MIT Software Predicts Music Hits

TaranT · Post by **TaranT** » Sat Nov 12, 2005 4:23 am

...Brian Whitman and Tristan Jehan have devised a computer program that listens to a song, then predicts how humans will react to it....

The goal is to pinpoint trends in pitch, rhythm and cadence that are driving consumer spending habits. However, the MIT researchers believe they've taken the science to another level.

"Some people really care about instrument sounds and complexity of the music," Mr. Whitman said. "But the 14-year-old teenage girl could care less, as long as her friends are listening to it."

The MIT method, developed at the school's renowned Media Laboratory, also takes into account social responses to hit music that are fed into the algorithms.

The researchers pull data from weblogs, chat rooms and music reviews -- anywhere a song is being discussed -- and feed it into the computer, which allows the software to gauge the popularity of a certain sound.

Once all the information is tabulated, the computer can listen to an entirely new album and predict how people will respond based on what it knows about the latest reactions to the music it has already heard.

If it sounds far-fetched, consider this: the system has been predicting Billboard hits with surprising accuracy over the past several months. While people may think their musical tastes are unpredictable and whimsical, they are actually quite traceable, Mr. Whitman says.

source

Org Profile · Post by **Otohiko** » Sat Nov 12, 2005 1:06 pm

I demand it listens to Captain Beefheart!

rubyeye · Post by **rubyeye** » Sat Nov 12, 2005 1:20 pm

Interesting, but I still think it's bullshit. If you can "quantify" music, then you can "quantify" ART. (that debate has been going on for centuries)

I'd like to see their program compute the popularity of the latest Death Metal Bands, or Progressive/Power Metal ~ you know, things most "mainstream" public doesn't know exists because they are rarely Billboard Chart material.

Forget about the ones whose names people know about, Dream Theater, Queensryche ~ fuck 'em. I'm talking about the really obscure bands, such as oh...

Stride - Imagine
Cloudscape - S/T
Vision Divine - The Perfect Machine
Place Vendome - S/T
Redemption - The Fullness of Time
Nocturnal Rites - The Grand Illusion

TaranT · Post by **TaranT** » Sat Nov 12, 2005 3:25 pm

I think they're aiming to predict the few songs that make it to the very top of the popularity charts. This is a "simple" problem compared to scaling and rating the huge mass of other music - which would be much more difficult; i.e. impossible.

downwithpants · Post by **downwithpants** » Sat Nov 12, 2005 3:26 pm

i can imagine this as a useful (though potentially dangerous) economic tool for its predictive powers on popular tastes, but i can't imagine it has good predictability on individual musical tastes, because the enjoyment of music is largely context-dependent, and context can't be accurately represented by a formula in an algorithm.

Org Profile · Post by **Otohiko** » Sat Nov 12, 2005 3:39 pm

Well, in fairness, it HAS been proven that certain melodic patterns are inherently 'catchy' and whatnot - I've seen other ways that's been quantified before, so it's not that surprising.

But they're obviously exaggarating a bit; going by quantities alone, I don't think there's any possible way that any brain would find something along the lines of semi-atonal [my beloved King Crimson's usual] riffs in 11/8 anything but frustrating

Unpronounceable_Symbol · Post by **Unpronounceable_Symbol** » Sat Nov 12, 2005 5:04 pm

Please realize that the machine isn't doing anything label execs haven't been trying to do strictly by ear since the 1940s; it's just proving that there is a certain quantifiable element behind it. Aside from working in the musical medium, it's got little to do with music, and everything to do with potential record sales. There are probably label execs who listen to Beefheart and KC but, if confronted with the idea of selling them on a world tour, would just chuckle sadly, because as Otohiko mentioned, nobody in their right mind could find it "catchy."

Org Profile · Post by **Otohiko** » Sat Nov 12, 2005 6:43 pm

Unpronounceable_Symbol wrote:There are probably label execs who listen to Beefheart and KC but, if confronted with the idea of selling them on a world tour, would just chuckle sadly, because as Otohiko mentioned, nobody in their right mind could find it "catchy."

Sort of OT, but here's an ironic case to illustrate that:

I consider the 70's prog band Gentle Giant to be pretty much the most talented group of people I ever heard. The band went under mostly because it became unsupportable. However, one of their primary players, Derek Shulman, later went on to become vice-president of PolyGram, and then president of ATCO records and then Roadrunner records, and signed massively popular bands like BonJovi, AC/DC and Slipknot. Which, in my view, are even at their best not half as talented as Shulman's own musical career was.

badmartialarts · Post by **badmartialarts** » Sun Nov 13, 2005 2:06 am

I assume this program is using some sort of goal-seeking neural net to listen to music and equate it with a popularity level as defined by mainstream music culture. In which case, this is highly likely to be accurate. Why? Because people as an aggregate are scarily easy to predict. Now, people as individuals are far more difficult. Asimovian psychohistory, anyone?

Scintilla · Post by **Scintilla** » Sun Nov 13, 2005 2:19 am

<b>COULDN'T</b> CARE LESS