"Cinavia"/Verance BluRay audio watermark protection

[Ghitulescu]

Im not sure if the player have H/W to detect Cinavia, they just have to have powerfull enough Processor to do it on the fly without running out of steam to do playback.
Cinavia is a detection, not a processing of audio. It can be done with extremely low resources.

[rotty]

Cinavia is a detection, not a processing of audio. It can be done with extremely low resources.

It has to decode the audio to analogue and then look for the watermark short term echos, and then match these patterns with its algorithm, (iffy patterns at that), this is processor intensive.

For example, the Panasonic BD35 could never impliment this detection, CPU no where near good enough to do that and its main task.

[omegaman7]

It has to decode the audio to analogue and then look for the watermark short term echos, this is processor intensive.

For example, the Panasonic BD35 could never impliment this detection, CPU no where near good enough to do that and its main task.

Interesting. So it sounds like you know a lot about this watermark. Care to crack it? :p

[rotty]

Unfortunately Sony keeps re-directing me when I try to use their U.S. or European Sites.

Hi Hello_hello

I am in europe (UK). If there are any sites/downloads you want me to try to get for you let me know.

I know that for DVD we are region 2 and you are region 4.

I Think for Blu Ray you are the same region as we are (region B). There may of course be other blocks within your player s/w that would stop a European firmware from being accepted.

[rotty]

Interesting. So it sounds like you know a lot about this watermark. Care to crack it? :p

Hi omegaman

LOL I am very flattered that you would say that, I would love to say I could, would be very rich if I could eh LOL.

Just to put my penny worth in though, I think that removal or dissruption of the watermark would be more destructive to the audio than would be acceptable.

After all, the trick is to remove this whilst keeing the patient alive, the patient of course in this case being the audio.

I can remove Cinavia easily, remove the audio stream, the patient is of course then dead.

Avoiding is probably the best option, a stand alone player or PC s/w player than doesent care if the watermark is present.

[omegaman7]

I have a basic understanding of the watermark. I've been putting it off, and putting it off, but I'd really like to take a CLOSE look at a segment of the audio(perhaps 20 min or so). Unfortunately, time is something I can't seem to set aside for this. Perhaps this weekend is finally the time. But without knowing what I'm looking for, It'd be nearly a shot in the dark. There is no doubt a way to suppress the watermark/frequency-ies, without noticing. After all, they put it there, and "we can't hear it".

I won't hold my breath of course. I do have a way of noticing patterns though. Although when they're highly complex...

[rotty]

I have a basic understanding of the watermark. I've been putting it off, and putting it off, but I'd really like to take a CLOSE look at a segment of the audio(perhaps 20 min or so). Unfortunately, time is something I can't seem to set aside for this. Perhaps this weekend is finally the time. But without knowing what I'm looking for, It'd be nearly a shot in the dark. There is no doubt a way to suppress the watermark/frequency-ies, without noticing. After all, they put it there, and "we can't hear it".

I won't hold my breath of course. I do have a way of noticing patterns though. Although when they're highly complex...


Well it probably isnt that we cant hear it, its that we dont notice it. Short term echos are not that disturbing to us, we only seem to notice once they delay steps over a certain time, and its very sudden strangley enough.

The Audio watermark for DVD/BD is probably different to the Watermark in Cinema's.

The Audio watermark in Cinema's will be well within the audible spectrum and quiet blatent. After all, we dont care too much if the audio in Cinema isnt perfect, people eating / talking / sweet wrappers etc in background and we are not going to replay and replay THAT audio at home, once only at the Cinema so we dont care too much.

We know that the watermark in Cinema has to be at the analogue level and in the middle to lower part of the audio spectrum so as to be picked up and survive microphone recordings from people making illegal recordings (CAM). Also, it needs to be in this part of the spectrum to survive the accostics of the building.

BUT, is this the same for DVD/BD, just out of interest, has anybody made a bad microphone recording of a BD with the watermark. I mean really done this.

