Choir Transformer: Generating Polyphonic Music with Relative Attention on Transformer

TL;DR

This paper introduces the Choir Transformer, a neural network model utilizing relative attention for improved polyphonic music generation, achieving state-of-the-art accuracy and realistic harmony close to Bach's music, with flexible style adaptation.

Contribution

The paper presents a novel Transformer-based model with relative positional attention and a new music representation for polyphonic music generation, surpassing previous methods in accuracy.

Findings

Achieved 4.06% higher accuracy than previous state-of-the-art.

Generated music with harmony metrics close to Bach's compositions.

Demonstrated style and input-based customization of generated music.

Abstract

Polyphonic music generation is still a challenge direction due to its correct between generating melody and harmony. Most of the previous studies used RNN-based models. However, the RNN-based models are hard to establish the relationship between long-distance notes. In this paper, we propose a polyphonic music generation neural network named Choir Transformer[ https://github.com/Zjy0401/choir-transformer], with relative positional attention to better model the structure of music. We also proposed a music representation suitable for polyphonic music generation. The performance of Choir Transformer surpasses the previous state-of-the-art accuracy of 4.06%. We also measures the harmony metrics of polyphonic music. Experiments show that the harmony metrics are close to the music of Bach. In practical application, the generated melody and rhythm can be adjusted according to the specified input, with different styles of music like folk music or pop music and so on.