“BlueJam: A heuristic-based approach to evolutionary music generation”
Abstract
Visions of a comprehensive algorithmic composer have been pursued over the years. Recent attempts at this have incorporated evolutionary algorithms, a notion whereby populations of candidate solutions compete to make it to the next generation and be used as the seed for further solutions. These algorithms have performed well where the problem is well defined but only an approximate solution is achievable. BlueJam is a program written in Java and Pure Data, that implements algorithms inspired by evolutionary techniques, along with heuristics to guide the evolution in the hope of producing more salient output. We attempt to discern if this heuristic-based approach can improve upon previous efforts to generate music that is not purely deterministic.
Download the paper and appendices below.
