Dissertation/Thesis Abstract

The Status of AI in Music: A Study of the Musical Metacreation Conferences, 2012-2018
by York, Bruce, M.M., Western Illinois University, 2019, 138; 13901476
Abstract (Summary)

The purpose of this paper is determine the status of Artificial Intelligence (AI) in music generation by reviewing the Musical Metacreation (MUME) conferences. I conducted a review of the development of AI and the different approaches to creating AI systems to better understand the difficulties in the development of any AI system. I found that developers have used four approaches in the creation of AI systems. They are: the Turing or acting humanly approach, the cognitive or thinking humanly approach, the logic or thinking rational approach, and the agent or acting rational approach. To better understand musical AI systems presented at MUME, I conducted a review of the significant papers on AI musical systems prior to 2012. The review identified that the AI system in music had progressed from agent and logic approaches to more of a cognitive approach. The developers applied theories and practices from psychology, linguistics and epistemology to the generation of AI music systems. A major barrier identified through both of the reviews was how to create a computer function to evaluate the output of the AI musical creation system as either good, acceptable or bad.

I reviewed the founding and organization of the MUME conference to determine the breadth and depth of the organization. MUME conferences are not dominated by any group, organization, approach or country. As I reviewed 98 of the 100 papers presented in the six conferences (two of the papers did not have links in the MUME or the conference sponsor’s websites), I found examples of AI in music that went beyond the simple composing of music. There were systems that created music from pictures, video feeds, animal sounds, and the movement of people. Developers presented systems that acted as full partners in improvisational music generation in real time. Presenters offered alternative AI music generation systems for other genres but typically such systems reflect the bias of the creator or user of the system. Despite their success for specific genres, the development of a universal evaluation function to determine what is good music remains to be solved.

Indexing (document details)
Advisor: Hardeman, Anita
Commitee: Locke, Brian, Chin, Hong-Da
School: Western Illinois University
Department: Music
School Location: United States -- Illinois
Source: MAI 81/2(E), Masters Abstracts International
Source Type: DISSERTATION
Subjects: Music history, Artificial intelligence
Keywords: Artificial intelligence, Metacreation, Music
Publication Number: 13901476
ISBN: 9781085702911
Copyright © 2019 ProQuest LLC. All rights reserved. Terms and Conditions Privacy Policy Cookie Policy
ProQuest