News for hackers. Stuff that matters.
Re:Every Noise at Once by svantana on Friday April 02, 2021 @05:11AM (#26670056)
For various reasons they are secretive about this, but the genre system is inherited from EchoNest which they acquired in 2014. EchoNest's main idea was to combine audio analysis with natural language processing and web scraping to categorize music. In fact the two founders were PhD's in each respective field.

But nowadays I'm guessing they are utilizing their huge dataset of playlists to group artists under tags. Obviously such systems will work better the more popular an artist is.

Re:Every Noise at Once by spoctrial on Friday April 02, 2021 @07:15AM (#26670515)
They are not that secretive, it's been written about several times. Long story short: a lot of it is still Glenn McDonald, the person behind everynoise, that was acquihired when they bought echonest.

Here are a few articles about it (there are several more): https://artists.spotify.com/blog/how-spotify-discovers-the-g... https://www.thefader.com/2020/12/02/escape-room-genre-spotif... https://www.thestar.com/entertainment/2016/01/14/meet-the-ma...

Re:Every Noise at Once by svantana on Friday April 02, 2021 @11:05AM (#26671990)
Well, ideally your spotify artist page should contain the genres you're tagged with, along with some info of how that came to be, and even a way to request a change to those. But they don't want people to game the system, and also not let competitors know how they do things (I'm guessing).
Re:Every Noise at Once by spoctrial on Saturday April 03, 2021 @09:35AM (#26680922)
At least you get the genre info is in the web API, just not present on the artist page (you need a token for the query to work, but you can create that on dev page): https://developer.spotify.com/console/get-artist/?id=10YNQq8...
Re:Every Noise at Once by Cthulhu_ on Friday April 02, 2021 @06:00AM (#26670237)
I wonder / kinda hope that they have a bunch of music experts working for them that manually categorize and tag music, at least to kickstart their algorithms.