Modeling the rhythmic complexity of professional drumming with an oscillation-driven reservoir computer

Area of Science:

• Computational Neuroscience
• Cognitive Science
• Music Cognition

Background:

• Musical performances, especially drumming, exhibit complex rhythmic patterns with subtle variations in timing and amplitude.
• The cerebellum and basal ganglia are known to play crucial roles in motor timing and rhythm production.
• Understanding the neural mechanisms behind learning and generating complex rhythms is a significant challenge.

Purpose of the Study:

• To propose and evaluate a neural-inspired computational model for learning and internalizing complex rhythmic patterns in drumming.
• To investigate the potential of oscillation-driven reservoir computing in simulating human-like expressive musical performances.
• To analyze how specific model parameters, like high-frequency oscillators, contribute to replicating drumming characteristics.

Main Methods:

• Utilized an oscillation-driven reservoir computer, a type of recurrent neural network, to model temporal learning.
• Trained the model to replicate specific drumming patterns, including Jeff Porcaro's hi-hat patterns and multi-genre drum kit performances.
• Incorporated high-frequency oscillators (50-100 Hz) within the model architecture.

Main Results:

• The model accurately reproduced the characteristic fluctuations and microtiming patterns of professional drumming.
• Outputs closely matched original drumming performances in terms of timing deviations and audio features.
• The inclusion of high-frequency oscillators was key to replicating expressive complexity.

Conclusions:

• Oscillation-driven reservoir computing effectively replicates the rhythmic complexity found in professional drumming.
• This computational approach offers insights into the brain's principles for motor timing and rhythm generation.
• The model provides a framework for further research into the neural processing of intricate rhythmic patterns.