Perception of string quartet synchronization

Alan M. Wing, Satoshi Endo, Tim Yates, Adrian Bradbury

Timing variation in small group musical performance results from intentional, expressive, and unintentional, error components in individual player timing. These timing fluctuations produce variability in between-player note asynchrony and require timing adjustments to keep the ensemble together. The size of the adjustments relative to the asynchrony (correction gain) affects the amount and nature of asynchrony variability. We present new listening tests to estimate thresholds for perception of between-player asynchrony variability and to determine whether listeners use differences in the nature of the variability, as well as in its magnitude, to judge asynchrony. In two experiments, computer-simulated ensemble performances of a 48-note excerpt from Haydn Op. 74 No. 1 were generated. Between-player note asynchrony was systematically manipulated in terms of level of within-player timing variability (Experiment 1) and correction gain (Experiment 2). On each trial, participants listened to two samples, one ("target") with more between-player asynchrony variability than the other ("test"), and reported which was "less together." In both experiments, the test sample correction gain was fixed at the statistically optimal value of 0.25 and the within-player timing variability was minimal (zero except for random variability in the initial note). In Experiment 1 the target correction gain was fixed at 0.25 and the timing variability was adjusted over trials by a staircase algorithm designed to converge on the level of asynchrony variability giving 75% correct identification. In Experiment 2 the timing variability in the target was set at half that in Experiment 1 and the correction gain was varied to converge on 75% correct identification. Our results show that the between-player asynchrony variability giving 75% correct identification in Experiment 2 was significantly lower than in Experiment 1. This finding indicates that people are sensitive to both the degree of variance and the micro-structure of the time-series of the asynchronies caused by differences in correction gain when judging lack of togetherness in quartet performance.