Digital sound synthesis via parallel evolutionary optimization (Paralel evrimsel eniyileme ile sayısal ses sentezleme)

摘要

In this research, we propose a novel parallelizable architecture for the optimization of various sound synthesis parameters. The architecture employs genetic algorithms to match the parameters of different sound synthesizer topologies to target sounds. The fitness function is evaluated in parallel to decrease its convergence time. Based on the proposed architecture, we have implemented a framework using the SuperCollider audio synthesis and programming environment and conducted several experiments. The results of the experiments have shown that the framework can be utilized for accurate estimation of the sound synthesis parameters at promising speeds.