This code contains a show case of the paper “Jazz Solo Instrument Classification with Convolutional Neural Networks, Source Separation, and Transfer Learning” that was accepted in the ISMIR 2018 conference.
The code contains wav examples and the trained ConvNets to plot segment- and clip-wise predictions of the instruments in the an audio mixture.
The model has been trained to recognize the following instruments: soprano saxophone (ss), alto saxophone (as), tenor saxophone (ts), clarinet (cl), trombone (tb), and trumpet (tp).
The audio contains 6 jazz solos:
- Ornette Coleman - Ramblin (as)
- Buddy DeFranco - Autumn Leaves (cl)
- John Coltrane - My Favorite Things (ss)
- Frank Rossolino - Moonlight in Vermont (tb)
- Lee Morgan - The Sidewinder (tp)
- Michael Brecker - African Skies (ts)
You can check the code out here.
You can read the paper here.