Methods for processing and analyzing passive acoustic monitoring data: An example of song recognition in western black-crested gibbons

Zhou, Xiaotao; Hu, Kunrong; Guan, Zhenhua; Yu, Chunjiang; Wang, Shuai; Fan, Meng; Sun, Yongke; Cao, Yong; Wang, Yijie; Miao, Guangting

doi:10.1016/j.ecolind.2023.110908

Cited by 5 publications

(2 citation statements)

References 58 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The best performing model (F score: 0.87) exceeded previously reported SVM models for detecting gibbon vocalisations e.g., Hylobates funereus detector, F score: 0.78 (Clink et al 2023). It also performed comparatively to other CNN models, e.g., Nomascus hainanus detector, F score: 0.91 (Dufourq et al 2021), Nomascus concolor detector, F score: 0.92 (Zhou et al 2023).…”

Section: Discussionmentioning

confidence: 99%

“…For these reasons, gibbons are model organisms for developing and testing guidelines for automated detection. So far, this has been done for a handful of species, including the Hainan gibbon (Nomascus hainanus) (Dufourq et al 2021), Western black-crested gibbon (Nomascus concolor) (Zhou et al 2023), Northern grey gibbon (Hylobates funereus) (Clink et al 2023), and southern yellow-cheeked crested gibbon (Nomascus gabriellae) (Clink et al 2024).…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Automated detection of Bornean white-bearded gibbon (Hylobates albibarbis) vocalisations using an open-source framework for deep learning

Owens,

Hockings,

Imron

et al. 2024

Preprint

View full text Add to dashboard Cite

Passive acoustic monitoring is a promising tool for monitoring at-risk populations of vocal species, yet extracting relevant information from large acoustic datasets can be time-consuming, creating a bottleneck at the point of analysis. We adapted an open-source framework for deep learning in bioacoustics to automatically detect Bornean white-bearded gibbon (Hylobates albibarbis) "great call" vocalisations in a long-term acoustic dataset from a rainforest location in Borneo. We describe the steps involved in developing this solution, such as collecting audio recordings, developing training and testing datasets, training neural-network models, and evaluating model performance. Our best model performed at a satisfactory level (F score = 0.87), identifying 98% of the highest-quality calls from 90 hours of manually-annotated audio recordings. We also found no significant difference in the distribution of great call detections over time between the manual annotations and the model's output, and greatly reduced analysis times when compared to a human observer. Future work should seek to apply our model to long-term acoustic datasets to understand spatiotemporal variations in H. albibarbis' calling activity. With additional information, such as detection probability over distance, we demonstrate how our model could be used to monitor gibbon population density and spatial distribution on an unprecedented scale.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%