Automated aural classification used for inter-species discrimination of cetaceans

Abstract

Passive acoustic methods are in widespread use to detect and classify cetacean species; however, passive acoustic systems often suffer from large false detection rates resulting from numerous transient sources. To reduce the acoustic analyst workload, automatic recognition methods may be implemented in a two-stage process. First, a general automatic detector is implemented that produces many detections to ensure cetacean presence is noted. Then an automatic classifier is used to significantly reduce the number of false detections and classify the cetacean species. This process requires development of a robust classifier capable of performing inter-species classification. Because human analysts can aurally discriminate species, an automated aural classifier that uses perceptual signal features was tested on a cetacean data set. The classifier successfully discriminated between four species of cetaceans-bowhead, humpback, North Atlantic right, and sperm whales-with 85% accuracy. It also performed well (100% accuracy) for discriminating sperm whale clicks from right whale gunshots. An accuracy of 92% and area under the receiver operating characteristic curve of 0.97 were obtained for the relatively challenging bowhead and humpback recognition case. These results demonstrated that the perceptual features employed by the aural classifier provided powerful discrimination cues for inter-species classification of cetaceans.

Repertoire and classification of non-song calls in Southeast Alaskan humpback whales (Megaptera novaeangliae)

“…Similarly, foundational studies of bowhead whale vocalizations rely almost exclusively on aural-visual classification despite a growing number of unique sound types (Clark, 1982;Clark and Johnson, 1984;W€ ursig et al, 1993;Stafford et al, 2008;Delarue et al, 2009), many of which are similar to humpback whale vocalizations. Automated detection of vocalizations is appealing, particularly with large data sets; however, even robust detectors are dependent on training sets (Binder and Hines, 2014). The three-part method described here may be an effective tool for developing representative training sets that can be applied to larger databases.…”

Impacts of environment-dependent acoustic propagation on passive acoustic monitoring of cetaceans

“…Since many PAM systems rely on the time-frequency characteristics of the vocalizations for detection and classification [1,4,19,20,21,22,23,24], signal distortion has the potential to negatively affect the accuracy of PAM systems [10,22,25]; however, little research has been directed towards this problem. Some authors acknowledge that propagation effects likely impact the accuracy of PAM systems, but do not analyze how their system is affected.…”

“…First, a general automatic detector is implemented with detection parameters set to achieve a high detection rate, while recognizing that this will likely generate many false detections. Then an automatic classifier is used to considerably reduce the number of false detections and identify the cetacean species [19]. This philosophy for detection/classification is employed to ensure that cetacean presence is noted, which is of particular importance when monitoring for at-risk species, or those that vocalize infrequently.…”

Fuzzy information transmission analysis for continuous speech features

“…Examples of acoustic classification problems apart from vowel recognition that satisfy the requirements for the applicability of the fuzzy FITA include automatic emotion classification (Ooi et al, 2014), animal vocalization classification (Clemins et al, 2005;Chen and Maher, 2006;Binder and Hines, 2014), aircraft classification (S anchez Fern andez et al, 2013), and musical genre classification (Tzanetakis and Cook, 2002). Where some of these studies do not exactly satisfy all the applicability requirements, they can be adjusted to fit the requirements by adding background noise or increasing the number of classes.…”