This dataset is extracted from the MTG-Jamendo Datase
We applied the following transformation:
Convert audio to spectrogram diagram.
Join tagging list into one-line text.
For example, there is a genre list that has three tags: chillout, downtempo, and easylistening.
The raw text of this example will be "chillout downtempo easylistening".
