Project details and research

Origination: Exploring Dolphin Intelligence

What began as a quest for two-way acoustic communication between dolphins and humans has led to DeepAcoustics, an AI software model that detects a broad range of marine mammal call types in noisier, underwater environments. DeepAcoustics is built upon DeepSqueak, with modifications developed at Ocean Science Analytics.

In DeepAcoustics, bioacousticians and other researchers now have a research tool that allows detection of a broad range of call types, targeting calls produced by select species. In addition, DeepAcoustics can characterize tonal whistles by automatically extracting whistle contours from recordings. Contours are sorted by shape into an optimal number of whistle clusters yielding a set of discrete whistle types that characterize the repertoire produced by a group of animals.

Peter presents his latest dolphin whistle technology: DeepAcoustics, at the 2022 Society for Marine Mammalogy Conference poster session.

Automated detection and categorization of whistle contours for characterizing the repertoire of bottlenose dolphins

Authors

  • Peter Sugarman

  • Kevin R. Coffey, Ph.D.

  • Elizabeth Ferguson

  • Jennifer Pettis Schallert

  • Gabriela Alongi

  • Heidi Lyn, Ph.D.

Connect with us

If you’d have specific questions about our work with DeepAcoustics, are looking for a speaker on this subject, or want to explore an opportunity to collaborate, please connect with us here.