Nielsen and Digimarc Announce Watermark-Based TV Content Identification Service

"Nielsen, the media measurement company, has teamed up with Digimarc to introduce an online content identification scheme for the US television industry based on watermarking technology. The two companies announced that they are collaborating on a scheme called Nielsen Media Manager, based on watermarking technology for video content that Nielsen has been using for four years in its media measurement business. Digimarc, the principal holder of patents in the watermarking space, is contributing technical assistance as well as IP licensing to the scheme.

Nielsen Media Manager is aimed at roughly the same set of customers as video fingerprinting technologies from companies like Audible Magic, Philips, Thomson, and Vobile: user-generated content sites, social networking sites, file-sharing services, and so on. Projected availability of the technology is mid-2008, and the companies intend to expand it beyond television content to movies, games, and other content types in the future.

The availability of a scheme like Nielsen Media Manager takes watermarking to a new level of viability for online content identification. Thus it begs a comparison with fingerprinting. Vendors of these technologies -- particularly those, like Philips and Thomson, that offer both watermarking and fingerprinting -- like to say that the two technologies are complementary rather than competitive, which is true to some extent. But they both apply to the conceptually simple task of identifying a file as a copyrighted work.

The most basic difference between the two technologies with respect to the generic task of content identification is that someone -- a content owner or service provider -- must embed a watermark in a file for it to be detected by a downstream entity (such as a file-sharing service), whereas this is not necessary in fingerprinting. Furthermore, the downstream entity must use a watermark detection tool based on the same scheme as the one used to embed the watermark. Yet watermarking is more accurate than fingerprinting, and it is more efficient to detect watermarks in video than it is to compute fingerprints, meaning that -- all else being equal -- it is cheaper for downstream entities to implement.

The other advantage that watermarking has over fingerprinting is that any data at all can be embedded as a watermark (subject to size limitations), whereas if fingerprinting works correctly, the same content will yield the same fingerprint. The same content can contain different watermarks; a good example of this is Universal Music Group's use of watermarks in unprotected MP3 music files to denote the different retailers (e.g., Amazon, Wal-Mart) that sell them. Different watermarks can be used for various purposes, such as to signal different usage terms (e.g., promotional vs. paid download) or different ads that should run against the content.

Yet this flexibility requires that standards be set for identifiers that determine what to do with content, as well as rules for interpreting those identifiers, in order to avoid the chaos of different schemes for content owners, retailers, advertisers, etc. And because watermarks typically can only contain a few dozen bytes of data, it would be necessary to maintain online directories that connect watermarks to information about how to interpret them.

That's exactly what Nielsen is developing. It already has a database of watermarks for a large portion of US television content, as well as audio signatures (fingerprints) that it uses as backups in case watermarks cannot be detected. It also has a proprietary identifier scheme. Nielsen intends to open up its database to third-party schemes, including those based on identifier standards such as ISAN. That way, an application can use a Nielsen-embedded watermark to look up (for example) an ISAN number, which in turn can lead to other information about the content. This falls short of the ideal of a totally open standard for globally unique content identifiers; it positions Nielsen as a gatekeeper.

Still, many dots have to be connected in order for watermarking to fulfill its potential in enabling business models for unencrypted online content, and Nielsen Media Manager is the first serious effort toward making this happen. If Nielsen and Digimarc are successful in getting a variety of online entities to adopt the technology, it bodes well for the future of watermarking in general."

By Bill Rosenblatt, DRM Watch