Simple library to work with the IEEE AASP CASA Challenge 2013 dataset in Python
This package requires:
Download and run the setup:
python setup.py install
python setup_post_install.py
This will download the database from the official website of the challenge, reorganize with different folders for each class and downsample everything to mono / 8kHz
With this all you need to do is to define two functions:
The first that will extract a descriptor for each file of the dataset
The second that will fit a classifier on a train set and return predicted results on a test set
Everything else (iterating over the files in the dataset and 5-fold cross-validation according to the specifications so you can compare with published results) is handled by the auxiliary functions
See example.py for more details
Giannoulis, Dimitrios, et al. "Detection and classification of acoustic scenes and events: An ieee aasp challenge." Applications of Signal Processing to Audio and Acoustics (WASPAA), 2013 IEEE Workshop on. IEEE, 2013.
Giannoulis, Dimitrios, et al. "A database and challenge for acoustic scene classification and event detection." Signal Processing Conference (EUSIPCO), 2013 Proceedings of the 21st European. IEEE, 2013.