"Cinavia"/Verance BluRay audio watermark protection

[setarip_old]

@kurkosdr

Hi!

As I posted both in my Post #1 with which I started this thread, and Post #14, the specific "Verance"/"Cinavia" technique appears to date back to at least 2001:

Post #1 It was first licensed to Universal Studios in 2003, for use with DVDs and VHS tapes - but likely never used

Post #14 If anyone with more technical knowledge than I have is interested, here's a link I came upon, to a "challenge" regarding audio watermarking, dating back to 2001:

http://www.usenix.org/publications/library/proceedings/sec01/craver.pdf

[SamuriHL]

Then there must be more to this or else a "simple" notch filter at 15kHz oughta remove it, right? I for one would not likely tell such an excision has been made from the original.

Let me put it this way. Someone pointed a camcorder at a wall with a Cinavia title playing in the background. The analog recording was then played back on their PS3 and the error 3 Cinavia message came up. :) I have just made an MP4 with stereo crappy AAC audio and the PS3 picks that up, as well.

[mrr19121970]

@Klendathu

I'd like to see more graphs like this on different audio codecs & line in too, if it's not too much trouble for you.

Thanks.

[mike20021969]

Anyway, I thought Cinavia is a waveform based protection.

[jdobbs]

Seems like a logical course would be to extract the audio from a non-cinavia DVD, extract audio from a Cinavia BD disc, and do a difference analysis. Anyone done that?

[k-c-ksum]

Seems like a logical course would be to extract the audio from a non-cinavia DVD, extract audio from a Cinavia BD disc, and do a difference analysis. Anyone done that?

yeah, extract the core audio and demux the audio into seperate channels, preferable 1 with not much going on and play it on the ps3 to see if its detects the cinavia, if it does then demux the same audio channel from the dvd and compare.

2011 the year cinavia become's a pain in the ass :(

[SamuriHL]

2011 the year cinavia become's a pain in the ass :(

Only if your player supports it. :)

[Klendathu]

I am reasonably sure that anything that survives the psycho-acoustic requantization of MP3 is rather low-tech in nature.

Especially when this technology is old and was supposed to be used in a VHS recorder. I can't imagine them putting an expensive DSP into such a machine. A 10 cent notch filter? More likely.

I am sure testing will reveal something. First make a short recording that triggers the PS3, then dissect it by cutting anything away in 1khz steps, so you have 15 files each with a different range of 1khz. See which ones trigger, see which ones don't. Rejoin those which do not trigger, see if they still do not trigger. Try the same with another movie. What's left then holds the key.

[ChiDragon]

I'm no audio wizard, so perhaps someone else will find these useful. 54 second samples, one with Cinavia and one without, both sourced from DTS-HD MA tracks. It's possible they are mixed slightly differently.

http://img13.imageshack.us/img13/3969/cinaviaflacs.png (http://chidragon.thedessie.com/CinaviaFLACs.zip)

[setarip_old]

Has anyone determined whether "Cinavia" MUST read ALL audiostreams and determine that they ALL contain the incriminating watermark before "doing its thing?

[SamuriHL]

Has anyone determined whether "Cinavia" MUST read ALL audiostreams and determine that they ALL contain the incriminating watermark before "doing its thing?

Well, I made a BD-RE of just the main DTS-HD MA track on The Losers and it detected it from that.

[setarip_old]

@SamuriHL

Hi! Well, I made a BD-RE of just the main DTS-HD MA track on The Losers and it detected it from that.If you're saying that the disc you made contained ONLY ONE audiostream then you misunderstand me. I'm wondering what might happen if one added an ADDITIONAL, perhaps blank, audiostream to a backup of a "CINAVIA"-infested Blu-ray - so that if it originally had 4 audiostreams, it would now have 5)...

[SamuriHL]

You're correct. I misunderstood. No, I didn't try that. I'm 99.9% sure it will have no affect at all. But someone should try it.

[Peer van Heuen]

I am reasonably sure that anything that survives the psycho-acoustic requantization of MP3 is rather low-tech in nature.


Well, the "predecessor" of this watermark, the one used on several DVDA discs, already was way more sophisticated than that. It was not a "signal added" to the audio on some frequency band, but more like a modification of existing information.
Its playfield was on certain frequency bands, yes, but a) were they slightly varying and b) removing the frequencies would have removed the very core of the audio.

Also, I'm not sure why people seem to be so convinced, that it's an "old" watermarking technology. Just because some papers that allegedly may be the basis of the tech were marked "2001" doesn't mean much. Even if the watermark is based on these, it doesn't mean it hasn't evolved. On top of that, I think I remember that at least one of them (referred to somewhere either here or on our forum) that was from 2008.

Anyway - our own findings so far indicate that it is definitely not as easy as cutting out some frequency and be done with it.

