Joho the BlogYouTube's automated copyright filter - Joho the Blog

YouTube’s automated copyright filter

YouTube uses a Content Identification tool that enables copyright holders to find instances of copyrighted material so that they can then issue a takedown notice, “moneytize” it, or track it. This includes identifying copyrighted material on a video’s audio track and automatically muting it. There’s no mention in Google’s discussion of Fair Use exemptions.

Mark Smitelli has poked around at the system, uploading copies of the copyrighted song “I Know What Boys Like,” sonically altered in various ways: compressing or expanding the time, lowering or raising the pitch, adding noise, etc. Mark runs the complete results, but to roughly summarize: Altering parameters more than 5% often seems to fool the Identifier, and using less than 30 seconds also seems to let the clip slip through the rule-bound robot’s shiny little nets. Playing clips in reverse confused the Identifier, but stripping out everything except the vocals did not.

Using a clip for as satire [later: I probably got this wrong] or political commentary undoubtedly wouldn’t keep it from the Identifier’s snares, although such use is likely protected and non-infringing. The Identifier, unsurprisingly, seems to be a poor reader of human intention. [Thanks to David Abrams for the tip.]

20 Responses to “YouTube’s automated copyright filter”

  1. […] it.” Here is the EFF complaint. David Weinberger has posted some details on Google’s Content Identification tool, which is being used in the shotgun […]

  2. My understanding is that ContentID isn’t as innocent as I read your blog to suggest. It’s not just that it “enables copyright holders to find instances of copyrighted material so that they can then issue a takedown notice…” If that were the case, at least there would be some human review before the takedown notice was sent. Instead, ContentID can be set to automatically block the content – no notice, just automatic filtration. YouTube says it pretty plainly here: (“Fully Automated. Once you’re set up, Audio ID and Video ID identify, claim, and apply policies to YouTube videos for you.”) and also at the link your blog provides to the Google Content Identification Tool site.

    I appreciate you blogging on this important issue. I think it’s important to get the details right exactly because the lack of human involvement in automated blocking is a significant difference from a human-sent DMCA notice.

    Thanks for your post.

  3. I think that, in the end, the Downfall parodies will remain a part (And a significant part) of YouTube’s culture. If people want them up, they will find a way. It is only a matter of time before a DMCA counter notice challenge makes it clear that we have the right to do this sort of remixing.

    YouTube is tarnishing its brand image by not correcting its Content ID system.

    I offer an analysis of the political and cultural significance of the Downfall parodies in my book, Watching YouTube: Extraordinary Videos by Ordinary People (University of Toronto Press, 2010).

    Dr. Strangelove
    University of Ottawa

  4. It’s not clear in what year the tests of “I Know What Boys Like” were done (the article references the ’90s), but all the videos were removed for infringement, as well as the account, and it’s entirely possible their process has been improved (as they purport to do).

    Also a correction, you say “There’s no mention in Google’s discussion of Fair Use exemptions.” In fairness to them, in the first Google link you provided, they talk both about if the match is wrong (human intervention) as well as the DMCA process (with a link to the counter-notification).

  5. You say: ‘Using a clip for as [sic] satire or political commentary […] is likely protected and non-infringing.’

    I was under the impression that, unlike parody, arguing for fair use exemption on the grounds of satire makes for a very weak case. Can you cite any actual cases where a satire as fair use defense has been used successfully?

  6. Misterfric, nope, I’m not a lawyer, so I very likely got it wrong. Thanks for the correction.

  7. BSatLaw, I didn’t mean to imply the process is innocent. On the contrary, the fact that it’s robotic and automated is a real problem since fair use cannot accurately be ascertained by robots; I meant the snarky language about robots to imply that. Sorry that I failed to make that clear.

    RJFerret, I’m standing by what I said. The pages I linked to tell you what to do if the material you posted was u>misidentified. But, fair use covers cases where the identification was accurate – it really was a snip of “I know what boys like” – but your use of it is not an infringement. I would prefer that Google’s materials help educate us about our rights.

  8. Thanks for your reply and clarification, davidw.

  9. […] Joho the Blog » YouTube’s automated copyright filter This probably explains why many videos, particularly songs contain bits of dialogue/fillers before & after the primary content. (tags: youtube copyright filtering tools google) […]

  10. […] this it is dubious how long these clips will continue to circulate… var a2a_config = a2a_config […]

  11. […] Joho the Blog » YouTube’s automated copyright filter […]

  12. […] YouTube’s automated copyright filter “Using a clip for as satire or political commentary undoubtedly wouldn’t keep it from the Identifier’s snares, although such use is likely protected and non-infringing. The Identifier, unsurprisingly, seems to be a poor reader of human intention.” […]

  13. YouTube’s content identification tool is not perfect. The proper route for people to take is to use legal music to begin with. People who use music with proper permission from sites like don’t have to worry about how to fool the system.

  14. […] Joho the Blog » YouTube's automated copyright filter […]

  15. […] YouTube’s automated copyright filter […]

  16. @Scott — AAMOF, users have had legit, royalty-free music, especially music they have paid to license, blocked via Content ID too. No music is completely safe, particularly in the case of warhorse classical recordings that may sound the same, or may have been re-released by multiple labels.

    In the long term this may prove an embarrassment to a number of record labels that have sold the same music “exclusively” more than once, but for ordinary YouTube users it’s more like a game of roulette, where the only eventual good outcome will happen if you have perfect record-keeping of your own videos, and are articulate enough and lucky enough to submit a coherent DMCA counterclaim that is not merely ignored by the label or other body that made the original DMCA takedown claim by robomatching.

  17. You really don’t have a right to steal something you didn’t create actually. The law may have given you the right but you are still stealing. You lack the creativity to come up with something original and have the arrogance to stake a claim on the right to take what someone else has created and use it how you wish, all in the name of comedy. That’s pretty sad.

    In the end, you are still a thief.

  18. Yeah, how about this? I have just encountered an issue with a classical recording. A sheet music publisher has somehow convinced Youtube that it is the copyright holder of a particular 19th century classical work. Now, this particular publisher does have an edition, but so do a dozen other publishers. Also, the work is in the public domain (and public domain sheet music is freely available). I posted MY OWN recording of such a work. I did not use their edition. This scummy company is POSTING ADS and getting REVENUE ILLEGALLY from MY recording!! How dare they even file a copyright with Youtube. Sure, they’re entitled to revenue from the sale of their sheet music, and legitimately hold a copyright on any introductory or analytical prose in their edition, but the work is in the public domain. Furthermore, the actual performance is copyrighted by ME!!

  19. JuneLaw couldn’t be more wrong. The original post (didn’t read it, I’m guessing) pointed out that a mainstream news program discussed the contents and merit of a song. In order for viewers to know which song, a short excerpt had to be played. That doesn’t make MSNBC thieves. No one said stealing is right.

    In my case I used two royalty-free tracks with Creative Commons attribution rights. I posted the rights. YouTube didn’t even review the video, they just declined to monetize it pending proof of my rights to the songs. I of course provided them immediately, and have since never heard back.

    I understand they have to cover their asses, but it’s gone way too far.

  20. Terrific post however I was wondering if you could write a litte more on this subject?
    I’d be very grateful if you could elaborate a
    little bit more. Thank you!

Web Joho only

Comments (RSS).  RSS icon