Finger Printing The Web Content Revolution

Finger Printing
Finger Printing the Web Content Revolution

A growing number of major players in Web video and music are turning to “finger printing” content protection technology and services supplied by Audible Magic to secure their assets. In mid-March, Viacom International’ MTV Networks acknowledged it was working with Audible Magic, joining 20th Century Fox-owned MySpace, Sony Corp.-owned Grouper, GoFish,,, NBC Universal, the four major music labels, hundreds of independent content owners and other social networking and user-generated content (UGC) sites in the use of Audible Magic’s CopySense technology.
 Founded in 1999, Audible Magic offers technology and services that could help settle a burgeoning showdown over monetization of commercial content on UGC sites. Audible Magic encapsulates and stores core content elements – finger prints – along with content rights rules to provide a ready means of identifying instances of forbidden usage. With this technology media companies enjoy some greater assurance that UGC sites can prevent abuses by tapping into the finger print identifier and DRM database online and to block unauthorized uploads of copyrighted content “into the wild.” Further, the company has developed a CopySense Plug-In for P2P developers; CopySense Network Appliance for business, education and government networks; and RepliCheck for the media manufacturing industry.

Less noticed, the company has the potential to shake up assumptions about DRM (digital rights management)  by enabling content owners to register unique fingerprints for each piece of video or audio along with associated DRM permissions and other business rules in a database that now counts more than five million works. This model allows DRM to effectively reside in the Internet cloud, rather than in DRM wrappers enveloping each piece of content.

In this broad discussion with ScreenPlays senior editor Peter Lambert Audible Magic co-founder, president and CEO Vance Ikezoye explains that his company’s intent is not only to enforce copyrights online, but to facilitate unfettered business model experimentation that he believes is destined to create “the next cable TV.”

ScreenPlays: Can you give us a short description of what your technology does?

Ikezoye: At a 30,000-foot level, what our technology does is called finger printing. As opposed to watermarking, finger printing measures things resident in the content itself, either the music or the video, rather than embedding some kind of information in the file. So the way we operate is we take known content, and we take these measurements of the content that are robust through format changes – in the case of video, whether it’s 16 by 9 or 4 by 3, or in audio, whether it’s an MP3 file or a CD.

We measure some of the essence of how that video or song is perceived. We take those known samples and put them into a database along with the information about that piece of content. Then later on, with any piece of unknown content, we can take those same measurements and look them up in the database to compare it with known content.

SP: What does it mean to measure the essence of how a piece of content is perceived?

Ikezoye: If you think about MP3 as a compression algorithm, MP3 seeks to preserve the characteristics of a piece of audio and what people hear, and MP3 encoding throws away information that humans don’t hear. They may throw out really high or low frequencies. So they get to the things that represent most of the information, and that’s why when you have MP3 compressed audio, the file can be one-tenth the size of the uncompressed file. And you can really tell the difference quality-wise, but in essence you know it’s the same song.

So just take that down to the absolute extreme, and think about it in terms of the least amount of information that still can be contained in a compressed file and still communicates that this is that song. That’s what we do. A song that may be 30 megabytes as a full WAV file may be 3 megabytes as an MP3 file. Our finger print is probably more like 10 to 20 kilobytes. So it’s very small, yet it can very much uniquely identify a piece of music so uniquely that we can distinguish between a Pearl Jam performance of a song in a Berlin concert from the L.A. concert performance of the same song.

SP: Given your company name, does your technology use only the audio component to track video?

Ikezoye: Yeah, the name is kind of misleading these days. We have implemented the use of our audio fingerprinting to identify the soundtrack of video, but we’ve also had in development video imaging finger printing technology that we’re conducting tests on right now. So I guess we should be changing our name to Content Magic or Media Magic.

SP: But for the most part, I imagine that, in video, when you say ‘soundtrack,’ you refer to dialogue as well as music?

Ikezoye: Yes, and in fact, we believe that the soundtrack is probably 80 percent of the problem. You could identify most pieces of content uploaded that have both the video and the soundtrack. Now it doesn’t cover the place where you have a different image overlaid on the soundtrack or where people put a different soundtrack over their home movie, as an example.

But from the first pass of monetization, and from a copyright point of view, you want to define where the video and the audio are synched up.

SP: Your Web site suggests that, with your technology, there’s no need for DRM. Why is that? Because isn’t this an after-the-fact misuse of content tool, as opposed to rules that allow use?

