Agent-based graphic sound synthesis and acousmatic composition

  • Stephen Pearse

Student thesis: Doctoral Thesis


For almost a century composers and engineers have been attempting to create systems that allow drawings and imagery to behave as intuitive and efficient musical scores. Despite the intuitive interactions that these systems afford, they are somewhat underutilised by contemporary composers. The research presented here explores the concept of agency and artificial ecosystems as a means of creating and exploring new graphic sound synthesis algorithms. These algorithms are subsequently designed to investigate the creation of organic musical gesture and texture using granular synthesis. The output of this investigation consists of an original software artefact, The Agent Tool, alongside a suite of acousmatic musical works which the former was designed to facilitate. When designing new musical systems for creative exploration with vast parametric controls, careful constraints should be put in place to encourage focused development. In this instance, an evolutionary computing model is utilised as part of an iterative development cycle. Each iteration of the system’s development coincides with a composition presented in this portfolio. The features developed as part of this process subsequently serve the author’s compositional practice and inspiration. As the software package is designed to be flexible and open ended, each composition represents a refinement of features and controls for the creation of musical gesture and texture. This document subsequently discusses the creative inspirations behind each composition alongside the features and agents that were created. This research is contextualised through a review of established literature on graphic sound synthesis, evolutionary musical computing and ecosystemic approaches to sound synthesis and control.
Date of AwardOct 2016
Original languageEnglish
Awarding Institution
  • University of Sheffield
SupervisorAdrian Moore (Supervisor) & David Moore (Supervisor)

Cite this