[asarian]

@SamuriHL

I'm wondering what might happen if one added an ADDITIONAL, perhaps blank, audiostream to a backup of a "CINAVIA"-infested Blu-ray - so that if it originally had 4 audiostreams, it would now have 5)...

Even if that would work, which I'm 99.9% sure it won't, Sony and co. would simply, and easily, fix that on the next firmware update.

Also, I have always had great respect for the SlySoft folks. Nonetheless I doubt Cinavia will ever be successfully removed; if for nothing else, because I don't want it removed, even. Which is to say, even if someone could very cleverly remove (or disrupt) the markers that trigger the Cinavia detection, the resulting HD audio would still have to be resampled -- at quality loss (for the same bitrate). Consequently, I think the simplest -- and probably best -- solution will be for someone to 'brick' their PS3 one day, and remove or disable the Cinavia detection itself at firmware level. Or we should go to simply building our own HTPC's.

[Peer van Heuen]

Consequently, I think the simplest -- and probably best -- solution will be for someone to 'brick' their PS3 one day, and remove or disable the Cinavia detection itself at firmware level. Or we should go to simply building our own HTPC's.

...chiming in on this - I think Cinavia will basically make non-conforming players more popular and HTPC solutions with a combination of on-the-fly protection removal and a simple player will be given a push.

SlyPlayer is already doing nicely and will be available soon... say February or so? ;)

[k-c-ksum]

Patching the firmware for your standalone might be posible? ps3 might be out. Are Firmwares for standalones alot less robust? is it possible to compare a pre cinavia player firmaware to an infecteds one.?

[rotty]

Has anyone determined whether "Cinavia" MUST read ALL audiostreams and determine that they ALL contain the incriminating watermark before "doing its thing?

Lets put it this way, a mono track, if marked, will also trigger Cinavia.

[setarip_old]

@rotty Lets put it this way, a mono track, if marked, will also trigger Cinavia.Let's put it this way, if an extra track is added that is NOT marked is what I'm discussing...

[rotty]

@rotty Let's put it this way, if an extra track is added that is NOT marked is what I'm discussing...

If you mean, if you’re watching the movie and listening to an unmarked secondary track then I don’t know, it would depend at what stage in the player the info is taken for the Cinavia detection circuit.

I take it you were interested in another track with same audio as the master track but without the watermark to see the difference, reckon they would have thought of that but you never know.

P.S.
I’m sure I remember reading before in a post that in the first appearance of Cinavia, people have muxed an audio track from a DVD (obviously same movie) onto their burned blu ray disk and that did not have Cinavia. So there are probably plenty of examples of duplicate tracks, one with and one without Cinavia for comparison. If you knew where the marks were I bet you would actually be able to hear the difference.

Butchering audio like this with Cinavia make a nonsense of having HD Master tracks.

[Klendathu]

Anyway - our own findings so far indicate that it is definitely not as easy as cutting out some frequency and be done with it.

No matter how much it butchers the audio, nothing but cutting frequencies can help locate it. I am not overly concerned with loss of audio quality, since finding the Cinavia is the more interesting problem. Once I do know exactly where it idles around, the question is to find out how it works and how rhythmic the signal needs to be.

If that doesn't help, I am sure there are interesting things to do with audio resolution. Sure, Cinavia can trigger at 16bit sampling rate, but will it still trigger at 8bit? at 4bit? What is the minimum resolution required for it to work? How can you mess up a 16bit signal so that the resolution is messed up in the right places? What happens if the signal runs through a stereo with DSP effects simulating different room sizes? I am confident, I will have fun for a while.

Beyond that I am not thinking about removing the watermark, but rather imprinting my own watermark on top of it. Slightly less inaudible maybe and 100% corrupting the underlying watermark. That is what you do with an analog signal, you do not remove it, you simply jam it. If layering a stronger signal on top of it still triggers Cinavia, then we still know where to weaken the sinal. Weakening an already inaudible signal? I can't imagine a lot of quality loss happening there.

[rotty]

No matter how much it butchers the audio, nothing but cutting frequencies can help locate it. I am not overly concerned with loss of audio quality, since finding the Cinavia is the more interesting problem. Once I do know exactly where it idles around, the question is to find out how it works and how rhythmic the signal needs to be.

If that doesn't help, I am sure there are interesting things to do with audio resolution. Sure, Cinavia can trigger at 16bit sampling rate, but will it still trigger at 8bit? at 4bit? What is the minimum resolution required for it to work? How can you mess up a 16bit signal so that the resolution is messed up in the right places? What happens if the signal runs through a stereo with DSP effects simulating different room sizes? I am confident, I will have fun for a while.

Beyond that I am not thinking about removing the watermark, but rather imprinting my own watermark on top of it. Slightly less inaudible maybe and 100% corrupting the underlying watermark. That is what you do with an analog signal, you do not remove it, you simply jam it. If layering a stronger signal on top of it still triggers Cinavia, then we still know where to weaken the sinal. Weakening an already inaudible signal? I can't imagine a lot of quality loss happening there.


