Zaugg, S., van der Schaar, M., Houégnigan, L., André, M. Real time classification of sperm whale clicks and shipping impulses from a deep sea observatory 4th International Workshop on Detection, Classification and Localization of Marine Mammals Using Passive Acoustics, Sep 2009 Résumé: The automated acoustic detection of cetaceans in real time is an important tool to study their behaviour and distribution in the field and for mitigating human activities that are potentially harmful to them. However the classification in a fully automated way is challenging due to the diversity of acoustic events and background noises. Acoustic data from the NEMO ONDE deep sea antenna (-2000 m) indicated that impulsive ship noise was present in 10 % of the recordings. These anthropogenic impulses pose a serious challenge to the detection of sperm whale clicks, since they often share similar time frequency properties and hence could be the cause of many false positive detections. As part of an integrated classification system, we present a classification module aimed at the automated and real time classification of clicks from sperm whales and click-like sounds produced by shipping. The module is composed of two stages: A first stage, detects segments that contain impulsive sounds within a specifiable frequency band and returns their location. A second stage uses a feed forward neural network to classify impulses that have been detected in the first stage as sperm whale clicks or shipping impulses. The module’s reliability was tested on data from the NEMO ONDE antenna (including the dataset made available for the workshop). The achieved classification performance indicates that the module reliably separates a large proportion of sperm whale clicks from shipping impulses. Projet: ESONET, Network of Excellence Projet: LIDO, Listening to the Deep-Ocean Environment