A portfolio of compositions based on trajectories through musical state spaces

This project attempts to develop a compositional approach where electro-acoustic music can be produced in a way that can easily integrate techniques and processes from different types of musical practices. The method adopted is inspired by an approach developed in Computer Science, to integrate the various fields of Artificial Intelligence. In this portfolio, music composition is viewed as creating the conditions so that the resultant music is an emergent phenomenon of the interactions of a collection of autonomous musical agents. To create a composition, a composer assembles a collection of musical agents and then delegates the task of creating the low-level musical material to each agent, specifying how each agent behaves. Every agent can exist in a number of states and their current state is reflected in their current contribution to the musical texture. By determining the ways an agent can travel through its state space, the composer can shape the composition. This state transition strategy can be implemented in such a way as to emulate the various idiomatic strategies a musician might apply in different musical contexts. Encapsulating compositional processes inside agents provides a way to potentially combine disparate compositional approaches. To create the works in this portfolio a general compositional framework was devised based on this idea, and three composition models were developed: the Painting Approach, the Pruning Approach, and the Goal-Led Approach. Several compositions were made with each model, varying the details of the different implementations. The portfolio includes interactive and fixed form works for computer-based instruments, analog synthesizers and acoustic instruments. Almost all of the works have some improvisational component. The approach of encapsulating generative processes in agents, so as to guide them as they explore their state spaces proved to be a successful means of producing music in a way that enables free mixing strategies for generating musical material. The ability to choose each agent’s state space traversal strategy gives a composer high-level tools to effectively determine the overall shape of the music. Most of the portfolio works, created with this approach, have been performed either at curated international conferences or festivals. Additional material(s) submitted with thesis.