If it survives that, then we know it is well within the middle/lower end of the spectrum and that it is at the analogue level. If it does not survive this then we know that is NOT the same as Cinema watermark.

This really has to be verified to get a starting point.

All this above or similar will I would have though have been done by the proffesionals, but we need to do this to gain the knowledge of the beast for ourselves.

Now I dont want to be a kill joy, but lets say we know exactly what to look for, and we know what to remove, will the patient (the audio) survive.

[omegaman7]

BUT, is this the same for DVD/BD, just out of interest, has anybody done a bad microphone recording of a BD with the watermark. I mean really done this.

That's it! I gotta play with this now! :) You give me an interesting idea.

[ChiDragon]

I guess it depends on what you consider a bad microphone recording? Someone mentioned testing it in post #11.

[rotty]

I guess it depends on what you consider a bad microphone recording? Someone mentioned testing it in post #11.

Hi ChiDragon

Bad Mic recording I mean cutting top end, just to see where we are, i.e. try several recordings.

Ending up with very very muffeled recording.

Obviously, cutting this top end is NOT the way to beat it, but at least we know where we are.

Record the audio crudely, i.e. in a room with a Mic, at least 10 feet away (with gain enough to be at normal level), so you are recoding with a Mic with the sound comming from a speaker(s), yes that crudely.

As I say, this is NOT a way to beat Cinavia, but it will let us know what we are looking at.

If Cinavia CAN be recorded with a GOOD Mic recording then Cinavia is NOT at the Digital level, it IS at the same level as the Cinema watermark, i.e. Analogue.

It could of course be at 2 levels, BUT if it fails the BEST Mic recording it probably is at the Digital level only (which means that what we have been led to beleive is wrong and that for DVD/BD NO analogue system for the Watermark on physical media exists (or at least at present)).

[rotty]

That's it! I gotta play with this now! :) You give me an interesting idea.

Hi omegaman,

Sounds interesting fella, would love to know what results you get.

[SamuriHL]

Mic recordings have been well tested. It survived all of them. Someone used a camcorder pointed at a wall while they played a cinavia movie in the background. The recording tripped the cinavia detector upon playback. It has been well understood that the signal is embedded at the analog level since the beginning.

Sent from my Xoom using Tapatalk HD

[rotty]

Mic recordings have been well tested. It survived all of them. Someone used a camcorder pointed at a wall while they played a cinavia movie in the background. The recording tripped the cinavia detector upon playback. It has been well understood that the signal is embedded at the analog level since the beginning.

Sent from my Xoom using Tapatalk HD

Hi Samuri

Yes had always thought it was, but, it occurred to me is this the ONLY level a marking has been placed in the audio stream.

Anyway, I agree with what you said some time ago, re avoidance.

[omegaman7]

I imagine that not all mic's will pick up the signal though. Especially, if they've been tweaked to filter out certain frequencies, along the audio spectrum. Which wouldn't require the use of a mic to accomplish ;) With use of appropriate software.

[rotty]

I imagine that not all mic's will pick up the signal though. Especially, if they've been tweaked to filter out certain frequencies, along the audio spectrum. Which wouldn't require the use of a mic to accomplish ;) With use of appropriate software.

Well yes your right, a microphone will not have a flat response across the audio spectrum.

But that wont be of any consiquence for this experiment.

[apgood]

I imagine that not all mic's will pick up the signal though. Especially, if they've been tweaked to filter out certain frequencies, along the audio spectrum. Which wouldn't require the use of a mic to accomplish ;) With use of appropriate software.

It might be interesting to see what happens if say the frequencies above say 15kHz are filtered out and/or low level subsonic frequencies below as 20Hz.... Assuming of course it isn't interleaved through the whole spectrum and doesn't need to be an ”intact” signal....

Sent from my Nexus 4 using Tapatalk 2

[IanB]

We already know the manifestation of a Cinavia data bit is a very short period echo. An effect chosen to be virtually inaudible.

I suspect these 'echoes' make a very distinctive pattern when you look at a Fourier transform of the data, i.e. in the frequency domain.

