|"Hack-Proof" DVD-Audio watermark cracked at Princeton – but under legal duress|
|Home Theater News Music - Technology News|
|Written by Richard Elen|
|Friday, 27 April 2001|
Professor Edward Felten of Princeton’s Computer Science Department must be breathing a sigh of relief round about now.
He and his team have been under considerable music industry pressure to refrain from revealing the results of their research last year in which they appear to have cracked the majority of the industry’s SDMI (Secure Digital Music Initiative) audio copy-protection and watermarking schemes, including, apparently, the Verance scheme that is already being put into use for DVD-Audio and SDMI Phase I products. Felten and his team have finally given in and withdrawn their paper.
The Princeton team, attending the Fourth International Information Hiding Workshop in Pittsburgh this week, had been set to present the contents of their paper in which they reveal how the codes were cracked and how they work. Now they’ve bowed to industry demands to "withdraw the paper… assure that it is removed from the Workshop distribution materials and destroyed, and avoid a public discussion", as Matthew Oppenheim, Secretary of the SDMI Foundation, put it in a letter to Professor Felten. How do I know the contents of that letter? Well, as you might expect, the letter, and the paper itself, have already been widely published on the Internet.
Last year the SDMI announced a "public challenge", inviting members of the public to try to break a number of SDMI’s data-encoding technologies, almost certainly those being considered for SDMI’s Phase II, designed to watermark compressed audio clips to enable tracing of illegal copies, or preventing such copies being made. Of course, the first two rules of any Intelligence service are: "There is no code that can’t be cracked," and "To any measure, there’s a counter-measure," so it is perhaps no surprise that Felten’s team, among others, claims to have succeeded.
The SDMI Challenge was a game to be played by very strict rules. Rather less information was given to the "public" than would be available to a pirate. Most of the challenges involved sets of three files, one unwatermarked, the same with watermark added, and a third file containing another watermarked piece. The challenge in these cases was to produce a version of File 3 without the watermark. Other challenges were designed to test technologies designed to keep a track from being separated from its original album source.
To test their results, researchers had to email their file to an SDMI "oracle" that would tell them, by means untold, whether their file passed or failed.
Felten’s team claims to have "reverse-engineered and defeated all four of their [SDMI’s] watermarking technologies." They didn’t go for the two remaining challenges as the SDMI oracle appeared to be broken, and not enough information was supplied. They also did not go the whole way in the challenge as to do so would have required them to keep the research confidential, a condition that they, at the time, did not want to be bound by – although ultimately, officially at least, they did not go public.
Do Felten and his team believe they can defeat any audio protection scheme? "Certainly," they say, "the details of any scheme will become known publicly through reverse-engineering." Using their techniques, they believe, "no public watermark-based scheme intended to thwart copying will succeed." They conclude, "…if it is possible for a consumer to hear or see protected content, then it will be technically possible for the consumer to copy that content." If not the consumer, then certainly a well-organized illegal copying industry.
We have no intention of publishing extracts of the paper itself, of course, or even telling you where it is. But it’s easy to find. I was sent two links to what I take to be copies of the paper on different servers, but when I accidentally deleted the emails, it took merely a short search on Yahoo to find one of them again. Whether or not Felten presents his paper, its contents are already known, along with Oppenheim’s letter.
Hopefully now with the supposedly hack-proof watermarking code broken, the powers that be will give DVD-Audio a chance to become a more mainstream format by removing the restriction that forces end users to connect their DVD-Audio players via 5.1 analog inputs as opposed to digitally as you would connect a CD player, a DVD-Video player or any other modern digital audio-video source thus removing the cost barrier of having to buy not just a new DVD-Audio player and software but also making DVD-Audio compatible with millions of existing AV systems with digital inputs and 5.1 processors.
Source: undisclosed internet news agencies.