Slashdot Mirror

Cambridge Consultants demonstrated something similar a few years ago. It's called Sprint and there's a great big picture of it here.

Cambridge Consultants demonstrated something similar a few years ago. It's called Sprint and there's a great big picture of it here.

Cambridge Consultants demonstrated something similar a few years ago. It's called Sprint and there's a great big picture of it here.

This cell phone tracking actually gives civil libertarians an argument as to why these raids cannot possibly be justified in most cases because the police can figure out where the person is going, and ambush them when they have the advantage (something they don't have when assaulting a home).

Or alternatively, they can purchase a high-tech hand-held radar system to see through walls

Why couldn't they have simply waited for him to leave his house when he was going out to work?

:) I think it works... there are many companies which have products for this purpose. Do a google search for "digital audio watermark" to find examples (including technical papers on citeseer). What is termed "robust" watermarks (as opposed to "fragile") can survive the usual audio processes (D/A, A/D, dsp, lossy compression, etc.).

e.g.:
a press release
verance
Overview of Felten's Attack on SDMI
Template Matching

IANAExpert. In some cases, you may be correct: if the watermark detection can be reverse engineered, it *may* be possible to remove it... the only question is how much damage must be done to the audio to obscure the watermark, and I think this is where the research is being emphasized.

Also, I don't think it is as simple as your secret key analogy: detection of the watermark does not necessarily mean you can properly remove it. As I understand it, there are many possible sources within the audio which aid in re-constructing the watermark and various tolerances for their recombination (all depending upon the particular algorithm), so there may be a google of ways to remove the known watermark, but only a few of the ways actually maintain reasonable audio quality (as compared to the original).

e.g. in some aspects, the original audio might already contribute to the watermark's existence, and so no modification of that aspect must occur when embedding the watermark. Hence, when attempting to remove the watermark, it is difficult to determine whether that aspect was originally part of the song or whether it was added by the watermarking technique.

On a side note, the other interesting possibility is the existence of multiple watermarks, wherein a second watermark (which is not detected or required for playback) is an ID which is unique to each copy of the song released, potentially allowing the distributor to track a "pirated" song back to the "pirate" (e.g. if it was legally purchased they might track records of its sale, and if it wasn't legally purchased, they could at least track it back to the last known legal distributor).

A couple of simple and brief pages I found using google "digital audio watermark":
a press release
Overview of Felten's Attack on SDMI