Markov-switching state space models for uncovering musical interpretation

For concertgoers, musical interpretation is the most important factor in determining whether or not we enjoy a classical performance. Every performance includes mistakes—intonation issues, lost note, an unpleasant sound—but these are all easily forgotten (or unnoticed) when performer engages her audience, imbuing piece with novel emotional content beyond vague instructions inscribed on printed page. In this research use data from CHARM Mazurka Project—46 professional recordings of Chopin’s Op. 68 No. 3 by consummate artists—with goal elucidating musically interpretable decisions. We focus specifically each performer’s tempo examining interonset intervals note attacks recording. To explain decisions, develop switching state space model and estimate it maximum likelihood, combined prior information gained music theory practice. estimated parameters to quantitatively describe individual decisions compare recordings. These comparisons suggest methods for informing instruction, discovering listening preferences analyzing performances.