Ikezoye: I think, first, in the grand scheme, we’re complementary to DRM.
DRM is a mechanism where you can wrap a piece of content and have within that wrapper some business rules pertaining to use and to enable business models. The downside of DRM is having to wrap, and once you’ve unwrapped, it’s out in the wild and you no longer have any ability to apply any kind of business rules.

As you know, there’s a lot of content out there that’s not wrapped. In fact the majority of content out there is not wrapped. So what we can do is apply those same kind of business rules to content that’s in the wild. If a piece of content shows up on MySpace, a copyright holder could have the ability to say, ‘When that shows up, we don’t want it published on whatever that site is.’ We can apply the same business rules after the fact, or where content may be user generated. 

People think about DRM in two aspects: from an anti-piracy point of view and from a desire to apply business rules. We can play both of those roles on the work itself, rather than the file. What I mean by that is, if a file is DRM wrapped, you could have that same file that’s not DRMed, and we could apply the rule to both, whereas with DRM wrapping, the content has to go out of the box with DRM on it.

SP: So in the case where you’re applying those rules to the unwrapped content, in terms of enforcement, can you actually keep the piece of content from getting distributed? Or is it a matter of catching it after it’s distributed, so that, like watermarking, it’s a forensics tool?

Ikezoye: It depends. Where we can be implemented in a peer-to-peer client like iMesh, for example, if content shows up, post-release – post-emergence into the wild – we can apply that rule that says, ‘This is a copyrighted song; the copyright holder doesn’t want this to be out in the wild; so block it from being copied or downloaded or uploaded.’

It’s similar in a MySpace case. Say someone had a pre-release song, and they wanted to put it on their own MySpace page, we could block it because we’d say, ‘Oh, this is pre-release, it’s owned by EMI, don’t let this be published on the MySpace platform.’

So you can implement some of the same types of controls as DRM without having to deal with a wrapper. To me, the reality of the world is that everything won’t be DRMed, and there will always be content that has been unlocked.

SP: Now, you provide services as well as products, and I gather a key part of that is registration of content in your database.

Ikezoye: Yep. We allow copyright holders – either studios or music labels or publishers – to register their content in the database along with their business rules. Today, on the music side, we have all the four major labels registering content. Most of it is pre-release. We have hundreds of independents who are registering.
And by the way, registering a business rule could be, ‘Let it be used. Let it be distributed.’ It doesn’t always have to be ‘Take down.’ Small garage bands could just say, ‘Whenever you see this piece of content, let it be up on every MySpace page.’

And then we’re allowing studios to register film or television content in the database as well. 

We then provide services to sites like MySpace and Grouper to identify content being uploaded, to filter content being uploaded, and to apply these business rules.

SP: How are you helping MySpace, and at what stage are you in with that pilot or deployment?

Ikezoye: I believe we are in production there, doing screening and uploading. The only thing pilot about it is the process of building the database. Right now it’s only NBC Universal and UMG, so we haven’t put in all the content from the other labels. We’re starting small and building so that we make sure this all works within their infrastructure.

So the system and the service are ready to scale. I think we’re just working with the copyright holders on designating what goes into the database.

SP: So what is the process that happens with screening and uploading? How does the service work as part of your offering?

Ikezoye: We give some software to MySpace, and as part of their content ingestion process, when users upload files, they process it with our software, and then that software calls home to the database and says, ‘What is this?’ We then tell them what it is in a business rule. We host all of the databases, but the content never leaves the site.

SP: So I’d imagine you’re building a substantial data center presence yourself.

Ikezoye: Yes, if you’d see my collocation bill every month. We’re buying servers and building a good presence in a collocation facility. It needs to be Class A, fully fault-tolerant, because there are a lot of files going up as part of the process for a MySpace, so you can’t slow it down. From the consumer’s point of view, if you upload something, MySpace doesn’t want it to sit there for two days before it shows up on the page.

SP: What other notable customers or partners have you garnered on the portal side—the MySpaces or any of these emerging peer-to-peer distributors?

Ikezoye: Some of announced places – we’re in the peer-to-peer space with iMesh ( On the Web 2.0, community front, we’ve announced MySpace, Grouper, and we have just announced GoFish. There are a few others we’ve signed but haven’t announced, and there are a lot more in process.

