Automated detection of Hainan gibbon calls for passive acoustic monitoring

Abstract

1AbstractExtracting species calls from passive acoustic recordings is a common preliminary step to ecological analysis. For many species, particularly those occupying noisy, acoustically variable habitats, the call extraction process continues to be largely manual, a time-consuming and increasingly unsustainable process. Deep neural networks have been shown to offer excellent performance across a range of acoustic classification applications, but are relatively underused in ecology.We describe the steps involved in developing an automated classifier for a passive acoustic monitoring project, using the identification of calls of the Hainan gibbon (Nomascus hainanus), one of the world’s rarest mammal species, as a case study. This includes preprocessing - selecting a temporal resolution, windowing and annotation; data augmentation; processing - choosing and fitting appropriate neural network models; and postprocessing - linking model predictions to replace, or more likely facilitate, manual labelling.Our best model converted acoustic recordings into spectrogram images on the mel frequency scale, using these to train a convolutional neural network. Model predictions were highly accurate, with per-second false positive and false negative rates of 1.5% and 22.3%. Nearly all false negatives were at the fringes of calls, adjacent to segments where the call was correctly identified, so that very few calls were missed altogether. A postprocessing step identifying intervals of repeated calling reduced an eight-hour recording to, on average, 22 minutes for manual processing, and did not miss any calling bouts over 72 hours of test recordings. Gibbon calling bouts were detected regularly in multi-month recordings from all selected survey points within Bawangling National Nature Reserve, Hainan.We demonstrate that passive acoustic monitoring incorporating an automated classifier represents an effective tool for remote detection of one of the world’s rarest and most threatened species. Our study highlights the viability of using neural networks to automate or greatly assist the manual labelling of data collected by passive acoustic monitoring projects. We emphasise that model development and implementation be informed and guided by ecological objectives, and increase accessibility of these tools with a series of notebooks that allow users to build and deploy their own acoustic classifiers.

A predictive flight-altitude model for avoiding future conflicts between an emblematic raptor and wind energy development in the Swiss Alps

“…Artificial neural networks gained popularity in many fields of biology during the last decade [60]. For example, they have been used for behavioural classification from tri-axial acceleration data [61–63] or from GPS data [64,65] and to model datasets with high temporal resolution [63,64,66].…”

BioCPPNet: Automatic Bioacoustic Source Separation with Deep Neural Networks

“…While the human speech separation problem is a competitive area of work, the bioacoustic CPP has received comparatively less attention, as current bioacoustic research often emphasizes other ML-based tasks such as automated detection and classification of bioacoustic sounds [30][31][32] . However, recent work has implemented both semi-classical and deep ML-based approaches to address bioacoustic source separation, employing time domain and TFR-based algorithms.…”

Improving automatic processing of wildlife sound recordings

“…While for clean, loud calls accurate species-level recognition has been possible for several years now, the performance dramatically drops when moving to noisy and sparse recordings (Goëau et al, 2016), and has remained a challenge since (Kahl et al, 2020). Some recently explored approaches to overcome this are: testing different network architectures and meta-structures (Ibrahim et al, 2021;Dufourq et al, 2021;Merchan et al, 2020), adding separate networks for noise learning and removal , optimising spectrogram pre-processing for CNNs (Knight et al, 2019;Merchan et al, 2020), learning more robust features directly from the waveform (Sanchez et al, 2021). These studies have reported good performance on their respective target species.…”

“…However, achieving good results with CNNs so far has required thousands of labelled examples for each species; see e.g. Zhong et al (2020); Dufourq et al (2021). Collecting such datasets is difficult and expensive in bioacoustics.…”

“…This is compatible with occupancy or presence-absence models, but any inference based on individual call timestamps will require additional post-processing, as in e.g. Dufourq et al (2021). Procedures for this step have received less attention, and are commonly ad hoc, even in tasks that specifically evaluate timestamp accuracy .…”