The description of the Cinavia watermark as "inaudible" in this case does not mean you cannot hear it, means un-noticeable, it will be audible.

It does survive really bad recording with a microphone, and I do mean bad recording.

You mention a 15K signal present, a cheap microphone will not be able to detect those frequencies, put a 100N cap across an eletret mic and try recording the marked audio, it still survives, the cap will limit top end to about 4K.

Because we don’t know where in the time line these marks are we are not carefully listening out for it. Once these mark points are identified, we will be able to probably identify by ear where they are once when we get to hear what we are listening out for.

And yes your totally right, disrupting will be the best option. Removal will cause us to really notice where the marks used to be.

Dont be fooled by Cinavia being quickly detected thereafter the inititial detection. This does not mean the mark is continuous.

[setarip_old]

If, as I've read either at the Doom9 Forums or elsewhere, "Cinavia" must first ascertain that ALL audiostreams of a Blu-ray disc contain the watermarking (to avoid being sued for erroneously turning off a legitimated Blu-ray disc). If that is the case, perhaps it's as simple as adding (NOT replacing) 1, 2 or even 5 blank audiostreams...

[asarian]

If, as I've read either at the Doom9 Forums or elsewhere, "Cinavia" must first ascertain that ALL audiostreams of a Blu-ray disc contain the watermarking (to avoid being sued for erroneously turning off a legitimated Blu-ray disc). If that is the case, perhaps it's as simple as adding (NOT replacing) 1, 2 or even 5 blank audiostreams...

That would be an incredible loophole! Therefore I can hardly believe it to be true. The logical thing would be to shut off playback the moment you detect Cinavia anywhere on the disc (and the stream it's in is not on a pressed Blu-Ray). Someone with a new firmware in his PS3 should try it out, though.

Also, wasn't Cinavia only present in HD audio? It seems to make little sense for SD audio, as people could then simply burn a BD-R with the audio track of a lowly DVD release.

[asarian]

Weakening an already inaudible signal? I can't imagine a lot of quality loss happening there.

Quality loss is always going to occur, even if you were able to remove Cinavia 100% accurately, without any collateral damage. And simply because you'd have to resample the resulting stream. Now, if the original were 16-bit, you'd be in luck, and you could upsample to 24-bit, and the loss would be extremely minimal (to absent, perhaps); but at the same bitrate, quality loss will ensue; that's a given. Perhaps you don't mind, but I consider a non-lossless fix (with loss of quality, that is) not a satisfactory permanent solution.

Also, how are you going to resample? Is, say, SlySoft going to incorporate a DTS-MA encoder? Likely not. So, a resampled stream will probably be LPCM. Which is fine when used for a custom software player; but if you want to reburn your Blu-Ray, that extra lengthy LPCM track may cause you to run out of disc space.

Because we don't know where in the time line these marks are we are not carefully listening out for it. Once these mark points are identified, we will be able to probably identify by ear where they are once when we get to hear what we are listening out for.

And yes your totally right, disrupting will be the best option. Removal will cause us to really notice where the marks used to be.

Dont be fooled by Cinavia being quickly detected thereafter the inititial detection. This does not mean the mark is continuous.

Not continuous, perhaps, but certainly continuously spread out over the entire stream.

Let's say a PS3 triggers at X times detection of the marker; then, typically, you would be inclined to disrupt at each, say, X-1 marker. But you can bet the distance between these markers is random (within a certain window, of course). And you can also bet that X itself is not a fixed number, either. So, in practice you'd probably have to dampen any and all markers you find.

[rotty]

Quality loss is always going to occur, even if you were able to remove Cinavia 100% accurately, without any collateral damage. And simply because you'd have to resample the resulting stream. Now, if the original were 16-bit, you'd be in luck, and you could upsample to 24-bit, and the loss would be extremely minimal (to absent, perhaps); but at the same bitrate, quality loss will ensue; that's a given. Perhaps you don't mind, but I consider a non-lossless fix (with loss of quality, that is) not a satisfactory permanent solution.

Also, how are you going to resample? Is, say, SlySoft going to incorporate a DTS-MA encoder? Likely not. So, a resampled stream will probably be LPCM. Which is fine when used for a custom software player; but if you want to reburn your Blu-Ray, that extra lengthy LPCM track may cause you to run out of disc space.



Not continuous, perhaps, but certainly continuously spread out over the entire stream.

Let's say a PS3 triggers at X times detection of the marker; then, typically, you would be inclined to disrupt at each, say, X-1 marker. But you can bet the distance between these markers is random (within a certain window, of course). And you can also bet that X itself is not a fixed number, either. So, in practice you'd probably have to dampen any and all markers you find.

