AI x Acoustic Data

Oceans are a tumult of sounds coming from human activities, marine life and geological phenomena. Hydrophones can record a lot of different signals including engine roar, artificial sonar, seismic blasts, iceberg melting, whale vocalizations, marine mammals clicks and other sounds which show that, contrary to the image of the sea as a silent world portrayed by Jacques Cousteau in the 1950’s, the sea can be described as a cacophony of sounds. Sometimes the noise pollution make it impossible for marine creatures to hunt or communicate and concerns about anthropogenic noise in the ocean have lead some countries to develop underwater noise regulation. For marine systems, acoustic data represent information characterizing the proper functioning of the systems and identifying the source of the sound is essential for their monitoring.

To master noise emission and its effects, it is essential to characterize the sound and identify its source. The sound can be produced intentionally or not. It may be characterized by continuous or intermittent signals with a short or a long duration.

The objective is to find new ways to process underwater acoustic data and to analyze it as efficiently as possible. This implies being able to separate the sound sources and to classify or even identify them. New AI algorithms can revolutionize the current capabilities of systems.