Contour Detection

• Call statistics, as well as unsupervised clustering, are calculated on spectrotemporal contours.

• Each USV’s contour is constructed by first calculating the frequency at maximum amplitude for each time point in the spectrogram, and then cleaned by removing non-tonal features such as silence or broad spectrum noise. The tonality of each audio signal is calculated by dividing the geometric mean of the power spectrum by the arithmetic mean and subtracting from 1.

• The tonality and amplitude percentile thresholds may be changed in "Tools > Change Contour Threshold".

• Higher values of tonality and amplitude percentile will result in more conservative contours.