Large-scale automatic acoustic monitoring of African forest elephants' calls in the terrestrial acoustic recordings

An automatic detection system for African forest elephants' rumble calls, developed and tested using 432 hours of acoustic recordings from eight locations in Gabon, achieved a true positive rate of 79.19% with just 5.70 false positives per hour. The system's F1-score reached 0.77 when a high score threshold was applied, demonstrating its effectiveness in processing large-scale acoustic data. Applied to a dataset spanning 420 days over three years from Ivindo National Park, the system revealed distinct seasonal and daily patterns of call activity, as well as temporal variations, offering valuable insights into the communication behavior of these elusive animals.

Shiu, Y., Wrege, P. H., Keen, S., & Rowland, E. D. (2014). Large-scale automatic acoustic monitoring of African forest elephants’ calls in the terrestrial acoustic recordings. The Journal of the Acoustical Society of America, 135(4_Supplement), 2334. https://doi.org/10.1121/1.4877660