Towards the Automatic Classification of Avian Flight Calls for Bioacoustic Monitoring

Abstract

Automatic classification of animal vocalizations has great potential to enhance the monitoring of species movements and behaviors. This is particularly true for monitoring nocturnal bird migration, where automated classification of migrants’ flight calls could yield new biological insights and conservation applications for birds that vocalize during migration. In this paper we investigate the automatic classification of bird species from flight calls, and in particular the relationship between two different problem formulations commonly found in the literature: classifying a short clip containing one of a fixed set of known species (N-class problem) and the continuous monitoring problem, the latter of which is relevant to migration monitoring. We implemented a state-of-the-art audio classification model based on unsupervised feature learning and evaluated it on three novel datasets, one for studying the N-class problem including over 5000 flight calls from 43 different species, and two realistic datasets for studying the monitoring scenario comprising hundreds of thousands of audio clips that were compiled by means of remote acoustic sensors deployed in the field during two migration seasons. We show that the model achieves high accuracy when classifying a clip to one of N known species, even for a large number of species. In contrast, the model does not perform as well in the continuous monitoring case. Through a detailed error analysis (that included full expert review of false positives and negatives) we show the model is confounded by varying background noise conditions and previously unseen vocalizations. We also show that the model needs to be parameterized and benchmarked differently for the continuous monitoring scenario. Finally, we show that despite the reduced performance, given the right conditions the model can still characterize the migration pattern of a specific species. The paper concludes with directions for future research.

Bioacoustic monitoring of animal vocal behavior for conservation

“…Performance depends in part on extraneous sources of sound (e.g., other species' calls) and the overall noisiness of the environment (e.g., anthropogenic noise, wind, rain), as well as the acoustic structure of the vocalizations, which is important in the choice of algorithm (Brandes, 2008;Cragg, Burger, & Piatt, 2015;Priyadarshani, Marsland, & Castro, 2018;Salamon et al, 2016;Towsey, Planitz, Nantes, Wimmer, & Roe, 2012). Since bioacoustic monitoring typically generates very large volumes of sound data, algorithms to detect vocalizations from sound files, termed call recognizers, are critical to the success of bioacoustics as a wildlife monitoring tool.…”

“…The reliability of recognizers, however, has been mixed, performing well for some species and poorly for others. Performance depends in part on extraneous sources of sound (e.g., other species' calls) and the overall noisiness of the environment (e.g., anthropogenic noise, wind, rain), as well as the acoustic structure of the vocalizations, which is important in the choice of algorithm (Brandes, 2008;Cragg, Burger, & Piatt, 2015;Priyadarshani, Marsland, & Castro, 2018;Salamon et al, 2016;Towsey, Planitz, Nantes, Wimmer, & Roe, 2012). For instance, Towsey et al (2012) were able to successfully detect the characteristic "whip-crack" of the eastern whipbird, Psophodes olivaceus (100% recall; 67% precision; 82% accuracy) using syntactic pattern recognition.…”

Ecoacoustics can detect ecosystem responses to environmental water allocations

“…That said, to date, automated analysis for ecoacoustic event detection is only in its infancy. While particularly in bird conservation the application of automated annotation techniques is slowly advancing (Holmes, McIlwrick, & Venier, 2014;Salamon et al, 2016;Stowell & Plumbley, 2014), for the time, being most analysis will be based on manual annotation. This is, of course, time consuming.…”

Robust sound event detection in bioacoustic sensor networks

“…The main branch has exactly the same architecture as the one that reported state-of-the-art results in urban sound classification [75] (Urban-8K dataset [76]) and species classification from clips of avian flight calls [56] (CLO-43SD dataset [77]). Its first layer consists of 24 convolutional kernels of size 5x5, followed by a rectified linear unit (ReLU) and a strided max-pooling operation whose receptive field has a size of 4x2, that is, 4 logmelspec frames (i.e.…”