Recognises song slices genre with ~75% accuracy. Whole songs genres are recognised a liitle bit more accuratly, because we look at multiple slices, not just one. Seems to be working quite well.

1. Create folder tree in project root:
   -data //folder in which the training data will be stored
   -- //inside data create folders with named by genre ie. rock, blues etc.
   -spectrograms //folder in which the generated spectrograms will be stored
   -- //inside spectrograms create folders with named by genre ie. rock, blues etc. as you did with the data folder
   --test //folder in which the spectrograms of test songs will be generated
   -songs //in that folder paste any songs that you want to categorise by trained model

2. Create spectrograms by using spectrogramsCreator.createSpectrograms()

3. Train model using main.py

4. Uncomment lines 34-36 in test.py and use it to categorise your songs.