This style of DFT technology is very immune to analogue and digital domain filtering and distortion, which is precisely a characteristic the black hats in this game would strive for.

I am sure the design of the 'echoes' is also such that trying to simplistically remove them from a source would spoil it significantly, i.e. many false matches.

If I was designing this I would use some devious clock recovery scheme over these bits to extract a data stream of 1's and 0's, it needs to survive PAL speed up, it also needs to cope with long period of silence in the host audio track.

I would apply a very substantial forward error detection and recovery algorithm like a hamming code. In the end we only need 2 or 3 bits of data to trigger the detection flags and packing 1000's of 'echo' bits per minute is trivial.

[ChiDragon]

So with literal silence as the input would your algorithm just wait it out until you have a signal to alter?

[ChiDragon]

I did a "bad camcorder" recording with my camcorder pointed out the window and The Amazing Spider-Man BD playing approx 11 ft away on terrible all-in-one PC speakers, with the volume just loud enough to listen comfortably at my chair in front of it. Plenty of street noise in the recording and the dialogue in the movie is muffled to the point of being unlistenable.

Cinavia remains intact.

[omegaman7]

I did a "bad camcorder" recording with my camcorder pointed out the window and The Amazing Spider-Man BD playing approx 11 ft away on terrible all-in-one PC speakers, with the volume just loud enough to listen comfortably at my chair in front of it. Plenty of street noise in the recording and the dialogue in the movie is muffled to the point of being unlistenable.

Cinavia remains intact.

Thanks for this post. It helps a great deal!

[hello_hello]

Hi Hello_hello

I am in europe (UK). If there are any sites/downloads you want me to try to get for you let me know.

I know that for DVD we are region 2 and you are region 4.

I Think for Blu Ray you are the same region as we are (region B). There may of course be other blocks within your player s/w that would stop a European firmware from being accepted.

Thanks, but I was really just trying to visit Sony U.S. or Europe for a look to see how much of the current firmware is advertised as containing Cinavia.... given I only came across firmware for one player which is advertised as containing it in Australia (the cheapest model). Using Google's cache I could get to the U.S. site.
Even though the model numbers are the same, it appears players are quite different from region to region or country to country, at least when it comes to firmware, as they have different features (such as the ability to connect to different online services etc).

[rotty]

I did a "bad camcorder" recording with my camcorder pointed out the window and The Amazing Spider-Man BD playing approx 11 ft away on terrible all-in-one PC speakers, with the volume just loud enough to listen comfortably at my chair in front of it. Plenty of street noise in the recording and the dialogue in the movie is muffled to the point of being unlistenable.

Cinavia remains intact.

Hi ChiDragon

A great test, interesting that the sound was comming from the poor PC speakers (mid to hi mid range).

Is it possible for you to do a fairly good Mic recoding as WAV or MP3 or if not convert it to any sound file that itunes will accept. Then play the file in itunes and use the graphic equaliser in itunes to experiment with the file.
Obviously it will mean you then re recoding from itunes with camcorder but would be interesting.

Start with only low end outputting.

[ChiDragon]

I don't understand the purpose of doing a new "good" mic recording if the goal is to see at what point Cinavia stops being detected.

[rotty]

I don't understand the purpose of doing a new "good" mic recording if the goal is to see at what point Cinavia stops being detected.

You need a good recording to start with so that when you filter certain frequency ranges using the itune graphic equaliser you can see at what point it affects Cinavia detection.

The bad recording you made still has most of the lower mid range + mid range + upper mid range.

Your test was good because you now know that DVD/BD version of the watermark its quite robust and also is analogue for sure.

With a new good recoding being played back in itunes, you will be able to attenuate/lift certain ranges of frequencies using the itunes graphic equaliser.

You have to start with a good recording and then make it bad (as far as playback is concerned using the graphic).

