Comparative Analysis of the Vocal Repertoire of Eulemur: A Dynamic Time Warping Approach

Abstract

21The diversity of qualitative approaches and analytical methods has often undermined 22 comparative research on primate vocal repertoires. The purpose of the present work is 23 to introduce a quantitative method based on dynamic time warping to the study of 24 repertoire size in Eulemur spp. We obtained a large sample of calls of E. coronatus, 25 E. flavifrons, E. fulvus, E. macaco, E. mongoz, E. rubriventer and E. rufus, recorded 26between 1999 and 2013 from captive and wild lemurs. We inspected recordings 27 visually using spectrograms, then cut and saved high-quality vocal emissions to single 28 files for further analysis. We extracted the acoustic features of all vocalizations of a 29 species using the Hidden Markov Model toolkit, an application of dynamic time 30warping, then compared cepstral coefficients (a feature widely used in automatic 31 speaker recognition) pairwise. We analysed the results using Affinity Propagation 32 clustering. We found that Eulemur species share most of their vocal repertoire but 33 species-specific calls determine repertoire size differences. Repertoire size varied 34 from 9 and 14 vocalisation types among species, with a mean of 11. Group size is 35 thought to favour the evolution of vocal complexity at the species level but our results 36 suggest that this relationship should be reconsidered, as Eulemur rubriventer has the 37 largest vocal repertoire but shows a relatively small average group size when 38 compared to congeneric species.

Finding Meanings in Low Dimensional Structures: Stochastic Neighbor Embedding Applied to the Analysis of Indri indri Vocal Repertoire

“…Accordingly, in agreement with Peckre and colleagues [28], we expected to find a clearer clusterization of discrete calls and a weaker grouping accuracy of graded ones. Finally, in agreement with the “social complexity–vocal complexity hypothesis” [30] and the social complexity hypothesis for communicative complexity [28], we expected indris to possess a small repertoire size if compared to that of other lemurs [21] or other primates [36] living in larger social groups.…”

“…However, vocal repertoires may represent an ideal model for this kind of analysis. Indeed, the sounds investigation often implies the analyses of huge, high-dimensional datasets [21]. We used t-SNE to analyze the vocal repertoire of Indri indri , the largest living lemur and the only one producing coordinated vocal displays.…”

Human Non-linguistic Vocal Repertoire: Call Types and Their Meaning

“…We then examined the quality of the resulting clustering solution by measuring (1) the average Silhouette Index, which is a measure of compactness and purity of clusters, and (2) the similarity of the clustering solution to the clusters defined by the most common name of each sound chosen by the participants (cf. Gamba et al 2015 ).…”

Clarifying and expanding the social complexity hypothesis for communicative complexity

“…This bias, together with the diversity of quantitative methods applied to identify call types, currently results in substantial inhomogeneity among studies. For example, the vocal repertoire of the Mongoose lemur (Eulemur mongoz) has been characterized by visual inspection as composed of either 8 (Petter and Charles-Dominique 1979), 10 (Curtis and Zaramody 1999), or 14 (Gamba et al 2015) call types, whereas quantitative analyses suggested 9 (Gamba et al 2015) or 15 (Nadhurou et al 2015) different call types. Thus, there is currently no general agreement on how to identify different call types and, hence, how to assess vocal repertoire size objectively.…”