The Endless folk-rnn Traditional Music Session

Music generated by a recurrent neural network named folk-rnn,
trained on over 23,000 tunes posted to The Session

Compare to the Endless magenta Traditional Music Session
featuring tunes generated by magenta

Tunes with titles are generated by the first version of folk-rnn, which learned to generate titles

Tunes starting with numbers are generated by the second version of folk-rnn,
and come from the folk-rnn Session Book Volumes 1-10

Tunes starting with 'fv3' are generated by the third version of folk-rnn,
which was trained on a version of the training data involving transpositions of the material,
to have a root C and C# (over 46,000 transcriptions, trained on a Titan X Pascal GPU donated by the NVIDIA Corporation)

Playlist updated every 5 minutes

  1. Jump At The Sun 22183
  2. Meerack's 14156
  3. Ailie Fire's gatching 13173
  4. fv3_1798_30860
  5. Bluetanna 25394
  6. Kelfica Connor's 15373
  7. An Fairchie A Bula 23030

Total number of tunes synthesised now: 47747

Created by: Bob L. Sturm and João Felipe Santos. (More info.)