If you havnt seen the itunes graphic then take a look and you will see that the controls (10 of them + overall gain) either lift or attenuate frequencies from +12db to -12db giving you a 24db max difference between frequency ranges. These ranges are:

32HZ 64HZ 125HZ 250HZ 500HZ 1KHZ 2KHZ 4KHZ 8KHZ 16KHZ

Remember that for every 3db increase you double the power out (obviously, for every 3db decrease you halve the power out)

[ChiDragon]

You need a good recording to start with so that when you filter certain frequency ranges using the itune graphic equaliser you can see at what point it affects Cinavia detection.

Then start with the lossless file from the disc...

I have no interest in playing around with an equaliser. Not my thing.

EDIT: Did a 2nd gen recording using the same technique but with the first recording as the audio being played back on the PC. Cinavia is not detected. Dialog is recognizable as human speech but no words are decipherable.

[rotty]

Then start with the lossless file from the disc...

I have no interest in playing around with an equaliser. Not my thing.

EDIT: Did a 2nd gen recording using the same technique but with the first recording as the audio being played back on the PC. Cinavia is not detected. Dialog is recognizable as human speech but no words are decipherable.

The only way to know if Cinavia modulates the media sound in one frequency domain is to use a filter, and the easiest way to do this is with an equaliser.

itunes is free and easy to use.

I understand if you dont want to mess with this but its simplest way to find out.

As to your last point, if they muffled the audio across the band to a level where Cinavia was not detected then this still tells us nothing.

[kypec]

So the conclusion of ChiDragon's test would be that as soon as sound quality has deteriorated enough for Cinavia watermark being undetectable so did the overall legibility of audio for human speech recognition :)

[rotty]

So the conclusion of ChiDragon's test would be that as soon as sound quality has deteriorated enough for Cinavia watermark being undetectable so did the overall legibility of audio for human speech recognition :)

Yes but the whole spectrum was degraded.

It may well be that only part of the spectrum need to be degraded. Thats why we need to use specific targeted filtering to find out. If that is the case, then finding the watermark will be easier.

[ChiDragon]

Do you just not have a Cinavia player? If you send me degraded files I will play them.

[rotty]

Do you just not have a Cinavia player? If you send me degraded files I will play them.

Unfortunately I dont have a Cinavia enabled player.

[diogen]

It has to decode the audio to analogue and then look for the watermark short term echos, and then match these patterns with its algorithm...Cinavia is detected during playback.
All the heavy lifting - decoding audio/video, mixing, etc. - is done as part of playback.
Audio decoding uses under 10% of the resources (if you do it specifically for Cinavia detection).

Considering hidef is playable today (almost) by Raspberry Pi, the Cinavia algorithm itself uses literally zero resources...

Diogen.

[rotty]

Cinavia is detected during playback.
All the heavy lifting - decoding audio/video, mixing, etc. - is done as part of playback.
Audio decoding uses under 10% of the resources (if you do it specifically for Cinavia detection).

Considering hidef is playable today (almost) by Raspberry Pi, the Cinavia algorithm itself uses literally zero resources...

Diogen.

Where did you get the 10% figure from.

Even if that figure was correct, it cannot be correct for any CPU.

If it took 10% resource of a very fast CPU, it would take much much higher resource from a much slower CPU to do that task in the same time.

LOL can you imagine how long it would take to decode one frame of 1920 x 1080 using a Z80 at 100% resource.
(just to make the point)

Anyway, this has nothing to do with the main interest here.

[diogen]

Where did you get the 10% figure from.IIRC, this was reported at AVS about the very first HD-DVD player from Toshiba.
It was a pure software (Linux) player, the first and the last one.

Attempts to measure this number on a PC show even smaller numbers for audio decoding (I agree, a proper test is hard).

And btw, this number is irrelevant. See below.
Even if that figure was correct, it cannot be correct for any CPU.What does CPU have to do with it? Today everything is done in hardware. Ever opened a HD player?
It has one chip and some auxiliary silicon. That's it. Read KJack's posts starting 2007...
If it took 10% resource of a very fast CPU, it would take much much higher resource from a much slower CPU to do that task in the same time.Again, what does this have to do with it?
LOL can you imagine how long it would take to decode one frame of 1920 x 1080 using a Z80 at 100% resource.
(just to make the point)What point? The irrelevant one?

