Millea and Wakefield’s Article

Researchers at the University of Huddersfield have looked into automating the composition of popular music with evolutionary algorithms.


Automated composition may be regarded as a search within the space defined by a datatype representing musical compositions. We develop a hierarchical representation of popular music compositions with the aim of increasing the probability of finding potential hits. Musical variations are calculated as difference vectors between patterns extracted from a given set of existing compositions. These form the basis of the mutation operator within an evolutionary algorithm search.