Slashdot Mirror
Another Audio Watermark Scheme Wins TI DSP Contest
CaptainTylor writes: "Texas Instruments' DSP and Analog Design Contest Challenge is over, and the winner is a group of students from Technion, the Israel Institute of Technology, who presented yet another scheme for digital-audio watermarking, and got US$100,000 for it. Here is a Dallas Morning News article on the winners, which is of course light on the tech details. Abstracts of the winner and the other two finalists are available, but I couldn't find the full submissions. It's worth noting that the competition was not specifically about copyright protection, just about using the TI TMS320 DSP in interesting ways. Wonder how long it'll take before someone cracks this scheme..."
And speaking of schemes, cracking, audio and contests, Logic Bomb writes: "According to an article from the Associated Press, the United States National Archives are holding a contest of sorts to see if anyone can finally figure out what was erased on the infamous Watergate tape that pushed Nixon's downfall over the brink. It would be amazing to have this national mystery put to rest."
That this is the URL for the robot:
http://real.uwaterloo.ca/~robot/
The link below is not the URL for the robot:
It's not the tool that's bad, it's how it's used that can be bad...
"Actually, I heard the blank spot was just noise introduced by a top-secret CIA funded copy-protection scheme from the era. ;) "
Wouldn't that mean that the attempts to hear what's there would violate the DMCA?
I think we're in trouble.
Even if every single CD-R writer company did embed watermarking code in hardware, how long do you think it would take for someone to come up with a hardware solution? Think Playstation and modchips.
End of lesson. You may press the button.
Actually, that brings up an interesting question: Suppose someone decides today to use a copyright protection scheme which was cracked by researchers *before* the DMCA went into effect. Does it then suddenly become illegal to traffic in the so-called circumvention mechanisms? Does it become illegal to republish or redistribute the paper?
If so, a lot of back-issues of technical journals could be considered contraband under the law. Whee!
As far as I'm aware, there are only two options for a company intending to use a watermarking technology:
1. Embed the watermark in the actual audio (for example, a high-frequency series of hisses throughout the song. this would enable the watermark to be present, more or less, if translated into other audio file formats)
2. Embed the watermark in the file format
Consumers will most likely not stand for the first option, and the second option is worthless after the watermarking scheme is cracked.
er, so anyway, the point is (and i think this was on /. at some point) you can check this url - scroll down a bit to get to the meat of the crack - for more information on audio watermarking and its effectivity:
http://cryptome.org/sdmi-attack.htm
anyway, selling media online has always been something churning around in the back of my mind and every time i come up with some sort of idea that may or may not work i, i can pretty much figure out a way to crack it.
my conclusion is that there is no way you can encrypt this data in an uncrackable way due to the fact that at some point you have to send audio data to the sound card - and if anyone intercepts that stream or |'s it to another dev then all your effort is wasted.
surely a viable solution would be to use hardware-based decryption using PKI. like lets say my sound card had a mobile (cell phone) like SIM card slot in it. so i buy this sound card, register with some MSP (music service provider) who supple me with a SIM card that i slot into my sound card. then i can download encrypted media (like NOT destiny's child, please!) and, tada, a workable solution.
the first thought that has always come into my mind here is *ping* HARDWARE DONGELS. nooooo. what a success that was! but i feel this is slightly different. its not like you are having software that probes for the existance of something which can be decompiled and cracked.
i've got a number of other great ideas on this that i've been formulating for over 3 years now, if anyone out there is interested in getting together to maybe push something like this forward (i have a lot of ideas with regards to ownership of content and so on as well - basically giving you the same "freedom" of ownership that you have with physical media), give me a shout on michael_jw_bartlett at hotmail dot com.
of course you can always buy gold audio jacks, plug it into you sound card's output and plug the input into your cd writer... ;)
Nitpick:
cd sampling rate is 44100, thus nyquist says the max reproducible freq is 22050hz.
You'll see that all the settings on any 16bit audio software match those numbers (except dat, which is 48khz, iirc)
My opinions are my own, and do not necessarily represent those of my employer.
try burning a SafeDisc2 protected image on a new Plextor drive.
No problem. The PlexWriter PX-W2410A does perfect Safedisc 2 copies, at least with the firmware shipped to reviewers.
Of course, SD2 is pretty much broken if you take into consideration BetaBlocker, a program you can use to 'fix' a SD2-protected image prior to burning. Works with any burner.
Belief is the currency of delusion.
It's [EXPLETIVE DELETED] [EXPLETIVE DELETED] [EXPLETIVE DELETED] [EXPLETIVE DELETED] [EXPLETIVE DELETED] [EXPLETIVE DELETED] [EXPLETIVE DELETED] [EXPLETIVE DELETED] [EXPLETIVE DELETED] [EXPLETIVE DELETED]....
/. If the government wants us to respect the law, it should set a better example.
