Deep Learning for Ethoacoustics of Orcas on three years pentaphonic continuous recording at Orcalab revealing tide, moon and diel effects

• Main Authors: Poupard, Marion, Best, Paul, Schlüter, Jan, Prévot, Jean-Marc, Symonds, Helena, Spong, Paul, Glotin, Hervé
• Other Authors: Laboratoire des Sciences de l'Information et des Systèmes (LSIS), Centre National de la Recherche Scientifique (CNRS)-Arts et Métiers Paristech ENSAM Aix-en-Provence-Université de Toulon (UTLN)-Aix Marseille Université (AMU), DYNamiques de l’Information (DYNI), Laboratoire d'Informatique et Systèmes (LIS), Université de Toulon (UTLN)-Centre National de la Recherche Scientifique (CNRS)-Aix Marseille Université (AMU)-Université de Toulon (UTLN)-Centre National de la Recherche Scientifique (CNRS)-Aix Marseille Université (AMU), Représentations musicales (Repmus), Sciences et Technologies de la Musique et du Son (STMS), Institut de Recherche et Coordination Acoustique/Musique (IRCAM)-Sorbonne Université (SU)-Centre National de la Recherche Scientifique (CNRS)-Institut de Recherche et Coordination Acoustique/Musique (IRCAM)-Sorbonne Université (SU)-Centre National de la Recherche Scientifique (CNRS), Université de Toulon (UTLN)-Centre National de la Recherche Scientifique (CNRS)-Aix Marseille Université (AMU), PNRIA
• Format: Conference Object
• Language: English
• Published: HAL CCSD 2019
• Subjects: Ethoacoustics; Deep Learning; Convolutional Neural Networks; Orcas; Orcinus orca; Cetaceans; Bioacoustics; Environmental factors; Soundscape; Big data; [SDE]Environmental Sciences; [SDE.BE]Environmental Sciences/Biodiversity and Ecology; [INFO.INFO-AI]Computer Science [cs]/Artificial Intelligence [cs.AI]; Canada; Orca
• Online Access: https://hal.archives-ouvertes.fr/hal-02445426; https://hal.archives-ouvertes.fr/hal-02445426/document; https://hal.archives-ouvertes.fr/hal-02445426/file/poupard2019_orcalab_ocean.pdf

International audience One of the best ways of studying animals that produce signals in underwater environments is to use passive acoustic monitoring (PAM). Acoustic monitoring is used to study marine mammals in oceans, and gives us information for understanding cetacean life, such as their behaviour, movement or reproduction. Automated analysis for captured sound is almost essential because of the large quantity of data. A deep learning approach was chosen for this task, since it has proven great efficiency for answering such problematics. This study focused on the orcas (Orcinus orca) of northern Vancouver Island, Canada, in collaboration with the NGO Orcalab which developed a multi-hydrophone recording station around Hanson Island to study orcas. The acoustic station is composed of 5 hydrophones and extends over 50 km 2 of ocean. Since 2016 we are continuously streaming the hydrophone signals to our laboratory at Toulon, France, yielding nearly 50 TB of synchronous multichannel recordings. The objective for this research is to do a preliminary analysis of the collected data and demonstrate influence of environmental factors (tidal, moon phase and daily period) on the orcas' acoustic activities.