Artificial Intelligence as a creative tool and instrument within the practice of music production

Ben Clarke

Research output: Types of ThesisMaster's thesis

66 Downloads (Pure)

Abstract

This practice-based MRes engages with current music generation tools that use Artificial Intelligence (AI). These tools are viewed through a hauntological lens and integrated into music production practices for creating electronic dance music, whilst investigating different ways of conceiving human-AI music interactions. The research focused on the music generation tool Jukebox and culminated in a body of work that showcases the creative output of this system. When assessing research into AI’s impact on music production much of it focused on technical aspects, most commonly, sound generation, as opposed to the creative potential of these systems and their use as a device for music producers. This research bridges that gap, exploring the conceptual applications of AI in music production. To examine how AI could be deployed in music production practices I first surveyed the work of contemporary artists who were using AI tools. In doing this I gained knowledge of tools available for generating my own audio. For this I utilised Jukebox, using a Google Colab document which granted access to the equipment required to generate audio. I developed and trained my own model using this system to create two projects which showcase Jukebox’s capabilities. This is accompanied by this exegesis that explores the process of creating this body of work and contextualises it within the broader topic of AI in music production. Whilst researching this subject the conversation around AI and its implementation into artistic practices has developed at an ever accelerating pace. This study acts as a snapshot in time, where there is still a balancing act to be had when collaborating with the machine. As opposed to recent, more refined systems, which offer little to no collaboration with your artificial counterpart. This study shows how these technologies can be utilised in creative ways, highlighting the potential for them to be integrated into music production practices.
Original languageEnglish
QualificationMaster of Philosophy
Awarding Institution
  • London South Bank University
Supervisors/Advisors
  • Parkinson, Adam, Supervisor
  • Rietveld, Hillegonda, Supervisor
Award date27 Sept 2024
Publisher
Publication statusPublished - 27 Sept 2024

Cite this