Identification and Recognition of Bio-acoustic Events in an Ocean Soundscape Data Using Fourier Analysis

Growing research in the design and development of automated systems to identify and recognize bio-acoustic signals is hastily finding significance with the increase in the volumes of ocean soundscape data recorded from real-time ocean environment every year, especially in handling non-voiced unstructured animal calls. With the vastness in the species richness in the ocean, precisely recognizing and classifying a sound-producing source from long durational passive acoustic recordings is manually tedious. In this case, automated identification and classification systems serve as an effective alternative. In this chapter, an automated marine acoustic source recognition system is designed to identify and classify non-voiced acoustic sources in real-time marine soundscape data of the South Virgin Islands using deep learning. Better identification results are achieved by using Fourier analysis, which recognizes individual frequency components in an aperiodic soundscape data. Research efforts in this regard are a great boon in marine species conservatory research.