Minimum description length modelling of musical structure

Panayotis Mavromatis

Research output: Contribution to journalArticle

Abstract

This article presents a method of inductive inference whose aim is to build formal quantitative models of musical structure. The models are constructed by statistical extraction of significant patterns from a musical corpus. The minimum description length (MDL) principle is used to select the best model from among the members of a non-parametric model family characterized by an unbounded parameter set. The chosen model achieves optimal compromise between goodness-of-fit and model complexity, thereby avoiding the over-fitting normally associated with such a family of models. The MDL method is illustrated through its application to the Hidden Markov Model (HMM) framework. We derive an original mathematical expression for the MDL complexity of HMMs that employ a finite alphabet of symbols; these models are particularly suited to the symbolic modelling of musical structure. As an illustration, we use the proposed HMM complexity expression to construct a model for a common melodic formula in Greek church chant. Such formulas are characterized by text–tune association that is governed by complex rules. We show that MDL-guided model construction gradually ‘learns’ important aspects of the melodic formula's structure, and that the MDL principle terminates the process when nothing significant is left to learn. We outline how the musical applications of MDL can be extended beyond the HMM framework, and how they address general methodological concerns in empirical music research.

Original languageEnglish (US)
Pages (from-to)117-136
Number of pages20
JournalJournal of Mathematics and Music
Volume3
Issue number3
DOIs
StatePublished - Jan 1 2009

Fingerprint

Modeling
Markov Model
Hidden Markov models
Model Complexity
Model
Inductive Inference
Musical Structure
Length
Overfitting
Nonparametric Model
Terminate
Goodness of fit
Religious buildings
Music
Hidden Markov Model
Family
Framework

Keywords

  • Computational modelling
  • Greek chant
  • Hidden Markov Models
  • Inductive inference
  • Information theory
  • Machine learning
  • Minimum description length
  • Model selection
  • Pattern recognition
  • Text–tune association

ASJC Scopus subject areas

  • Modeling and Simulation
  • Music
  • Computational Mathematics
  • Applied Mathematics

Cite this

Minimum description length modelling of musical structure. / Mavromatis, Panayotis.

In: Journal of Mathematics and Music, Vol. 3, No. 3, 01.01.2009, p. 117-136.

Research output: Contribution to journalArticle

Mavromatis, Panayotis. / Minimum description length modelling of musical structure. In: Journal of Mathematics and Music. 2009 ; Vol. 3, No. 3. pp. 117-136.
@article{ed8e0e36f6df45358a5723302bd39510,
title = "Minimum description length modelling of musical structure",
abstract = "This article presents a method of inductive inference whose aim is to build formal quantitative models of musical structure. The models are constructed by statistical extraction of significant patterns from a musical corpus. The minimum description length (MDL) principle is used to select the best model from among the members of a non-parametric model family characterized by an unbounded parameter set. The chosen model achieves optimal compromise between goodness-of-fit and model complexity, thereby avoiding the over-fitting normally associated with such a family of models. The MDL method is illustrated through its application to the Hidden Markov Model (HMM) framework. We derive an original mathematical expression for the MDL complexity of HMMs that employ a finite alphabet of symbols; these models are particularly suited to the symbolic modelling of musical structure. As an illustration, we use the proposed HMM complexity expression to construct a model for a common melodic formula in Greek church chant. Such formulas are characterized by text–tune association that is governed by complex rules. We show that MDL-guided model construction gradually ‘learns’ important aspects of the melodic formula's structure, and that the MDL principle terminates the process when nothing significant is left to learn. We outline how the musical applications of MDL can be extended beyond the HMM framework, and how they address general methodological concerns in empirical music research.",
keywords = "Computational modelling, Greek chant, Hidden Markov Models, Inductive inference, Information theory, Machine learning, Minimum description length, Model selection, Pattern recognition, Text–tune association",
author = "Panayotis Mavromatis",
year = "2009",
month = "1",
day = "1",
doi = "10.1080/17459730903313122",
language = "English (US)",
volume = "3",
pages = "117--136",
journal = "Journal of Mathematics and Music",
issn = "1745-9737",
publisher = "Taylor and Francis Ltd.",
number = "3",

}

TY - JOUR

T1 - Minimum description length modelling of musical structure

AU - Mavromatis, Panayotis

PY - 2009/1/1

Y1 - 2009/1/1

N2 - This article presents a method of inductive inference whose aim is to build formal quantitative models of musical structure. The models are constructed by statistical extraction of significant patterns from a musical corpus. The minimum description length (MDL) principle is used to select the best model from among the members of a non-parametric model family characterized by an unbounded parameter set. The chosen model achieves optimal compromise between goodness-of-fit and model complexity, thereby avoiding the over-fitting normally associated with such a family of models. The MDL method is illustrated through its application to the Hidden Markov Model (HMM) framework. We derive an original mathematical expression for the MDL complexity of HMMs that employ a finite alphabet of symbols; these models are particularly suited to the symbolic modelling of musical structure. As an illustration, we use the proposed HMM complexity expression to construct a model for a common melodic formula in Greek church chant. Such formulas are characterized by text–tune association that is governed by complex rules. We show that MDL-guided model construction gradually ‘learns’ important aspects of the melodic formula's structure, and that the MDL principle terminates the process when nothing significant is left to learn. We outline how the musical applications of MDL can be extended beyond the HMM framework, and how they address general methodological concerns in empirical music research.

AB - This article presents a method of inductive inference whose aim is to build formal quantitative models of musical structure. The models are constructed by statistical extraction of significant patterns from a musical corpus. The minimum description length (MDL) principle is used to select the best model from among the members of a non-parametric model family characterized by an unbounded parameter set. The chosen model achieves optimal compromise between goodness-of-fit and model complexity, thereby avoiding the over-fitting normally associated with such a family of models. The MDL method is illustrated through its application to the Hidden Markov Model (HMM) framework. We derive an original mathematical expression for the MDL complexity of HMMs that employ a finite alphabet of symbols; these models are particularly suited to the symbolic modelling of musical structure. As an illustration, we use the proposed HMM complexity expression to construct a model for a common melodic formula in Greek church chant. Such formulas are characterized by text–tune association that is governed by complex rules. We show that MDL-guided model construction gradually ‘learns’ important aspects of the melodic formula's structure, and that the MDL principle terminates the process when nothing significant is left to learn. We outline how the musical applications of MDL can be extended beyond the HMM framework, and how they address general methodological concerns in empirical music research.

KW - Computational modelling

KW - Greek chant

KW - Hidden Markov Models

KW - Inductive inference

KW - Information theory

KW - Machine learning

KW - Minimum description length

KW - Model selection

KW - Pattern recognition

KW - Text–tune association

UR - http://www.scopus.com/inward/record.url?scp=77952893384&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77952893384&partnerID=8YFLogxK

U2 - 10.1080/17459730903313122

DO - 10.1080/17459730903313122

M3 - Article

VL - 3

SP - 117

EP - 136

JO - Journal of Mathematics and Music

JF - Journal of Mathematics and Music

SN - 1745-9737

IS - 3

ER -