Counterpoint by Convolution

TL;DR

This paper introduces a convolutional neural network model for music composition that allows nonlinear editing through blocked Gibbs sampling, improving sample quality over traditional causal models.

Contribution

It presents a novel approach combining orderless NADE with Gibbs sampling for more realistic, nonlinear music generation, aligning closer to human compositional processes.

Findings

Gibbs sampling significantly enhances sample quality.

Blocked Gibbs sampling outperforms ancestral sampling in log-likelihood and human evaluation.

The model is flexible and not tied to a causal direction of composition.

Abstract

Machine learning models of music typically break up the task of composition into a chronological process, composing a piece of music in a single pass from beginning to end. On the contrary, human composers write music in a nonlinear fashion, scribbling motifs here and there, often revisiting choices previously made. In order to better approximate this process, we train a convolutional neural network to complete partial musical scores, and explore the use of blocked Gibbs sampling as an analogue to rewriting. Neither the model nor the generative procedure are tied to a particular causal direction of composition. Our model is an instance of orderless NADE (Uria et al., 2014), which allows more direct ancestral sampling. However, we find that Gibbs sampling greatly improves sample quality, which we demonstrate to be due to some conditional distributions being poorly modeled. Moreover, we show that even the cheap approximate blocked Gibbs procedure from Yao et al. (2014) yields better samples than ancestral sampling, based on both log-likelihood and human evaluation.