Cinavia Copyright Protection

Holy shit niggers:

https://en.wikipedia.org/wiki/Cinavia

The watermarking and steganography facility provided by Cinavia is designed to stay within the audio signal and to survive all common forms of audio transfer, including lossy data compression using discrete cosine transform, MP3, DTS, or Ogg Vorbis. It is designed to survive digital and analogue sound recording and reproduction via microphones, direct audio connections and broadcasting, and does so by using audio frequencies within the hearing range. It is monaural and not a multichannel codec.

o Only a single channel of audio is required to detect the watermark
o The watermark is able to survive re-recording through a microphone
o The watermark can be detected through "the production, duplication, distribution, broadcast, and consumer handling of recorded content"
o Different copies of otherwise identical works can be distinguished

Let's brainstorm ways to extract the waveform pattern from this!

Sorry, >>1, I got too excited reading the specs, and forgot to answer your request.

It works like any hash cipher, but makes sure that it maintains signature harmonics, so that if the sound is reverbed/modified, because the harmonics are unique values throughout sample, the signature is maintained, even though it has been tampered. You would use matrix mathematics to determined that the sample has the equivalent signature as when the original sample was produced.

It combines block cyphers and vector calculus under matrices, which is awesome.

So about your request, that would be simple. Make a sample without the cipher, copy the sample, apply the cypher on the copy, then compare the two. You can use inverse or subtractively translation of your original sample, to null out the copy, and leaving the cipher-mark to play and re-record.

tl;dr It's possible, just O(4n) of work. And requires you have the original sample to derive its sound. The Advanced Encryption Standard is one prime example of what Cinavia does at the block level.

>>5
Ah, then that is stupid, especially if the encoding you are using specifically cuts out anything above the human auditory spectrum (e.g. MP3, Vorbis, etc). It should be using random harmonics under functions like vibratos, treble, mixed sonatas, etc. while using the block cipher, all at low amplitudes(volume). To make the sample/cipher more unique, it should allow for internal input variables or functions to add to the complexity (time, some key (e.g. PGP), GUID, salts, etc.).

But if you are saying it is audible, then they might need to update their algorithm, to make as inconspicuous as dither-marks (in image signing/watermarking). To the trained it would appear as re-recorded or lossy re-encoding noise.