Another Listening Post Wired Blog post from Eliot Van Buskirk had me going today:
Amateur musicians should eventually be able to use similar technology to create entire songs using only a vocal melody and an idea of which band -- or mix of bands -- would sound right playing the accompaniment. Want a backing track for your "Somewhere Over the Rainbow" cover that sounds like a mix of Radiohead and Green Day? Soon, you could be able to click a button and make it so.
Pretty cool stuff... the program will take songs you "feed" it from a particular artist and become that band in style - they call it "Automatic Style Specific Accompaniment" or ASSA. You can see this in action in this video which shows the ASSA success rate of 82% in coming up with the melod of Creep by Radiohead after being given 3 of the band's songs (High and Dry, Fake Plastic Trees and Airbag).
Who knows, eventually an artist/songwriter may license their "style" under ASSA? Will you be able to copyright a style? I guess if it can be analyzed and stripped to such a point that it can be copied and utilized to another songwriter's benefit, the answer may be "Why not?".
Right now the system is bits and bytes and MIDI-based, but theoretically, it should work on audio files as well in the future. As the author states in the article: "If all you need is a melody, lyrics and a concept of which band or bands you want your accompaniment to resemble, the bar to songwriting will be lowered."
Something to think about...