YouTube blocks recording of public domain Schubert/Liszt composition

Post by imihai1988 » Mon, Aug 26, 2019, 4:10 PM UTC

brokensegue wrote: post #55

crazygringo wrote: post #24

Content ID is never going to be perfect (so there always needs to be an appeal mechanism). But I'd always expected that false matches would be due to some kind of hash collision, not different recordings of the same piece. Because YouTube doesn't know what the composition is, it only compares to other performances (which are under copyright). And the fact that two different pianists could interpret and perform a work…

Hash collisions? No, I don't think you understand how these content matching systems work.

Post by driverdan » Mon, Aug 26, 2019, 4:12 PM UTC

dpcan wrote: post #20

YouTube seems to be taking the brunt of the criticism when it's the repercussions of copyright infringement lawsuits that are the problem. I'm sure YouTube would love to be more lenient. If they could say "anything goes" they'd be thrilled, but we know they cannot because of the LAWS that are protecting IP owners. IP should be protected, but to what extent? And the bigger problem is the amount, and rate, of which con…

Post by albertsondev » Mon, Aug 26, 2019, 4:17 PM UTC

danso wrote: post #37

I have a slightly tangential question: can someone link to articles and writeups that discuss how YouTube and other platforms efficiently disambiguate between a valid/allowable recording and a prohibited recording, when the content between valid/prohibited is very similar or near identical in the most obvious features (e.g. rhythm and key).

Post by nine_k » Mon, Aug 26, 2019, 4:23 PM UTC

cameronbrown wrote: post #33

dpcan wrote: post #20

YouTube seems to be taking the brunt of the criticism when it's the repercussions of copyright infringement lawsuits that are the problem. I'm sure YouTube would love to be more lenient. If they could say "anything goes" they'd be thrilled, but we know they cannot because of the LAWS that are protecting IP owners. IP should be protected, but to what extent? And the bigger problem is the amount, and rate, of which con…

I'd argue dotcom domain squatting has a big part in the lack of names available. Eventually, most startups that could use them will go bust and sell the domain 5-10 years down the line anyway, it's outliers like google.com that are always going to be taken. New namespaces like .app should help with the problem, but at the end of the day, the culture itself has a limited namespace. I've always gotten fairly lucky with…

Post by visarga » Mon, Aug 26, 2019, 4:27 PM UTC

stephen_g wrote: post #58

coliveira wrote: post #17

The big problem with classical music is that the compositions are all in public domain, however most performances are not. Given the subtle differences between performances, algorithms are not smart enough to figure out the difference. In fact, even humans would have a hard time to distinguish between two performances of certain classical pieces.

I don't think it's so much that the algorithms aren't smart enough to tell, it's that Google have had to make the algorithm still match in the case of people altering music with filters/effects, time stretching, pitch shifting etc. - so it's been made deliberately a lot less sensitive to most of the factors that a musician brings into their interpretation. That's fine - it's probably necessary for Google to not be co…