Digital Watermarking: Noise Only If You Can Hear It
If it ain’t signal, it’s noise. That’s the dismissive adage that computer programmers, technologists and musicians use to disparage information that is considered intrusive or in any way a disturbance in a communications channel. The concept of introducing data – such as a digital watermark – into the signal stream of one’s recorded music is, at first blush, an affront to both the musician’s art and the sound engineer’s science.
Just how can digital watermarking contribute to the purest exposition of the artist's composition?
In the most literal sense, it can’t, although the question assumes that the technological goal of digital watermarking is exactly in accord with the artist’s creative aspirations. Digital watermarking’s greatest service to the recording artist and composer begins when the muse has been sated and the artist’s work recorded and rendered into digital formats. In this way, it has more in common with radio, the phonograph or the CD player – or even the signalling tags used by record stores to control theft, all of which have served the artist and his distributors over the decades.
In the Internet Age, digital watermarking will become no less vital to the success of the artist than the technological infrastructure such as radio and CDs that comprise essential elements of the artist’s overall economic milieu.
Digital watermarking is, in the artist’s context, a technology that supports the commercial and curatorial functions that are unavoidable aspects of the vocation, as important as economic structure that supports his art, i.e. recording contracts or promotional programs.
Bottom line: artists need to be able to claim title to their work in order for them or their distributors to organize markets for it. As well, they need to be able to audit the distribution chain.
Because it is difficult at best to clearly define how value will reside in a world of instant access to music recordings, title must first exist for the music itself, at the very least to verify its authenticity. In a way, digital watermarking’s utility parallels that of the invisible marking schemes, such as invisible ink on apparel items that indicate its point of origin, that some consumer goods manufacturers are using to monitor and audit their distribution systems. [Check out interviews with artist Boggs and musician Lars Ulrich on the Blue Spike Arts page.]
Artists, consumers, distributors and presenters of all kinds need mechanisms to identify, locate, sort and collate the music in their own collections.
Without the instrumentation provided by digital watermarking, organizing these kinds of functions will become increasingly problematic as more recorded music comes online. For marketing efforts as well, watermarking will become ever more vital as distributors use it to distinguish bogus copies from legitimate copies and reward consumers who hold authentic ones.
The Napster, which represents for some the ultimate liberation of music from impeding media like records and tapes, is already breaking down. Copyright activists and pranksters are already interfering with the Napster’s indexing functions by mislabeling songs or linking up their collections of adulterated music files. (Some gadflies are mounting song files on the service that have embedded into them recorded tirades against music piracy.) The necessity of tags and labels for carrying data now contained on CD labels and jewel cases becomes ever more apparent as the volume of recorded music increases and crowds the hard drives of music fans who, no doubt, are missing a satisfying part of the music consumer experience with raw, unpackaged, hobbyist-quality download files.
For all the important functionality that digital watermarking holds out, there is growing artistic anxiety that watermarking technology could negatively influence the reproduction of recorded music. It's a valid concern.
No advance in sound reproduction has arrived without some consequences. Initially, radio had limited range and fidelity. Records came with their crackle and cassette tapes came with ubiquitous hiss. Digital watermarking has the fortune of being developed in an age in which superior fidelity of commercial recordings is the norm. As such, audio engineering and retailing groups have already put recording industry body, the Secure Digital Music Initiative (SDMI), on notice that they are concerned about digital watermarking’s potential impact on the quality of recordings.
At least one member of the Audio Engineering Society (AES) in May 2000 went on record with concerns that watermarking technology chosen by SDMI may introduce audible artifacts into recorded music. That engineer, Tony Faulkner, owner of Green Room Productions in London, questioned whether the testing of the watermarking systems for SDMI screening technology would include testing in the sampling rates typically used by premium-quality audiophile recordings – generally ranges between 192 kHz and 176.4 kHz – to assure that it did not produce audible artifacts.
In late July, 2000, at a demonstration of the technology, record producers, according to a report in Audiofile.com, were “astonished” when they discovered that the watermark left audible artifacts in recorded music. After that event, record producers and audio engineers will be keeping an even closer watch on SDMI’s activities. Still, musicians need not be passive players in this industrial development. Blue Spike’s digital watermarking schemes as well as that of competitors are available today for inspection by discerning musicians.
Blue Spike's Giovanni® in test after test has proven inaudible in a broad range of music.
Blue Spike invites and encourages A/B tests of our own technologies and comparisons with any other watermarking technology in the field. The double blind test is as good as any. Take a recorded music sample. Mark it with Giovanni and/or any competitor’s product. Have a friend play one and then the other without telling you which is which. Then record your impressions. Look at the play sequence and see if you were able to detect any differences between the original recording and the watermarked copy. After all, it really is noise only if you can hear it.