Yes the mark will be repeated throughout. But not continuous and of course is not possible at all in silent passages.

I get the impression from Peer's comments that Slysoft is going down the s/w player route to circumvent the watermark rather that remove or dissrupt it.

We still have the ultimate weapon, dont buy watermarked disks, even if that means buying none for a while until it is removed.

[asarian]

Yes the mark will be repeated throughout. But not continuous and of course is not possible at all in silent passages.

Well, silent is rarely truly silent, of course (think static and/or background noise). Though, granted, embedding Cinavia in a 'silent' passage would be kinda pointless (I mean, from the perspective of the copyright holder, who cares whether I can listen to silence or not?), and it would make it easier for people to detect Cinavia.

I get the impression from Peer's comments that Slysoft is going down the s/w player route to circumvent the watermark rather that remove or disrupt it.

And I have to admit I didn't fully grasp that remark of his, in that software solutions, like XMBC, to play without Cinavia on a HTPC, already exist. But perhaps he has referring to alternatives to PowerDVD and ArcSoft TMT and such.

EDIT: Another way would be to come up with a way to hack into Windows' underlying Media Foundation structure and such, so as to manipulate audio streams at low-level. Then you could continue to use your regular playback software. Although I think that would probably only work for pure MS products, like Windows Media Player and Windows Media Center.

[rotty]

Well, silent is rarely truly silent, of course (think static and/or background noise). Though, granted, embedding Cinavia in a 'silent' passage would be kinda pointless (I mean, from the perspective of the copyright holder, who cares whether I can listen to silence or not?), and it would make it easier for people to detect Cinavia.



And I have to admit I didn't fully grasp that remark of his, in that software solutions, like XMBC, to play without Cinavia on a HTPC, already exist. But perhaps he has referring to alternatives to PowerDVD and ArcSoft TMT and such.

EDIT: Another way would be to come up with a way to hack into Windows' underlying Media Foundation structure and such, so as to manipulate audio streams at low-level. Then you could continue to use your regular playback software. Although I think that would probably only work for pure MS products, like Windows Media Player and Windows Media Center.

Cinavia needs to modulate existing audio and thats what I meant by not possible in silent or near silent passages.

Yes I think thats what Peer meant, an alternative to existing s/w players.

[SamuriHL]

EDIT: Another way would be to come up with a way to hack into Windows' underlying Media Foundation structure and such, so as to manipulate audio streams at low-level. Then you could continue to use your regular playback software. Although I think that would probably only work for pure MS products, like Windows Media Player and Windows Media Center.

That wouldn't work. The player controls the audio stream as it's read off the disc (or wherever it comes from if it's reauthored). Internally it's checking for the presense of Cinavia and if detected, it stops outputting the audio stream.

As for what Peer is saying, he very clearly stated that he believes the future of software playback is going to be SlyPlayer(tm). This is a full featured, *freeware* software player that SlySoft is developing. It'll play all your unprotected backups even if they contain Cinavia. It'll just ignore that part. Of course, there are many other solutions, as well. MPC-HC, VLC, XBMC, etc will all never support Cinavia so they can be used for playback. I personally backup all my BD's to MKV format and use MPC-HC to play them back with perfect bitstreaming. So, there are ways around the Cinavia problem that don't require any work or removal. SlyPlayer isn't out yet, and Peer alluded to the running joke that all new SlySoft products are released in "February". LOL! But it is in development.

[jdobbs]

That wouldn't work. The player controls the audio stream as it's read off the disc (or wherever it comes from if it's reauthored). Internally it's checking for the presense of Cinavia and if detected, it stops outputting the audio stream.

As for what Peer is saying, he very clearly stated that he believes the future of software playback is going to be SlyPlayer(tm). This is a full featured, *freeware* software player that SlySoft is developing. It'll play all your unprotected backups even if they contain Cinavia. It'll just ignore that part. Of course, there are many other solutions, as well. MPC-HC, VLC, XBMC, etc will all never support Cinavia so they can be used for playback. I personally backup all my BD's to MKV format and use MPC-HC to play them back with perfect bitstreaming. So, there are ways around the Cinavia problem that don't require any work or removal. SlyPlayer isn't out yet, and Peer alluded to the running joke that all new SlySoft products are released in "February". LOL! But it is in development. A new software player wouldn't do anything for people how use their SAP, though... which is by far the vast majority. The only real solution is removal. Firmware mods would also work -- but could become a fulltime job for somebody.

The bottom line is that if Cinavia can be added to an audio stream, it can also be removed. The hard part is finding what to remove.

On-the-fly audio analysis and removal could be a real headache for a driver-based anti-protection mechanism, though. It may be something more easily done after a complete rip. Ughhh... I'll leave all that up to the experts.

