Do Music Generation Models Encode Music Theory?

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Wei, Megan, Freeman, Michael, Donahue, Chris, Sun, Chen
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914961837522944
author Wei, Megan
Freeman, Michael
Donahue, Chris
Sun, Chen
author_facet Wei, Megan
Freeman, Michael
Donahue, Chris
Sun, Chen
contents Music foundation models possess impressive music generation capabilities. When people compose music, they may infuse their understanding of music into their work, by using notes and intervals to craft melodies, chords to build progressions, and tempo to create a rhythmic feel. To what extent is this true of music generation models? More specifically, are fundamental Western music theory concepts observable within the "inner workings" of these models? Recent work proposed leveraging latent audio representations from music generation models towards music information retrieval tasks (e.g. genre classification, emotion recognition), which suggests that high-level musical characteristics are encoded within these models. However, probing individual music theory concepts (e.g. tempo, pitch class, chord quality) remains under-explored. Thus, we introduce SynTheory, a synthetic MIDI and audio music theory dataset, consisting of tempos, time signatures, notes, intervals, scales, chords, and chord progressions concepts. We then propose a framework to probe for these music theory concepts in music foundation models (Jukebox and MusicGen) and assess how strongly they encode these concepts within their internal representations. Our findings suggest that music theory concepts are discernible within foundation models and that the degree to which they are detectable varies by model size and layer.
format Preprint
id arxiv_https___arxiv_org_abs_2410_00872
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Do Music Generation Models Encode Music Theory?
Wei, Megan
Freeman, Michael
Donahue, Chris
Sun, Chen
Sound
Artificial Intelligence
Computation and Language
Machine Learning
Audio and Speech Processing
Music foundation models possess impressive music generation capabilities. When people compose music, they may infuse their understanding of music into their work, by using notes and intervals to craft melodies, chords to build progressions, and tempo to create a rhythmic feel. To what extent is this true of music generation models? More specifically, are fundamental Western music theory concepts observable within the "inner workings" of these models? Recent work proposed leveraging latent audio representations from music generation models towards music information retrieval tasks (e.g. genre classification, emotion recognition), which suggests that high-level musical characteristics are encoded within these models. However, probing individual music theory concepts (e.g. tempo, pitch class, chord quality) remains under-explored. Thus, we introduce SynTheory, a synthetic MIDI and audio music theory dataset, consisting of tempos, time signatures, notes, intervals, scales, chords, and chord progressions concepts. We then propose a framework to probe for these music theory concepts in music foundation models (Jukebox and MusicGen) and assess how strongly they encode these concepts within their internal representations. Our findings suggest that music theory concepts are discernible within foundation models and that the degree to which they are detectable varies by model size and layer.
title Do Music Generation Models Encode Music Theory?
topic Sound
Artificial Intelligence
Computation and Language
Machine Learning
Audio and Speech Processing
url https://arxiv.org/abs/2410.00872