Although cymbal performance requires fine motor control, its techniques have relied mainly on the player's intuition rather than concrete methodology. This study employs a music data science approach to analyze the playing motion of pair cymbals using an optical motion capture system. An expert percussionist participated in a recording session with a 9-camera OptiTrack setup (120 fps), and the author extracted six motion parameters: relative distance, relative velocity, 3-D angle between cymbals, and rotational angles (rolling, yawing, and pitching). These were temporally aligned with acoustic onsets derived from waveform analysis. Results showed that key motion events—such as angle minima and peak velocity—occurred within 200 ms of sound onset, with distinct timing patterns depending on dynamic level (f, mf, p) and mute condition. Muted strokes required longer wrist control durations, highlighting increased motion demands. Furthermore, 14 combined motion and audio features were used to map playing samples using multidimensional scaling (MDS), revealing clustering patterns by dynamics, especially for soft strokes. These findings demonstrate the potential of music data science to provide quantitative insights into nuanced performance techniques, supporting pedagogical feedback, performer training, and future computational studies in musical expression.
Miura et al. (Wed,) studied this question.