstract In this article we address the role of machine learning (ML) in the composition of two new musical works for acoustic instruments and electronics through autoethnographic reflection on the experience. Our study poses the key question of how ML shapes, and is in turn shaped by, the aesthetic commitments characterizing distinctive compositional practices. Further, we ask how artistic research in these practices can be informed by critical themes from humanities scholarship on material engagement and critical data studies. Through these frameworks, we consider in what ways the interaction with ML algorithms as part of the compositional process differs from that with other music technology tools. Rather than focus on narrowly conceived ML algorithms, we take into account the heterogeneous assemblage brought into play: from composers, performers, and listeners to loudspeakers, microphones, and audio descriptors. Our analysis focuses on a deconstructive critique of data as being contingent on the decisions and material conditions involved in the data creation process. It also explores how interaction among the human and nonhuman collaborators in the ML assemblage has significant similarities to—as well as differences from—existing models of material engagement. Tracking the creative process of composing these works, we uncover the aesthetic implications of the many nonlinear collaborative decisions involved in composing the assemblage.