The idea is simple. After generating a music library, the player allows for a mood-based play function. With a keylogger and some sort of a device to scan any visible text or tags on the screen. The songs in your library are tagged (or auto-tagged, with title and artist info, or imported from Last.fm). The tags are then matched and the dynamic library is generated at a rapid refresh rate.

The result? If I’m reading a page on war, I’ll hear Montségur or Paschendale. If I’m entering a lot of expletives in my input, an angry, dark song will play. With calmer, easy pages we have calmer, easy songs.

May be a lot of effort, though and I doubt if anyone will trust a music player with a keylogger and an on-screen information cache.