Development of advanced signal processing tools to describe, decompose and quantify complex underwater soundscapes and eventually define environmental acoustic descriptors
Soundscape decomposition includes:
- The automatic segmentation of long (several years) databases using machine learning techniques.
- The description and quantification of close, individually identifiable sounds and distant choruses of the different soundscape components (biophony, geophony and anthropophony)
- The detection and classification of close sound sources, including frequency-modulated and short transient deterministic broadband sounds using time-rhythm algorithms, recurrence plot analysis (RPA), etc.
- The development of 3D localization and tracking algorithms of biological sounds
- The spatial characterization of biological and anthropogenic sound production through large-scale acoustic maps