CHIASMA is a prototype, representing the initial results of my project on composing with genetic algorithms and evolutionary computation. Genetic algorithms provide means of numerical and combinatorial optimization by generating more promising ideas from a specifiable set of components and processes. Individuals of an initial population with a random set of genes “mate”, and produce “offspring”, which would evolve under the process of natural selection over the course of generations.
CHIASMA uses an intelligent machine listening system to evaluate fitness values of generated individuals, by using fast Fourier transform as its translation method. The initial population is created from two random arrays, representing their random genomes. They mate to create offspring with the successful traits highlighted. Each chromosome array is compared to the target one through a simple linear distance function. individuals would then have an importance weight factor, used as a multiplication value when added to the mating pool, to assure that more desirable genes are passed on to the next generation. A mutation rate of one percent is considered to increase the chance of fit individuals being generated.
The research project focuses on employing the process of genetic evolution as the composition structure, as changes in parameters of the algorithm such as allowed population per generation, mutation rate, search environment, lifespan, etc. influence the events and speed of the system reaching its target.
The algorithm is written in Supercollider.
Credits
Released September 4, 2021
Artwork by Arash Akbari
Mix/Mastering by Remmy for Duality Sound Lab