/tonal_detectors

Tonal detectors for low frequency vocalization of whales

Primary LanguageMATLABMIT LicenseMIT

Tonal detectors repository

Tonal detectors for low frequency vocalization of whales DOI

Author: Léa Bouffaut, Ph.D. Personal website | Researchgate

This work was conducted during my Ph.D. financed by the french Naval Academy (Institut de Recherche de l'Ecole Navale - Brest, France) and was also developed during my visit to the Center for Conservation Bioacoustics at the Cornell Lab of Ornithology - Cornell University, Ithaca (NY, USA).

Methods are described and compared for the detection of low-frequency whale signals (Antarctic Blue Whale) in:

L. Bouffaut, S. Madhusudhana, V. Labat, A. Boudraa and H. Klinck, “A performance comparison of tonal detectors for low frequency vocalizations of Antarctic blue whales,” accepted for publication for J. Acoust. Soc. Am. on Dec 31, 2019.

Abstract: Extraction of tonal signals embedded in background noise is a crucial step before classification and separation of low-frequency sounds of baleen whales. This work reports results of comparing five tonal detectors, namely the instantaneous frequency estimator, YIN estimator, harmonic product spectrum, cost-function, and ridge detector. The comparisons, based on a low-frequency adaptation of the Silbido scoring feature, employ five metrics which quantify the effectiveness of these detectors to retrieve tonal signals having a wide range of signal to noise ratios (SNRs) and the quality of the detection results. Ground-truth data were generated by embedding 10 synthetic Antarctic blue whale (Balaenoptera musculus intermedia) calls in randomly-extracted 30-minute noise segments from a 79~h-library recorded by an Ocean Bottom Seismometer (OBS) in the Indian Ocean during 2012-2013. Monte-Carlo simulations were performed using 20 trials per SNR, ranging from 0 dB to 15 dB. Overall, the obtained tonal detection results show the superiority of the cost-function and the ridge detectors, over the other detectors, for all SNR values. More particularly, for lower SNRs (≤ 3 dB), these two methods outperformed the other three with high recall, low fragmentation, and high coverage scores. For SNRs ≥ 7 dB, the five methods performed similarly.

Methods implemented (Matlab functions)

  1. Instantaneous frequency estimator, based on Boashash, "Estimating and interpreting the instantaneous frequency of a signal. II. algorithms and applications," Proc. of the IEEE 80(4), 540-568 (1992) doi: 10.1109/5.135378.

  2. YIN estimator, based on A. De Cheveigné and H. Kawahara, "YIN, a fundamental frequency estimator for speech and music," J. Acoust. Soc. Am. 111(4), 1917-1930 (2002) doi: 10.1121/1.1458024.

  3. Harmonic product spectrum, A. M. Noll, "Pitch determination of human speech by the harmonic product spectrum, the harmonic sum spectrum, and a maximum likelihood estimate," in Symposium on Computer Processing in Communication, ed., University of Brooklyn Press, New York, Vol. 19, pp. 779-797 (1969).

  4. Cost-function-based detector, based on M. F. Baumgartner and S. E. Mussoline, "A generalized baleen whale call detection and classification system," J. Acoust. Soc. Am. 129(5), 2889-2902 (2011) doi: 10.1121/1.3562166.

  5. Ridge detector, based on S. K. Madhusudhana, "Automatic detectors for underwater soundscape measurements," Ph.D. thesis, Curtin University, 2015.

  • An additional function is added to allow SNR-control on simulations.

Illustrations of the detectors on ABW calls