Characterizing acoustic features of marine animal sounds

Abstract

Software tools were designed to characterize the acoustic features of marine animal sounds. These have resulted in a set of calculated measurements that summarize particular aspects of sound sequences. The specificity of these measurements was enhanced by adjusting calculations to compensate for ambient noise. The sound measures included statistics for Aggregate Bandwidth, Intensity, Duration, Amplitude Modulation, Frequency Modulation, Short-term Bandwidth, Center Frequency, and Amplitude Frequency Interaction. The efficacy of noise compensation was tested for each statistic. Then, the sound measures were tested on a subset of 200 sequences of marine animal sounds, including sequences from 20 species: six baleen whales, 13 toothed species, and one seal. The statistics were reviewed for each species and a graphical comparison of all species was generated using principal components analysis. Preliminary results confirm that such sounds can be classified by means of relatively simple statistical algorithms, and we are encouraged to continue toward a system for automatic classification of marine animal sounds.