Acoustic detector for multiple vocalizing marine mammal individuals

Marine mammals play an important role in the marine ecosystems but are increasingly threatened by human activities. To better protect these species it is helpful to understand their spatio-temporal visiting patterns and to estimate their population sizes. In this work we present a non-invasive method to aid conservation efforts, using a multi-channel passive acoustic monitoring system. We test the system in Singapore’s noisy coastal waters and show that the processing developed herein can automatically classify marine mammal whistles depending on shape, estimate their direction of arrival and indicate whether a set of detected whistles that occur temporally nearby are likely to have arisen from multiple vocalizing individuals.