Symbolic Music Loop Generation with VQ-VAE

TL;DR

This paper presents a method for generating 8-bar musical loops using VQ-VAE, emphasizing discrete representations and manipulation of latent features to enhance musical structure and diversity.

Contribution

It introduces a novel approach leveraging VQ-VAE for symbolic music loop generation, focusing on discrete latent features and explicit rule-based extraction.

Findings

Discrete representations effectively model symbolic music sequences.

VQ-VAE captures musical properties better than other models.

Manipulating latent features enhances musical structure and repetition.

Abstract

Music is a repetition of patterns and rhythms. It can be composed by repeating a certain number of bars in a structured way. In this paper, the objective is to generate a loop of 8 bars that can be used as a building block of music. Even considering musical diversity, we assume that music patterns familiar to humans can be defined in a finite set. With explicit rules to extract loops from music, we found that discrete representations are sufficient to model symbolic music sequences. Among VAE family, musical properties from VQ-VAE are better observed rather than other models. Further, to emphasize musical structure, we have manipulated discrete latent features to be repetitive so that the properties are more strengthened. Quantitative and qualitative experiments are extensively conducted to verify our assumptions.