Real-time performance of energy based underwater acoustic event detectors embedded in a single-board computer

Summation of energy, commonly known as Energy Detectors, is a class of signal processing methods that can be used to detect underwater acoustic events in the recorded signal data. The key in Energy Detectors is the construction of a Detection Function that can differentiate the acoustic event from the background. In this work, we evaluate the performance of some classical approaches that develop detection functions based on energy summation, as well as their performance when running in a single-board computer for real-time analysis. A set of underwater acoustic recordings (pcm wave files) containing some bioacoustics events, with different size and frequency sampling, were used to construct signals envelopes. In each case, a set of methods (root mean square, moving average, absolute square value, fast Fourier transform and Hilbert transform) with different parameters values (window size and overlap) were tested on a Raspberry PI B+ (RPI) using Python as programming language. Results shows that RPI is able to perform post-processing tasks to construct the detection functions of the recorded signals, while acquiring data at the same time. This characteristic demonstrates their potential in developing real time detection systems on single board computers as RPI.