Watermarks in libraries and plugins

[tauben]

Hello,

I've only found this thread from 2012 so far.

viewtopic.php?t=356442

The general consensus in that thread is that audio watermarks would harm the quality of the samples so they would not implement them in the wav files. But they would implement serial numbers and account ids in the "header" or exe files etc.

I remember, some years ago, I am not sure what brand and I can't remember exactly how they wrote this, I think it was a Symphobia library. They wrote something like you'd have to register all projects you are using the library in at the library creator (in this case this would have been Sam Projects). They also clearly declared that if you purchase the product you agree that they are using watermarks in their sample library blabla, I can not remember how they were terming this. But the statement was pretty clear that they would definitely use watermarks. They propably found out that such statements do not lead to more customers and changed strategies or laws changed such that they do not have to declare it anymore, not sure.

Couple of questions regarding this:
- Anyone remember if the above was the case?
- Is it possible to implement reidentifiable watermarks in audio libraries without harming the quality of the library?
- Is it possible to implement reidentifiable watermarks even in plugins (namely in the resulting track that went through the plugin) without harming the quality of the result?
- Are they really identifiable even after mastering process?
- Do they have to declare that they are doing this by law or can they do it without declaring it?

From my technical understanding, I would tend to answer 2, 3 and maybe even 4 with a yes. Technically, why shouldn't it be possible to implement a non audible signal somewhere at 19.8 khz that can be decoded? Noone would hear it and cut it out anyway. The question is just do they do it, just my thoughts.

What do you think?

Best Regards

[BertKoor]

I've only found this thread from 2012 so far.

How did you search? I found a bunch more, eg:
https://audiowatermarking.info/customers.php

It is in the nature of watermarking that even the fact of watermarking must be kept secret.

Can it be boycotted? Is it perhaps even a placebo product? Interesting questions. I have my doubts, but it's not my problem.

[audiojunkie]

Watermarking is great! I would take watermarking over any other sort of copy protection (except serial number or keyfile) if possible. It is entirely possible to watermark samples and even software in such a way there there is absolutely no effect on the sound. For samples, the easiest way would simply place the water mark outside of the rang of human hearing. For software, you would simply add the information prior to compiling. Either of these ways will allow the software (or samples) to work better than any form of DRM, and it wouldn't hurt the honest at all. Of course, pirates would be able to find ways to get around the watermarks, but they don't purchase the software anyway, so why make it harder for the honest? Also, watermarking works better in Linux than any of the Challenge/Response or ILOK DRM crap. It's a great all around choice.

[mjolnir]

during the mid nineties there was an JASA or AES article about frequency hopping (almost like FM or FSK) watermarking of audio within the actual audio data (not the header) it was some kind of narrowband echo or something like that, but it jumped around to different frequencies so that it couldn't be easily detected or erased. If i remember correctly, the algorithm and technique used was resilient against DSP / gain attacks somehow.

I can't remember if the article was in the Journal of the Acoustical Society of the Americas (JASA) or if it was in the journal of the Audio Engineering Society. Somebody at GearSpace.com might know.

i think one or all of the inventors was Japanese. I think there were four collaborating inventors.

[audiojunkie]

during the mid nineties there was an JASA or AES article about frequency hopping (almost like FM or FSK) watermarking of audio within the actual audio data (not the header) it was some kind of narrowband echo or something like that, but it jumped around to different frequencies so that it couldn't be easily detected or erased. If i remember correctly, the algorithm and technique used was resilient against DSP / gain attacks somehow.

I can't remember if the article was in the Journal of the Acoustical Society of the Americas (JASA) or if it was in the journal of the Audio Engineering Society. Somebody at GearSpace.com might know.

i think one or all of the inventors was Japanese. I think there were four collaborating inventors.

Oh, clever!! Similar to WiFi signals... I wonder how they made it undetectable to the human ear...

[mjolnir]

i think this might be helpful: (keyword search: frequency hopping audio watermark)

https://www.bing.com/search?q=frequency ... +watermark

notice the results in bing.com
top result today is from IEEE from 2010-2011 or something like that.

[DSmolken]

Putting a watermark in dither by making some bits of the dither specific to a customer is totally doable. You'd just need to do it in one sample in a library.

[BertKoor]

Putting a watermark in dither by making some bits of the dither specific to a customer is totally doable.

But such a watermark will surely be lost after mixing / mastering. Even a simple DA-AD conversion will render it away.

[HAL76]

You can also place the watermark in a small area after the end of the loop that is not played at all.

Very small watermarks are also possible.

The biggest problem with this is the effort it takes for the manufacturer to test and the ability to do so in general.

I already took a closer look on some Samples from Roland in Wavelab. In this case they were exported directly from Fruityloops and there was no data in the header. No file informations at all. I think I know why they don't put so much emphasis on it there.

From my perspective, the question arises as to whether copy protection such as watermarks, tracking and, if necessary, the enforcement of interests makes any sense from an economic point of view.

[DSmolken]

Putting a watermark in dither by making some bits of the dither specific to a customer is totally doable.

But such a watermark will surely be lost after mixing / mastering. Even a simple DA-AD conversion will render it away.

Yeah, this would be more for finding out whose copy ended up in a pirate copy etc, not for finding out whether a specific sample library was used in a track.

[Jaylewisham]

If I spend my money on a sound product, I do not want any water marks in my music.

[burkek]

Watermarking is absolutely pointless. You don't have the time, the money, or the technical ability to check every published audio file and video, then locate contact information for the person(s) who used the sample or VST output, then check validity of them being a licensee, nor the time, money or lawyers to litigate misuse - sometimes internationally.

[Shabdahbriah]

Found this interesting though:

https://audiowatermarking.com/blog/defe ... -emerging/

[HAL76]

Watermarking is absolutely pointless. You don't have the time, the money, or the technical ability to check every published audio file and video, then locate contact information for the person(s) who used the sample or VST output, then check validity of them being a licensee, nor the time, money or lawyers to litigate misuse - sometimes internationally.

At the end of the day it would probably turn out that, especially in legal disputes at an international level, the full amount of damages would go to the lawyer - who would then be happy about this job creation measure, if one can be found at all.

[seafire]

If I spend my money on a sound product, I do not want any water marks in my music.

Don't worry, there aren't. It's fake news to scare gullible users and crooks.