Expression in Live Coding: Gestural Interaction for Machine Musicianship

Doctoral thesis, 2023

This thesis is centered on the performance practice of musical live coding, which can be described as on-the-fly decision-making for computer music performance and blurs the lines between programming languages and computer interfaces. I specifically focus on live coding as a human activity based on serial skilled actions, and I discuss how we can interact gesturally with interfaces that are modified dynamically. I develop a system to address how to live code using continuous gestural interactions, and examine an unusual strategy of information processing which centers on a bottom-up approach. I then present a theoretical account of sensorimotor control in live coding, which examines pre-reflective processes during performance. Two more theoretical contributions address bodily gestures in various performance systems by presenting a conceptual framework, and an additional study presents a conceptual tool for developing agent-based systems in live coding. Observations of performance practices and systems are employed throughout the thesis to examine how different practitioners may use the systems, and what conceptual abstractions can be inferred. The question of how to facilitate creativity is discussed from both a theoretical and a practical perspective, touching on how the resultant newly produced knowledge may be transferred to practitioners. A particular focus is on live performance, visualization, and both human and machine listening.

The thesis findings present (i) visual and conceptual representations of live coding validated by cognitive mechanisms and cognitive paradigms, (ii) extensive reflections on practice and systemic modes of knowing, (iii) technical contributions for building of performance systems and performance structural analysis and (iv) theoretical accounts that contribute to the live coding literature. This work address both human modalities – audition, motor skills – and system modalities – notation, and the presence of software agents. I present a conception of various kinds of interactivity variations, a term used to designate various manners of gestural interaction that may arise in performance systems. I argue that musical live coding should incorporate radical experimentation with craft practices, pointing to a future practice wherein risk becomes clearly apparent in our gestural expressions.