This project is meant to explore the possible application of acoustic fingerprinting for detecting playback traps used in poaching of song birds. The algorithm was based on the Shazam algorithm for music recognition.
In order to run the scripts in this repository, you can create a virtual environment and install the dependencies from requirements.txt using the following commands:
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt
Under experiments/ you can find different scripts that are purely for the testing and evaluation of the fingerprinting algorithm.
The test_data.csv file contains a compiled list of bird songs from the xeno-canto online archive that can be used for testing purposes.
The _config.yaml file can be used to specify the parameters for any number of experiments for both indexing songs and recognition.
If you want to run full experiments including creating a database, fingerprinting, comparing other songs and outputting a csv containing the match for each song as well as a confidence score, use: python run_tests.py
If you only want to test creating the database and populating it, run:
python setup.py
python add_songs_to_db.py
If you only want to test the detection of certain audio files using an existing database, run: python tests_db.py
For testing the web application of a live detector add songs using scripts/add_songs_to_db.py and run:
python app.py
 | 
|