The word on Free Software needs to go out to these kids, to show them that their admirable skill can help make the world a more free place. If they can program a DSP for watermarking, maybe they can help create better speech synthesis for the Stephen Hawkings of the world...
Carefree highway, let me slip away on you.
Why bother with speakers and a microphone? Connect your sound card's (stereo system's, whatever) sound output jack directly to the microphone or input jack. No fuss, no mess.
Just hold the audio up to the light, it's right there.
I think I'll stop here.
Is it possible to manufacture a player that is completely programmable? Like that first personal computer that came in a kit. People could buy a "blank" player and write their own firmware for it. The firmware images could be downloaded and installed so my kit player could be a cd/dvd/cdr/cdrw/something completely different. This would place TOTAL control in the hands of the "consumer" (don't you hate being called that?).
Samsung took back my unlocked bootloader because Google wants me to rent movies. They're both evil.
All digital watermarking does is represent digital data (actually, any type of data, but in this case it they say digital) in an analog form, which is then grafted onto a larger source (the most common example being a music file).
So, just like a modem signal, a water mark is digital data that has been turned into sound. It has to be fairly robust to withstand degradation of the original signal yet still be readable. To be effective as a watermark (in music) it also has to be unobtrusive, i.e. you don't notice it. But it is an audio signal that (if you want it to be effective for things like SDMI) resides within the range of human hearing, or more accurately within the range of common recording devices.
(Very) Simple example: I spell my name in Morse code with a dog whistle that has a frequency just above what the human can hear but within the range of common recording equipment (be they digital or analog) while recording "Push Push in the Bush." I have now crudely watermarked that track.
Now, I can't hear the watermark on the recording that I made, and I make copies of this track I still won't be able to hear the watermark (because it is still just slightly above the audible range of humans, although my dog may howl), but its still being passed on from copy to copy.
If I serially copy it from tape to tape to MP3 and back to tape, assuming that there wasn't too much signal degradation (and if there was the music would sound like shit too) if I played that 4th generation copy and fed the signal through an oscilloscope, you'd still be able to see the Morse code, and my dog will still howl.
Now, how does this thwart copying and protect copyrights? It doesn't. That's not the point. All a watermark does is merely encode data into an analog system that will stay intact from analog copy to analog copy assuming the signal hasn't degraded to much (and like I said a good, robust watermark should be able to withstand a fair bit of degradation). Where the copy protection comes in is that the SMDI folks want to make devices that will recognize a watermark in an audio file and allow or not allow it to be played.
Check out:
.F. Tilik, A.A. Beex. "Encoding a hidden digital signature onto an audio signal using psychoacustic masking", in Proc. 1996 7th International Conf. on Signal Processing Apps and Tech.. pp 476-480
for a much better explanation.
If you don't have anything nice to say, say it often.
- Ed the Sock
Anyone who recovers what Nixon erased from the tape will be sued by his estate under the DMCA for circumventing his encryption scheme. Sorry.
you are correct sir.
No matter how unrealistic it is as a business model, copy protection is still an interesting problem, especially through watermarking.
Inclusion of watermarking code into DSPs is inevitable. CD-R companies, for example, have been eager to embrace similar methods... try burning a SafeDisc2 protected image on a new Plextor drive. Even a perfect data source can just be blocked by hardware, by detecting patterns.
Obviously, some companies will see a way to make a profit by getting around this. Educated consumers will buy hardware without locks.
The question is, will Congress permit anyone to create CD-R writers (for example) in the future that do not have firmware copy protection.
I hope the DMCA was an anomaly, and not an example of things to come.
...that canada's still represented...although the university of waterloo is getting weirder and weirder. take a bunch of nerds, put them in a small town with a beer festival and this is the result, i guess.
anybody (from UW or otherwise) got an abstract for this one?
Canada
University of Waterloo
Intelligent Motion Control Using the TMS320LF2407 Applied to a Six-Legged Walking Robot
Maybe they actually wrote the crack first, then reverse engineered it.
A feeling of having made the same mistake before: Deja Foobar
These things should be sponsored by universities or non-profit organizations like the ACM, not companies. I'd imagine the winning solutions are worth far more than the prizes offered.
No, Thursday's out. How about never - is never good for you?
nope.
First - digital audio contains VERY little frequencies that can't be heard by humans - what would be the point of encoding audio that can't be heard? For dogs? c.f. Nyquist sampling limit. Compact discs are sampled at 45Khz, so 22.5Khz is the upper limit of the sound that they can repro. That is outside of human range, but not by much.
Second - generally these watermarking schemes are very sophisticated. A simple band limiting/passing filter isn't going to do it. You need to do know the specifics of how it's encoded to remove it.... not that it can't be done (c.f. the guys that broke the SDMI challenge). Watermarks contain 'echo's' of themselves, time dependant bits, time independant bits, using one sound to mask another (a human wouldn't hear it), etc etc etc. They're pretty sophisticated.
