Different programs entirely - that one you read about recognized similar trends in existing hits in new songs. It would be entirely incapable of writing music. Also - algorithms like that have likely done real damage to the quality of mass produced music and Emily did not, it wrote some beautiful pieces that can only be considered a net gain for humanity, in my eyes.
The short story is that David Cope fed an algorithm a whole bunch of Bach and it learned to write more Bach (this was called EMI). He then destroyed it and kept working after that producing something less specific, and thus, Emily.
I wonder what he’s up to lately - but I’m kinda crap at google-stalking and he hasn’t gotten any major news since he released Emily’s album. It’s kinda a bummer - this dude is a major stone for the history of AI music (and composition, generally) when it comes to start really writing that in a decade or two, and there’s far too high of a chance he’ll be largely forgotten.
But people really hated EMI, especially after it passed a musical Turing test.
DeepMind has gotten started on music now, too. What they’ve shown of it so far isn’t very good at all - but the thing to remember is that EMI and Emily just wrote the music, a human or midi player had to actually play it. DeepMind is building the actual performance of music, not writing it - it can better be thought of as a specific kind of ‘singing’ than any kind of ‘writing’. Will be interesting to see where it goes with that - but oh man, from what I’ve seen, the musicians are PISSED.