I just read up on Cinavia, DRM system. The description of it left two questions which I'd like to forward to some VH tech pros.
1. What would that "watermark" be that can withstand re-encoding, re-recording etc.?
2. How can that resistence / persistence be achieved?
+ Reply to Thread
Results 1 to 11 of 11
Thread: Cinavia technicals
Das Leben ist eine Nebelwand voller Rasierklingen. (C. Bukowski)
@pandy... I read your suggestion and came across http://forum.doom9.org/showthread.php?p=1419651#post1419651. This brings up another question: Is that "embedded signal" always the same, or are there variations? Besides, which un-altered audio formats are affected by Cinavia? DTS only, AC3 also? And, referring to SamuriHL from the Doom9 forums, if
a - the signal is always the same,
b - the signal is embedded in AC3 streams on BDs,
c - the signal is not embedded in AC3 streams on DVDs,
couldn't it be possible to work out a workaround thru bitwise comparison or something like that?Das Leben ist eine Nebelwand voller Rasierklingen. (C. Bukowski)
a - the signal doesn't have to be the same. There may be more than one watermark.
b - the signal fingerprint could theoretically be coded in any track if it's that resilient. It depends on licence costs and onboard software.
c - Not yet. There is no point. The chipset and licence is likely more than the cost to manufacture players.
Manufacturers would refuse to implement
Not unless the key can be deciphered. Who's to say how many watermarks are embedded in a track. Your assumption is it's just one.
So i must admit that i was never curious to study Cineavia enough to understand all aspects of this protection scheme (no time to watch movies thus no interest).
However Cineavia works on frequency domain (phase-frequence) thus bit2bit corellation will not provide usefull info (bit2bit for decoded PCM audio).
IMHO to fight with Cineavia first You need to be able detect Cineavia thus You need exactlyu same implementations as Cineavia protection in players, second You need to compensate phase-frequency changes in oposite to Cineavia modulation.
In thery at least implement similar to Cineavia modulation but with random(pseudo) sequence to spread Cineavia data - in other words to jam Cineavia (which itself is based on some detection and need proper SNR to be able detected) - there is lot papers that describing Cineavia scheme robustness.
AFAIK Cineavia can be used to stroe almost any type of data - with relatively low bit rate but anyway - currently Cineavia is used only to store DRM but it can be used to almost anything.
From human ear phase-frequency changes are small and random like thus subjectively Cineavia should be transparent.
To compare protected and unprotected audio You need translate both audios to frequency domain (DFT/FFT, DCT) then compare spectrum with focusing on particular areas (over 8kHz) - search for phase/frequency fluctuations - this is classical spread spectrum transmission.
To be honest i never have opportunity to see Cineavia in action - but it can be curious to see where Cineavia is stored - first - lowpass audio signal to establish in which part of the audio spectrum protection is stored, to recreate at least partially audio spectrum similar to SBR techniques can be used.
IMHO, the only way it will be deciphered is if the ROM code is reverse engineered and someone finally understands the method.
patents for Cineavia and theoretical background seems to be available - IMHO ROM REI is not necessary.
I think that commercial Cineavia detector use only DCT (AC3) coefficients to detect thus whole detection not involve additional signal processing thus is completely transparent - ideally same AC3 with and without Cineavia + bit2bit comparison to detect correlation and based on this recreate Cineavia. IMHO on player side Cienavia is more like logical bit operation than signal analysis thus implementation of Cineavia should be not to complex from required processing power point of view.
Well...if anyone's looking for a list of titles containing Cinavia protection, you can look
here: http://en.wikipedia.org/wiki/Cinavia#List_of_Known_Releases_with_Cinavia_Watermarking or
here: http://blog.dvdfab.com/cinavia-protection.htmlDas Leben ist eine Nebelwand voller Rasierklingen. (C. Bukowski)
Has anyone noticed if you load an audio file into audacity that has Cinavia and expand the envelope view all the way. there is a tiny "blip" at a 21khz interval. none of the filter sin the program will remove them. I thought of building an analogue lo pass filter and re-recording it through it.-Any thoughts?
Perhaps techniques like spectrum band replication can be used to restore a bit more HiFi sound (similar to used in HE AAC) or perhaps everything about 8kHz can have jamming (parasite) phase modulation that can override cinavia phase changes.
Last edited by pandy; 14th Jan 2013 at 04:38.