The DeepSqueak tool identifies marine mammal challenges

Hiding under the surface of the ocean, marine mammals use sound to navigate, detect prey and a wide range of natural behaviors. Passive acoustic data from underwater environments can provide valuable information about these animals, such as their presence or absence in the area, their density and quantity, and their vocal response to anthropogenic noise sources.

As the size and number of acoustic data sets increase, accurate and rapid matching of bioacoustic signals with relevant sources becomes more complex and important. This is especially difficult in noisy, natural acoustic environments.

Elizabeth Ferguson of Ocean Science Analytics will talk about how DeepSqueak, a deep learning tool, can classify underwater acoustic signals at the 182nd meeting of the Acoustic Society of America during her presentation “Developing Deep Neural Networks to Detect Marine Mammal Challenges Using open sourcehandy tool. ”The session will take place on May 23 at 11:25 a.m. in the eastern United States as part of a conference at the Sheraton Denver Downtown Hotel.

Spectrograms show how acoustic signals different frequencies change over time. They look like heat maps, with brighter areas indicating higher sound intensity at this frequency and time. DeepSqueak uses deep neural network image recognition and classification techniques to identify important characteristics in spectrograms and then map these features to specific sources.

“Although we used DeepSqueak to detect underwater sounds, this handy open source tool would be useful for a variety of terrestrial species,” Ferguson said. “Call detection capabilities extend to frequencies below the ultrasonic sounds for which it was originally intended. Thanks to this and DeepSqueak’s ability to detect variable types of calls, the development of neural networks is possible for many species of interest.”

DeepSqueak was originally designed to classify ultrasonic signals from rodents, but its neural network allows technology to adapt to detect sounds at other frequencies. Ferguson and her team used a method and data from hydrophones on a coastal hardy array of Ocean Observatory initiatives to detect teal whalesdolphins and finvalshaving very variable calls with a wide range of frequencies.