SP: I’d imagine there’s a two-way process: the more you establish yourself with the MySpaces and Groupers, the more ready the content owners are to look at your technology?

Ikezoye: Yeah, and I think it’s more than just ‘look at’ now. We’ve reached the point where people have confidence in the technology and confidence in our role and how we provide these services – a trusted third-party status. Now it’s a matter, frankly to me, more of the mechanics of ensuring that there’s a mechanism to get content into the database with the business rules.

The business rules are just as important, because you might want to vary the business rules from site to site. You might have a video that you allow an exclusive on MySpace and a month later on Grouper or somebody else. You’re going to start seeing more of that in music and film I think.

As you’ve probably seen, there’s quite a bit of creativity on some of these sites. You could easily see, at some point, that you’d have some sponsorship by some of the copyright holders of some mashup, and you give some exclusivity for using that content to one site versus another. And frankly, you start registering that kind of mashup in the end too.

[Several days after this interview, CBS Sports verged on this kind of arrangement, sponsoring an NCAA Basketball channel on YouTube, providing highlight clips and encouraging fans to interact around them.]

SP: Can you say whether you are in talks with Google/YouTube?

Ikezoye: We don’t generally comment on things that happen or don’t happen, so you need to talk to them about that.

SP: A February New York Times article quoted YouTube as saying that “On YouTube, identifying copyrighted material cannot be a single automated process.” Are you making that claim in the first place?

Ikezoye: I don’t think I’m making that claim. What it always comes down to is helping to enable business models between the copyright holders and these new sites and services, especially UGC (user-generated content) sites. And getting down to some business arrangement, there will always be discussions of guidelines for what’s allowable and what isn’t, what you can get paid on and what you can do and can’t do. That is primarily a business discussion.

Once that business discussion occurs, then I think the technology is in a place where it can help enable those business models and the tracking and computation that’s required to make the models occur. Now, complete, thorough identifying of every piece of content coming up and where it comes from, that may happen in the future, but I don’t think that’s the issue today.

Today the issue is how do we prevent straight out piracy and how do we prevent taking a movie and putting it up on one of these sites, and more importantly, how do you enable the copyright holder, if they choose, to get paid. From that point of view, we solve a lot of that problem. A lot of people in the technology space think that you have to be, out of the chute, a system that solves every problem. The detractors always say, ‘You don’t solve this or that,’ but there are very few systems that solve everything.

Stepping back, I think our existence allows businesses and these UGC sites to actually do a business deal, instead of ending up in a world of legal actions and takedown notices.

SP: The Electronic Frontier Foundation has published some critiques of your CopySense appliance and charges that it’s easily defeated with SSL. How are you responding to that claim?

Ikezoye: That’s with our network appliance rather than the implementation we have on sites like MySpace. We have a network appliance that can be installed on a university network that can control file sharing to filter and prevent copyrighted works from being traded. EEF says that if it were an encrypted peer-to-peer, we can’t do that copyright filtering, and that’s a correct statement. However, most P2P today is not encrypted, and secondly, it goes back to what I was saying about addressing a lot of problems with technology, as opposed to resorting to pure legal issues. That’s kind of my theme: now we can start talking about how to take some proactive measures, instead of sitting back and being reactive.

SP: Joost, Amazon Unbox, Netflix, BitTorrent and others are creating a wave of online video service launches, a lot of them P2P. a lot of them offering high-value content. Is Hollywood at a tipping point in terms of rights protection confidence, or is all this just more experimentation for the time being? 

Ikezoye: I think, if you look back, there is real interest in trying to work with these new business models to see if there can be a win-win. I don’t know if anyone on either side is completely convinced of its success, but we need to applaud people for trying and having a dialogue about it.

If you think about what happened with the recording industry six years ago when P2P first came out, it was all legalities and trying to shut them down. Today there’s a lot of discussion about trying to cut deals and share revenues and use advertising as a mechanism to get paid. I think those are all going in the right direction toward willingness and openness to be innovative. You see the studios especially involved in a lot of different content delivery networks, Web 2.0 sites. It’s very proactive in trying to explore and co-develop new biz models.

SP: This may go to what you said about no one solution fixing it all, but should anybody even be looking for a tipping point in terms of aggregate technologies – DRM, encryption, finger printing, watermarking – that together make the marquee copyright holders confident that they can really unleash their content?

