Bat detective—Deep learning tools for bat acoustic signal detection

Abstract

Passive acoustic sensing has emerged as a powerful tool for quantifying anthropogenic impacts on biodiversity, especially for echolocating bat species. To better assess bat population trends there is a critical need for accurate, reliable, and open source tools that allow the detection and classification of bat calls in large collections of audio recordings. The majority of existing tools are commercial or have focused on the species classification task, neglecting the important problem of first localizing echolocation calls in audio which is particularly problematic in noisy recordings. We developed a convolutional neural network based open-source pipeline for detecting ultrasonic, full-spectrum, search-phase calls produced by echolocating bats. Our deep learning algorithms were trained on full-spectrum ultrasonic audio collected along road-transects across Europe and labelled by citizen scientists from www.batdetective.org. When compared to other existing algorithms and commercial systems, we show significantly higher detection performance of search-phase echolocation calls with our test sets. As an example application, we ran our detection pipeline on bat monitoring data collected over five years from Jersey (UK), and compared results to a widely-used commercial system. Our detection pipeline can be used for the automatic detection and monitoring of bat populations, and further facilitates their use as indicator species on a large scale. Our proposed pipeline makes only a small number of bat specific design decisions, and with appropriate training data it could be applied to detecting other species in audio. A crucial novelty of our work is showing that with careful, non-trivial, design and implementation considerations, state-of-the-art deep learning methods can be used for accurate and efficient monitoring in audio.

Making inference with messy (citizen science) data: when are data accurate enough and how can they be improved?

“…The first data quality concern associated with estimating species distribution is that although professionals, volunteers, crowdsourced aggregates, and machine-learning algorithms commonly identify species accurately overall (>95%, e.g., McClintock et al 2010, Norouzzadeh et al 2018, overall species identification accuracy is often weighted by a few very common and easily identified species and the accuracy of individual species is highly variable. The range of misclassification we considered here (3-30%) is not unique to our study; similar rates of misidentification are documented across a range of methodologies for classifying trail camera images (McShea et al 2016, Norouzzadeh et al 2018, Tabak et al 2018 or recorded calls , McClintock et al 2010, Farmer et al 2012, Mac Aohda et al 2018, Priyadarshani et al 2018. This suggests that despite the overall accuracy of many data sets processed by humans with limited training (volunteer or not) or automated algorithms, there is a non-trivial risk of substantially overestimating the distribution of many species using many commonly used data types.…”

AudioMoth: Evaluation of a smart open acoustic device for monitoring biodiversity and the environment

“…Some contemporary sampling methods achieve coverage with semi‐automated monitoring technology, for example camera traps triggered by infra‐red sensors. Much of the methodology used in acoustic monitoring lags behind this trend, tending to feed large quantities of captured data through detection software after deployment (Mac Aodha et al., ). Despite the heavy demand on memory storage, passive acoustic monitoring (PAM) has proved useful for estimating ground‐level biodiversity abundance and occurrence, particularly of smaller and more cryptic species (Newson, Bas, Murray, & Gillings, ), and it is often employed for analyses of soundscapes (Towsey et al., ).…”

Emerging opportunities and challenges for passive acoustics in ecological assessment and monitoring

“…Deep convolutional neural networks (CNNs) are particularly promising, since these can learn discriminating spectro‐temporal information directly from annotated spectrograms (bypassing a separate feature extraction stage), improving their robustness to sound overlap and caller distance (Goeau et al., ) (Figure d). In recent tests, CNNs have markedly outperformed alternative methods on detection and classification of biotic and anthropogenic sounds in urban recordings (Fairbrass et al., ; Salamon & Bello, ) and animal calls in noisy monitoring datasets (Goeau et al., ; Mac Aodha et al., ; Marinexplore, ). Their performance in more complex tasks that involve distinguishing multiple overlapping vocalisations (e.g., songs in the dawn chorus) has not yet been tested, although their success in similarly challenging computer vision and individual human voice recognition tasks is a promising sign (e.g., Lukic, Vogt, Dürr, & Stadelmann, ).…”

“…Some studies have partially addressed this issue by augmenting training data with background noise to simulate different distances and acoustic environments (Salamon & Bello, ). Online data labelling projects such as Bat Detective (http://www.batdetective.org) and Snapshot Serengeti (http://www.snapshotserengeti.org) have also involved citizen scientists in annotation of CNN training data (Mac Aodha et al., ; Norouzzadeh et al., ).…”

“…animal calls in noisy monitoring datasets (Goeau et al, 2016;Mac Aodha et al, 2018;Marinexplore, 2013). Their performance in more complex tasks that involve distinguishing multiple overlapping vocalisations (e.g., songs in the dawn chorus) has not yet been tested, although their success in similarly challenging computer vision and individual human voice recognition tasks is a promising sign (e.g., Lukic, Vogt, Dürr, & Stadelmann, 2016).…”