We've sent a verification link by email
Didn't receive the email? Check your Spam folder, it may have been caught by a filter. If you still don't see it, you can resend the verification email.
Started July 20th, 2008 · 35 replies · Latest reply by dobroide 15 years, 8 months ago
Hey guys,
Interesting. This was of course bound to happen at some point. I used to regularly find people selling my audio effect plugins on ebay (i.e. they are free on my site!).
People letting 3rd parties sell content on their site have no excuse. I.e. if AudioSparx has content on their pages which has been put there by a 3rd party and sell that content, they have to be 100% sure that content is not illegal.
I have contacted the website, and will contact creative commons as well.
- Bram
Were the owners of audiosparx really concerned about the issue one might think that they would delete the whole profile of USAMA and all sounds attributed to them. Perhaps a petition signed by freesound users would help? I realize this won't in any way prevent them from creating a new nick and repeating the procedure, still IMO this would be a clear message to the ripper and the community concerned.
Dobroide,
You're right. After reading your post I come around to your point of view. Audiosparx is stealing those sounds just as much as the vendor is because they are making money from them and also extending their license to cover the vendor.
A couple of possible remedies.
They require their vendors to post $100 bond in order to be registered. If a copyright holder finds their work being sold by someone else, they get the $100 and any revenues that have been charged for that work. The vendor has to repost the $100 bond. This makes it in the financial interest of both audiosparx and the vendor not to sell materials they do not have copyright to. And the copyright holder gets some compensation out of it.
Another option is to leave all works on freesound as they are, but also automatically register them on audiosparx as well. If downloaded from freesound, they are under the creative commons license. If purchased from audiosparx, they are under the audiosparx license and the copyright holder gets the revenue. Then people who take the trouble to look can use the work under creative commons, but those who don't at least pay the copyright holder instead of a thief.
I think it is ironic that the people shouting the loudest about copyright breaking, the movie studios, might have broken copyright themselves by using sounds from audiosparx.
wisslgisse.
.
.I think it is ironic that the people shouting the loudest about copyright breaking, the movie studios, might have broken copyright themselves by using sounds from audiosparx.
Thank you, wisslgisse. Yeah, not only ironic, it also gives you something to think about. Years ago a guy said all property is theft. Of course the statement is contradictory (as "If no property is rightfully owned, that is, if nothing is property, there can be no such concept as “theft”; http://en.wikipedia.org/wiki/Property_is_theft ), but we all get the point, I guess And by the way, I was thinking of this kind of problems when, several months ago, I wrote about digital watermarking of sounds in this same forum. People prone to claim authorship on other's samples should know they are taking risks
Anyway, I for one will take this lesson home: let's all keep flooding the internet with free content of good quality, that's our best move. At present, each time I watch some amateur video from Youtube being reproduced on TV, I smile to myself. In a few years time, who knows?
D
I wonder how hard it would be for the freesound site to add such a digital watermark that said freesound, or any other symbol, over and over again in the 20 KHz to 40 KHz range, or the 0 to 20 Hz range to every sample uploaded.. Small effect on the sound but any spectrograph shows it clearly. Done right, it would probably be difficult to remove too.
I vaguely recall those posts. I'll have to go back and read them.
http://www.audiosparx.com/sa/play/play.cfm/sound_iid.32552
An orchestra playing Rimsky-Korsakow's Flight of the bumble bee. Artist: Usama. Yeah right. 30 seconds cost only 179 dollar. What a bunch of crooks.
Thanks Bram, you show you care for your users.
Other sites that allow selling user's content have manual review (such as revver.com, videosite). After all they too make a profit of the content. If they make profit of illegal content, they're also responsible for that. They review for their own well-being. Audiosparx should do so too (btw, YouTube should too, even though it's an excellent source to watch some old music videos).
Anyway, I for one will take this lesson home: let's all keep flooding the internet with free content of good quality, that's our best move. At present, each time I watch some amateur video from Youtube being reproduced on TV, I smile to myself. In a few years time, who knows?
Thank you so much for saying that, Dobroide! And watermarking is sensible, too.
wisslgisse
I wonder how hard it would be for the freesound site to add such a digital watermark that said freesound, or any other symbol, over and over again in the 20 KHz to 40 KHz range, or the 0 to 20 Hz range to every sample uploaded.. Small effect on the sound but any spectrograph shows it clearly. Done right, it would probably be difficult to remove too.I vaguely recall those posts. I'll have to go back and read them.
