Sparse coding for efficient bioacoustic data mining: Preliminary application to analysis of whale songs

Joseph Razik (1) , Hervé Glotin (1, 2) , Maia Hoeberechts (3) , Yann Doh (1) , Sébastien Paris (1)

1 LIS - Laboratoire d'Informatique et Systèmes
2 IUF - Institut Universitaire de France
3 Ocean Networks Canada

Bioacoustic monitoring, such as surveys of animal populations and migration, needs efficient data mining methods to extract information from large datasets covering multi-year and multi-location recordings. Usually, the study of the humpback whake songs is based on the classification of sound units, notably to extract the song theme of the singers, which might signify the geographic origin and the year of the song. Most of these analyses are currently done with expert intervention, but the volume of recordings drive the need for automated methods for sound unit classification. This paper introduces a method for sparse coding of bioacous-tic recordings in order to efficiently compress and automatically extract patterns in data. Moreover, this paper proposes that sparse coding of the song at different time scales supports the distinction of stable song components versus those which evolve year to year. It is shown that shorter codes are more stable, occurring with similar frequency across two consecutive years, while the occurrence of longer units varies across years as expected based on the prior manual analysis. We conclude by exploring further possibilities of the application of this method for biopopulation analysis.

Domains: Machine Learning [cs.LG]; Signal and Image Processing

Joseph Razik, Hervé Glotin, Maia Hoeberechts, Yann Doh, Sébastien Paris. Sparse coding for efficient bioacoustic data mining: Preliminary application to analysis of whale songs. International Conference on Data Mining Workshops, 2015, Atlantic city, United States. ⟨hal-02993300⟩