Abstract
A large variety of sound sources in the ocean, including biological, geophysical, and man-made, can be simultaneously monitored over instantaneous continental-shelf scale regions via the passive ocean acoustic waveguide remote sensing (POAWRS) technique by employing a large-aperture densely-populated coherent hydrophone array system. Millions of acoustic signals received on the POAWRS system per day can make it challenging to identify individual sound sources. An automated classification system is necessary to enable sound sources to be recognized. Here, the objectives are to (i) gather a large training and test data set of fin whale vocalization and other acoustic signal detections; (ii) build multiple fin whale vocalization classifiers, including a logistic regression, support vector machine (SVM), decision tree, convolutional neural network (CNN), and long short-term memory (LSTM) network; (iii) evaluate and compare performance of these classifiers using multiple metrics including accuracy, precision, recall and F1-score; and (iv) integrate one of the classifiers into the existing POAWRS array and signal processing software. The findings presented here will (1) provide an automatic classifier for near real-time fin whale vocalization detection and recognition, useful in marine mammal monitoring applications; and (2) lay the foundation for building an automatic classifier applied for near real-time detection and recognition of a wide variety of biological, geophysical, and man-made sound sources typically detected by the POAWRS system in the ocean.
Highlights
A large-aperture densely-populated coherent hydrophone array system typically detects hundreds of thousands to millions of acoustic signals in the 10 Hz to 4000 Hz frequency range for each day of observation in a continental shelf ocean via the passive ocean acoustic waveguide remote sensing (POAWRS) technique [1,2]
The marine mammal vocalization data, that include fin whale vocalizations obtained from POAWRS sensing, were partially processed at sea and further analyzed in post-processing
The set of 12 features for each acoustic signal detection is automatically calculated as a subroutine in our POAWRS processing software right after signal detection from beamformed spectrogram analysis, and available for input to the classification algorithms
Summary
A large-aperture densely-populated coherent hydrophone array system typically detects hundreds of thousands to millions of acoustic signals in the 10 Hz to 4000 Hz frequency range for each day of observation in a continental shelf ocean via the passive ocean acoustic waveguide remote sensing (POAWRS) technique [1,2]. We focus our efforts on developing automatic classifers for fin whale vocalizations detected in the Norwegian and Barents Seas during our Norwegian Sea 2014 Experiment (NorEx14) [1]. The fin whale vocalization signals have been previously detected, identified and manually labeled via semiautomatic analysis followed by visual inspection. A total of approximately 170,000 fin whale vocalizations have been identified and extracted from the coherent hydrophone array recordings of NorEx14 [1]. The main types of fin whale vocalizations observed were the 20 Hz pulse, the 18–19 Hz backbeat pulse, the 130 Hz upsweep pulse, and the
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call