Who shall I say is calling? Validation of a caller recognition procedure in Bornean flanged male orangutan (Pongo pygmaeus wurmbii) long calls

Abstract

Acoustic individual discrimination has been demonstrated for a wide range of animal taxa. However, there has been far less scientific effort to demonstrate the effectiveness of automatic individual identification, which could greatly facilitate research, especially when data are collected via an acoustic localization system (ALS). In this study, we examine the accuracy of acoustic caller recognition in long calls (LCs) emitted by Bornean male orangutans (Pongo pygmaeus wurmbii) derived from two data-sets: the first consists of high-quality recordings taken during individual focal follows (N = 224 LCs by 14 males) and the second consists of LC recordings with variable microphone-caller distances stemming from ALS (N = 123 LCs by 10 males). The LC is a long-distance vocalization. We therefore expect that even the low-quality test-set should yield caller recognition results significantly better than by chance. Automatic individual identification was accomplished using software originally developed for human speaker recognition (i.e. the MSR identity toolbox). We obtained a 93.3% correct identification rate with high-quality recordings, and 72.23% with recordings stemming from the ALS with variable microphone-caller distances (20-420 m). These results show that automatic individual identification is possible even though the accuracy declines compared with the results of high-quality recordings due to severe signal degradations (e.g. sound attenuation, environmental noise contamination, and echo interference) with increasing distance. We therefore suggest that acoustic individual identification with speaker recognition software can be a valuable tool to apply to data obtained through an ALS, thereby facilitating field research on vocal communication.

Application of a semi-automated vocal fingerprinting approach to monitor Bornean gibbon females in an experimentally fragmented landscape in Sabah, Malaysia

“…Although some of the groups were still present in the same location, it is possible that there was movement of groups around the area that could have influenced our results. Also, the distance between the caller and the microphone has been shown to influence the accuracy of caller identification in orangutans (Spillmann et al 2017), and it is possible that differences in recording distances of the focal females between subsequent seasons could have influenced our results. Despite these limitations, our mean classification accuracy across seasons (78%) was substantially better than chance (25%) and it seems likely that with a larger data-set our classification accuracy would improve.…”

“…Passive acoustic monitoring is a non-invasive technique that utilizes sound recording devices to monitor vocal animals (Merchant et al 2015). e potential for passive acoustic monitoring to improve conservation efforts for terrestrial animals is widely recognized (Blumstein et al 2011;Wrege et al 2017), and bioacoustics techniques are being used to identify individuals in a wide variety of taxa including owls (Grava et al 2008), orangutans (Spillmann et al 2017) and tigers (Ji et al 2013), as well as for occupancy detection of primates (Heinicke et al 2015;Kalan et al 2015) and monitoring of primate group ranging and territory use (Kalan et al 2016). e use of bioacoustical methods to address ecological and conservation questions has become increasingly popular due to the increase in data storage capabilities and battery life, a decrease in size and cost of recording devices and the development of new methods for automating acoustic analyses (Blumstein et al 2011).…”

“…e use of Mel-frequency cepstral coefficients (MFCCs) provides a fully automated method of feature extraction that is standardized and reproducible (Mielke and Zuberbühler 2013). MFCCs have been used successfully in human speech recognition (Picone 1993;Han et al 2006), bird song classification (Chou et al 2008;Lee et al 2006), classification of species, call type and caller identity in blue monkeys (Mielke and Zuberbühler 2013) and discrimination between male orangutan individuals (Spillmann et al 2017).…”

Automatic acoustic identification of individuals in multiple species: improving identification across recording conditions

“…Also, song birds may change territories in subsequent years or even within a single season [27]. Some other studies of individual acoustic identification, on the other hand, provided evidence that machine learning acoustic identification can be robust in respect to possible long-term changes in the acoustic background but did not provide evidence of being generally usable for multiple species [30,32]. Therefore, the challenge of reliable generalization of the machine learning approach in acoustic individual identification across different conditions and different species has not yet been satisfactorily demonstrated.…”

“…Classical approaches that are more flexible include Gaussian mixture models (GMMs) and hidden Markov models (HMMs), previously used extensively in human speech technology [30,42]. These do not rely on a strongly fixed template but rather build a statistical model summarizing the spectral data that are likely to be produced from each individual.…”