|Nov 13, 2005 | Articles|
Understanding what makes pop music popular
by Eric Bangeman
[from: ars technica] Back in the day, hits were tracked by record sales and predicted by how they sounded to music industry veterans. If you really wanted to test the waters, you could let Dick Clark play a soon-to-be-released single on American Bandstand and see if people moved to it. ("It's got a good beat and you can dance to it. I'll give it a 93, Dick.") Even today, the tastes of the record-buying public are something of a mystery to the labels. A couple of PhDs at MIT may change that with a program that mimics the musical tastes of the public.
The application, written by Brian Whitman and Tristan Jehan of MIT's Media Laboratory, "listens" to music, analyzes elements of a song (e.g., pitch, beat, tempo, melody) and gives it a thumbs-up or thumbs-down. When compared to the Billboard charts, the software is "surprisingly accurate."
Predicting which music will get listeners on the dance floor, and more importantly, to buy CDs or individual tracks from online music stores has become big business. One company, HitPredictor, scored big in 2002 when it hit gold with its advice on how to stagger releases from a Christina Aguilera album to maximize sales. Unlike Whitman's and Jehan's application, HitPredictor uses a combination of focus groups and other market data to determine how the public will respond to new music.
Music retailers of all stripes would love to have reliable data on consumers' musical tastes. All of us have had recommendations thrust at us on Amazon or one of the online music stories. Those are hit and miss, and in my case, more often miss. Having an application that is able to analyze the songs or CDs in your shopping carts and then use a reliable algorithm to come up with suggestions that you would actually like would thrill retailers.
Will software that can nail the musical tastes of the public lead to even more homogenized-sounding airwaves? If Whitman and Jehan have their way, it won't. Their goal is to actually broaden the musical tastes of the public by using data gleaned from the application to get a wider variety of music on the radio. Anything that results in less bubblegum pop and whiny rock is fine by me.
Source URL (ars technica): http://arstechnica.com/news.ars/post/20051113-5560.html