Abstract

The A-Life approach to Music is a promising new development. The vast majority of existing A-Life systems for musical composition employ a Genetic Algorithm (GA) to produce musical melodies, rhythms, and so on. In these systems, music parameters are represented as genotypes and GA operators are applied on these representations to produce music according to given fitness criteria. We have identified two methodological limitations of such GA-based systems: one relates to the fact that composition should not be driven by a constant set of fitness criteria and the other is to do with the fact that music is largely a cultural phenomenon driven by social pressure and this is cumbersome to model with standard GA alone. An approach improve this scenario is to build systems with A-Life algorithms designed primarily to address musical issues, rather than using algorithms that were not designed for music in the first place. The work presented in this paper contributes to this line of thought by proposing the design of algorithms that consider music as a cultural phenomenon whereby social pressure plays an important role in the development of musical conventions. We introduce three algorithms: popularity, transformation and complexity algorithms, respectively. The algorithms were implemented in the context of a system for composition of rhythms, where the user can explore their potential to generate rhythmic sequences and also monitor their behavior. Finally, we explore the composition capabilities of the system by using the rhythms developed by the agents during the simulations in a collective performance environment. This bottom-up approach automatically defines an implicit metric structure.

Citations