Re: SDMI - The way I would make the water mark.

[Ben Galehouse <bgalehou () PACBELL NET>]

Steve Mosher wrote:

>         Given two copies of the same audio, but with different watermarks,
> you can still (at least) corrupt the watermark by averaging or what have
> you. I'm not as good with DSP as I'd like to be, but I'm sure you'd also
> have to be wary of harmonics and other fun stuff this way. It might also
> bugger up hi-fi sound, binaural sound and sound sourcing effect,
> depending on exactly how it's implemented.


Suppose that the watermark invloved changing every 30th second of the
music.  (i.e., 1 second of every 30 was marked).  Suppose that the
initial offset was random. Then averaging would just merge the marks.
The mark becoems weaker, eventually becoming noise, but if your number
of copies were limited (i.e. you had to pay for each), it might not be
easy to defeat that way.

Alright, that particular system would be pretty easy to defeat with
three differently marked filed and a good binary diff program, but there
are probably better systems with an analogous property.