Bird audio recordings (foreground and background) to accompany the work:

"Automatic acoustic identification of individuals: Improving generalisation across species and recording conditions"
by Dan Stowell, Tereza Petrusková, Martin Šálek, Pavel Linhart

Published September 2018.


This dataset contains labelled recordings of individuals from three different bird species:

* Little owl (Athene noctua)
* Chiffchaff (Phylloscopus collybita)
* Tree pipit (Anthus trivialis)

Recordings were made in the Czech Republic and Hungary. Each sound file represents a vocalisation from one labelled individual, or (for "bg" recordings) the ambient sound shortly before/after vocalisation. For further details please refer to the paper.

The dataset takes approx 11 GB of disk space after the ZIP files have been uncompressed.


WAV files:
=====================

The audio files are 44.1 kHz mono WAV files, grouped into subfolders according to species and whether a foreground ("fg") or background ("bg") recording.


CSV annotation files:
=====================

Each "training set" and "test set" is represented in a separate CSV file. For example, to run a standard chiffchaff automatic classification test you could train a system using

  chiffchaff-withinyear-fg-trn.csv

and then test it using

  chiffchaff-withinyear-fg-tst.csv

The CSV files refer to one audio file per line, with a simple "one-hot" labelling of which bird is present: the first column is the filename, and the remaining columns each represent one possible bird, and there will be a "1" in the column corresponding to the correct individual.


Research publication:
=====================

"Automatic acoustic identification of individuals: Improving generalisation across species and recording conditions"
by Dan Stowell, Tereza Petrusková, Martin Šálek, Pavel Linhart
(submitted in 2018)


Acknowledgments:
=====================

DS was supported by EPSRC Early Career research fellowship EP/L020505/1.
PL was supported by the National Science Centre, Poland, under Polonez fellowship reg. no UMO-2015/19/P/NZ8/02507 funded by the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 665778.
TP was supported by the Czech Science Foundation (project P505/11/P572).M\v{S} was supported by the research aim of the Czech Academy of Sciences (RVO 68081766).


Copyright:
=====================

Copyright (c) is held by the individual recordists.

These data are published under the Creative Commons Attribution licence (CC BY):
 https://creativecommons.org/licenses/by/4.0/
This licence allows to to re-use the data for almost any purpose (follow the link for more information), as long as you give credit to the original source.

For academic reuse, we ask that you do this as a citation to the research paper, given above.


We would also like to hear about it, so please feel free to contact us to let us know.