I shed some light on the mystery of Watermarking...

[Thomas_J]

Using a 16-bit 44khz .wav format file of approx. 8 seconds of silence, I wrote a piece of software that generates a pattern based on the length of the file, and scatters out bits and pieces of the watermark in the file, according to that pattern.
The serial number was 7 digits and was embedded randomly into the file, using a randomized pattern, later stored in a \"key-file\". This way, the only way to decipher the watermark within the sound file, is to use the keyfile which holds the randomized pattern data used in that particular .wav.

I then mixed some music I\'ve done with the watermark datafile. The result is inaudible. And even in mp3-format I\'m able to extract the watermark data from the .wav and read it.

I imagine this must be how it is done. Probably a bit more complicated, but the general idea seems to be the same.

Here are the sound files I used in my test:

http://finjaco.home.online.no/Test-Watermark.mp3
(this is the generated digital watermark)

http://finjaco.home.online.no/Test-Mix_And_Watermark.mp3
(this is the sound sample with the embedded watermark in - can you hear it?)


Thomas

[Ned Bouhalassa]

Interesting... in theory (for me). You see, I can\'t recognize the embeded watermark... because I can\'t hear the original watermark in the first place! [img]images/icons/shocked.gif[/img] Is it above 16k?

[MartinL]

Interesting...

But what happens if you add reverb, compression, EQ, filters, etc. to your mix? The data containing the watermark will probably be changed or disappear...

Martin

[Bruce A. Richardson]

Originally posted by MartinL:
Interesting...

But what happens if you add reverb, compression, EQ, filters, etc. to your mix? The data containing the watermark will probably be changed or disappear...

Martin

<font size=\"2\" face=\"Verdana, Arial\">People who have developed watermarking schemes say this is not the case--that the watermark will even survive a D/A/D conversion. I\'m sure that one could indeed devise some complete insanity of processing to eventually break it, but at that point, the marketability of a derivative product would be eliminated anyway. In the case of an orchestral sample, it\'s pretty unlikely that a user (or mis-user) would process every single note to that extent.

The major argument against watermarking has not been the detection of the watermark, but the difficulty of prosecuting it as a criminal matter. More likely, the property owner\'s case would land in a civil court, where it would be time consuming (read: $$$$) to mount.

However, I still think it is worthwhile, even if it is simply to force accountability. Right now, many libraries trade in the retail channel with voluntary registration. I think any library in the retail channel should have a tear-off card that the retailer fills out upon purchase, requiring valid I.D. from the purchaser. All internet sales should require full identification. ANY receiver of libraries should be documented, period.


From my personal perspective, it is difficult for me to accept more invasive copy protection as a norm, when these kinds of basic identification methods are ignored. There IS accountability in a person knowing he possesses a library with a discrete watermark, and that he has had to provide contact/ID information in order to execute the purchase.

There has to be some real or at least perceived consequence to sharing. That\'s why it is rampant. In the retail channel, where you were likely to have been offered \"sounds\" when purchasing a hardware sampler in the 90s, you have a lack of education. In the user channel you have an even more dramatic lack of understanding. There is a class of user who \"shares\" because he is genuinely surprised he cannot do as he wishes with what he \"bought.\"

On average, people here are a highly educated group of consumers, and look at the different responses to licensing agreements even here!!

When you demonstrate and demand accountability at every step in the chain, you create an awareness of the licensing structure and its importance. People need this education. They need to know that if they share with a friend, that friend might share with another one, who might share with five, then suddenly YOUR copy will start showing up on the \"crack\" channels. A person should be willing to accept the fact that he will be sued out the ying-yang if his ID\'ed copy of a library shows up on Al Gore\'s fine internet. And that person SHOULD have the ever living snot sued out of him, for the retail price per every download that can be traced of the pirated product. The internet is a double-edged sword. Anonymity allows mischief, but the transactional nature of file transfer then becomes the other edge of that blade.

Will a sample vendor ever be able to collect damages for 1000 copies of VSL, for example? Probably not. But a successful lawsuit would certainly ruin that musician for life, not to mention blackballing his candy butt in our ever so tiny community.