[Sharc]

How about
1. Convert (recode) a Cinavia protected audio from the HD stream (DTS or whatever) to ac3 / 448
2. Compare with the corresponding extracted audio of a SD (DVD) ac3 / 448 original

e.g. synchronize the 2 streams in time, subtract 2 from 1 and analyze the difference ....

[SamuriHL]

A new software player wouldn't do anything for people how use their SAP, though... which is by far the vast majority. The only real solution is removal. Firmware mods would also work -- but could become a fulltime job for somebody.

The bottom line is that if Cinavia can be added to an audio stream, it can also be removed. The hard part is finding what to remove.

On-the-fly audio analysis and removal could be a real headache for a driver-based anti-protection mechanism, though. It may be something more easily done after a complete rip. Ughhh... I'll leave all that up to the experts.

Right, I agree that removal is definitely the end goal. However, I don't think it's going to prove all that simple to just remove it. It's going to take time. For now we need to come up with a solution that works today. For those with standalones, there really isn't one unfortunately.

[rotty]

How about
1. Convert (recode) a Cinavia protected audio from the HD stream (DTS or whatever) to ac3 / 448
2. Compare with the corresponding extracted audio of a SD (DVD) ac3 / 448 original

e.g. synchronize the 2 streams in time, subtract 2 from 1 and analyze the difference ....

A man after my own heart.

The best way to do this is to convert the Cinavia infested track(s) to mono (this make everything much simpler)
It doesent really matter at what quality, just a reasonable quality.

Do the same for the non infested audio track(s)

As you said, it is of course important for them to be totally synced. And I mean really in sync, we are looking at part cycle differences using a differential op amp.

The two tracks do have to be exactly the same (apart from Cinavia of course)

Connect say the Cinavia track into the + input of the amp, use an LM380 for ease.
Connect the non Cinavia track into the - input.

You must have potentiometers on both inputs.

Turn completely down pot 2, adjust pot 1 so the sound output is a good listsening level.
Now adjust pot 2 up slowly, if your tracks are in sync, then as you increase pot 2 the output should become lower and lower.

As I say, if the tracks are the same and in sync, you should get to the point when you hear nothing or very very little out of the amp. Turnng the pot higher from this point you will start to hear the output again.
Adjust pot 2 for the lowest output you can get

You will only hear an output if there is a difference in the inputs, either a difference in amplitude, phase or of course if the info is not the same at any point.

Sit back and listen out for the differences.
Cinavia will be heard.

P.S. forgot to say, re mux the two mono tracks, (the one with and the one without Cinavia) into a single stereo track.
Listsneing (not from the op amp) to the sound using your normal player using your normal audio system, if in sync the sound should appear in center of field and solid, no spread. If it sounds like that then you have the tracks in phase and synced up and can continue with the test.

Put say left output into the + input of the op amp and right channel into the - input of the op amp.
and set up as before the P.S.

[asarian]

A man after my own heart.

The best way to do this is to convert the Cinavia infested track(s) to mono (this make everything much simpler)
It doesent really matter at what quality, just a reasonable quality.

Do the same for the non infested audio track(s)

As you said, it is of course important for them to be totally synced. And I mean really in sync, we are looking at part cycle differences using a differential op amp.

The two tracks do have to be exactly the same (apart from Cinavia of course)

Connect say the Cinavia track into the + input of the amp, use an LM380 for ease.
Connect the non Cinavia track into the - input.

You must have potentiometers on both inputs.

Turn completely down pot 2, adjust pot 1 so the sound output is a good listsening level.
Now adjust pot 2 up slowly, if your tracks are in sync, then as you increase pot 2 the output should become lower and lower.

As I say, if the tracks are the same and in sync, you should get to the point when you hear nothing or very very little out of the amp. Turnng the pot higher from this point you will start to hear the output again.
Adjust pot 2 for the lowest output you can get

You will only hear an output if there is a difference in the inputs, either a difference in amplitude, phase or of course if the info is not the same at any point.

Sit back and listen out for the differences.
Cinavia will be heard.

While laudible, I fear such attempts are if not futile, then doomed to fail. Even if you could 'diff' the Cinavia superimposing with exact certitude, then it's by no means certain you will have extracted useful markers. For all we know, Cinavia adds a lot of extra, random noise, just to throw us off track.

My point being, the only truly successful way, IMHO, will be to reverse-engineer the firmware of a standalone player, and watch it, step-by-step, to see what actual markers are being used -- instead of just the entire diffed signature. Only that will give us a real insight as to what needs to be removed/dampened.

[Peer van Heuen]

And I have to admit I didn't fully grasp that remark of his, in that software solutions, like XMBC, to play without Cinavia on a HTPC, already exist. But perhaps he has referring to alternatives to PowerDVD and ArcSoft TMT and such.


I was not saying, that new players are needed, I merely pointed out that alternative players will gain popularity - those existing and those to come...

