a tool to intelligently categorize music tracks by genres
The dataset used to train the algorithms can be found here Audio files are of 22050Hz Mono 16-bit .wav format. Support for different audio formats will be added in the future.
In order to train the Multilayer Perceptron and SVM add the following program arguments:
-train genres/jazz genres/classical genres/metal genres/pop
-train - sets the program to extract features from audio files found in the folders given as arguments genres/jazz - folder given as argument, relative to the directory the tool is run in
The genres will be extracted from the last folder name. (e.g. for genres/jazz it will be jazz)
If no arguments are given, the program will attempt to classify all the audio files found in genres/? folder. The output result of both the MLP and SVM will be displayed in the console.