Computational Approaches to Music Motor Performance: Clustering of Percussion Kinematics Underlying Performance Style

The present study investigated motor kinematics underlying performance-related movements in marimba performance. Participants played a marimba while motion capture equipment tracked movements of the torso, shoulders, elbows, wrists, and hands. Principal components analysis was applied to assess the movements during the performance related to sound production and sound preparation. Subsequent cluster analyses sought to identify coupling of limb segment movements that may best characterize performance styles present in the performance. The analysis revealed four clusters that were thought to reflect performance styles of expressive performance, postural sway, energy efficiency, and a blend of the former styles. More specifically, the expressive cluster was best characterized by limb movements occurring along the vertical z-axis, whereas the postural sway cluster was characterized by forwards and backwards motions of the torso and upper limbs. The energy efficient cluster was characterized by movements of the body moving left to right along the marimba, whereas the blended style demonstrated limited delineation from the alternate styles. Such findings were interpreted as evidence that performance styles occur within a framework of biomechanical constraints and hierarchical stylistic factors. Overall, the results provided a more holistic understanding of motor execution in percussion performance.