Microsoft was awarded U.S. Patent 7,266,697, entitled "Stealthy audio watermarking," on Sept. 4, for the duo's work.
As the patent's abstract explains it: "The watermark identifies the content producer, providing a signature that is embedded in the audio signal and cannot be removed. The watermark is designed to survive all typical kinds of processing and malicious attacks."
The stuff is probably the most thorough and complicated technology ever to be applied to 99-cent music files. It's robust enough to resist all attempts to remove the watermark from the clip, including changes in time and frequency scales, pitch shifting, and cut/paste editing.
I should note that it's important to make a distinction between what Microsoft is doing here--watermarking--and what's commonly thought of as DRM, which is copy protection or encryption. Watermarking isn't encryption and it doesn't necessarily prevent unauthorized playback. On the other hand, it can serve in place of any other type of DRM, if the playback system (i.e., the MP3 player or online music store) requires the presence of the watermark before it'll let you listen to your file. (I'm putting this paragraph here in anticipation of all the "Wolfe, you don't know Jack about DRM" comments I'll probably get anyway.)
The watermarking scheme evolved by the Microsoft scientists is so robust that, if it's used properly, it can indeed serve as an uncrackable DRM scheme. Keep in mind that Kirovshi and Malvar aren't just proposing a single watermarking method. Their patent outlines three (count 'em) different ways to apply spread-spectrum to the task of locking-down audio files.
Check out this description of three-fold DRM they're doing, from the patent:
"[The] watermarking system employs chess spread-spectrum sequences to improve the balance of positive and negative chips in the watermarking sequences. . . In another implementation, a watermarking system employs an energy-level trigger to determine whether to skip encoding of a portion of a watermark within a given time span of an audio clip. . . In a [third] implementation, a watermarking system begins encoding of a watermark at a variable position after the beginning of an audio clip."
Here are a few flow charts, from the patent itself, which give a more definitive idea of how the watermarking technique works in practice. (Click on the first one to get to the image gallery, which will let you view the other five.)
Microsoft's watermarking patent diagram. (Click to see the full flowchart, and to view the rest of the Microsoft patent pictures.)
Another advantage of Microsoft's watermarking method, as opposed to conventional encryption, is that it plugs the so-called analog hole, under which hackers with a lot of time on their hands can play a digital file, record the analog output via onto a cassette, and then re-digitize it. (That's a lot of work for a 99-cent song. I like Chrissie Hynde's pre-Internet-era take on this question, where she said, and I'm paraphrasing: "What do I care if some 14-year-old girl is taping my song off the radio?") Re-tapers can still get busted because the watermark remains intact, even after analog conversion.
A month ago, I said that the battle between the big bad content producers (record companies) who keep banging their heads against the DRM wall, and the freedom- (and theft-) loving consumers who want completely unprotected music files, had been won by the latter. (See: "DRM Scorecard: Hackers Batting 1000, Industry Zero".)
Now it's time to revise that tally, and score one for Microsoft.
P.S. A coda to my discussion above is that, in addition to having come up with a stronger way to do DRM, with this patent Microsoft may have secured some licensing claims on other companies doing watermarking.
Audio representation, showing positioning for watermarking. (Click to see the Microsoft patent diagrams.)