The AviaNZ project can be found at https://github.com/smarsland/AviaNZ. For more information about the project, see http://www.avianz.net.
This fork is designed to provide a version of AviaNZ that can run on a headless Linux device in real-time. That is, audio is directly sent to the software and the software returns a confidence for that audio snippit. Some library API calls have been updated to work with later versions of the packages. This fork has been developed with Python 3.10.
git clone https://github.com/ysims/AviaNZpip3 install -r requirements.txt- Build the Cython extensions by running
cd util/ext; python3 setup.py build_ext -i; cd ../.. python3 AviaNZ.py -f <path/to/file> -r <model_name>
Where <path/to/file> is the path to a wav file and <model_name> is the name of the recogniser in the filters folder you want to use.
An example is python3 AviaNZ.py -f sound_files/AustralasianBittern_5min.wav -r Bittern.
Please navigate to the original repository to find the authors of this software. The software website is http://www.avianz.net.
If you use the AviaNZ software, please credit the original authors in any papers that you write. An appropriate reference is:
@article{Marsland19,
title = "AviaNZ: A future-proofed program for annotation and recognition of animal sounds in long-time field recordings",
author = "{Marsland}, Stephen and {Priyadarshani}, Nirosha and {Juodakis}, Julius and {Castro}, Isabel",
journal = "Methods in Ecology and Evolution",
volume = 10,
number = 8,
pages = "1189--1195",
year = 2019
}
AviaNZ is based on PyQtGraph and PyQt, and uses Librosa and Scikit-learn amongst others.
Development of this software was supported by the RSNZ Marsden Fund, and the NZ Department of Conservation.
The work done in this fork is supported by a NSW Trust Grant and the University of Newcastle.