Artificial Funklord

[ deep-learning  machine-learning  udacity  wwe  ]

As a kid, I loved to play Toe Jam & Earl with my brother. The game’s music was epic and inspirational, at least as far as we were concerned. Later on, in the future–now the past!–we would learn some musical instruments and jam out funk-rock style improvs largely inspired by TJ & E sounds.

Little did guitarKev know that futureKev (currently known as nowKev) would one-up him: I’m going to show you some brand-funking new TJ & E music generated in the depths of an artificial intelligence whose style is so funky TJ & E would surely recognize this interplay of man and machine as a Funk Lord, or at least a Rap Master.

A little bit about RNNs

Basically, easiest approach is just considering input/output of an RNN cell. The internals can be confusing at 1st, 2nd, or 3rd glance, but understanding the input/ouput is not so bad. Consider for a moment that your mind is an RNN. At each moment you have certain memories and thoughts, and these lead to certain actions. Any actions taken become but a memory, which may be something you wish to forget – or something you wish to tell and re-tell, and maybe embellish a little bit! As an RNN, at each moment you consider your current thoughts/memories and what action you’ve just taken; this leads to taking a new action and impacting your thoughts/memories; rinse and repeat.

More on all this later… Have to get back to work.

References

These links will probably be helpful:

  • https://medium.com/@shiyan/understanding-lstm-and-its-diagrams-37e2f46f1714
  • https://magenta.tensorflow.org/2016/06/10/recurrent-neural-network-generation-tutorial/
  • https://github.com/tensorflow/magenta/tree/master/magenta/models/melody_rnn
    • translate MIDI file into TF data format
    • simple example of music generation
  • https://www.youtube.com/watch?v=4DMm5Lhey1U
  • https://github.com/llSourcell/How-to-Generate-Music-Demo
Written on May 11, 2017