Kinematics Feature Selection of Expressive Intentions in Dyadic Violin Performance
Paper in proceedings, 2017
There is evidence that bodily movement plays a crucial role in regulating expressivity in music performance. Advances in technologies related to human movement research (e.g. motion capture using infrared cameras) give us the opportunity to study bodily motion with millimeter precision. Consequently, we can extract fine-grained kinematic characteristics and perform statistical learning techniques in order to identify similarities and differences in spatial accuracy of intended expressive movements. In this study, we applied feature extraction and feature generation algorithms to identify the kinematic characteristics that better predict expressive intentions. The results suggest that musical expressivity is not physically rendered in similar movement patterns during perception and during production of dyadic musical performance. We propose that future studies should focus on the interaction between motor experience and visual perception of expressivity.