That said - I haven't seen any alternative BD player that got me too excited, yet. Most have drawbacks, a choice from no GPU acceleration, unreliability, not supporting all streams, not all audio codecs, only play individual .m2ts files.
Usability issues on top of all that.

Some, like XMBC, are quite advanced and a good choice for an HTPC, but you can't just shove in a blu-ray disc and watch it.

Anyway - certainly, we are working on removing Cinavia. So, no, we're definitely not "going down the s/w player route to circumvent the watermark", we also didn't launch the SlyPlayer project because of Cinavia.

I'm just saying, that with Cinavia being added to more and more discs, buyers will have more incentive to not touch hardware players, but instead go for an HTPC with such an alternative player. It's simply less trouble.

Not that it really needs much more of that: I was on a trip to Europe recently, had a bunch of my DVDs along and was so used to not worrying about region codes, that I forgot, that these discs were going to be utterly useless on a region 2 DVD player in the hotel room... :rolleyes:
That's part of the "collateral damage" you buy into, when using certified devices.

My discs. Bought. Paid for. Can't watch them. Screwed.
I felt - apart from feeling a little stupid, as I really didn't think of that beforehand - abused.

[xenex]

@Peer

You should have brought along a laptop with AnyDVD installed.

Seriously, though, until there is some sort of 'consumer outrage' most people are just going to believe "that's just the way it is." Heck, I expected an outcry when DVDs started using the Macrovision APS to stop VHS recording. Nope. I expected complaints when DVDs abused PUOs to force watching previews, etc. Nope. I really thought the hue and cry would begin with HDCP requirements. Nope.

The average 'consumer' writes "INSERT HERE" on a Post-It note (copyright 3M), pulls down his pants, bends over, attaches the note to his posterior, and smiles.

Such is the state of the world.

[rotty]

While laudible, I fear such attempts are if not futile, then doomed to fail. Even if you could 'diff' the Cinavia superimposing with exact certitude, then it's by no means certain you will have extracted useful markers. For all we know, Cinavia adds a lot of extra, random noise, just to throw us off track.

My point being, the only truly successful way, IMHO, will be to reverse-engineer the firmware of a standalone player, and watch it, step-by-step, to see what actual markers are being used -- instead of just the entire diffed signature. Only that will give us a real insight as to what needs to be removed/dampened.

Hi Asarian
Whatever Cinavia adds to the track, it will be heard using the op amp, every part of the audio that is the same as the original will produce silence or next to no output from the amp if wired as I stated in last post.

Added noise or whatever is added will NOT be on the original non Cinavia track, therefore difference between the two tracks and only differnces will be output from the LM380N and therefore you will hear them.

Certainly not futile.

As I say, sit back and listen, mostly you will hear nothing, but every time Cinavia has altered the track you will hear it, no matter what Cinavia has added.

[Ghitulescu]

The best way to do this is to convert the Cinavia infested track(s) to mono (this make everything much simpler)
It doesent really matter at what quality, just a reasonable quality.

Do the same for the non infested audio track(s)

As you said, it is of course important for them to be totally synced. And I mean really in sync, we are looking at part cycle differences using a differential op amp.

The two tracks do have to be exactly the same (apart from Cinavia of course)

Connect say the Cinavia track into the + input of the amp, use an LM380 for ease.
Connect the non Cinavia track into the - input.

You must have potentiometers on both inputs.

Turn completely down pot 2, adjust pot 1 so the sound output is a good listsening level.
Now adjust pot 2 up slowly, if your tracks are in sync, then as you increase pot 2 the output should become lower and lower.

As I say, if the tracks are the same and in sync, you should get to the point when you hear nothing or very very little out of the amp. Turnng the pot higher from this point you will start to hear the output again.
Adjust pot 2 for the lowest output you can get

You will only hear an output if there is a difference in the inputs, either a difference in amplitude, phase or of course if the info is not the same at any point.

Sit back and listen out for the differences.
Cinavia will be heard.

Nice testing procedure, however ..... :)

The existence of a comparable non-cinavied audio defeats the purpose of implementing Cinavia in the first place, because everyone would use this one and not the cinavied one :)

So I have to assume that either all editions would be cinavied or bar none. I'm referring here to the corresponding track only (it makes no sense to compare the EN track with the DE or FR or ES one) and to US, RU, AU, JP etc. editions.

[rotty]

Nice testing procedure, however ..... :)

The existence of a comparable non-cinavied audio defeats the purpose of implementing Cinavia in the first place, because everyone would use this one and not the cinavied one :)

So I have to assume that either all editions would be cinavied or bar none. I'm referring here to the corresponding track only (it makes no sense to compare the EN track with the DE or FR or ES one) and to US, RU, AU, JP etc. editions.

