An annotated library of underwater acoustic recordings for testing and training automated algorithms for detecting Antarctic blue and fin whale sounds

Brief description

This annotated library contains both a data set and a data product.

The data set contains a sub-sample of underwater recordings made around Antarctica from 2005-2017. These recordings were curated and sub-sampled from a variety of national and academic recording campaigns. Recordings were made using a variety of different instruments, and sub-samples span 11 different combinations of site and year. Spatial coverage of the recordings includes sites in the Western Antarctic Peninsula, Atlantic, Indian, and Pacific sectors. Temporal coverage of recordings covers a representative sample throughout each recording year for the years of 2005, 2013, 2014, 2015, and 2017. The focus is on low-frequency sounds of blue and fin whales, so curated recordings have been downsampled to sample rates of either 250, 500, 1000 or 2000 Hz. Recordings are all in 16-bit wav format. The file name of each wav file contains a timestamp with the date and time of the start of that file. Recordings are contained in the /wav/ subfolder for each site-year (e.g. Casey2014/wav).

The data product is in the form of annotations that describe the times within each WAV file that contain detections of blue and fin whale sounds. Each annotations are stored as a row in a tab-separated text file (with descriptive column headers), and each text file describes a particular type of sound. These annotation text files are formatted as Selection Tables that can be directly imported into the software program Raven Pro 1.5 (Cornell Bioacoustics Laboratory).

Full description of the details of the creation and use of this dataset are described in the draft manuscript contained in the documentation folder.

Lineage

Progress Code: completed

Statement: Full description of the particulars of this dataset are described in the draft manuscript contained in the documentation folder.

Notes

Purpose
Since 2001 hundreds of thousands of hours of underwater acoustic recordings have been made throughout the Southern Ocean south of 60°S. Detailed analysis of the presence of the distinctive calls of Antarctic marine mammals in these circumpolar recordings could provide novel insights into their ecology, but manual inspection of the entirety of all recordings by experts would be prohibitively time consuming and expensive. Automated signal processing methods have now developed to the point that they can be applied to these data in a cost-effective manner. However training and evaluating the efficacy of these automated signal processing methods still requires a representative library of sounds that have been annotated to identify the true presence and true absence of different sound types. This work presents such a library of annotated recordings for the purpose of training and evaluating automated detectors of Antarctic blue and fin whales. Creation of the library has focused on the annotation of a representative sample of recordings curated by the Acoustic Trends Working Group of the International Whaling Commission’s Southern Ocean Research Partnership (IWC-SORP) in order to ensure that automated algorithms can be developed and tested across a broad range of instruments, locations, and years. Specifically, we have annotated low frequency sounds (<150 Hz) produced by blue and fin whales, and we have qualitatively noted the presence of common Antarctic noise sources such as ice, wind, seismic surveys, and ships.