Adaptive Energy-Based Acoustic Sperm Whale Echolocation Click Detection

Abstract

Sperm whales (Physeter macrocephalus) generate a series of broadband impulsive echolocation signals called clicks when diving in search of prey. These acoustic signals can be divided into usual clicks and creak clicks. They have a frequency spectrum that lies primarily between 2 and 18 kHz. In this paper, an adaptive energy-based click detector called the adaptive Teager energy operator (ATEO) is developed. A windowing technique is used to account for the time-varying characteristics of these signals, and the window length is adapted by estimating the interclick interval (ICI). The usual clicks and creak clicks are classified via ICI differences. The proposed method is compared with several well-known techniques using real data provided by the Atlantic Undersea Test and Evaluation Center (AUTEC). Performance results are presented which show that the proposed technique is a robust and efficient detector that can improve the click detection true positive rate (TPR) as well as reduce the false positive rate (FPR).