Automatic prosodic clustering of humpback whales song (original) (raw)
Related papers
Subunit definition and analysis for humpback whale call classification
Applied Acoustics, 2010
Songs of humpback whales (Megaptera novaeangliae) have been studied for several years to gain a deeper insight on the intraspecific social interactions. Such a complex acoustic display is indeed thought to play an important role in both the mating ritual and male to male interaction. Hence, the need to classify the unit constituents of a song objectively and systematically has become crucial to allow processing large data sets. We propose a new approach for song segmentation based on the definition of subunits. Songs of humpback whales collected in Madagascar in August 2008 and 2009 were segmented using an energy detector with a double threshold and classified automatically with a clustering algorithm using MFCCs: the results, which were checked against a manual classification, showed that the use of subunit as the basic constituent of a song rather than the unit produces a more accurate classification of the calls. Such results were expected given that subunits are generally shorter in duration and less variable in terms of their frequency content and so their characteristics are more easily captured by an automatic classifier. Analysis of songs from other years and various areas of the World is necessary to corroborate the repeatability of the method proposed.
Bioacoustics, 2020
Humpback whale song is comprised of well-structured distinct levels of organization: combinations of sounds, repetition of combinations, and a sequence of repetitions, which have no clear silent intervals. This continuous sound output can be hard to delimit, rather, it could be interpreted as a long series of states of a system. Recurrence plots are graphical representations of such series of states and have been used to describe animal behavior previously. Here, we aim to apply this tool to visualize and recognize structures traditionally used in inferences about behavior (songs and themes) in the series of units manually extracted from recordings of humpback whales. Data from the Abrolhos bank, Brazil were subjected to these analyses. Our analytical tool has proven efficient in identifying themes and songs from continuous recordings avoiding some of the human perception bias and caveats. Furthermore, our song extraction is robust to errors coming from both manual and automated transcriptions, constructing a level of description largely independent of the first stage of analysis. HIGHLIGHTS • The complex hierarchical and cyclical structure of humpback whale song can be visually represented in recurrence plots • Recurrence plots of humpback whale singing provide insights into the intraindividual variation in song structures of a male • Our tool can help extracting structures, from long recordings, removing some human perceptual caveats
Understanding the structure of humpback whale songs (L)
The Journal of the Acoustical Society of America, 2012
This Letters section is for publishing (a) brief acoustical research or applied acoustical reports, (b) comments on articles or letters previously published in this Journal, and (c) a reply by the article author to criticism by the Letter author in (b). Extensive reports should be submitted as articles, not in a letter series. Letters are peer-reviewed on the same basis as articles, but usually require less review time before acceptance. Letters cannot exceed four printed pages (approximately 3000-4000 words) including figures, tables, references, and a required abstract of about 100 words.
A Quantitative Technique to Compare and Classify Humpback Whale (Megaptera novaeangliae) Sounds
Ethology, 2010
In an attempt to minimize observer bias, numerical taxonomy methods were used to describe and classify humpback whale sounds. The spectrograms (N = 1255) were digitized into a 16 × 21 binary matrix. The rows were 16 frequencies selected on a logarithmic scale (0.12–8 kHz). The columns were 21 time samples taken every 0.1 s. Each point of the matrix was coded 1 if it lay over part of the sound. Other binary variables were added to code for relative intensity within a sound, frequency modulation and amplitude modulation. The sounds were then compared using the Jaccard similarity coefficient for binary data, and classified with average linkage cluster analysis. This technique produced 115 clusters, which were compared with my aural and visual impressions of the sounds. I agreed with most major categories identified by cluster analysis, but many small clusters had to be fused to other categories. This was partially due to the technique used, and to the complexity of the repertoire under study. Improvements are proposed to further reduce observer bias in classification of sounds, and thus make studies of animal communication performed by different researchers or on different species more easily comparable.
Sound production by singing humpback whales
The Journal of the Acoustical Society of America, 2010
Sounds from humpback whale songs were analyzed to evaluate possible mechanisms of sound production. Song sounds fell along a continuum with trains of discrete pulses at one end and continuous tonal signals at the other. This graded vocal repertoire is comparable to that seen in false killer whales ͓Murray et al. ͑1998͒. J. Acoust. Soc. Am. 104, 1679-1688͔ and human singers, indicating that all three species generate sounds by varying the tension of pneumatically driven, vibrating membranes. Patterns in the spectral content of sounds and in nonlinear sound features show that resonating air chambers may also contribute to humpback whale sound production. Collectively, these findings suggest that categorizing individual units within songs into discrete types may obscure how singers modulate song features and illustrate how production-based characterizations of vocalizations can provide new insights into how humpback whales sing.
Marine Mammal Science, 2013
Consistent and well-defined criteria for the classification and measurement of humpback whale song features are essential for robust comparisons between investigators. Song structure terminology has been well-established and used by many authors, though at times inconsistently. This review discusses the development of the nomenclature describing humpback song and explores the potential significance of the often-overlooked variation in song patterns. Within the hierarchical definition of humpback song, the most problematic issues arise from the inconsistent delineation of phrase types, and the use of the metric of song duration without regards to variability in thematic sequence. With regards to the former, a set of guidelines is suggested to facilitate consistent delineation of phrases. With regards to the latter, current research demonstrates that the "song duration" metric has resulted in the disregard of variability at this level, which is more widespread than traditionally reported. An exemplar case is used to highlight the problem inherent in defining and measuring song duration. Humpback song is evaluated within the framework of avian songbird research, and a shift in analysis paradigm is recommended, towards phrase-based analyses in which sequences of phrases are treated as a salient feature of song pattern.
1998
Calls of killer whales, Orcinus orca, were analyzed using comuted sound features to classify sound patterns and identify call similarities. Calls were classified and separated according to the pod/family group within clans identified previously by John Ford (U. BC) in the Vancouver whale populations. Acoustic characteristics of the same call type from different individuals were extremely similar, so that discrimiting these different sounds was the goal. The WHOl AcoUStat pro and associated database systems were used to define numrical statistics for each call, and then, these were comed to sort and classify the sounds. The results were in agreemet with Ford i s descriptions of the calls derived from visual inspection of sound spectrograms of calls. The classification analyses demonstrated that although specific shaed calls fro different killer whales were much alike, they could be sorted by the pod/ subpod of the whales producing the calls. A typical analysis, for example, of the N4 call from Clan A (Vancouver, BC), classified 97% of the calls correctly accordig to the po/famly of the whales producing the calls. Remaing calls were vaiant, and likely a result of individual differences in call sounds. Similar classification analysis were tested on unorted, unanalyzed recordings from different populations of whales, and these too could be distinguished, with 98.5% correct separation of the calls.
The Journal of the Acoustical Society of America, 2015
The songs of the male humpback whales have traditionally been associated with breeding activities at low latitude breeding grounds during winter. This study provides the first detailed analysis of humpback whale songs recorded in the subarctic waters of Iceland using passive acoustic recorders. Recordings were collected during three winter seasons: 2008-2009, 2009-2010, and 2011 during which singing was detected in all seasons. Peak song occurrence was during January-February in all years; this coincides with the timing of the peak breeding season of humpback whales in the Northern hemisphere. A total of 2810 song units from all years were measured and statistically divided into 14 groups, which constructed 25 phrases. The song unit repertoires included stable song unit types that occurred frequently in songs during all years while the occurrence of other song unit types varied more between years. Around 60% of the phrases were conserved between the first two study seasons, while th...