Our entry for the 2015 Facebook hackathon at EPFL

snooki-dj is a small proof of concept implementation for procedural music generation

As we all know famous djs such as Snooki or Paris cost a lot and "produce" less than average music. That's why we created a music generator that will help the earthlings replace these overpriced "artists". Our app will provide strength for the restoring music on planet earth and make the world a better place.

At the moment snooki creates 3 simultaneous tracks, Drum, Bass and Harmony. For the harmony track we create a markov chain from the chord changes in several midi files we got from mididb.com. Bassline and Drum tracks are generated automatically from a discrete probability distribution for each eigth note in a 4/4 measure. For basslines these probabilities signify how probable it is that the root node, the octave above or any random note from the chord is played (probability of rests are implicit). Similarly for the drumtrack we have probabilities for bass, snare, hihat and rest at each eigth note.

• python 2.7
• mingus (0.4) with this patch: bgr/mingus#24

the visualization code for the demo uses the following additional libraries

• tkinter
• matplotlib
• networkx

from Snooki import Snooki s = Snooki() ... for chord in s.play(): print chord