Sounds Unheard of: Evolutionary algorithms as creative tools for the contemporary composer

Doctoral thesis, 2004

Evolutionary algorithms are studied as tools for generating novel musical material in the form of musical scores and synthesized sounds. The choice of genetic representation defines a space of potential music. This space is explored using evolutionary algorithms, in search of useful musical material. In this way, the computer becomes a creative tool that is integrated into the artistic process. Several implementations of these ideas are presented, based on interactive evolution, genetic algorithms and artificial life models. A number of different representations of music and sound are discussed, with a focus on parameterized sound synthesis and the representation of scores as recursively described trees. The dynamic behavior of a class of sound synthesis engines, consisting of networks of modulating oscillators, is also investigated. Finally, a number of compositions that have been composed with the tools described in the thesis are presented.