Let’s Remix Everything
Microsoft has — accidentally — unleashed the next musical era upon us:
(Here is the original, if you need a refresher)
The harbinger and bringer of this revolution is Microsoft SongSmith, which automatically generates a cheesy, MIDI style accompaniment to any vocal track you care to chuck at it. Naturally this has led to people generating deliciously awry cover versions of famous tracks, like that kids-TV-show version of the Beastie Boy’s Intergalactic, above.
Some other favorites are the Cheers style rendition of Eye of the Tiger:
And this Eurodancey version of Hotel California:
For all its silliness, I seriously think this is the near future of pop music.
DRM free downloads are standard now, and before long lossless audio will be commonplace too. I think the next big step is for bands (and orchestras) to start offering individual instrumental channels. Of course, you could still download the band-preferred mix of the instruments (i.e., something like the regular tracks you download now), but in addition you could download individual tracks for voice, drums, guitar, etc. and mix them yourself. The ex-consumers get to partake in the creativity.
And this is why software like SongSmith is so important. People will need tools to interact with these musical components at a level way above notes and chords. We need to seriously abstract our musical manipulation. I think the fact that people have gotten so excited about playing around with even this relatively simple tool demonstrates that there is a huge and hungry market.
Unfortunately, right now the raw materials — that is, a clean vocal track — are hard to get unless you happen to have an a-capella version of a piece, or know a dodgy friend of a friend who has access to original studio tapes.
Geek moment: this totally reminds me of the evolution of programming languages. Back in the beginning everything was programmed using machine code and assembly language, sets of indecipherable instructions like:
MOV ah,0x0
INT 16h
This is exactly like learning to read music in order to manipulate it: a long, hard and unintuitive learning curve. Something you have to devote a good chunk of your life to. Compare this to the graphical wires and functions I now drag around on a screen in lab every day, producing in five minutes programs which would take years to write in assembly or C.
I want ultra-high-level, powerful and clever tools to manipulate music.
But while we’re waiting, will someone with access to a voice-only version of the Ode To Joy, please — pretty please! — run it through SongSmith?




January 30th, 2009 at 8:26 am
Thanks for the interesting insight on this trend — as the creator of most of the Songsmith videos on YouTube (including two of the ones you linked) I’ve thought a lot about how this reflects on music in general.
I’ll see what I can do about getting you an Ode To Joy :-)