Automatic accompaniment for improvised music

TitleAutomatic accompaniment for improvised music
Publication TypeMaster Thesis
Year of Publication2009
AuthorsMartín, D.
preprint/postprint documentstatic/media/Martin-Daniel-Master-Thesis-2009.pdf
AbstractModern music students practice improvisation playing on top of a software that generates accompaniment, but it does not listen to the musician's input. Automatic accompaniment systems (AAS) provide appropriate musical accompaniment to a human player in a given musical context by listening to them. We are interested in developing an AAS. Therefore, the purpose of this project is to build a system that generates accompaniment for a given melody, emulating a jazz pianist accompanying a human soloist improvising in jazz style. The method used is based on re-using stored fragments to create the accompaniment. Two musical fragment representation models have been studied: a tree representation model and a statistical model, from which the latter has been implemented.