Digital watermarking (of audio, pictures, video, etc) has been an active research area in computing security and cryptology for the last decade or so. Unfortunately, unless I'm mistaken (and I could well be; I've not been doing research in the area for a couple years), the current state of the art has not yet produced a robust audio watermarking scheme that's both inaudible and infeasible for an attacker to remove. It's a very hard problem. My sense is that the advantage will tend to go to the attacker here. Google "information hiding workshop" to get a sense of current work in this area.
And I strongly agree that the most effective and practical countermeasure to the kind of "piracy" being discussed here is indirect rather than technological or legal. Flood the 'net with high quality, freely usable content (that's why I'm here, for one...)
-matt
mab wrote:
Digital watermarking (of audio, pictures, video, etc) has been an active research area in computing security and cryptology for the last decade or so. Unfortunately, unless I'm mistaken (and I could well be; I've not been doing research in the area for a couple years), the current state of the art has not yet produced a robust audio watermarking scheme that's both inaudible and infeasible for an attacker to remove. It's a very hard problem.
I read the articles in the forum discussion of digital watermarking, and followed many of the links. Bram had posted a google search there. I followed it and came to this link that seems to be a robust scheme. I am *not* an expert in this area so I might have missed the obvious but it seemed to cover the bases very well. The neatest thing was that the watermark varied in amplitude with the audio and the amplitude of the watermark was calculated so that the tone was masked by the audio. They have an ingenious method for preventing people from stealing a watermarked work and calling it their own as well. All in all, it looked very good. They tested it pretty hard and the results were impressive..
research.microsoft.com/users/binzhu/Papers/audio2.pdf
That technique, like virtually all audio watermark techniques, uses a common secret keystream for both watermarking and watermark recovery. That is, in such schemes, you can't tell someone how to see if the watermark is present without also telling them everything they need to know to remove it. That's fine for some purposes, but probably not for the purposes under discussion here.
Also, these watermarking schemes exploit perceptual coding techniques (as MP3 does) to be inaudible only under normal listening conditions. They are not guaranteed (and are unlikely) to remain inaudible under the kind of heavy post-processing that samples here are intended for.
Unfortunately, it's just a hard problem.
-matt
soundhead
Most thieves are likely to be too lazy to remove watermarks...
Not that I'm advocating watermarking every sound.
I agree, too lazy. Watermarking probably is overkill, but some form of label would be nice. Like the stamps used to mark books in many libraries - I think they call this an 'ex libris' ?
Something that can be added automatically at uploading, some form of hash, a unique combination of size-date-time appended at the start/end, or... I don't know
Thank you for the feedback
D
Although it is not infallible AT ALL (adding one sample at amp zero will make it fail), we calculate md5 sums of all audio files on freesound. This is to prevent people from uploading exactly the same file twice. You have probably seen this happen to you.
But there are other ways to detect verbatim (or very similar) copies. This is called fingerprinting rather than watermarking and is a much more robust system. Basically it works like the "search similar" on freesound. Say we would have access to the audiosparx database, we could calculate the most similar sound on freesound for all the sounds on audiosparx. If the distance between that sound, and a sound in freesound is VERY small, then it's probably a copy. This even works when the sound is compressed in some way.
I will ask audiosparx to give us access to their database and see if we can use this process to eradicate all freesound source material from their site.
- Bram
Bram
Although it is not infallible AT ALL (adding one sample at amp zero will make it fail), we calculate md5 sums of all audio files on freesound. This is to prevent people from uploading exactly the same file twice. You have probably seen this happen to you.But there are other ways to detect verbatim (or very similar) copies. This is called fingerprinting rather than watermarking and is a much more robust system. Basically it works like the "search similar" on freesound. Say we would have access to the audiosparx database, we could calculate the most similar sound on freesound for all the sounds on audiosparx. If the distance between that sound, and a sound in freesound is VERY small, then it's probably a copy. This even works when the sound is compressed in some way.
I will ask audiosparx to give us access to their database and see if we can use this process to eradicate all freesound source material from their site.
- Bram
Thank you for the input, Bram, that would be great.
Now, psychology is very important, you know... so the simple knowledge that Freesound samples are (or can be) fingerprinted would have a double positive effect: that of deterring lazy thieves, and that of comforting honest users (which otherwise can be discouraged to contribute).
Certainly, it would do no harm reminding users/visitors at the main freesound page that samples are/can be marked