Hi Ghitulescu

I have read in some previous posts that someone had remuxed track from non Cinavia DVD to BD to replace the Cinavia track on the blu ray. If an NTSC DVD then prob no re syncing required.

I think that there are examples of duplicate tracks out there for comparison, espcially from the early days of Cinavia.
I hope so.

all the best fella.

[Ghitulescu]

Hi Ghitulescu

I have read in some previous posts that someone had remuxed track from non Cinavia DVD to BD to replace the Cinavia track on the blu ray. If an NTSC DVD then prob no re syncing required.

I think that there are examples of duplicate tracks out there for comparison, espcially from the early days of Cinavia.
I hope so.

all the best fella.

My point was that cinavia is applied to HD lossless audio and no other source has this audio.

[asarian]

Added noise or whatever is added will NOT be on the original non Cinavia track, therefore difference between the two tracks and only differnces will be output from the LM380N and therefore you will hear them.

Certainly not futile.

As I say, sit back and listen, mostly you will hear nothing, but every time Cinavia has altered the track you will hear it, no matter what Cinavia has added.

Agreed. I merely meant to say, that 'diffing' two signals, for the purpose of finding the Cinavia markers, is likely fail (because of the random noise they may have added and such).

As for purely hearing the difference, sure, your method would work. I'm just not entirely certain on what merely hearing the Cinavia signal would accomplish towards making an algorithm to remove the scheme.

[rotty]

Agreed. I merely meant to say, that 'diffing' two signals, for the purpose of finding the Cinavia markers, is likely fail (because of the random noise they may have added and such).

As for purely hearing the difference, sure, your method would work. I'm just not entirely certain on what merely hearing the Cinavia signal would accomplish towards making an algorithm to remove the scheme.

Hi Asarian

To fight this we need to know what it actually is, this will tell us. At the moment we are fighting an invisible enemy.
If we do this we wont be.

[mike20021969]

SlySoft are obviously working on, and confident of removing, Cinavia.

More 'tips' from Joe Bloggs on what to do about removing it...surely are no longer relevent?
It seems to just be going round in cirlcles now after 18 pages.

[SamuriHL]

I was not saying, that new players are needed, I merely pointed out that alternative players will gain popularity - those existing and those to come...

That said - I haven't seen any alternative BD player that got me too excited, yet. Most have drawbacks, a choice from no GPU acceleration, unreliability, not supporting all streams, not all audio codecs, only play individual .m2ts files.
Usability issues on top of all that.

Some, like XMBC, are quite advanced and a good choice for an HTPC, but you can't just shove in a blu-ray disc and watch it.

Anyway - certainly, we are working on removing Cinavia. So, no, we're definitely not "going down the s/w player route to circumvent the watermark", we also didn't launch the SlyPlayer project because of Cinavia.

I'm just saying, that with Cinavia being added to more and more discs, buyers will have more incentive to not touch hardware players, but instead go for an HTPC with such an alternative player. It's simply less trouble.

Not that it really needs much more of that: I was on a trip to Europe recently, had a bunch of my DVDs along and was so used to not worrying about region codes, that I forgot, that these discs were going to be utterly useless on a region 2 DVD player in the hotel room... :rolleyes:
That's part of the "collateral damage" you buy into, when using certified devices.

My discs. Bought. Paid for. Can't watch them. Screwed.
I felt - apart from feeling a little stupid, as I really didn't think of that beforehand - abused.

This would be why I can't wait for SlyPlayer. I know it wasn't created for Cinavia, but, in the end it will provide a nice solution to the problem. Removing Cinavia is a good end goal, but, it's going to take a fair bit of work. In the meantime, it behooves us that use HTPC's to make sure we always have a player that doesn't have Cinavia detection. This, of course, doesn't help the standalone people. For them, they have no choice but to wait for Cinavia to be removed. However, those of us that use HTPC's use them for the very reason that we have options. That's now proving to be a really good thing.

(And I'm shocked! You didn't bring your backups instead?? :D)

[Peer van Heuen]

It seems to just be going round in cirlcles now after 18 pages.

The going round in circles is basically because the ideas coming up here are the very obvious ones that the engineers of the watermark came up with first.
It is understandable, of course, but the architects of Cinavia are signal processing specialists, and you really need a little background in order to be able to get into their minds.

Things just aren't as simple as many think they are.

Hi Asarian
Whatever Cinavia adds to the track, it will be heard using the op amp, every part of the audio that is the same as the original will produce silence or next to no output from the amp if wired as I stated in last post.

Added noise or whatever is added will NOT be on the original non Cinavia track, therefore difference between the two tracks and only differnces will be output from the LM380N and therefore you will hear them.

Certainly not futile.

As I say, sit back and listen, mostly you will hear nothing, but every time Cinavia has altered the track you will hear it, no matter what Cinavia has added.

