PartyTime, I'd be glad to offer a viable alternative to what EDIG offers with regard to watermarking. But you'll have to help me first by pointing me to some material on what they offer with regard to watermarking.
Without knowing what they are doing, I will offer what may be an alternative. (But, since I DON'T know what they offer, it could just be the same thing.)
Now, mind you, I've not studied this, and am only vaguely aware of how watermarking of images is done. (Which suggest another approach - study how watermarking of images is done and apply a similar technique.)
If somebody asked me how to watermark music, I would suggest creating a unique code for each file that is to be watermarked, and signing it using standard public-key techniques. (Thus preventing forgery). This would be a small number of bits (perhaps 128 or 256), which I would embed in the audio (probably repeatedly) through a form of "audio stegnography". Of course, multiple distributed copies of the same track might be watermarked with different codes depending on distribution source, or even with a unique code for each consumer it is distributed to.
I'd suggest adding inaudible soundwaves to the track, in a detectable pattern, which would act as a kind of "carrier" for the watermark. These soundwaves would be inaudible to the human ear because they are "masked" by louder sounds at nearby frequencies. (Yes, similar to the way that MP3 and some other lossy compression schemes work.) Some variation of this detectable pattern (for example, variations in the length of tones).
The challenge would be to come up with inaudible sound waves which are NOT expected to be removed by MP3 or other lossy compression schemes currently existing or expected to be developed.
A nice advantage of this approach is that, assuming the lossy compression problem can be licked, the watermarks could survive transcoding and even analog recording. (Indeed, surviving analog recording would probably be less challenging than surviving transcoding using lossy codecs.)
A disadvantage is that it would probably not be possible to avoid hackers being able to write software that could detect the "carrier" (after all, it has to be detectable) and then remove it (and thus, the watermark) through digital signal processing.
Give me a little time, and I could probably dream up a dozen other different schemes, but this is just the first one that came to mind.
Just to give you an idea, though, here's another one: instead of embedding inaudible (by the human ear) tones or sounds, you could introduce inaudible (by the human ear) but detectable (by electronic circuits or digital filters) variations in pitch, which would encode the bits. I'm not sure how practical this would be, though, because the human ear is actually quite sensitive to variations in pitch, but I do think it might be a possibility.
I'm sure, though, that the industry has already though of these and a few more, and doubt that EDIG has anything that the industry needs that badly. |
