Humans keep on changing the natural habitat of marine animals, including dolphins.
Thus, the need to monitor the population and behavior of dolphin species is increasing ever more. Counting the number of encounters in a location with each species, is a leading method to evaluate the population of those marine animals in that area.
Classifying dolphins by sight as was done so far, pose many problems. However, underwater recordings tend to have decent quality, despite changes in water conditions. Also, due to the better propagation of sound waves in water, sound waves typically cover great distances.
In this project we constructed a software which can manage and access big amounts of data. Later we extracted basic features of the whistles for both dolphin species, which will be regarded as type 1 and type 2 from now on. We classified them achieving 82.5%, 76.2% and 73% percent of correct classification, using k-NN, SVM and LDA algorithms accordingly.