July 28, 2004

More Selling Out

Sigh... I thought of a CD-audio copy protection scheme today. This one has at least one advantage to users over the standard "disc-breaking" technique: it allows you to listen to CDs on the PC. There's another flaw in this method: I don't believe it can protect against more than one type of compression at a time. It also may be cripplingly difficult to implement. Anyway, it seems feasible, so I'll say it anyway.

The idea was inspired by the wonderful CodeProject, specifically the latest in Corinna John's series of .NET steganography articles and an interesting article about "selection watermarks" (amusingly, the latter only works if the user is using MSIE).

My thought is this: with close enough study of MP3/etc. compression, it might be possible to "encode" an audio file so that some of the audio contained is removed. Compression tries to remove only inaudible parts of an audio file, so how can this help us?

[I do not guarantee this paragraph - I need to find my sources. Take it with a salt shaker.] Auditory science has shown that in the human ear, through a process we don't quite understand, lower-frequency sounds can be synthesized from high-frequency waveforms. I don't believe it has anything to do with beat frequency between two waves, FWIW. For decades, it was thought that it only worked the other way around, IIRC.

It struck me that if suitable "inaudible noise" were injected into the audio and if MP3 encoders (say, at 160kBps or less) could be made to strip it all out, it might be possible to get some kind of annoying distortion going -- a sawtooth or something. The biggest trick, of course, would be making it truly inaudible when played at full quality. You see why it's a single-scheme solution -- if MP3s are prevented from encoding it, something like Windows Media might work.


Post a Comment

<< Home