Our basic research focuses on two problems: the problem of musical evolution and the problem of representation of musical experience. Our innovative approaches to these problems make our research agenda unique and adventurous.

Broadly speaking, systems for musical composition can be of two types: abstract algorithmic or knowledge-based. In general terms, the former employs non-musical algorithms to generate music. By “non-musical algorithms” we mean algorithms that were not originally designed for music; for example, fractals, cellular automata, sieves, probabilities, etc. The latter are those systems that generate music based on music theory, often by means of rules learned automatically from a set of musical examples. The common characteristic of these two types of systems is that they assume the existence of compositional rules a priori, which must either be programmed manually or learned from existing examples. We propose an evolutionary approach to composition systems. This approach seeks to answer the following question: Would it be possible to program machines to evolve their own musical compositional rules from scratch? In order to address this question, we are using Evolutionary Computation and Artificial Life techniques to model the evolution of music in surrogate societies of artificial agents. These agents are being programmed with the cognitive and physical abilities deemed necessary to evolve music, rather than with compositional rules or procedures.

By representation of musical experience we mean representation of how the brain perceives music in physiological terms. Until recently, the issue of musical representation had focused primarily on symbolic notation of musical information and structure, and on the representation of musical performance. Research on how we represent musical experience is emerging as a rich area of investigation thanks to ongoing advances in brain-scanning technology such as EEG (electroencephalogram) and fMRI (Functional Magnetic Resonance Imaging). A new area of research is emerging, which is a natural progression from (or complement to) Psychology of Music: Neuroscience of Music. This new area is of particularly interest for its potential application in the development of interactive music systems. The brain has to generate internal representations of the auditory input permitting the stimulus to be segregated from its background to be acted upon. Recent work on auditory processing has shown that the auditory system, far from being a passive receptor of sound, is constantly adjusting its processing to reflect the current acoustic context and task demands. Perception therefore involves a process of prediction. The need for making better predictions is what drives the development of representation and cognitive structures. An understanding of how the brain predict events on the basis of musical experience is a fundamental requirement for the design of interactive music systems; for example, for musical improvisation.

For further information please visit the Computer Music Research website.