Re: [dvd-discuss] A TPM without use limitations -- thoughts?

[daw(at)mozart.cs.berkeley.edu (David Wagner)]

John Schulien  wrote:
>Having read the original article that started this thread, it appears to me
>that what they are talking about is not adding any sort of identifying
>information at all, but devising an algorithm to make an acoustic
>"fingerprint" of existing, unmodified songs.  In other words, this
>"fingerprint" isn't something that is added to a song -- it is a digital
>summary of the song -- used for recognizing a song.  

Well, ok, but we can still imagine stripping the fingerprint from a MP3
file even if no fingerprint was explicitly added to the original file.

In general, I'm skeptical of any system that claims to be able to
reliably identify a MP3 track even in the presence of a sophisticated
adversary that is trying to fool you.  This is a very hard problem.
There are many human-imperceptible transformations an adversary could
apply to the MP3 track to try to fool the system.