Again, in bulleted form:
- a blu-ray player has to be able to decode both audio and video (at least the mandatory codecs).
If it can't do that, it can't be called a blu-ray player!! With me so far?
- whether you bitstream for decoding in your receiver or not, the decoding is done anyway (in Cinavia-able players)
- a comparator with a few KB of RAM to cache a few ms of the audio is doing the Cinavia checking.
The last can be done within the SOC by now.

Once again, where does this stuff about Cinavia being resource hungry come from?

Diogen.

EDIT: The "cleanest" way of beating Cinavia would be disabling the analog audio in the player alltogether and bitstream to a (dummy) receiver...:)

[rotty]

My God Diogen

You really do just like to argue about anything dont you.

Like I said, who cares about this point.

Obviously, avoiding the detection is the easiest and probably best like I have said as well.

I hope that this will always be possible.

[diogen]

You really do just like to argue about anything...Not really, but I'm developing some taste for it...:sly:

As mentioned before, Cinavia discussion on doom9 is a form of masochism.
For people that know anything and everything about audio/video, how to rip, transcode,
(re)encode, how to use every OS under the sun, etc. - typical doom9 forum member - it is
much easier to rip the bloody BD-thing and play it on anything but the few Cinavia-detecting BD players.

But this simple solution hasn't stopped this discussion for almost 100 pages.
So, yes, I want in on the fun of it...

And since we are in penny-pinching mode, bypassing the analog stage isn't avoiding detection, but ignoring it...

Diogen.

[rotty]

Not really, but I'm developing some taste for it...:sly:

As mentioned before, Cinavia discussion on doom9 is a form of masochism.
For people that know anything and everything about audio/video, how to rip, transcode,
(re)encode, how to use every OS under the sun, etc. - typical doom9 forum member - it is
much easier to rip the bloody BD-thing and play it on anything but the few Cinavia-detecting BD players.

But this simple solution hasn't stopped this discussion for almost 100 pages.
So, yes, I want in on the fun of it...

And since we are in penny-pinching mode, bypassing the analog stage isn't avoiding detection, but ignoring it...

Diogen.

It sure is masochism but speaking for myself, I don’t really take the attempts to do anything about it by any of us too seriously, it’s just a bit of fun I suppose.

Also, and I may be totally wrong, I think that if it could be removed, that the cure would be worse than the disease.

I suppose, that's its strength.

[omegaman7]

I haven't forgotten :S Believe me... fact of the matter is, I have a rather full plate at the moment. I guess the real problem, is I'm not affected by Cinavia. I have an insatiable hatred for $ony. So I don't employ their hardware. The PS3 gets nothing but glares from me. I'm not exactly sure what brought it on. Their business practices for one...

I have dreamed up some ideas for attempting to either discover the pattern, or inadvertently removing it. When I FINALLY attempt it, I will post here. This seems to be a worthy thread for posting findings :) I'm currently testing a cutting edge piece of hardware, for review. When that's done, I should have much more time for this.

[SeeMoreDigital]

Given that Cinavia is mostly confined to the 'English' audio stream. I wonder what would happen if you: -

* De-mux the English language surround sound stream and separate it into mono audio tracks
* Save the English 'centre' audio track and discard all the others
* De-mux one of the foreign language surround sound streams and separate it into mono audio tracks
* Discard the 'centre' foreign language audio track and save all the others
* Re-combine the English 'centre' audio track with the foreign language surround sound tracks

The question is whether Cinavia would still be detected when it's confined to just one track within the surround sound stream?


Cheers

[ChiDragon]

I can tell you that the surround tracks on their own are detected as Cinavia.

Even if your method worked, dialogue isn't confined to the centre track (ex. room echo).

[SeeMoreDigital]

I can tell you that the surround tracks on their own are detected as Cinavia. Indeed, English language surround sound tracks contain Cinavia. But from what I understand, the foreign language surround sound channels do not !


