A Selection of Machine-Generated NES Game Titles

I’ve begun learning how to do machine learning. It is extremely complicated, but I feel after just a few days of trying I understand it much better. This python wrapper for tensorlm has made generating results much easier for me. I’ve got my little machine trained on the full NES US release list, and it’s giving me back things like this…

  • Super Super Hank
  • Doobley Dragon
  • Chanter
  • Class
  • Cartlevanina I: The Antiration
  • Kiggt’s Burtman
  • Jet
  • Rome Alone
  • Rock Hart’s Cave
  • Rock Master
  • Masher: II Lands
  • Greeado A World II
  • Pet of Tower
  • Baseball Loaded III

They can’t all be winners.

TransProse

Creating music from text:

Using natural language processing, TransProse reads in text and determines densities of eight different emotions (joy, sadness, anger, disgust, anticipation, surprise, trust, and fear) and two different states (positive or negative) throughout the novel. The musical piece chronologically follows the novel. It uses the emotion density data to determine the tempo, key, notes, octaves, etc. for the piece depending on different rules and parameters.

via Steen’s newsletter. You can buy the first album of generated music on Bandcamp.