![]() ![]() Applying the model Fn 1⁄4 of a uniform tube closed at one end (Fitch and Reby 2001), we 4 approximated L formant frequencies as: where n are formant numbers, L is VTL and c is the speed of sound in air (350 ms − 1 ). The mean VTL values of dissected specimens served to establish LPC parameters 2 n for 1 c creation of formant tracks. Weenink, University of Amsterdam, Netherlands, ). The first four formants were measured using linear predictive coding (LPC) with the Praat DSP package, version 4.3.21 (P. All measurements were exported to Microsoft Excel (Microsoft Corp., Redmond, WA, USA). This option recognizes periodic components in the sound signal and measures their period with 0.25-ms precision, which allowed us to calculate the f0 with 1-Hz precision. With “ Autocorrelation ” option of Avisoft, we measured the mean f0 of the entire call. From the mean power spectrum of the entire call, we measured the maximum amplitude frequency (fpeak) and the upper, medium and lower quartiles (q25, q50 and q75), covering respectively 25%, 50%, and 75% of call energy. We measured the duration with the standard marker cursor in the spectrogram window (sampling frequency 48 kHz, Hamming window, FFT 1024 points, frame 50% and overlap 93.75%) using Avisoft SASLab Pro software (Avisoft Bioacoustics, Berlin, Germany) ( Fig. 1 and 3), we measured the same 10 variables (duration, f0, 4 power variables and the first 4 formants) and for each snort 5 variables (duration and 4 power variables). ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |