Comment Re:computers CAN make good music (Score 3, Insightful) 295
If I remember correctly, Cope's methods were vaguely similar to Mister Whitman's in the very broadest sense, in that EMI created a large probabilitive database to analyze the connections between melodic and harmonic events. Obviously, this database would be 'calibrated' to the style of the music given as input. I never studied EMI, so I don't know how deep the simulation of pseudogrammar went, but it certainly produced some interesting results. One year in orchestra we actually performed a faux-Mozart overture reconstituted from Don Giovanni. I remember that the one composer EMI was able to simulate with uncanny plausibility was Scott Joplin (insert predictably caustic remark of your choice).
The problem with Eigenradio is that it lacks even EMI's limited contextual awareness. It would be akin to writing a book by analyzing the most common positions of ink secretions on the page. And so, not surprisingly, the "music" produced is simply uniformly limpid collages of sound without any underlying form, direction, or meaning.
The problem with Eigenradio is that it lacks even EMI's limited contextual awareness. It would be akin to writing a book by analyzing the most common positions of ink secretions on the page. And so, not surprisingly, the "music" produced is simply uniformly limpid collages of sound without any underlying form, direction, or meaning.
