Cognitive factors in generative music systems

This research aims to inform the development of generative music algorithms with principles drawn from research into music perception and cognition. Research has provided insights into the ways humans mentally organise musical sound and resulted in development of complex theories of musical expectation. Implementing these theories in generative music systems has the potential to produce music which is perceptually more meaningful for the listener. A detailed description of a prototype generative music algorithm is given. The algorithm aims to automatically create music that displays tonal and metric hierarchies in a Western tonal style, grouping structure and encourages the formation of musical expectations. The extent to which the system goals have been achieved is evaluated by objective analysis. Subjective analysis of the system output is suggested as a crucial aspect of further work.