In short (ha [img]images/icons/rolleyes.gif[/img] ), I think if we as users do not demand accountability, in each other as well as in the development community, we are ALL screwing ourselves. We\'re on the same side. Any responsible person wants authors of sampleware and software to get the profits they deserve, because that keeps the wheels turning and the state of the art advancing. At the same time, any responsible developer wants his tools and content to be as freely usable by legitimate licensees as is possible. I am suggesting that we have not nearly explored the potential of aggressive identification techniques. And without any evidence of how those work, how can we know what impact they might hold? Can we really, honestly look at the balance sheets, and see proof that aggressive identification of licensees wouldn\'t pay as much in sales as encryption/challenge-response methods?

I think so much of this is a perception problem. I think there is a disconnect, and it is that people really don\'t understand that the copy they shared with a friend IS the copy that turns up on the web. The \"casual pirate\" is a total myth. There is no such thing. Until the user population is educated and inundated with this reality, and until that causality is acknowledged by the mainstream of the user base, this copy protection \"war of wills\" is just like an escalating cold war.

OK, sorry. Two cups of coffee, and I\'m already wearing the ends off my fingers.

I will really end with this: Ernest Cholakis would be a great resource for a watermarking discussion. I think he has researched it mightily. I\'m glad Thomas raised the point, though, because I truly do like the idea of identification-based tracking of sampleware/software. It has no downside, and any time you increase accountability, you increase respect on all fronts.

[Alaia]

Watermarking news:

http://www.iis.fraunhofer.de/amm/techinf/water/index.html

[Hans Adamson]

As I see it, there are many practical questions to address with a \"serial number system\" on samples.

1. Sample creation is, believe it or not, a very small niche. The cost of investing in watermarking technologies will have to be in proportion to the revenue of sample sales.

2. The actual process of giving each burned CD an individual number, and keeping track of which CD has which number, has to be automated in some way. Does this process require re-writing a 1-2Gb instrument file to embed the individual serial number for each copy? If this cannot be an automated process, a person would have to be hired just to do to perform this task.

3. If the two issues above are resolved and each CD has now an individual number, how can I use it to prevent pirating? - In the pirate scenario, you would find your sample available from a pirate\'s website. In order to establish the serial number of the pirate\'s copy we order a copy from the pirate to an inconspicuous address. The pirate is located in, for example, Russia. The sample ID tells me that the sample was sold to Mr. X in Hong Kong. Which international jurisdiction will empower me to take action against Mr. X? Presume there is an judicial jurisdiction like this. Who gives me access to this procedure? The cost of hiring a lawyer versed in international copyright to pursue the case will probably cost as much as the total revenue from sales of the library up to date. Presume that I mortgage the house to get the guy. It turns out he is a 17 year old hacker with no assets. Where does that leave me?

4. How to stop the sample from wandering from one user to the next by sample sharing? Which bureau can I hire to monitor the flow of music in radio and in movies? And what would be the cost of a monitoring apparatus like this? If there is an easy answer to all this, there is still the question of what standard of evidence is required in court to convict someone of using a sample not registered to him. And in the end it comes down to whether this effort even will be possible to pay for with the revenue from creating and selling samples for GigaStudio.

When there are clear convincing answers to these questions about procedure and logistics, it may start to look like an alternative. But then again there is the question: When pirating is bringing the largest media companies to their knees, with all the resources they assign to fight this - Why would I be successful going down the same road?

Hans Adamson
Art Vista Productions
http://www.artvista.net/

[Bruce A. Richardson]

Hi Hans,

I don\'t think you go for the person using the pirated samples. Too much monitoring. You don\'t go for the 17 year-old hacker punk. No payoff. You take your time, and wait until someone with a huge media name\'s \"copy\" lands on the pirate channel, then you hire a cutthroat lawyer and take him down so hard he can\'t make noise when he farts. You simply download every crack until you find your poster child, and you make an example.

After all, don\'t you want the the guy who does the initial \"sharing?\" That\'s the root of the problem.

[Nick Batzdorf]

1. Sample creation is, believe it or not, a very small niche. The cost of investing in watermarking technologies will have to be in proportion to the revenue of sample sales.

<font size=\"2\" face=\"Verdana, Arial\">I would have thought the same technology would be applicable to songs and albums.