Automated detection of white whale (delphinapterus leucas) vocalizations in St. Lawrence estuary and occurrence pattern

Catherine Bédard, Yvan Simard

Abstract


A detailed behavioral and habit study on the vocalization of the white whale population, is presented. White whales are known for their high degree of acoustic activities and their vocalization are variable in time and frequency. An automated method using a sequence of signal processing algorithms is developed to detect sound of white whale. A threshold is applied to transform the spectrogram into a binary image on which residual noise is cleaned using two specific image filters. The frequency band of the vocalizations is relatively stable over the seven days of sampling, while the vocalization rate is variable from day to day. This method can detect all the diverse white whale calls and pulsed tones emerging in the signal after noise filtration. The vocalization rate intensity of the false detections is very low compared to that of the detected calls. The prime frequency band used by white whales in recorded data is slightly lower than the estimated frequency rate.

Keywords


Acoustic signal processing; Acoustic surface wave filters; Acoustic wave propagation; Algorithms; Data recording; Speech analysis; Image filters; Signal processing algorithms; White whales

Full Text:

PDF

Refbacks

  • There are currently no refbacks.