Comments on: Stealthy Audio Watermarking: DRM

By: Method

Method — Sun, 16 Sep 2007 10:59:50 +0000

1) Most of us will stick with our trustworthy encoder apps and codecs that we’re already happy with.

2) With the prevalence of rootkits, trojans, viruses, etc. – It doesn’t take a big leap of imagination to imagine that someone could get their hands on your own personally-encoded media and distribute it… all without your knowledge.

Fortunately, they’re a long way off ever having a 100% solid case against anyone, especially all those on Windows boxes where Microsoft’s slack-ass security can make you a file-sharer or even password-sharer without having ever known about it.

By: Reader's Write

Reader's Write — Sun, 16 Sep 2007 04:09:36 +0000

It does not matter if they crack this or not. People want to be able to do as they please with their music. Once any form of DRM is found on it, it makes that track much less for people to want to purchase. And this IS just another form of DRM!

Look at itunes. That is the biggest market the industry has for selling music….yet it is only a FRACTION compared to the people that use p2p.

So if they ever want to create a market that sells to the masses and turn them away from p2p, this will fail.

By: Your friendly EE student

Your friendly EE student — Sat, 15 Sep 2007 15:58:27 +0000

The human ear is very sensitive to changes in frequency, however it is very insensitive to changes in magnitude. We would never notice if the note at 11kHz was multiplied by 1.0001, 1.0002 or 1.0003 much less if one instance of the note was multiplied by this value. However a machine looking for this note and comparing it to the original magnitude would have no problems determining the slight alteration. Us, without the original recording would never notice nor would we have anything to compare it to. It is more than possible to imperceptibly mark a sound recording so that none would ever know.(without information about the cipher) The question becomes who cares?

If the music is prepared for mass market on a physical medium then the usefulness of 100,000 individual watermarks is nil. One could give away, sell, or otherwise no longer be in possession of the media.

If it was originally distributed as a download, then as stated earlier, AM noise could be introduced that would also need to be small enough as to be imperceptible to the human ear. Normalization would also be effective but that would almost certainly degrade the music quality.

The only people this would effect are those given non-transferable copies directly traceable to them.

Here’s the kicker, this is only one possible scheme for an undetectable watermark. There are many, many ways to encrypt information within a data stream. Like it or not, the only reason restrictive DRM doesn’t work is because we are already provided with the decryption key. Here, they will never release the key.

By: Reader's Write

Reader's Write — Fri, 14 Sep 2007 17:40:48 +0000

âI donât believe taht a water marking will be totally inaudibleâ yes it can we humans can only hear a small range of frequencyâs the watermark may be heard by your pet dog but mot you.

No because only high end equipments ecxeed (slighthly) the human hearing range. if you temper with this near human hearing range region of the spectrum the watermarking itself will be inaudible but it will generate distortion of modulation that you will be able to notice on a full size stereo. This will appear like a more “muddy, fuzy sound” similar to what you get when sound is compressed in a lossy format. (Just pluq your ipod to your full size stereo to see what I am talking about.) Moreover in this case It would be easy to filter it out without affecting the sond too much while getting ride of the distortions. They could play with slight changes in amplitude that will be unoticable but there again it will generate distortions.

The bottom line is that the water marking need to be detectable to be useful overwise what is the point of it. If someone can detect it anyone can. Computers are double edge swords because their speed can be used to increase the level of encryption but the same speed can be use for the decription at the same level. With hudred of thousand of minds out there for whom cracking stuff is an hobby no wonder cracking anything take a most few months most often few days.

By: Reader's Write

Reader's Write — Fri, 14 Sep 2007 16:36:48 +0000

This is rather more technical than any other scheme I have read about. Yes, of course the first thing to do would be to apply a mid-pass filter to the digital song file that only passes frequencies between 40Hz and 18kHz. But from the concepts written above, it seems they are *slightly* modifying the actual musical tones themselves. I’m not so sure the ‘golden ear’ test would fail in this instance. For example, if a song had a 2 second guitar riff with several fast attack/decay cycles, followed by a long ramp up of a singer’s voice, their technique would *barely* extend some of the peak notes on the riffs (or would add a slight echo) by a few microseconds and embed a few bits of data in that sequence. Their algorithm could be based on a statistical sequence of certain “loud” and “soft” passages passing a preset value, thus detecting software/equipment would know where to look for the data. Attempting to modify those attack/decay values without changing how the music sounds would be quite difficult. I doubt it is so dumb as to use the inaudible spectrum to embed data since that is trivial to filter out. Note their mentioning of “digital noise”, almost an oxymoron, but very real (less dramatic but more correct would be to call it “unused digital bits” instead). This “noise” may (but usually doesn’t, depending on the decoder software) affect the sound quality of a song, and statistically manipulating it in known and predictable ways would be one method to use if you want to watermark a file. To completely remove their proposed scheme one would have to digitally-reconstruct the music file completely. Yuck. Perhaps a better approach would be to apply several filters on top of each other that did different things, such as very slightly extend or reduce the attack/decay values or add a very slight echo to notes of a certain frequency, etc. Each user would need to “play” with the files using their own ears to determine how much messing around is still acceptable. All that is required is to prevent the complete detection of the watermark, not remove it.

By: Reader's Write

Reader's Write — Fri, 14 Sep 2007 16:02:20 +0000

I guess you could introduce your own AM noise across the entire spectrum to scramble it but that might degrade the music quality.

By: Reader's Write

Reader's Write — Fri, 14 Sep 2007 13:39:16 +0000

Fortunately, impossible to crack DRM is not possible. However, for the same reason, impossible to detect watermarks are possible.

By: Reader's Write

Reader's Write — Fri, 14 Sep 2007 13:24:31 +0000

There could also be very slight amplitude modulation within the audible range…
I cannot actually think of a good way to remove that without comparing it to the original recording….
Oh well, I am sure there are more clever electrical engineers than I.

By: Reader's Write

Reader's Write — Fri, 14 Sep 2007 10:46:53 +0000

So basically all thats needed to crack this DRM is recording equipment (or software) that simply ignores any frequency outside the range that we can hear?

It’s just too easy. The snake oil merchants are getting more and more blatant and the recording industry more and more desperate obviously.

By: Reader's Write

Reader's Write — Fri, 14 Sep 2007 07:11:06 +0000

“I donât believe taht a water marking will be totally inaudible” yes it can we humans can only hear a small range of frequency’s the watermark may be heard by your pet dog but mot you.

By: Reader's Write

Reader's Write — Fri, 14 Sep 2007 04:31:19 +0000

Beside we want more quality not less. I don’t believe taht a water marking will be totally inaudible at least not with the curent equipement. Moreover if the equipment can remove the water marking so can anyone else.

This is another fraud. But if the parasites pay for that it will be few hunderd millions less to sue the citizen. Good!

By: Reader's Write

Reader's Write — Fri, 14 Sep 2007 04:27:14 +0000

I am sure that if Microsloth show up with their water marking in public It will be cracked in less than a month.

Beside that who is going to use the microslosh crap to messup their audio files? May be Shitman?

By: Reader's Write

Reader's Write — Thu, 13 Sep 2007 21:08:12 +0000

to be honest this is the best form of DRM they could have. it lets people do whatever they want with the digital file but if illegally shared is traceable like a post mark.

whether it could be broken yes but so can any analog output.

By: Patrick

Patrick — Thu, 13 Sep 2007 20:18:45 +0000

SDMI made claims about the robustness of their watermarking technology as well, and even went so far as to challenge people to break it.
Ed Felten and a few CS grad students did just that. Anything that can be added can be added to a file can be removed or scrambled.