About 4 years (?) ago at MIT i saw a demonstration of a watermarking scheme where they played music out of some speakers. They held a microphone up to the speaker, and their computer spit out what the watermark was. So the people who keep saying that they'd re-record watermarked audio this way can only do it if they have a recording device that doesn't obey/enforce the watermark. The fact that they've gone Digital->Analog and possibly back to digial doesn't matter a single bit.
Brought to you by the people that gave you the Infinite Sharpness filter for Photoshop (find those completely sharp faces buried in fuzzy films) and the Internet Welcome Screen.
but, the piracy that they're trying to fight wouldn't normally be created with analog equipment...the recording industry realises that people have been taping stuff from the radio or cd/tape to tape for years, the fact that it's so easy to create CD or nearly CD quality recordings with cheap equipment seems to have them running scared...
how about mp3 compression, would the compression destroy a simple hide-it-in-the-highs type of watermark, or are there ways to get around it?
if the signal does remain in the audible range, the only ways I can see of injecting a message without altering the original song would be playing "tricks" with amplitude or duration...both of which would be totally useless for digital-analog recordings and would probably send epileptic or tell every tenth kid that "it's time to kill"......hmmmmm, maybe they've just modified the old subliminal message tricks? hehehe...be prepared to buy every crappy record that comes out.
Their presentation to the judges was very impressive (we presented first, and all I could think while watching theirs was "we're screwed..."). They demonstrated both the addition and detection of a watermark for pre-recorded and live audio. A couple of times, they played out loud just the watermark. It was pretty garbled, but you could definitely make out the content of the original audio. They did it for both vocal and instrumental music; both times, you could make out the lyrics and melody of the original sound. And when combined with the original, the watermark was inaudible (as promised).
But the most impressive part of all is that it was all done in real-time. They watermarked the audio from a local radio station as it was being broadcast, playing both the resulting watermark and watermarked audio with virtually no delay.
I told them in Dallas, but I'll say it again- Congratulations on winning. You guys were definitely worthy competition. With any luck, we'll face each other again next time around.
-patrick
I just figured out how to crack it! Take great quality audio reproduction equipment (Speakers) and then some very high quality audio collection devices (Microphones) and put it all in a proper sound room, or box.
For right and left channel sound you will need to have at least two speakers and at least two microphones and some way of merging the two recordings.
It might be a little far from perfect, but so are MP3's and the sound from the "Cracked" watermarked music.
Wait a minute... Since Sklyarov was arrested for creating a tool to break a protection scheme, does this mean that speakers and microphones are now illegal?
I suppose it is all how you interpret the DMCA...
--
.sig seperator
--
If you ignore the other uses of a tool, does that make the tool less useful, or you less useful?
what was erased on the infamous Watergate tape that pushed Nixon's downfall over the brink. It would be amazing to have this national mystery put to rest
;)
Actually, I heard the blank spot was just noise introduced by a top-secret CIA funded copy-protection scheme from the era.
Hey, at least we got some of the tape! If all this copy-protection shit had been introduced 30 years ago, we wouldn't have the tape AT ALL.
Heehee.
"Old man yells at systemd"
And they want us to PAY for that? I think not!
1) Get a team of 10 audio researchers
2) Get one persone working on a watermarking technique
3) Get the other 9 try to break the technology
4) If (technology broken) then goto 2) with another researcher
...and loop forever until you realize that you can't make a watermark that cannot be broken.
Opus: the Swiss army knife of audio codec
"Wonder how long it'll take before someone cracks this scheme.."
I wonder how long it'll take before someone gets arrested for cracking this scheme...
Lets be clear here....... these watermarks are in the AUDIO - not in the specific digial representation. They are designed to survive thru generation loss, going from digital->analog->digital again (and maybe going form CD -> MP3 encoding back to CD) or any of a thousand variations.
If i make a recording that is a voice saying "DO NOT DUPLICATE" and every recording device on the market 'listens' for that sound, and if it hears it refused to record, the fact that you play via analog an re-record won't matter. The only way around it would be to remove the "DO NOT DUPLICATE". Now what these schemes do is exactly that - except they do lots of complicated and tricky ways to hide "DO NOT DUPLICATE", along with trying to encode it in the most robust fashion even if the signal gets modified. And in the same vein, they try to hide the signal in such a way that removing it would cause too much undesireable in the underlying music. The Stanford guys that broke the SDMI challege showed that it was possible to remove, but removing it ISN'T simple.
In your above 'cracking' example, it would only work if your recording device (not the microphones, the actual machine that persisted the music coming in) didn't respect/obey the watermark.
You haven't cracked anything. You've only demonstrated that you haven't a clue on how this technology works, nor a simple basic understanding of signal processing theory. If you ask Santa nicely he might bring you a spectrum analyzer for chrismas.....
Why go that far, just tap it off the output of the D/A converter, or better yet (if you're a pro) tap it off the input to the D/A converters. 8)
A feeling of having made the same mistake before: Deja Foobar