Vocalization based Individual Classification of Humpback Whales using Support Vector Machine

Mazhar, Suleman; Ura, Tamaki; Bahl, Rajendar

doi:10.1109/oceans.2007.4449356

Cited by 19 publications

(11 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Various approaches for vocalization classification have been applied to birds, land and marine mammals, including the application to acoustic censusing [62]. In cetaceans, various techniques have been employed in vocalization classification from their cepstral features including dynamic time warping [63], neural network [64], Gaussian mixture models, hidden Markov models [65], multi-class support vector machine model [66], and multivariate discriminant analysis [67]. Here we employ pitch-tracking to extract key features of humpback D-moan vocalizations and apply the centroid-based K-means [68] method to classify them.…”

Section: Discussionmentioning

confidence: 99%

Diel and Spatial Dependence of Humpback Song and Non-Song Vocalizations in Fish Spawning Ground

Huang¹,

Wang²,

Ratilal³

2016

Remote Sensing

View full text Add to dashboard Cite

Abstract:The vocalization behavior of humpback whales was monitored over vast areas of the Gulf of Maine using the passive ocean acoustic waveguide remote sensing technique (POAWRS) over multiple diel cycles in Fall 2006. The humpback vocalizations comprised of both song and non-song are analyzed. The song vocalizations, composed of highly structured and repeatable set of phrases, are characterized by inter-pulse intervals of 3.5 ± 1.8 s. Songs were detected throughout the diel cycle, occuring roughly 40% during the day and 60% during the night. The humpback non-song vocalizations, dominated by shorter duration (≤3 s) downsweep and bow-shaped moans, as well as a small fraction of longer duration (∼5 s) cries, have significantly larger mean and more variable inter-pulse intervals of 14.2 ± 11 s. The non-song vocalizations were detected at night with negligible detections during the day, implying they probably function as nighttime communication signals. The humpback song and non-song vocalizations are separately localized using the moving array triangulation and array invariant techniques. The humpback song and non-song moan calls are both consistently localized to a dense area on northeastern Georges Bank and a less dense region extended from Franklin Basin to the Great South Channel. Humpback cries occur exclusively on northeastern Georges Bank and during nights with coincident dense Atlantic herring shoaling populations, implying the cries are feeding-related.

show abstract

Section: Discussionmentioning

confidence: 99%

Diel and Spatial Dependence of Humpback Song and Non-Song Vocalizations in Fish Spawning Ground

Huang¹,

Wang²,

Ratilal³

2016

Remote Sensing

View full text Add to dashboard Cite

show abstract

“…Automatic classifiers were further developed in [24] to distinguish humpback whale song sequences from nonsong calls in the Gulf of Maine by first applying Bag of Words to build feature vectors from beamformed time-series signals, calculating both power spectral density and MFCC features, and then employing and comparing the performances of Support Vector Machine (SVM), Neural Networks, and Naive Bayes in the classification. Identification of individual male humpback whales from their song units was investigated in [27], via extracting Cepstral coefficients for features and then applying SVM for classification. In [28], blue whale calls were classified using neural network with features derived from short-time Fourier and wavelet transforms.…”

Section: Introductionmentioning

confidence: 99%

Comparing Performances of Five Distinct Automatic Classifiers for Fin Whale Vocalizations in Beamformed Spectrograms of Coherent Hydrophone Array

et al. 2020

View full text Add to dashboard Cite

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.

show abstract

“…The bandwidth of the song lies from 50 to 4000 Hz [22]. We opted for this natural vocal as a carrier for this research due to its frequency and duration [23]. Low frequency allows for a larger transmission distance, which benefits the accomplishment of furtive communication in a bigger region.…”

Section: Introductionmentioning

confidence: 99%

Symmetry Oriented Covert Acoustic Communication by Mimicking Humpback Whale Song

et al. 2019

View full text Add to dashboard Cite

To meet the increasing demand of covert underwater acoustic communication, biologically inspired mimicry communication watermarking the data in symmetrical humpback whale song is presented. Mimicry is an entirely different approach from traditional covert communication where data are transmitted by spreading the waveform at a low signal to noise ratio. In this innovative technique, the carrier signal is imitated symmetrical to the ocean background noise, which can be shipping noise, anthropological noise, or the vocals emitted by sea animals. The eavesdropper can detect the communication signal, but will assume it to be real ocean noise due to its symmetry. It excludes the mimicked signal from recognition, which makes the communication covert. In this research, we watermarked the covert information in humpback whale song using discrete cosine transform in the frequency domain. The mimicked symmetrical signal provided excellent imperceptibility with the real song and an outstanding camouflage effect was calculated. We validated the novel concept by simulation and underwater tank experiment. 10−4 BER was achieved in the underwater tank experiment, which was diminished to zero error by using matching pursuit estimation and virtual time reversal equalization. This novel bionic covert communication technique is feasible for clandestine underwater acoustic communication in the presence of an eavesdropper with better imperceptibility.

show abstract

Vocalization based Individual Classification of Humpback Whales using Support Vector Machine

Cited by 19 publications

References 6 publications

Diel and Spatial Dependence of Humpback Song and Non-Song Vocalizations in Fish Spawning Ground

Diel and Spatial Dependence of Humpback Song and Non-Song Vocalizations in Fish Spawning Ground

Comparing Performances of Five Distinct Automatic Classifiers for Fin Whale Vocalizations in Beamformed Spectrograms of Coherent Hydrophone Array

Symmetry Oriented Covert Acoustic Communication by Mimicking Humpback Whale Song

Contact Info

Product

Resources

About