Ikezoye: I’d like to claim credit for our technology playing that role. I think it’s a bit of that: there’ a sufficient level of confidence in the technology and this kind of solution. The other part of this tipping point, I believe, lies with these new mechanisms for distribution. They’re both converging in a way that’s a positive. It’s not one by itself. Maybe it’s the content owners’ willingness to experiment with these things that has caused more people to create more business models and new mechanisms. I don’t know which is the cart or horse.

All I can say is that I see it coming together right now. You know that some of these business models are not going to succeed, but some of them are going to be really successful. Who would have guessed in the early days of cable TV when it was community antenna retransmission of broadcast stations that it would become what it has?

SP: You’ve focused on UGC sites, but there’s a lot of innovation in P2P, and Joost and others will have Web 2.0 characteristics, with posting and sharing and communicating. Do you have a role there, as well as in the user-generated, social sharing networks?

Ikezoye: Potentially. It’s not the first place we’re focused, but any time content is delivered over a pipe, there will be some needs for either compliance or auditing or even revenue generation from advertising, and to know what the content is and to track what’s being used. That’s the role we’re playing: what is it, and how many times is it being used? I can’t say we have any answers for a download model like a BitTorrent.

But over time, user-generated, user-manipulated content is just too powerful not to go mainstream. It’s going to change the way the business is.

SP: Because it’s so personalized, because it embodies so much user control?

Ikezoye: First, the move from physical and analog content to digital means everybody’s PC can become a video workstation. Secondly, I think there is a tremendous amount of creativity. You’ve seen Chevy hold a contest for commercial [advertising] production. Third, this whole phenomenon of communities is one that has legs, and community is about identifying communities. The more you can use content to form and serve communities, that’s what you’re trying to accomplish.

SP: There also seems to be an explosion in turnkey video publishing services, and that seems to be spawning all kinds of new online video producers and programmers with niche content and channels. How much interest and business are you getting from that realm, as opposed to established old media or the bigger video or social networking portals?

Ikezoye: It is a mix. We’ve got 75 to 100 people talking to us in the active funnel right now. There are sites that are trying to cultivate the new creator communities. They’re interested in the problem of one of these creators trying to upload commercial, copyrighted content. Secondarily, where it is unique, commercial content, those creators want to make sure it’s protected, tracked and identified later on other sites.

SP: Do you believe there’s a sense among old media that, even without completely settled protection issues, this new democratized creativity and competition means they have to be there or be left behind?

Ikezoye: I think if you go through our list, we have lot of old media companies talking to us. Fox owns MySpace. Sony Pictures bought Grouper. That’s old media recognizing the power of these new areas as creative platforms, as well as distribution platforms.

SP: Are consumer electronics companies as big a target for you as Web service providers and content owners, and how do you work with CE companies?

Ikezoye: We haven’t focused on that a lot up till now, but clearly there’s some opportunity for us. The first place in CE will probably be the evolving [mobile] phones, where it’s a community network and where multimedia content is being consumed or traded or copied. That’s an area where we’ll likely be active. Once you build that capability into business rules in the database, that’s an easy transition.

SP: Is Gracenote a direct competitor to Audible Magic? And are there others?

Ikezoye: I think their focus has been in the CE device, identification services for radios and PC applications, and we really haven’t played there. There are people who use fingerprinting to do other things.

The point is that it’s not the technology that Audible Magic has, though it’s an important component, but it’s the solution and role we provide for both the sites like MySpace and the copyright holders. We provide an independent third party that can be an intermediary or bridge between the two, and the technology is just a piece of that.

If our technology is being used to track usage for compensation, you frankly don’t want that role sitting with the site, and equally you don’t want it owned and run by the copyright holders. You don’t want either fox to be responsible for counting the chickens.

SP: One last question for you. Is there something overlooked, something you find yourself evangelizing about?

Ikezoye: I’m always of the position that people tend to overestimate the amount of change over the next two years and underestimate it over the next five or 10 years. I think we’re just at the beginning of new businesses and new business models around content that we can’t even imagine today. We’re just at the start of our ability to identify and recognize and classify and count content, and that is going to be a real critical piece of things going forward.

Over the next few years, a lot of models will blossom. Some of them are going to fall away. But what’s the next cable TV? That’s the question. There’s one lurking somewhere. Who knows what it is. I just hope to be part of it.


Please enter your comment!
Please enter your name here