an blog | AJBlog Central | Contact me | Advertise | Follow me:

Just in: EMI apologises for harassing piano boy

David Feldman, father of the young man whose piano videos got taken off Youtube when EMI claimed they were Daniel Barenboim’s, is a happy man.

Thanks to help offered by Slipped Disc readers, he managed to locate the right legal executive in EMI, who apologised for the error and removed the block. The legal eagle blamed the problem on new software (heard that one before?) and acknowledged that this kind of pursuit was less appropriate to classical music than it is to pop, where copyright violations are rife. He held out the possibility that classical music infringements might be removed altogether from the search program – which would be a very good and sensible result.

Well done, our readers!

Here, by way of reward, is another of Anthony Feldman’s videos, now restored to Youtube.

 

Related Posts Plugin for WordPress, Blogger...

Comments

  1. Super, good job! great result.

    Now can we argue about the definition of “classical music” please? ;-)

  2. What a relief! And thank you, Norman, for highlighting this particularly obnoxious “feature” of YouTube for everybody.

    It’s ironic that in the thumbnail links displayed (to me) at the end of Anthony’s Chopin clip, one can click on “FREDERIC CHOPIN – NOCTURNES (complete)” played by none other than Maurizio Pollini on DGG as if there were no copyright issues at all associated with that kind of upload. This is what makes me angry at YouTube! One can find just about everything on YouTube which was published at one time or another in commercial format, be it LP, CD, or DVD. How do these people do it without getting caught by YouTube’s Byzantine content ID system?

    • bratschegirl says:

      Not to mention the videos obviously shot with a hand-held device from a seat in the audience at a recital. YouTube make it very clear that they don’t want to hear from third parties pointing out that someone else’s rights have been infringed, so unless the artist protests, there they stay.

    • You can thank the music industry for these issues, remember all those high profile lawsuits against YouTube? This is the result. The system actually works surprisingly well… 90% of the time, but it certainly results in far too many false claims on ownership.

      For all those Pollini videos and all, the ads that appear on the videos generate money for the copyright owners, not to the kids uploading the music/videos. The music industry has in general agreed to this arrangement.

      If a video is flagged or removed, an individual can dispute that. I’ve had to do that a number of times for my own music. It is a pain for sure, but I don’t place the blame on YouTube, more like Warner, ASCAP, etc.

      • Yes, you can file the usual YouTube dispute form. Sometimes it works. But what do you do if the supposed “owner” reiterates their claim, all without ever examining the video in question? I have had this happen all too many times for it to be simple bad luck. Then I have to spend a significant amount of time tracking down a way to get someone inside of the corporation to do the right thing. All the while, I’m under some vague legal threat issued via YouTube.

        If you email YouTube directly, the YouTube “Team,” as they call themselves, sometimes responds. But often the “Team” is a barrier more than a help. They too often don’t even look at the video. They refer you to a FAQ section so you can file a formal dispute, which is what you have already done! On occasion, someone on the team actually cares enough to read my email. And on occasion they take action, or they let slip the email address of someone at UMG or Orchard who can. But this too is a very time-consuming and hit-or-miss process.

        I’m involved in one of those processes right now. Several of my videos cannot be monetized. There is a little exclamation mark in the monetization box that tells me I must submit proof of third party permission. Since I own the entire video, I do not need any third party permission. Yet my account shows no third party claims on those videos. It’s very hard to resolve an issue of this sort when you are shadowboxing against claimants who don’t show up on your account.

  3. http://digitaljournal.com/blog/14455
    EMI making a habit of offensive, dishonest business practices. Virtuosi can bash back too, ask the late Sviatoslav Richter.

  4. David Feldman here again. I would like to note that I don’t begin from the perspective that musical rights groups like EMI, UMG, Orchard Music etc. are in league with the devil. There is a strong case for effective intellectual property protection. My beef is over process.

    Like all the other rights holders groups, EMI is working in an environment in which the technological and legal mechanisms for dealing with everything from accidental use to outright theft are still evolving. The trick is to figure out cost-effective ways of offering people something that looks like due process. All parties have to see the system of adjudication as basically fair.

    If the bulk of the false positives (pieces falsely identified as belonging to EMI) are on the classical side of the recorded library, then the remedy may indeed lie in separating classical from the other genres and working out a separate method for dealing with potential infringements.

    From what I’m told, the bulk of these cases arise when people add pop music sound tracks to videos they create. The audio is caught by the matching program and the violation notice is sent out by YouTube. The “accused” in this case may dispute the charge, not realizing that the audio clip is the violation, not the video itself.

    In any case, the key here is for a person who is accused of a copyright violation to be able to reach a human being for a decision, without having to file an expensive court case. If the supposed violator chooses to contend the notice, EMI should not be presumed correct based solely on a crude computer sound identification.

  5. I’d love for someone to prove me wrong, but I have the feeling that this ‘sophisticated’ content id software merely matches some text keywords and very exact timing information. With the amount of material being posted every second, I can’t imagine that there is enough storage for youtube to actually match sound clips to sound clips, much less have an army of musical literates(?) checking for false matches.

    • I’ve a technical background so I can answer a bit the “how does it work”…

      Initially I would have assumed, like you, it to be checking the text values that the user has entered – not just the title but also the description, maybe also the comments. There isn’t much “merely” about it, it’s going to be quite a sophisticated thing to do with the amount of data YouTube has. You can’t just, for example, run a regular expression on all user-entered values every time the user changes them:
      - what about mis-spellings?
      - how do you maintain your enormous list of keywords?
      - how do you stop the whole thing grinding to a halt as it checks millions of text values per day, each against thousands or more keywords?
      - how do you tell the difference between “sounds like Barenboim” to “is Barenboim”
      - different languages?

      And most obviously, “what if there is no text description?”

      So it’s highly unlikely they match text. But please read on!

      When my own music got caught in this trap, my video was pulled some time after I uploaded it. This led me to suspect they do random sampling and then do the matching (my text would have not matched much except perhaps George Benjamin or Mr Britten!). The matching I supposed was on a waveform analysis of some sort.

      I think that when your video is uploaded, a waveform signature is taken of it at the same time as they take the video snapshot preview. This would be straightforward and an obvious time to do it. Then that could be added to a queue for later analysis offline, perhaps on a random sampling basis.

      Now this latter speculation looks correct after a bit of Googling; the thing they compare your file to is a Content ID which is supplied by the rights holder. Read more here: http://www.youtube.com/t/contentid

      To the person above who said “why don’t they just license Soundhound?” well, that’s because they licensed Audible Magic’s tech to do this.

      Read about what one keen user has discovered about how Content ID works here, if you’re a bit technically minded and curious:
      http://www.csh.rit.edu/~parallax/

      (This last link is quite fascinating by the way and I’d recommend it to everyone who’s shown an interest in this thread)

    • I thought so, too, until I saw this report (possibly on a different website, though) — “A copyright claim on chirping birds highlights the flaws of YouTube’s automated system” (from Feb. 2012; there was a similar incident back in 2008 or 2009, I believe):

      http://thenextweb.com/google/2012/02/27/a-copyright-claim-on-chirping-birds-highlights-the-flaws-of-youtubes-automated-system/

      Believe me, YouTube (Google) has plenty of storage space for doing this kind of audio matching.

  6. If they are doing random sampling of uploaded videos, then the ones that I have posted seem to be wildly oversampled. Random bad luck for me? That seems unlikely.

  7. Mistaken for Barenboim? What a put-down!

an ArtsJournal blog