Audio Fingerprinting
Name That Tune
As the brouhaha over Napster heightens, a simple solution may wait in the wings

Oddball modern classical operas aren't exactly at the height of popularity right now, but that hasn't got composer Erling Wold down. He feels pretty good about his prospects and those of other obscure musicians like him — and thanks Napster for it. Although his songs aren't actually available on the file-swapping site (by his choice), he feels better just knowing they could be. So it pleases Wold that his handiwork stands to rescue Napster from almost-certain death at the hands of the big bad record labels.

No, a Wold-original aria isn't going to save Napster's neck. But a little piece of technology he crafted just might.

The technology in question is a nifty little doodad called audio fingerprinting. It's a tech wizard's riff on Name That Tune: Given the smallest snippet of a song, it can make the important distinction between Jill Scott and Jessica Simpson. Because the current stalemate in Napster's legal battle is a result of its alleged inability to comply with the record labels' battle cry for police-state-like protection of their copyrighted songs, this seems to be the perfect solution. Moreover, as of March 27, the record labels are pointing to audio fingerprinting as a potential remedy for Napster's policing problems. Strangely, Napster's not buying it, both literally and figuratively.

Wold, who by day is an engineer with software maker Audible Magic Corp, developed the technology in a research lab and patent factory called Muscle Fish, which he started with four fellow engineers. (Muscle Fish was acquired by Audible Magic last year and now functions as its research and development arm.)

The genius of Wold's invention is that, unlike so many other proposed salves, audio fingerprinting is content-based. At a bare minimum, Napster needs to be able to identify songs so it can block copyrighted material or track downloads and then compensate the labels and artists. Napster's current solution — tracking songs through file names — is weak at best. It's the moral equivalent of trying to catch tennis balls with a teaspoon: A lot more get by than get caught. But unlike file names, the music itself can not be manipulated, meaning that audio fingerprinting should put an end to sneaky file sharers slithering past policing efforts by misspelling Limp Bizkit.

Wold may be disappointed in Napster's lack of enthusiasm for the audio fingerprinting technology, but he's not altogether bewildered by it. "In some ways, Napster just doesn't want it," explains Wold, "because they want us to believe that it's just too hard to [police the site]. But the truth is, they'd just rather not remove copyrighted material from the site because that's what gives them their value."

Vance Ikezoye, CEO of Audible Magic, concurs. Things are seldom what they seem, he warns, and in this instance, both sides are guilty of lip service. "The complexity [of the Napster case] is to understand everyone's real motivations," he says.

Both Ikezoye and Wold point out that Napster is fully aware of the fingerprinting technology's existence. "Napster would say that our technology isn't mature enough and not good enough," says Ikezoye. The real problem, he says, is that it is too good. It would allow Napster to police its site in earnest — a compromise that Ikezoye claims Napster doesn't want to make just yet. Napster officials did not respond to a request for an interview.

Despite their supposed alarm over the filtering issue, the record labels didn't say boo about audio fingerprinting as a possible solution until March 27, many months after they became aware of the technology. In fact, Audible Magic is in talks with one of the five major labels now. The record labels' suggestion to adopt this seemingly obvious solution only after multiple heated rounds in court may be evidence of what has been suspected all along: The labels are out for blood. "Any form of 'Hey, here's a solution!' coming from the record labels has them buying into the long-term existence of Napster," says Ikezoye, "and...they don't believe Napster has any right to even exist."

Of course, if Audible Magic's negotiations with a major record label are any indication, there could be a compromise on the horizon yet. Ikezoye reports that the label in question may be interested in fingerprinting its entire catalog of songs and in turn cutting a deal with Napster stipulating that the technology be used to calculate compensation. Moreover, should Napster move to a subscription model, Ikezoye predicts there will a public outcry for the fingerprinting technology. If they're paying for tracks, consumers may well want some assurance that the downloaded song is of good sound quality and is, in fact, the song requested — two things audio fingerprinting can do.

Disingenuous attempts at policing digital downloads aside, neither Napster nor the record labels may be able to escape the rise of audio fingerprinting. As Geoff Schmidt, CEO of rival audio-fingerprinting developer Tuneprint, points out, in a sense Napster's fate is a moot point.

Live or die, a huge volume of digital music will one day be policed, if not by Napster by someone else. "The music is out there and it is going to need to be identified," says Schmidt. "That's where fingerprinting comes in. Essentially it's too late to turn back. The genie's already out of the bottle."