Genetic algorithm produces music from cacophony
JUN 19, 2012
DOI: 10.1063/PT.5.026111
Science : Robert MacCallum of Imperial College London and his colleagues can turn cacophony into music in a process akin to evolution. The computer-based procedure starts with eight-second sequences of randomly generated sounds. Random mutations, determined by computer, occur in those sounds and either improve or worsen the sound’s musicality. As with sexual reproduction, pairs of sequences (“DNA”) then exchange the mutated sounds (“genes”) to create new sequences. To assess whether a sequence was fitterâmdash;that is, more musically pleasantâmdash;than its parents, MacCallum enlisted human volunteers. Within about 2500 generations, strains of inoffensive, jingly music had evolved. A paper describing the project appears today in the Proceedings of the National Academy of Sciences.
© 2012 American Institute of Physics
