Towards a Relative-Pitch Neural Network System for Chorale Composition and Harmonization

Presenter Information

Sam Goree, Oberlin CollegeFollow

Location

King Building 321

Document Type

Presentation

Start Date

4-28-2017 1:30 PM

End Date

4-28-2017 2:50 PM

Abstract

Computational creativity researchers interested in machine learning approaches to computer composition often use the music of J.S. Bach to train their systems. Working with Bach, though, requires grappling with the conventions of tonal music, which can be difficult for computer systems to learn. In this paper, we propose and implement an alternate approach to composition and harmonization of chorales based on pitch-relative note encodings to avoid tonality altogether. We then evaluate our approach using a survey and expert analysis, and find that pitch-relative encodings do not significantly affect human-comparability, likability or creativity. However, an extension of this model that better addresses the criteria survey participants used to evaluate the music, such as instrument timbre and harmonic dissonance, still shows promise.

Keywords:

artificial intelligence, machine learning, neural networks, music composition

Notes

Session I, Panel 3 - Artistic | Transformations
Moderator: Jan Cooper, John Charles Reid Associate Professor of Rhetoric & Composition

Link to full text thesis at OhioLINK ETD Center:
http://rave.ohiolink.edu/etdc/view?acc_num=oberlin1495578351469519

Major

Musical Studies; Computer Science

Advisor(s)

Robert Geitz, Computer Science

Project Mentor(s)

Benjamin Kuperman, Computer Science
Adam Eck, Computer Science

April 2017

This document is currently not available here.

Share

COinS
 
Apr 28th, 1:30 PM Apr 28th, 2:50 PM

Towards a Relative-Pitch Neural Network System for Chorale Composition and Harmonization

King Building 321

Computational creativity researchers interested in machine learning approaches to computer composition often use the music of J.S. Bach to train their systems. Working with Bach, though, requires grappling with the conventions of tonal music, which can be difficult for computer systems to learn. In this paper, we propose and implement an alternate approach to composition and harmonization of chorales based on pitch-relative note encodings to avoid tonality altogether. We then evaluate our approach using a survey and expert analysis, and find that pitch-relative encodings do not significantly affect human-comparability, likability or creativity. However, an extension of this model that better addresses the criteria survey participants used to evaluate the music, such as instrument timbre and harmonic dissonance, still shows promise.