Skip to main content
. Author manuscript; available in PMC: 2018 May 29.
Published in final edited form as: Anim Cogn. 2015 Nov 18;19(2):285–315. doi: 10.1007/s10071-015-0933-6

Figure 1. Extraction of the Predefined Acoustical Features (PAFs) and flow-chart showing the classification procedure using these parameters.

Figure 1

Five acoustical parameters were obtained from the temporal amplitude envelope of the sound (middle top, in blue), two parameters characterized the amplitude of the signal (middle top, in black), eight acoustical parameters were derived from the spectral amplitude envelope (middle center, in red) and seven acoustical parameters described the time varying fundamental (bottom center, in green). The fundamental is shown as a black line on the spectrogram. The fundamental is only extracted when the pitch saliency is greater than 0.5. These 22 acoustical parameters were then used to train two classifiers in vocalization category discrimination: a Random Forest and a Fisher Linear Discriminant Analysis. Performance was assessed by cross-validation. See Methods for more details on the calculation of the parameters and on the classification procedure.