Design and Evaluation of Human-Computer Rhythmic Interaction in a Tutoring System

A system for dynamic rhythmical interaction using hand claps, as well as a novel tutor system built on top of this interaction framework is studied. The interaction with the system in a training session, in which the system taught flamenco rhythms to people not familiar with them by providing synthetic hand clapping sounds and visual cues of the patterns, and providing feedback on the users' learning by listening to their clapping sounds, was evaluated. A detailed investigation of temporal aspects of clapping interaction revealed timing-related behavior on various levels of interaction. Task-related analysis showed a significant increase in tempo when clapping was no longer guided by a tutor. When clapping together with a tutor, timing is influenced by how well the system allows flexibility in timing individual beats. Allowing the virtual tutor's tempo to adapt slightly to the user's clapping enhances the level of correct accentuation during interactions with the system. The same flexibility liberates the clapper to spend slightly more time on accentuated than unaccentuated beats, and seems to facilitate more natural clapping interaction.