Back to search

The Open University

Reflexive Interactive Machine Learning for Timbral Exploration in Free Improvisation

Abstract

dc:description.abstract

This thesis investigates how an Interactive Reflexive Music System (IRMS) can be used to facilitate a free improviser’s exploration of timbre in their music-making. The first part of the thesis documents the design of TimbReflex, a small-data, timbrally focused IRMS for free improvisation. TimbReflex uses computational techniques, including audio segmentation and neural audio synthesis, to produce variations of a short musical input phrase. To interact with the system, a musician produces input phrases and responds to the system’s output phrases. TimbReflex produces its output phrases using a machine learning technique that transforms a timbral representation of the musician’s input phrases. The transformations facilitate the musician’s exploration of an expanded timbral space. Significantly, this technique allows TimbReflex to be trained on a small dataset, contributing to research on a small-data approach to creative AI systems. After introducing TimbReflex, the thesis presents a longitudinal study, structured over four sessions, conducted with ten musicians experienced in performing free improvisation. The results of this study present evidence that an IRMS can be successfully implemented to facilitate and elicit timbral exploration in a free improviser’s musical practice, with some limitations. The outcome of the study contributes to our understanding of how timbre is conceptualised among improvisers, and how this conceptualisation can impact the success of interactions with IRMSs for timbral music-making such as TimbReflex. The outcome also highlights the relevance of the reflexive component in TimbReflex for the participants, that is, the identification of their sound in the material generated by the system.

Degree

thesis:*
Name dc:type.qualificationname
phd
Level dc:type.qualificationlevel
doctoral
Grantor dc:publisher.institution
The Open University
Year dc:date.issued
2026

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Bolzoni, Andrea

Rights

Language dc:language
en

Chain of custody

source
Harvested from
The Open University
Base URL
oro.open.ac.uk/cgi/oai2
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
related terms
citation

Bolzoni, Andrea. Reflexive Interactive Machine Learning for Timbral Exploration in Free Improvisation. doctoral thesis, The Open University, 2026.