SDMI Watermarking Effort Rankles Engineers
Last year, SDMI conducted a series of blind listening tests on audio professionals in New York, Los Angeles, and Nashville to determine the audibility of the final contenders among several competing watermarking systems—signals imposed on the recordings in order to prevent their being copied, or to trace their history if they are. One of the requirements for watermarking audio recordings was that any identifier be robust enough to survive several generations of copying—including making cassette copies from AM radio broadcasts. Another was that it be innocuous enough to sound "transparent" to ordinary listeners. The outcome of the SDMI tests was the recommendation of a technology developed by San Diego–based Verance Corporation.
Verance's watermark was determined to be the least detectable—although trained listeners can identify it with near-100% accuracy. SDMI's belief was that if a watermark were basically undetectable to a select group of "golden ears," it would also be so to the general public. Despite the participation of some of its members in the listening tests, the Audio Engineering Society overall wasn't involved in the adoption of Verance watermarking, to the dismay of many. They are also concerned that an insufficient sample of listeners took part in the tests, and that not enough of them were ordinary music lovers with no professional ties to the industry.
A larger test sampling should include "a wide spread of musicians, producers, engineers, educated listeners, and uneducated listeners," according to classical producer and recording engineer Tony Faulkner, of London's Green Room Productions. "It seems that in-depth independent scrutinization has not been completed, and that the decision to adopt analog watermarking has been made in some haste, largely on the basis of the commercial interests of a small number of individual parties," Faulkner stated in a recent exchange of e-mails on the subject. The "individual parties" he refers to are the "Big Five" conglomerates of the music industry.
The SDMI listening tests were conducted in "an atmosphere of secrecy," Faulkner claims. Not consulted were smaller independent classical labels such as Telarc, Chandos, Hyperion, and Naxos—the last of which owns almost 20% of the classical market share in the UK, according to Faulkner. "None of these labels received any approach about analog watermarking of high-quality carriers (i.e., CD and better); they were as surprised as the rest of us when the news leaked out."
Testing was done at standard CD levels of resolution, Faulkner points out, not at higher bit rates or sampling frequencies such as 192kHz and 176.4kHz. Typically, if a watermark creates noise, the noise will get louder as the sampling rate rises, he says. Imposing noise on high-resolution recordings could defeat their whole purpose and scuttle their appeal for music lovers. Faulkner will chair a meeting on the watermarking issue at the 109th AES convention in Los Angeles, September 22–25. Present will be many of the big guns of the music industry, from such organizations as Sony, Warner Bros., the Recording Industry Association of America, and the International Federation of the Phonographic Industry.
The discussions in L.A. may be rancorous, as engineers will have all summer to discuss watermarking and associated issues on at least two online forums. As we approached deadline, we received a brief note from Faulkner alluding to "recent progress." Look here for updates as details emerge.
In a related development, the National Association of Recording Merchandisers has issued a position paper written by president Pamela Horovitz warning the SDMI to consider the rights and needs of consumers while developing digital commerce standards. The paper, "NARM's Baseline Principles for Online Commerce in Music," makes the accusation that some of SDMI's suggested technologies are misuses of copyright laws.