Bird species classification using their songs

Check out the ipython notebook ;)

The data should be downloaded from :

• Train data : http://sis.univ-tln.fr/~glotin/SABIOD/BIRD50/AmazonBird50_training_input.tar.gz
• Test data : http://sis.univ-tln.fr/~glotin/SABIOD/BIRD50/AmazonBird50_testing_input.tar.gz

And placed in subdirectories : AmazonBird50_training_input and AmazonBird50_testing_input respectively

Numpy, Scipy, Sklearn, python_speech_features :

git clone https://github.com/jameslyons/python_speech_features
cd python_speech_features
python setup.py install

openSMILE : https://web.stanford.edu/class/cs224s/hw/openSMILE_manual.pdf

The data comes from ... (TO be completed)

Détails des submits :

ypred2.csv : Welch features (1000:10000:500 -> 2 by 2) + RDF 800 estim, max_feat 5. 7-Cross-val score ~31% Leaderboard : 32%