Abstract
This thesis proposes the design and implementation of a multimodal system for human-machine music performances in real-time. The machine behavior is modeled under the concepts and paradigms related to an artificial Swarm of Autonomous Agents. The system used three advanced technologies as subsystems: Motion Capture, Spatial Audio, and Mixed Reality. These subsystems are integrated in one only solution that is evaluated regarding system measurements and music improvisation sessions. The system measurements determine the advantages and limitations in terms of effectiveness and efficiency; and the music improvisation sessions evaluate user interaction through the analysis of data recording and a survey. The results provide latency, jitter and other real-time parameters that are contrasted with user data. Moreover, the user analysis shows that the system is easy-to-use and highly enjoyable. These findings indicate that the strategy to conceive the system is validated and can be used for further investigation for autonomous agents and musicology aspects.