Fwd: Hit-song predictions get a scientific spin

From: Wade T. Smith (wade.t.smith@verizon.net)
Date: Fri 09 May 2003 - 11:36:36 GMT

  • Next message: Scott Chase: "Re: Fwd: Hit-song predictions get a scientific spin"

    Hit-song predictions get a scientific spin

    By Joan Anderman, Globe Staff, 5/9/2003

    So you think you've written a hit song? Guess again. Perfect hooks, killer beats, and powerhouse singing add up to nothing for many aspiring songwriters -- and for ambitious record executives, who have historically relied on good ears and gut instinct when it comes to finding talent. A new company is trying to take some of the guesswork out of the artwork, with help from science and supercomputers. Hit Song Science already is generating controversy, with some artists and record-label insiders saying it only highlights the desperation of a struggling music industry.

    Hit Song Science is a high-tech music analysis system that compares new songs to a massive database of chart-topping singles and predicts hit potential based on shared attributes.

    In other words, the more your song has in common with Usher's ''U Don't Have To Call'' or Santana's ''Smooth,'' the better your prospects for stardom.

    All five of the major record companies -- BMG, EMI, Sony, Universal, and Warner Bros. -- are currently using the service founded last year by Barcelona-based Polyphonic HMI. A modified online version, geared toward songwriters, was launched this week at www.hitsongscience.com.

    ''Our technology is to music what X-rays are to medicine,'' says Polyphonic HMI CEO Mike McCready. ''We help the record industry see their market and their music in a way they were previously unable to do.''

    Hit Song Science technology isolates sonic patterns in a song, ranging from tempo and chord progressions to melody, harmony, and pitch, and then compares the song to ''hit clusters'' gleaned from its database of 3 1/2 million songs. The system is updated weekly with new releases in order to effectively predict a song's potential for success in the current market.

    Ironically, HSS arrives at a moment when those in the music business face criticism that popular music is increasingly derivative and homogeneous.

    ''This is just another reason why the music industry is going down the tubes,'' says singer-songwriter Ellis Paul. ''We need to think on the edges, not down the middle.''

    But McCready says his company isn't trying to encourage cookie-cutter music.

    ''We hope we can help labels look at music that doesn't sound formulaic but will still return on their investment,'' he says. ''For example, we predicted the success of Norah Jones's [Grammy winning] `Don't Know Why.' Nothing in our data base actually sounded like it. Rather it was the combinations of patterns and properties that indicated hit potential.''

    But Jesse Harris, the New York songwriter who composed ''Don't Know Why,'' is skeptical. ''It sounds like a coincidence to me,'' says Harris. ''If they tell the label it's got hit potential and then the label puts lots of money into it, maybe that's why the song's a success.''

    In addition to using HSS to choose album singles, labels are also utilizing the technology to help screen music submissions from unsigned artists and short-list those identified by HSS as worthy of a closer look.

    Locally, the small label that's home to pop-rock band Elcodrive paid
    $3,000 for a 19-page analysis of songs on its debut indie album, which the band includes in a detailed submission package when trying to lure major labels to sign it. (Story, Page C1).

    But plenty of industry insiders and observers are worried that Hit Song Science reduces the artistic process to a stack of mathematical data, sacrificing creativity in the name of profitability.

    ''I think it's terrible,'' says Leigh Lust, senior vice president of A&R at Elektra Records. ''Look at the unique, inspiring artists that would never pop through if every label adopted this. It will make labels product-pushers even more than they already are.''

    ''It's a fascinating tool, and smart companies will use this as ancillary information,'' counters Jeff Fenster, senior vice president of A&R for Island/Def Jam Records. ''Especially in the present state of the industry.''

    ''But don't live by it,'' he warns. ''There's more involved in how music connects than what can be read in lines on a graph.''

    A Polyphonic album evaluation includes detailed analyses of each song as well as a numerical grade between 1-10. A rating over 7 means a song has a strong hit potential. The online service charges $49 per song and delivers a less in-depth report to songwriters within three days.

    Hit Song Science is hardly foolproof. It doesn't factor in such key elements as a song's lyrics and an artist's personal charisma, or intangibles like marketing and advertising campaigns.

    ''There are three requirements for a hit,'' says McCready. ''It has to sound like a hit to human ears, it has to have the right promotion, and it must have optimal mathematical patterns. That's where we come in.''

    Joan Anderman can be reached at anderman@globe.com

    This story ran on page A1 of the Boston Globe on 5/9/2003.
    Copyright 2003 Globe Newspaper Company.

    =============================================================== This was distributed via the memetics list associated with the Journal of Memetics - Evolutionary Models of Information Transmission For information about the journal and the list (e.g. unsubscribing) see: http://www.cpm.mmu.ac.uk/jom-emit



    This archive was generated by hypermail 2.1.5 : Fri 09 May 2003 - 11:45:42 GMT