Generative Statistical Models with Self-Emergent Grammar of Chord Sequences

TL;DR

This paper explores generative statistical models for chord sequences, demonstrating that hidden Markov and probabilistic context-free grammar models can learn syntactic categories that align with traditional harmonic functions, often outperforming conventional models.

Contribution

It introduces unsupervised learning of latent grammar models for chords, revealing their effectiveness and alignment with harmonic functions, which is a novel approach in music informatics.

Findings

Models outperform traditional Markov models in prediction.

Self-emergent categories correspond to harmonic functions.

Unsupervised learning induces meaningful syntactic categories.

Abstract

Generative statistical models of chord sequences play crucial roles in music processing. To capture syntactic similarities among certain chords (e.g. in C major key, between G and G7 and between F and Dm), we study hidden Markov models and probabilistic context-free grammar models with latent variables describing syntactic categories of chord symbols and their unsupervised learning techniques for inducing the latent grammar from data. Surprisingly, we find that these models often outperform conventional Markov models in predictive power, and the self-emergent categories often correspond to traditional harmonic functions. This implies the need for chord categories in harmony models from the informatics perspective.