Using Genetic Algorithms to Find and Evaluate Parameters for Adaptive Digital Audio Effects

Abstract

Cross-adaptive audio effects are audio effects whose parameters are controlled or influenced by features of audio signals other than the one being affected. In other words, there is a mapping from one or several analyzed features of some signal(s) to one or more parameters of an effect processing a different signal. This thesis aims to find and evaluate such mappings using AI methods, specifically a genetic algorithm, with the evaluation of a mapping s fitness being a measure of musical applicability.

A possible design and its implementation are presented, and experiments which show that the idea is feasible are carried out.