As for purely hearing the difference, sure, your method would work. I'm just not entirely certain on what merely hearing the Cinavia signal would accomplish towards making an algorithm to remove the scheme.

Absolutely. Actually, you won't even neccessarily "hear" even part of the watermark at all.
I dare anyone with some time on their hands to conduct the following experiment to prove how easy it is to make the "difference approach" totally useless (I did it a while ago out of curiosity):

Take any simple mono sound sample, feed it chunkwise into an FFT, phase shift the frequencies, then iFFT. You can even use different phase values per frequency band, without causing the sample any audible damage.
The difference of the input and output sample will more or less sound like a somewhat distorted version of the input signal.

Any "signal" that would have been planted will be just as hidden in this difference signal as it was in the original sound.

Now, you can try and re-align the phases, but they could also be slowly varying over time (slowly in these scales means within fractions of seconds).

And this is only one possible way to obfuscate a watermark signal.

(And I'm shocked! You didn't bring your backups instead?? :D)

Believe it or not - I hardly ever copy a disc at all.
I just use AnyDVD to watch discs in a dignified manner, without all the bending over.... :)

[SamuriHL]

Believe it or not - I hardly ever copy a disc at all.
I just use AnyDVD to watch discs in a dignified manner, without all the bending over.... :)

So....you use AnyDVD as how it was originally intended. How depressing! ROFLMAO! :D I can understand. I use it that way a lot, as well. Clearly for rented discs I only ever watch them but always with AnyDVD enabled. Yet another reason I want a player with no Cinavia support. Once the commercial software players add Cinavia, playing discs with Cinavia and AnyDVD enabled will no longer work. :(

[rotty]

[QUOTE=Peer van Heuen;1459474]

Hi Peer

You are of course right, if the phase of the marked track is slowly leading then lagging and worse, the speed at which it does it is random then a diffential would always be present. And as you say, a very slow change in phase would not be detected by us and therefore not destroy the our perceived quality.

In my defence, I did say to test the created stereo (from the Cinavia and non Cinavia mono tracks) by listening to the normal audio system and that it should sound central, solid and no spread at all. A changing phase in one of the tracks as you know would give a continuously changing positional shift in the stereo field and no point in continuing with the test.

Now this is just out of interest:
I dont know if you can answer this safely, but have you heard a watermaked portion of audio and the same portion without mark.

And I am very pleased (as im sure all of us are) that you are endevouring to remove this as well as creating the new player.
Look forward to both.

Thank you Peer.

[kurkosdr]

My point being, the only truly successful way, IMHO, will be to reverse-engineer the firmware of a standalone player, and watch it, step-by-step, to see what actual markers are being used
Yes, but how are we going to get the firmware out of the chip it's stored in inside the standalone? Have Conan the Barbarian squeeze the chip really hard till the firmware leaks out? Sure, you can try to get it from some firmware update disc, but there is a high chance it will be obfuscated, i don't think they will have left their firmware unprotected like that.
Unless of course hollywood makes a mistake and cinavia trickles down to some software player like PowerDVD/WinDVD//TMT3. Then it's hex editor time and we ll solve our problem for good. So basically we are counting on them making a mistake (allow the firmware to leak somehow). But i agree with you that reverse engineering a player is the only 100% sure way to reverse engineer cinavia. Difference analysis requires guesswork, which will fail if the watermarking method is too complicated, inserts random noise, or is dependent on the original signal in a weird way.
So, I think too we must have our eyes peeled for a reverse-engineer-able firmware, but what do we do till then?

Here is a crazy idea to solve the problem right now:
Let's assume you have that very precious bluray you must absolutely backup.
Now, how about buying the DVD release, copying the AC-3 or DTS audio track and put it under the video stream of 1080p mkv you just ripped? As another member said, if it's a NTSC DVD, it can be done. Syncing shouldn't be that hard if you have two identical video tracks (only the resolution differs). It can even be done on a frame by frame basis. Sure, you ll lose the lossless audio, but you can't always have it all. Till now, it's the only solution that works everywhere.

The other solution is to use i standalone mkv player (the ones without a dvd/bluray drive, just a usb port), but most of them read only FAT32, so your mkv's are limited to 4.35gb even if you have a 16gb thumbdrive.


P.S.: Has anyone tried a cinavia marked mkv with PowerDVD? Does PowerDVD show any cinavia messages?
P.S.: Has anyone actually tried to use the 15KHz notch filter and tried to play the resulting file on a cinavia aware player?

Please answer these questions, as to make sure we are not cutting out two easy solutions without actually trying them

[Sharc]

With all sophisticated signal processing one can imagine: The watermark must be darn robust as it seems to survive re-encoding into different formats, re-compression (non-linear), poor quality microphones etc.

[SamuriHL]

I've tested all the commercial players for Cinavia. *So far* none have it. But you can expect them all to get it eventually.