Robot to Pick Hit Songs

As if the record industry needed any LESS legitimacy...

http://www.abc.net.au/science/news/scitech/SciTechRepublish_805711.htm

<i>Music robot set to predict chart hits
ABC Science News
Thursday, 13 March 2003

If a lot of music by manufactured bands sounds tediously similar these days, things may soon get a lot worse.

A Spanish company says it has developed software to replace talent spotters that will "listen" to music and predict a song's chances of becoming a hit.

The software, Hit Song Science, has been developed by Polyphonic HMI of Barcelona to help record companies determine the chart potential of a song before deciding whether to invest in promoting it, British science weekly New Scientist says.

It works by matching a song against the musical traits of known hits, searching for identified patterns in beat, melody, pitch, chord progression, harmonic variation and fullness of sound.

Polyphonic HMI has built up a database of 3.5 million songs and discovered that even though their traits are scattered all over the map, the hits are concentrated into a tiny number of trait clusters.

"There are a limited number of mathematical formulas for hit songs," New Scientist quotes the company's chief executive, Mike McCready, as saying. "We don't know why."

Five major record companies have shown interest in the software because, says Polyphonic HMI, it picked out the music of jazz singer Nora Jones months before it became a success.

She walked away with eight Grammy awards last month for her first album.

Songs with matching traits do not always sound the same, though.
Nora Jones falls into the same match as hard rockers Van Halen, and Beethoven and U2 are in the same cluster.

But the Beatles and Elvis Presley are, comfortingly, in a distinctive niche all of their own.

Plans are afoot to sell shops a retail version of the software.
The idea is that shop assistants could recommend new songs and bands to customers who input details of their favourite music.

Critics fear the invention will only strengthen stereotypes.

"The music industry is not exactly renowned for its daring exploits," Peter Bentley of University College London said, who makes specialist musical software.

"If you rely on the computer too heavily, you will miss out on new things."
The report is carried in next Saturday's issue of New Scientist .</i>