Nonresonant stridulation - a challenge for digital signal processing [abstract]

Gunther Tschuch & Denis J. Brothers (2002). Nonresonant stridulation - a challenge for digital signal processing [abstract]. Bioacoustics, Volume 13 (2): 181
Abstract: 

Many insects and other arthropods use stridulatory organs without resonant structures. In such systems the resulting signal (sound pressure versus time) contains a few to several hundred small spikes. The repetition period of the spikes is 0.5 to 2 ms in many species while the spikes are sometimes as short as 0.1 ms. In time-frequency representations (TFRs) like sonograms, the problem is to find the optimal time window to analyse such signals. With a short-time Fourier transform it is not possible to get a sharp 2-dimensional image with high time and high frequency resolution simultaneously. For nonresonant stridulatory signals it is better to use TFRs based on the Wigner distribution. Unfortunately, this produces cross components that complicate a more detailed analysis. To avoid these problems, special methods based on general Cohen's class TFRs have been developed, for example the Adaptive Optimal-Kernel TFR by Baraniuk and Jones. The use of new and established methods of digital signal processing will be discussed.