Assessing the performance of a semi‐automated acoustic monitoring system for primates

A semi-automated acoustic monitoring system demonstrates promising efficacy for detecting primate vocalizations in the complex rainforest environment of Taï National Park, Côte d'Ivoire. The study, involving 12,851 hours of recordings from 20 autonomous units without manual preselection, achieved notable recall rates (50% and 42%) for Cercopithecus diana and Colobus polykomos, respectively. While performance was lower for Pan troglodytes and Procolobus badius, the system required minimal human intervention (3.5% time) compared to traditional manual data collection, highlighting its potential for large-scale, long-term monitoring. The approach leverages speaker segmentation, support vector machines, and Gaussian mixture models, offering a scalable solution for behavioral, ecological, and conservation studies.

Heinicke, S., Kalan, A. K., Wagner, O. J. J., Mundry, R., Lukashevich, H., & Kühl, H. S. (2015). Assessing the performance of a semi‐automated acoustic monitoring system for primates. Methods in Ecology and Evolution, 6(7), 753–763. Portico. https://doi.org/10.1111/2041-210x.12384