Gravy.fm is a long running experiment in audio analysis. Music is deconstructed into raw data and acoustic characteristics. Harmony, tempo, structure and other data points drive a custom playlist engine. Using listener statistics and social trends attached to time zones the engine attempts to evolve. This iteration is focused on ambient down-tempo electronica.There are others if you hunt hard enough, ask Susan. Technologies used include FMOD ,Python, Echo Nest, Last.fm .net, MySql and Shoutcast/Radionomy group.