Project DetailsWith the large increase in human marine activity, our rivers and seas have become populated with boats and ships projecting acoustic emissions of extremely high power that often affect areas of up to 20 square km and more. The underwater radiated noise (URN) level from large ships can exceed 100 PSI and is wideband, such that even at km distances of several kilometres from the vessel, the acoustic pressure level is still high. While evidence showed evidence for a clear disturbance impact on the hearing and behavior of marine mammals, there is still no systematic proof to the extent of this effect. Taking coastal dolphins as indicators to the impact of shipping URN, such a proof can be derived if changes in the dolphins' vocalizations are observed when vessels are around. In this project, we take a first step towards this end, and extract the features of dolphin's vocalization. Both statistical and geometrical features are explored. The former refers to measures such as the signal's duration, min and max frequency etc. The latter refers to polynomial fitting for the contour of the frequency-time whistle track. As a quotative measure, we choose the joint entropy metric, which, compared to a truly random reference set, represents the variation in the dataset. Analysis is performed over two datasets: a manually tagged dataset from France recorded for more than a year, and a manually tagged dataset from Eilat, Israel, with roughly 800 dolphins' whistles.
Our results showed that, based on normalized cross correlation between the original signal and its synthetic formation from the geometrical reconstraction, a polynomial of 4th degree yields the best result. Based on the entropy measure, statistical feature exploration yields larger variation than geometrical analysis. Principle componenet analysis (PCA) showed that the min, max, and start and final frequency of the whistle are the most important features.
