Musical performance is widely regarded to be one of the most breathtaking feats of human intelligence, but the nature of this creative act remains illusive and studies on expressive performance still lag behind those on composition and improvisation. However, with the advent of music in digital format, unprecedented computational prowess, and algorithmic development, it is now possible to probe scientifically and to mathematically model the creative work of performance.
Treating performance as a problem-solving task, a job of the performer is to find plausible units of coherence in the music at multiple time scales, and to exercise strategies to communicate convincingly such an analysis to the listener so as to influence the reception and parsing of the music. Over a series of studies, we show how computer analysis of performed music can reveal the shape and form of musical structures thus molded and projected in performance. By superimposing performed structures and score-based information, we show not only how these structures are made but also why, thereby enabling more nuanced understanding of the decisions that make for good, and perhaps great, performances.
Similarities between music and the human heartbeat has long been noted, but mainly in reference to normal heart rhythms. We show how abnormal heart rhythms exhibit behaviors akin to the variations introduced during performance, thereby opening the door to applying a host of analytical techniques hitherto reserved for music to cardiac arrhythmias. The novel musical view and description of abnormal heart signals has potential for facilitating personalized treatment and diagnoses. Finally, in a study with heart patients with biventricular pacemakers, we demonstrate that brain responses to structurally salient events in live music performance can have direct impact on cardiac electrophysiology.