One of many gloomy predictions made in Orwell’s 1984 was that all popular music would be computer-generated without human input. Orwell might say we’re one step closer to proving that prediction true.
Polyphinic HMI‘s Hit Song Science (HSS) technology, in a nutshell, 1) analyzes the acoustic properties of a new song, 2) finds similar songs released in the past, 3) extrapoloates sales performance of those ‘similar’ songs to statistically predict the expected level of success for the new song, and 4) provides potential suggestions to improve statistical “ratings.”
Recently, Polyphonic HMI had predicted that Aslyn‘s song, Be The
Girl, would break the Top 30. In less than two
months, it did. Further, Polyphonic also predicted the success for albums by Norah Jones and Maroon 5 before they were released, , creating a ton of press for the company.
Impressive. And sad? Does HSS mark a woeful day in songwriting history? HSS seems to prove what we’ve all known and feared: consumers prefer music that is “familiar.” In an industry already criticized for recycling old sounds, a technology that predicts success based on old hits seems liable to entrap artists in a perpetual repeating expression loop. Loyal customers of HSS now include many major labels; as well as independent bands trying to garner label attention with high HSS song ratings.
Jo Tatchell presents a series of interesting arguments from both sides of the fence. Jo concludes that, basically, the technology should be used as a “tool” by A&R reps; however, no technology can completely replace the talent required to write and perform a good tune. HSS is best used as a device to “fine tune” the structure and sound of a song, much like an audio engineer will tweak the EQ to get a recording just right.
Still, with the industry having become more and more risk-averse in terms of music selection, and moving towards mathematical, algorythm-based means of artist selection, Independent labels and artists will be more crucial than ever for music to evolve as an art form.
]]>
Hit Song Science Software Predicts Big Singles
by
Tags:
Comments
12 responses to “Hit Song Science Software Predicts Big Singles”
-
> consumers prefer music that is "familiar."
Selling music to "consumers" is the old bad way. The new better way to view recorded music is as a showcase for talent so that a *customer* (not a consumer) can be won over to then pry money from their wallets over a longer term. Whether the customer is an avid fan who will pay to come see you play live or a private company who will insert your song(s) in their commercial(s), these people are the keys to your success. By contrast the sheep tuning into FM radio and wandering the CD isles at Barnes & Noble who drive the large label mega sales don't matter to you. Or if they do matter to you, they shouldn't, because they don't give a crap about you or your music. Don't make music to be consumed, if you must make something to be consumed …write and release a cheeseburger.
-
The songs it predicts will be hits are probably ones I am not interested in anyway.
-
Hey, I bought a CD from B&N once..!
What you say, Pete, is true — music itself is becoming more of a promotional vehicle than a product. But…though long term revenue is a key financial goal, when you're just starting out, you will need to peddle your sound to strangers you have yet to meet. Your customers do not yet exist.
The early stage is the toughes for any band..!
-
Overreaction and doomsday predictions about new exciting technology seem to be the norm. Has anyone thought that a tool like HSS could open the door to a great line-up of independent talent who at the moment can't get their music heard by the major record companies. If their creations arrive with an HSS analysis and a high hit probability rating they may be listened to. Has anyone thought that HSS technology may be able to save the record companies millions of dollars in wasted marketing resources (promoting low hit potetial) and that this could lead to cheaper music for everyone. One spin-off from this may be less pirating of records. Come on guys get postive!!!!
-
Graham,
Your points definitely make a strong case for the positive side of HSS. Most songwriters believe in an innate talent to write good music. Collaboration and feedback can be cruicial for fine tuning a song. However, it's difficult for some creative artists (in cluding me 🙂 ) to accept artificial intelligence as a replacement for intelligence.
Maybe that's something we need to get over.
Many believe college degrees and graduate mostly evolved out of a need for individuals to earn a "stamp of approval", and hiring organizations to be able to flag qualified candidates with potential for advancement, despite the lack of actual education that took place.
HSS helps to provide a similar filtering service. And true, record labels would be more apt to sign more new bands if they could predict with greater certainty a band's success. So in that sense, I agree.
My main issue is with the use of past music to mold future songs. But then, so many bands already identify themselves with a sound as a "Band A" meets "Band B" or "Genre A" meets "Genre B", so we all in some sense already think in terms of the past when describing the current.
Anyway, I'm babbling a bit now… Who knows. The more I think about it, the more I want to try HSS just for the sake of sharing its results with GarageSpinners. (and any marketing initiative, for that matter). We'll see.
Thanks for your comments!
-Mike
-
I think i have got a hit on my hands hear. Run it through for me guys.
Here is a link to the song, “The Blues is Ok”
http://profile.myspace.com/index.cfm?fuseaction=user.viewprofile&friendid=25268126
-daddy rich
-
I’ll believe in this program but you have to do one thing for me first. Plug in the pertinent data (hits of the day) for Buddy Holly, Elvis, Pink Floyd and Led Zeppelin.
If it says they will sell millions, then I’ll be a believer.
-
You make a good point about this. If the industry uses this to predict pop potential there's a danger everything will sound alike after awhile if the analyzed songs become a part of the database. I think there is a danger there but I think people are impossible to predict even as crowd I aint worried, people like me will ignore the mainstream and write songs that we like.
-
You make a good point about this. If the industry uses this to predict pop potential there's a danger everything will sound alike after awhile if the analyzed songs become a part of the database. I think there is a danger there but I think people are impossible to predict even as crowd I aint worried, people like me will ignore the mainstream and write songs that we like.
-
You make a good point about this. If the industry uses this to predict pop potential there's a danger everything will sound alike after awhile if the analyzed songs become a part of the database. I think there is a danger there but I think people are impossible to predict even as crowd I aint worried, people like me will ignore the mainstream and write songs that we like.
-
Rob: And that's really key. Music as an art form will always be able to move forward, simply because, there are a lot of "rock star" musicians who have the guts and determination to create something unique and ground-breaking… Maybe not all — there are also a lot of hit-seekers that will just try to follow a formula. But there are quite enough indies and experimentalists to ensure music will continue to evolve.
-
Speaking of entertainment, the city is also full of shows and sporting events, too. The Circus Maximus Theater, for example, is home to many musical guests. Everything from rock to country is performed there. In fact, they often do comedy shows, as well.
Leave a Reply