Detection and species identification of baleen whale calls

Abstract

Baleen whales live over extensive home range and time scales. Study of how these animals use their vocalizations for communication requires massive data sampling over long periods. This paper describes a system for automating the sampling and analysis of baleen whale calls. Many species produce stable, homogeneous call structures which lend themselves to automated species identification. We have benchmarked a series of bioacoustical call identification algorithms against a set of blue and fin whale calls while systematically manipulating the signal-to-noise ratio. Blue (Balaenoptera musculus) and fin (B. physalus) whale calls are very stereotypical. Blue whale ‘‘A’’ and ‘‘B’’ calls have fundamental frequencies of approximately 17 Hz, narrow bandwidth, well-defined harmonic structure, and typical duration of 15–25 s. Fin whale ‘‘pulses’’ have fundamental frequencies of approximately 17 Hz, but are broadband in nature and short (approximately 1-s) duration. The results demonstrated a typical tradeoff of speed versus accuracy. The best algorithm was inserted into an underwater sound recording system and its signal-detection theoretic performance was quantified. Results will be discussed with respect to technological, ecological, and conservation aspects of baleen whale bioacoustics. [Project CS-1082 of the Strategic Environmental Research and Development Program.]