A System for Sonic Explorations With Evolutionary Algorithms

Abstract

The discovery of new sounds can inspire creativity, and various approaches have been explored in that effort. One approach is the application of evolutionary algorithms to sound synthesis. The work presented here is related to those efforts and focuses on the properties of pattern-producing networks and neuroevolution. The authors chronicle a journey of concept development and implementation iterations. A discussion of a browser-based environment facilitating such explorations is followed by an account of a migration to a more flexible execution technology stack, using the Web Audio API. As a result of that process, the authors introduce a system consisting of a software library and a command-line utility that facilitates further investigations. Those products are intended to serve as tools for further explorations of the applicability of evolutionary algorithms to the synthesis and discovery of inspiring sounds.