Echonest (startup in boston) has some libs on github for audio fingerprinting and retrievel. That solves part of it, but the labeling seems like it might be the tricky part. As far as how to quickly search, yeah maybe elasticsearch, but it might not really be needed as the number of songs is pretty finite.
True. Not only do ASCAP, BMI, and SESAC control a limited number of musical compositions, about 10 or 20 million at my last count, but the whole set of possible musical compositions is limited to a couple hundred million at most. If you want, I can explain further. (Hint: lawsuits alleging 8 note similarity, 14 possibilities for each note after the first, 14^(8 - 1))
But the number of recordings of these songs is effectively unbounded, as is the number of ways stations can distort any particular recording
They could cover or sample already existing background music. Or, since I would expect that instrumental background music would get a bit less scrutiny than top 40, songwriters could keep working in that area. (I would also like to note that I don't think all songwriters retiring would be a _good_ thing, just that that's about the only way to be sure that you won't infringe.)
Curious about the technology they use (Score:5, Interesting)
Re: (Score:2)
Re: (Score:4, Interesting)
the number of songs is pretty finite.
True. Not only do ASCAP, BMI, and SESAC control a limited number of musical compositions, about 10 or 20 million at my last count, but the whole set of possible musical compositions is limited to a couple hundred million at most. If you want, I can explain further. (Hint: lawsuits alleging 8 note similarity, 14 possibilities for each note after the first, 14^(8 - 1))
But the number of recordings of these songs is effectively unbounded, as is the number of ways stations can distort any particular recording
Re: (Score:2)
It's probably less songs than that, The record for a suit was 3 notes.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Re:Curious about the technology they use (Score:2)