Visual acoustic analysis techniques for North Atlantic right whale sounds

Passive acoustic monitoring (PAM) is becoming a more widely accepted tool in mitigating the potential impact of man-made noise on marine mammals. Many marine mammals, and in particular cetaceans (whales and dolphins), use sound to communicate, navigate, forage, and avoid predators. Automated vocalization detectors have been in development for many years. Validation is achieved through aural/visual verification by analysts examining acoustic data in both the time and frequency domains. However, many factors also influence the ability of the analyst to properly detect and classify marine mammal sounds. First, a priori knowledge of the acoustic signature and the analysts experience are critical factors. Assuming the a priori knowledge exists, the ability to correctly assess the presence of call is dependent on the signal processing and displays available. Frequency resolution, temporal integration, and normalization either enhance or inhibit the ability to make the assessment. The signal processing parameterization varies between species and vocalization types. Six types of vocalizations have been previously described for the North Atlantic right whale (Eubalaena glacialis); upcalls, gunshots, screams, downcalls, blows, and warbles. The parameterization required to optimally assess each of the vocalization types will be examined.