Comparing neural networks with conventional classifiers for fin whale vocalizations in beamformed spectrograms of coherent hydrophone array

Abstract

Listen

Translate article

A large variety of sound sources in the ocean, including biological, geophysical and man-made activities 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-sampled coherent hydrophone array. 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 a large training data set of fin whale and other vocalizations are gathered after manual inspection and labelling. Next, multiple classifiers including neural networks, logistic regression, support vector machine (SVM) and decision tree are built and tested for identifying the fin whale and other vocalizations from the enormous amounts of acoustic signals detected per day. The neural network classifier will use beamformed spectrograms to classify acoustic ...