A machine learning algorithm for analyzing audio clips, recognizing percussive content and then extracting individual hits as labeled samples.
For samples : 60% training 20% testing 20% cross-validation
Referenced papers and code can be found in the references_papers folder
To Do :
better onset detection code more features : now all in features.py with raise notimplementederror way to export 'samples' as files code to run training and testing