Even if your method worked, dialogue isn't confined to the centre track (ex. room echo). Agreed... It's just a suggestion...

[ChiDragon]

I meant that the surround channels on their own trigger detection.

And given that my camcorder test with muffled dialogue, folded-down surrounds, and the introduction of all sorts of other noises still triggered Cinavia, it seems to me that centre track + surrounds from a whole other movie would still trigger Cinavia.

[SeeMoreDigital]

I meant that the surround channels on their own trigger detection.
For which language?

[ChiDragon]

Ones that are Cinavia watermarked, of course.

[plasmacutter]

Has anyone tried a reference and diff attack on this?

I mean, it would be painfully easy if they made the mistake of cinavia-protecting a BD version without doing so for DVD, but even if they have not, there's a good possibility that diffing out a large number of tracks would give you what you need to extract.

[omegaman7]

I'll probably be attempting something like that this week(Finally on vacation!). There are multiple movies, where the Blu-ray has the "watermark", and the DVD does not. Pretty ignorant if you ask me, but hey, if they wanna make it easier for us ;)

But then, identifying it may not be difficult. It's removing it, or masking it without affecting the audio, that's the real challenge.

[plasmacutter]

Masking it is as easy as any filter for media... If they can embed it imperceptibly, it can be removed just as imperceptibly. Heck, there's a chance you wouldn't even need to isolate it, you could just "jam" it by simply applying the same imperceptible distortions to the whole thing, millisecond by millisecond.

As for the people earlier in the thread who worry this can't be broken, i've seen literature indicating the exact same scheme has been before on different media.

I also think the only reason it hasn't been cracked is the fact that it's simply not ubiquitous enough as an inconvenience yet. People are still able to choose a different player relatively inexpensively, or even build a cheap HTPC for 250 bucks.

[omegaman7]

Masking it is as easy as any filter for media... If they can embed it imperceptibly, it can be removed just as imperceptibly. Heck, there's a chance you wouldn't even need to isolate it, you could just "jam" it by simply applying the same imperceptible distortions to the whole thing, millisecond by millisecond.

As for the people earlier in the thread who worry this can't be broken, i've seen literature indicating the exact same scheme has been before on different media.

I also think the only reason it hasn't been cracked is the fact that it's simply not ubiquitous enough as an inconvenience yet. People are still able to choose a different player relatively inexpensively, or even build a cheap HTPC for 250 bucks.

That's exactly my thoughts. A lot of people still have first and second generation players, that don't support detection. But, if a lot of people make the switch to 3d players, we'll see a rise in frustrations. I'm not one to jump on that bandwagon. And I'm already protected for all time. I keep software backups like you wouldn't believe. And should they ever try to leash the PC (For all playback software), that'll be that. A lot of people would be quite upset by that.

It's days are numbered. Like all things, it's only a matter of time...

[ChiDragon]

I mean, it would be painfully easy if they made the mistake of cinavia-protecting a BD version without doing so for DVD

Howso? I posted samples of the same audio track with and without Cinavia in 2010 and no one came back with a solution. The audio was from different-region lossless BDs; no DVD compression making things more difficult.

[omegaman7]

I wouldn't say painfully easy. Perhaps for an algorithm given a specific set of instructions. But what instructions would we give it? LOL! Perhaps an algorithm could be told to look for a pattern, among 6 separate audio tracks. It would probably take even a powerful processor weeks to list all patterns it finds. But this really isn't my area of expertise. I honestly think the naked eye might have better advantages. Humans do have a gift for spotting things out of the ordinary at times. But this could more complex than that.

[plasmacutter]

they were probably approaching it programmatically (digital diff), which would produce nothing but junk, rather than running a diff against the sound itself, which would have to be done in an audio editor.

They say this is robust against compression, etc, so overlaying the waveforms in an audio editor should produce some pretty obvious results.

You'd need multiple samples to determine whether it's just one signal, or multiple variations, and the bands used.

If it's just one signal, you're looking at the easiest solution, if you isolate the bands it gives you a narrower range for "jamming".