In this paper, we propose a novel algorithm for score-informed tracking of the fundamental frequency over the duration of single tones. The tracking algorithm is based on a peak-picking algorithm over spectral magnitudes and ensures time-continuous f0 -curves. From a set of 19 jazz solos from three saxophone and three trumpet players, we collected a set of 6785 f0 -contours in total. We report the results of two exploratory analyses. First, we compared typical contour feature values among different jazz musicians and different instruments. Second, we analyzed correlations between contour features and contextual parameters that describe the metrical position, the in-phrase position, and additional properties of each tone in a solo.

Download