Massively Collaborative Composition of Music by Evolution

Robin H S Hawkins


Supervised by Kirill Sidorov; Moderated by Ralph Martin

Musical compositions are usually written by one person, or a small group of individuals. It is very difficult to involve large numbers of people in the writing process. For a large group of individuals to compose music together, each individual will only be able to make a small contribution to the whole. Non-musicians are not able to contribute to a composition with complex musical information. The input of all composers in such a scenario will have to be a series of simple, incremental refinements. This can be realised using an evolutionary algorithm.

Starting with a population of pseudo-randomly generated strings of notes (which may have very little inherent musical value), human participants in this project (composers) will rank the melodies against one another, serving as the fitness function. As the algorithm progresses and the generations advance, hopefully the average musicality of the population will rise. By the conclusion of the project, the individual melodies will be highly musical. Clusters of these melodies can then be proposed to the users/composers as a final composition. The users will be able to interact with the system via a web interface, which will allow them to listen to and compare melodies from the population.

Prerequisites: good capacity for abstract thought, experience in music, excellent programming skills, ability for independent thinking, and a degree of ingenuity.

Initial Plan (03/02/2014) [Zip Archive]

Final Report (06/05/2014) [Zip Archive]

Publication Form