GANs & Reels: Creating Irish Music using a Generative Adversarial Network

TL;DR

This paper introduces a novel approach for generating Irish traditional reel melodies using a non-recurrent GAN architecture that captures sequence information through dilated convolutions, enabling authentic melody creation.

Contribution

The paper presents a GAN-based method that models sequential musical data without recurrent layers, utilizing dilated convolutions to learn long-range dependencies in melodies.

Findings

Successfully generated Irish reel melodies

Demonstrated effectiveness of non-recurrent GAN architecture for music

Captured long-range dependencies in melodies

Abstract

In this paper we present a method for algorithmic melody generation using a generative adversarial network without recurrent components. Music generation has been successfully done using recurrent neural networks, where the model learns sequence information that can help create authentic sounding melodies. Here, we use DC-GAN architecture with dilated convolutions and towers to capture sequential information as spatial image information, and learn long-range dependencies in fixed-length melody forms such as Irish traditional reel.