Mardi 19 mars 2024 à 11h00 / Amphithéâtre François Canac, LMA
Abstract : The sperm whale, Physeter Macrocephalus, possesses the largest biosonar in nature. Made of multiple oil sacs, the sperm whale sonar is tailored to function from the sea surface down to a depth of 2 kilometers, emitting clicks as loud as 236 dB. It is multipurpose, as it produces clicks for either echolocation or socializing. However, the liquid wax that composes its sonar, made the sperm whales the target of whaling until 1986, when the remaining population was far too small to remain commercially viable, especially with the arrival of similar products from the petrochemical industry. The sperm whale population still faces some human threats, with the ingestion of plastic and collision with boats continuing to take a toll on their numbers. Studying sperm whales thus will have outcomes in multiple fields, in conservation, ethology, as well as in bioacoustics. Understanding the mechanism that governs the sperm whale sonar will help to study these other fields, as it is a key element in the sperm whale life. Aiming for this goal, we analyzes three databases with distinct characteristics, obtaining the trajectory of sperm whale dives. Clicks were also linked with the sperm whale that emitted them over multiple years of recording of the same population.
An efficient End-to-End deep learning classifier was trained to classify biosonar waveforms.
Finally, acoustically derived growth curves were obtain form those database.
Maxence Ferrari / chercheur LMA / équipe O&I