Automatic classification of delphinids based on the representative frequencies of whistles

Abstract

Classification of odontocete species remains a challenging task for passive acoustic monitoring. Classifiers that have been developed use spectral features extracted from echolocation clicks and whistle contours. Most of these contour-based classifiers require complete contours to reduce measurement errors. Therefore, overlapping contours and partially detected contours in an automatic detection algorithm may increase the bias for contour-based classifiers. In this study, classification was conducted on each recording section without extracting individual contours. The local-max detector was used to extract representative frequencies of delphinid whistles and each section was divided into multiple non-overlapping fragments. Three acoustical parameters were measured from the distribution of representative frequencies in each fragment. By using the statistical features of the acoustical parameters and the percentage of overlapping whistles, correct classification rate of 70.3% was reached for the recordings of seven species (Tursiops truncatus, Delphinus delphis, Delphinus capensis, Peponocephala electra, Grampus griseus, Stenella longirostris longirostris, and Stenella attenuata) archived in MobySound.org. In addition, correct classification rate was not dramatically reduced in various simulated noise conditions. This algorithm can be employed in acoustic observatories to classify different delphinid species and facilitate future studies on the community ecology of odontocetes.

Acoustic classification of dolphins in the California Current using whistles, echolocation clicks, and burst pulses

“…, ; Roch et al . ; Lin and Chou ) and thus, we had low expectations for our ability to differentiate Delphinus species from each other (or from S. coeruleoalba ) at the event level. Although most species had high correct classification scores in the event classifier, only 71% of Delphinus capensis (Dc) events were correctly classified, with one event misclassified as Delphinus delphis (Dd) and another as Stenella coeruleoalba (Sc).…”

“…, , Azzolin et al . ; Lin and Chou ). Nonetheless, there remains significant ambiguity in species classifications based solely on whistle characteristics.…”

“…Moreover, there is great overlap in whistle characteristics across many species , Azzolin et al 2014, Frasier et al 2016. Despite the high variation at an individual level and the extensive overlap in whistle characteristics among species, acoustic classifiers based on whistle features have shown reasonable success in acoustic species classification , Azzolin et al 2014Lin and Chou 2015). Nonetheless, there remains significant ambiguity in species classifications based solely on whistle characteristics.…”

Automatic classification of whistles from coastal dolphins of the southern African subregion

“…3. (Color online) The success rates from published whistle classification tests run with different numbers of species (red and blue squares show, respectively, the results of this study for the three species-classifier and the four-population classifier; Steiner, 1981;Wang et al, 1995;Matthews et al, 1999;Rendell et al, 1999;Oswald et al, 2003;Oswald et al, 2007;Roch et al, 2007;Gannier et al, 2010;Caillat, 2013;Gillespie et al, 2013;Azzolin et al, 2014;Parada and Cardenal-L opez, 2014;Lin and Chou, 2015).…”

“…Most often classification is based on discriminant function analysis (DFA; Rendell et al, 1999), classification and regression trees (CART; Gannier et al, 2010) or both methods (Oswald et al, 2007) using standard parameters extracted from the whistle contour, although spectral analysis has also been employed (Roch et al, 2007). Several whistle detection and classification programs have been developed (Oswald et al, 2007;Gillespie et al, 2013;Lin et al, 2012;Lin and Chou, 2015). Some, such as ROCCA (Oswald et al, 2007), require a human user to manually select high quality whistles suitable for classification, the fundamental frequency of which is then extracted and classified.…”