Talking Drums: Generating drum grooves with neural networks
This addresses the problem of automated music generation for drummers and producers, but it is incremental as it builds on existing sequence-to-sequence methods.
The paper tackled generating drum kit parts from kick-drum sequences using a neural network adapted from natural language translation, finding that a sampling technique selecting from the top three probable outputs performed best, but consistency varied by musical style.
Presented is a method of generating a full drum kit part for a provided kick-drum sequence. A sequence to sequence neural network model used in natural language translation was adopted to encode multiple musical styles and an online survey was developed to test different techniques for sampling the output of the softmax function. The strongest results were found using a sampling technique that drew from the three most probable outputs at each subdivision of the drum pattern but the consistency of output was found to be heavily dependent on style.