10 April 2019 in Journal of The Royal Society Interface
Many animals emit vocal sounds which, independently from the sounds’ function, contain some individually distinctive signature. Thus the automatic recognition of individuals by sound is a potentially powerful tool for zoology and ecology research and practical monitoring. Here, we present a general automatic identification method that can work across multiple animal species with various levels of complexity in their communication systems. We further introduce new analysis techniques based on dataset manipulations that can evaluate the robustness and generality of a classifier. By using these techniques, we confirmed the presence of experimental confounds in situations resembling those from past studies. We introduce data manipulations that can reduce the impact of these confounds, compatible with any classifier. We suggest that assessment of confounds should become a standard part of future studies to ensure they do not report over-optimistic results. We provide annotated recordings used for analyses along with this study and we call for dataset sharing to be a common practice to enhance the development of methods and comparisons of results.
13 August 2013 in PLOS ONE
Two important questions in bioacoustics are whether vocal repertoires of animals are graded or discrete and how the vocal expressions are linked to the context of emission. Here we address these questions in an ungulate species. The vocal repertoire of young domestic pigs, Sus scrofa, was quantitatively described based on 1513 calls recorded in 11 situations. We described the acoustic quality of calls with 8 acoustic parameters. Based on these parameters, the k-means clustering method showed a possibility to distinguish either two or five clusters although the call types are rather blurred than strictly discrete. The division of the vocal repertoire of piglets into two call types has previously been used in many experimental studies into pig acoustic communication and the five call types correspond well to previously published partial repertoires in specific situations. Clear links exist between the type of situation, its putative valence, and the vocal expression in that situation. These links can be described adequately both with a set of quantitative acoustic variables and through categorisation into call types. The information about the situation of emission of the calls is encoded through five call types almost as accurately as through the full quantitative description.
14 August 2015 in PLOS ONE
Humans as well as many animal species reveal their emotional state in their voice. Vocal features show strikingly similar correlation patterns with emotional states across mammalian species, suggesting that the vocal expression of emotion follows highly conserved signalling rules. To fully understand the principles of emotional signalling in mammals it is, however, necessary to also account for any inconsistencies in the way that they are acoustically encoded. Here we investigate whether the expression of emotions differs between call types produced by the same species. We compare the acoustic structure of two common piglet calls—the scream (a distress call) and the grunt (a contact call)—across three levels of arousal in a negative situation. We find that while the central frequency of calls increases with arousal in both call types, the amplitude and tonal quality (harmonic-to-noise ratio) show contrasting patterns: as arousal increased, the intensity also increased in screams, but not in grunts, while the harmonicity increased in screams but decreased in grunts. Our results suggest that the expression of arousal depends on the function and acoustic specificity of the call type. The fact that more vocal features varied with arousal in scream calls than in grunts is consistent with the idea that distress calls have evolved to convey information about emotional arousal.
7 June 2019 in Methods in Ecology and Evolution
Identity signals have been studied for over 50 years but, and somewhat remarkably, there is no consensus as to how to quantify individuality in animal signals. While there is a variety of different metrics to quantify individuality, these methods remain un‐validated and the relationships between them unclear. We contrasted three univariate and four multivariate identity metrics (and their different computational variants) and evaluated their performance on simulated and empirical datasets. Of the metrics examined, Beecher's information statistic (HS) performed closest to theoretical expectations and requirements for an ideal identity metric. It could be also easily and reliably converted into the commonly used discrimination score (and vice versa). Although Beecher's information statistic is not entirely independent of study sampling, this problem can be considerably lessened by reducing the number of parameters or by increasing the number of individuals in the analysis. Because it is easily calculated, has superior performance, can be used to quantify identity information in single variable or in a complete signal and because it indicates the number of individuals who can be discriminated given a set of measurements, we recommend that individuality should be quantified using Beecher's information statistic in future studies. Consistent use of Beecher's information statistic could enable meaningful comparisons and integration of results across different studies of individual identity signals.
9 October 2015 in Bioacoustics
1 November 2013 in Behavioural Processes