Potential of Complexity Science to Decoding the Bioacoustic Language of Sperm Whales

Tuesday, May 7, 2019

11.00 a.m.

ISI seminar room 1st floor

David Gruber Presidential Professor of Biology, City University of New York, Baruch College & The Graduate Center Research Associate, Invertebrate Zoology, American Museum of Natural History Explorer, National Geographic Society

Sperm whales, Physeter macrocephalus, possess the largest animal brain and also have one of the most complicated biological sound generators in the animal kingdom. We share the planet with these majestic creatures, who also live in family units and have complex vocal language, yet they remain largely mysterious. They produce sharp broadband pulses known as “clicks” at regular repetition rates of 1-2 seconds and with frequency reaching 30 kHz. Clicks are emitted at various repetition rates and patterns to produce sounds such as slow clicks, clangs, creaks, chirrups and codas. The sounds emitted by sperm whales have been studied since first being identified in the 1950s, yet a deep understanding of their sonic language remains to be deciphered.  This talk will discuss machine learning, computational and other modern techniques and technologies that may provide a deeper knowledge of their communication and behaviors. This includes examples of convolutional neural networks approaches to extract finer-scale details from cetacean spectrograms and long short-term memory deep recurrent neural networks to perform classification tasks such as vocal clan classification.