# High-Level Control of Drum Track Generation Using Learned Patterns of Rhythmic Interaction

## TL;DR

This paper introduces a deep learning model for conditional drum track generation that uses learned relational codes to control and transfer rhythmic patterns, demonstrating invariance to tempo and timing changes.

## Contribution

It presents a novel unsupervised approach to learn relational codes for controlling and transferring drum patterns in music generation.

## Key findings

- Codes enable diverse, plausible drum pattern generation.
- Model successfully transfers patterns between songs.
- Codes are invariant to tempo and time-shift.

## Abstract

Spurred by the potential of deep learning, computational music generation has gained renewed academic interest. A crucial issue in music generation is that of user control, especially in scenarios where the music generation process is conditioned on existing musical material. Here we propose a model for conditional kick drum track generation that takes existing musical material as input, in addition to a low-dimensional code that encodes the desired relation between the existing material and the new material to be generated. These relational codes are learned in an unsupervised manner from a music dataset. We show that codes can be sampled to create a variety of musically plausible kick drum tracks and that the model can be used to transfer kick drum patterns from one song to another. Lastly, we demonstrate that the learned codes are largely invariant to tempo and time-shift.