Skip to content

Research at St Andrews

Learning deep models from synthetic data for extracting dolphin whistle contours

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Author(s)

Pu Li, Xiaobai Liu, Kaitlin Palmer, Erica Fleishman, Douglas Michael Gillespie, Eva-Marie Nosal, Yu Shiu, Holger Klinck, Danielle Cholewiak, Tyler Helble, Marie Roch

School/Research organisations

Abstract

We present a learning-based method for extracting whistles of toothed whales (Odontoceti) in hydrophone recordings. Our method represents audio signals as time-frequency spectrograms and decomposes each spectrogram into a set of time-frequency patches. A deep neural network learns archetypical patterns (e.g., crossings, frequency modulated sweeps) from the spectrogram patches and predicts time-frequency peaks that are associated with whistles. We also developed a comprehensive method to synthesize training samples from background environments and train the network with minimal human annotation effort. We applied the proposed learn-from-synthesis method to a subset of the public
Detection, Classification, Localization, and Density Estimation (DCLDE) 2011 workshop data to extract whistle confidence maps, which we then processed with an existing contour extractor to produce whistle annotations. The F1-score of our best synthesis method was 0.158 greater than our baseline whistle extraction algorithm (~25% improvement) when applied to common dolphin (Delphinus spp.) and bottlenose dolphin (Tursiops truncatus) whistles.
Close

Details

Original languageEnglish
Title of host publicationInternational Joint Conference on Neural Networks (IJCNN)
PublisherIEEE Computer Society
Number of pages10
Publication statusAccepted/In press - 20 Mar 2020
EventIEEE World Congress on Computational Intelligence (IEEE WCCI) - 2020 International Joint Conference on Neural Networks (IJCNN 2020) - Glasgow, United Kingdom
Duration: 19 Jul 202024 Jul 2020

Conference

ConferenceIEEE World Congress on Computational Intelligence (IEEE WCCI) - 2020 International Joint Conference on Neural Networks (IJCNN 2020)
Abbreviated titleIJCNN
CountryUnited Kingdom
CityGlasgow
Period19/07/2024/07/20

    Research areas

  • Whistle contour extraction, Deep neural network, Data synthesis, Acoustic, Odontocetes

Discover related content
Find related publications, people, projects and more using interactive charts.

View graph of relations

Related by author

  1. Publisher Correction: Deep neural networks for automated detection of marine mammal species (Scientific Reports, (2020), 10, 1, (607), 10.1038/s41598-020-57549-y)

    Shiu, Y., Palmer, K. J., Roch, M. A., Fleishman, E., Liu, X., Nosal, E. M., Helble, T., Cholewiak, D., Gillespie, D. & Klinck, H., 30 Jun 2020, In: Scientific Reports. 10, 11000.

    Research output: Contribution to journalComment/debatepeer-review

  2. High resolution three-dimensional beam radiation pattern of harbour porpoise clicks with implications for passive acoustic monitoring

    MacAulay, J. D. J., Malinka, C. E., Gillespie, D. M. & Madsen, P. T., 25 Jun 2020, In: Journal of the Acoustical Society of America. 147, 6, p. 4175-4188 14 p.

    Research output: Contribution to journalArticlepeer-review

  3. Passive acoustic methods for tracking the 3D movements of small cetaceans around marine structures

    Gillespie, D., Palmer, L., MacAulay, J., Sparling, C. & Hastie, G., 29 May 2020, In: PLoS ONE. 15, 5, 16 p., e0229058.

    Research output: Contribution to journalArticlepeer-review

  4. Deep neural networks for automated detection of marine mammal species

    Shiu, Y., Palmer, K., Roch, M., Fleishman, E., Liu, X., Nosal, E-M., Helble, T., Cholewiak, D., Gillespie, D. M. & Klinck, H., 17 Jan 2020, In: Scientific Reports. 10, 12 p., 607.

    Research output: Contribution to journalArticlepeer-review

  5. Time of arrival difference estimation for narrow band high frequency echolocation clicks

    Gillespie, D. & Macaulay, J., Oct 2019, In: Journal of the Acoustical Society of America. 146, 4, p. EL387-EL392 7 p.

    Research output: Contribution to journalArticlepeer-review

ID: 267536251

Top