A Multi-Scale Attentive Transformer for Multi-Instrument Symbolic Music Generation

TL;DR

This paper introduces a multi-scale attentive Transformer model that enhances multi-instrument symbolic music generation by capturing diverse temporal and track dependencies, leading to improved quality over single-scale models.

Contribution

The paper proposes a novel multi-scale attentive Transformer architecture that models inter- and intra-track dependencies at various scales for better multi-instrument music generation.

Findings

Improves quantitative performance on SOD and LMD datasets.

Enhances qualitative musical coherence and harmony.

Outperforms single-scale models in multi-instrument generation.

Abstract

Recently, multi-instrument music generation has become a hot topic. Different from single-instrument generation, multi-instrument generation needs to consider inter-track harmony besides intra-track coherence. This is usually achieved by composing note segments from different instruments into a signal sequence. This composition could be on different scales, such as note, bar, or track. Most existing work focuses on a particular scale, leading to a shortage in modeling music with diverse temporal and track dependencies. This paper proposes a multi-scale attentive Transformer model to improve the quality of multi-instrument generation. We first employ multiple Transformer decoders to learn multi-instrument representations of different scales and then design an attentive mechanism to fuse the multi-scale information. Experiments conducted on SOD and LMD datasets show that our model improves both quantitative and qualitative performance compared to models based on single-scale information. The source code and some generated samples can be found at https://github.com/HaRry-qaq/MSAT.