YouTube is growing up. The video-sharing site, recently acquired by Google, is taking steps to become a more responsible corporate citizen. At the very least, it intends to avoid the fate that just befell Bolt.com and the Sony-owned Grouper, two similar, but smaller sites who are facing massive lawsuits for copyright infringement.
Universal Music, the same record label that led the legal charge against Internet music renegade Napster, is asking for damages of up to $150,000 for each instance of infringement on Bolt and Grouper—every streamed Mariah Carey video, for example. Grouper and Bolt, Universal said, “cannot reasonably expect to build their business on the backs of our content and the hard work of our artists and songwriters without permission.”
Lately, YouTube has been actively seeking that permission. The company recently struck ad-revenue-splitting partnerships with Universal Music, Warner Music and Sony BMG. But now comes the hard part. As a condition of those deals, YouTube has agreed to filter its site for the labels’ music and offer the music giants the opportunity to pull infringing content off the service if they don’t like how it’s being used. The technology that will help them do this is called “acoustic fingerprinting.” If it works, this model could help put the music industry back in the driver’s seat after years of watching their product enjoyed for free on the Web. If it doesn’t work, the newly minted “GooTube” and even MySpace may be joining Grouper and Bolt as defendants in court.
Think of acoustic fingerprinting as an automated version of the card game Go Fish. Record companies give their complete catalog of songs to acoustic-fingerprinting technology firms like Silicon Valley’s Audible Magic. Audible Magic’s software analyzes each song and assigns it a numeric identifier based on the unique properties of the sound file. Then the firm’s antipiracy filter monitors the songs uploaded to the media-sharing Web sites and attempts to find matches—digital files that share the same acoustic characteristics as the originals. If they find a match, the software filter allows the copyright holder to decide whether they want to yank the video or grant permission.
Filtering technology would be a significant move toward giving industry players a better idea of how much of their proprietary material is circulating on the Web and more control over how it is used. That’s not an easy feat. Sixty-five thousand new videos are being uploaded to YouTube each day. “Whenever you have a user-generated content system, the first thing you have to do is figure out what’s being stored on the network. If you don’t know what it is, there’s nothing you can do about it,” says Audible Magic CEO Vance Ikezoye.
Ironically, Audible Magic, a leader in the industry, owes its current privileged position to file-sharing pioneer Napster, the startup that initiated the music industry’s current troubles. Originally, the Los Gatos, Calif.-based firm, now seven years old, had a dot-com-flavored business idea: it was developing technology that could automatically recognize songs played over the radio—a sort of robotic, musical savant. When Napster suddenly took off, execs at Audible Magic decided that their song-recognition technology was better exploited for filtering the coming wave of copyrighted material that would be posted on the inevitable Napster knockoffs.
Today Audible Magic’s technology helps keep the major labels’ songs off one such peer-to-peer network: iMesh, which was sued by the music labels in 2003 but has since returned by promising to screen for copyrighted material. iMesh cofounder Talmon Marco says that the only reason the company is able to stay out of hot water is because acoustic fingerprinting is effective. “We haven’t seen any false positives, and we’re seeing a recognition rate of over 99 percent. If a company like YouTube wanted to check for users recording themselves dancing to a Madonna song, they can do that easily.”
Another major acoustic-fingerprinting player, SnoCap, was born in the flames of Napster’s legal woes. When the lawsuit-bedeviled start-up went bankrupt in 2002, several Napster employees, including creator Shawn Fanning, spun out the acoustic-fingerprinting technology that Napster itself was actively developing. “We started this company the day after Napster went bankrupt,” says Ali Aydar, a former Napster engineer who is now SnoCap’s chief operating officer. “We knew that if this marketplace was truly going to evolve, you had to be able to identify content.”
Unlike Audible Magic, SnoCap offers users an e-commerce buying opportunity when its software recognizes a rights-holder’s song. It recently signed a deal with MySpace to administer a music store on the social network and prevent users from selling songs they don’t own. Audible Magic, meanwhile, won the right to filter the YouTube service, at least for Warner Music content. Neither YouTube nor its new owners at Google have disclosed what technology they’ll use to give the other media companies control over their content.
One significant challenge in all this is that the technology only works for sound—the automated filters can’t yet understand the actual moving pictures that are the primary appeal of sites like YouTube. Audible Magic says it’s working on launching a video-recognition product next year, but SnoCap’s Aydar emphasizes the formidable challenges to developing such technology: for instance, software can’t easily match content if it appears in different video formats. For example, it might fail to recognize that a high-definition episode of the sitcom “The Office” is identical in content to a user-submitted copy of the show in normal resolution. And it might be entirely impossible to identify those jumpy, pirated copies of new movies that are filmed by bootleggers holding their camcorders in theaters.
That’s akin to being a very bad Go Fish player. And right now, the promise that they can play the game effectively is all that’s keeping online upstarts like YouTube out of the courtroom.