2019_programme: DEVELOPING A TREATMENT PIPELINE OF SUBMARINE ACOUSTIC DATABASES FOR DETECTING NOISE PHENOMENA OF INTEREST

Back


  • Session: 02. Underwater acoustic calibration, testing, facilities and standards
    Organiser(s): Robinson Stephen, Humphrey Victor, Linné Markus, Évora Victor
  • Lecture: DEVELOPING A TREATMENT PIPELINE OF SUBMARINE ACOUSTIC DATABASES FOR DETECTING NOISE PHENOMENA OF INTEREST
    Paper ID: 788
    Author(s): Achraf Drira, Medjber Bouzidi, Cedric Gervaise
    Presenter: Achraf Drira
    Presentation type: oral
    Abstract: Oceans suffer from particularly strong anthropogenic pollution. Passive acoustics play a fundamental role in this problem. Understanding the impact of radiated noise in the oceans on marine mammals in particular – and underwater life in general – is thus a pivotal task. We started from a logic of algorithm development that allowed us to identify sources of ambient noise including vessel noise, sounds emitted by invertebrates, and marine mammals. The objective was to obtain rates of presence of each source in the study areas and, thus, a clearer image of the impacts of anthropogenic sources on underwater life.\nOur goal has already been in existence for at least a decade. Nowadays, there are several methods that allow automatization of these analyses. Among these, time-frequency methods allow the detection of acoustic signatures of sources in the spectrogram and, given their high scores obtained in detection contests, they are frequently applied. The differences between these methods are mainly: a) the choice of parameters that must be adjusted with calibration measurements, b) the adaptation of the method to the medium as well as to the source type, and finally c) the difference in computing time.\nHere, we test one of these time-frequency methods that requires a relatively reduced number of adjustment parameters on real data. We evaluate the efficiency of the method, for example for detecting marine mammals clicks, by comparing our results with other types of detectors such as C-pod. We present the implementation of this method, and the end ideas for improvements such as the adaptation of the method to detect sounds of invertebrates and vessels.\n
  • Corresponding author: Mr Achraf Drira
    Affiliation: SINAY
    Country: France
    e-mail:

Back