École Nationale Supérieure de
Techniques Avancées

Abstract

The development of musical robots which create musical performances or assist musicians during their performances has been an active subject of research in the last few years. During this internship we aimed at designing a multi-agent system for the improvisation of rhythmic patterns, where agents compete to take the lead and play musical solos. This competing mechanism is based on evolutionary algorithms. As such a natural evolution of this system, is to investigate the ways in which separating these agents into multiple subpopulations, where some agents occasionally migrate from one subpopulation to the other and try to bring with them rhythmic knowledge of their past experience, affects the musical dynamics of these subgroups.