Sabara, RaziSoares, CristianoZabel, FriedrichValente de Oliveira, JoséJesus, Sergio2021-09-222021-09-2220200197-7385http://hdl.handle.net/10400.1/17139Ocean noise has been a topic of research for many years, for its impact in sonar detection, underwater communications and ocean acoustic observation in general. Recently, ocean sound has been designated as an Essential Ocean Variable (EOV) and is therefore, becoming increasingly recorded and monitored, along with other oceanic and meteorological variables. The research projects EMSO-PT and SUBECO aim at deploying ocean observatories along the coast of Portugal for long term ocean variables monitoring, among which ocean sound. Unlike other ocean variables, ocean sound allows for feature detection, characterisation and possibly identification with known patterns. This work shows the results obtained with current machine learning algorithms for feature detection and extraction on a two days recording of ocean noise obtained on a offshore buoy deployed under the SUBECO project, on the west coast of Portugal. Preliminary results show the possibility of improved event detection, followed by classification and clustering, that foresee a rapid and accurate analysis of large observatory acquired acoustic data sets.engOocean soundMSFDMachine learningAcoustic detectionconference object10.1109/IEEECONF38699.2020.9389067