[4110] in cryptography@c2.net mail archive
Re: Liquid Audio & MP3
daemon@ATHENA.MIT.EDU (decius@bleeding.edge.net)
Mon Feb 1 16:41:30 1999
From: decius@bleeding.edge.net
To: sameer@bpm.ai (Sameer Parekh)
Date: Mon, 1 Feb 1999 12:59:27 -0600 (CST)
Cc: gnu@toad.com, Markus.Kuhn@cl.cam.ac.uk, cryptography@c2.net
In-Reply-To: <199901291751.JAA27334@acid.bpm.ai> from "Sameer Parekh" at Jan 29, 99 09:51:36 am
> Is anyone on this list familiar with what is going on with
> Shoutcast/MP3/Liquid Audio? I heard that Liquid Audio was releasing
> their watermarking technology into the open MP3 standard. Anyone know
> about the technology, how strong it is, etc?
>
According to Liquid Audio their system works by watermarking any music you
buy with your name, address, and credit card number. If you redistribute
the music and the copyright owner finds it somewhere, they can check the
watermark to determine who distributed it first.
Because the watermark is going to be different for every copy of a
particular song this suggests that if you get three copies of a song with
different watermarks and do bit voting with them you can produce a fourth
file that contains all the information that is the same in the first three
(the song) but does not include any of the differences (the marks). One of
the papers on Ross Anderson's site talks about this (in the context of a
cable system). (Sorry, I don't have an exact reference, its been more than
a year since I read it.) His paper descibes an encoding system for a sort
of watermark such that any three copies of a TV broadcast are going to
have enough bits in the mark that are the same that if a copy is
redistributed the perp can be narrowed down to one of the three
responsible. Of course this requires enough space in the mark for a much
large code and some sort of upper bound on the number of copies
distributed, however it is possible that Liquid Audio is doing something
like this. (This is all educated theory. I haven't had time/motivation to
actually sit down and break Liquid Audio or any competing system. If they
are openning their standard, analyzing it might be less of a chore.)
If memory serves the mark Anderson describes is sort of steganographic in
nature. Most of the marks that I've seen have been FIR filters. This
complexity is needed to survive analog reproductions. As Liquid Audio
claims that their mark will survive analog reproduction its likely to be a
filter. Can you build a filter that removes the mark? Probably. Most of
the watermarks I've seen are designed to handle printing/scanning/
photocopying or in the case of music, playback/record/broadcast. They are
not designed to withstand an attack from an adversary armed with a
knowledge of the mark and a copy of MATLAB. In playing around with the
watermarks included with Photoshop a friend of mine was able to destroy
the marks simply by using the standard image filters without doing too
much damage to the image. This being the case certainly such marks could
be removed easily with no damage to the image by designing a custom
filter. Some of these marks exist in a totally different frequency range
than the image/song and can be cut out with a band pass filter very
easily. I have yet to see any papers on attacking marks with filters or on
defending marks from such attacks. I'd be interested in any references
anyone might have for that.
There is a LOT more to say here about the big picture but its probably off
topic for this list so I'm going to spare you the philosophical ranting.
The short term is that I think advertising is going to be a more effective
revenue stream for music publishers online than technical copyright
protection systems.
Tom Cross
