Music: Lackluster Similar Artists. Options?

Plex Media Server Metadata & Adding Files
,
1

The new agent uses Allmusic for “Similar Artists” This works great for popular artists, but is severely limited when it comes to certain genres of music or smaller artists.

For example lets look at similar artists to the DJ: Metrik.
Plex metadata refresh: 2 similar artists (Royalston and Rolanoid)
Allmusic.com’s similar artists: 4 (Royalston, Rolanoid, 1911, Sub Focus)
Tidal’s Related artists: 50
Last.fm’s Similar artists: 25 pages worth

This leads to a terrible experience if you try to use artist radio, or the new mixes from plexamp. If I did an artist radio on Metrik it would literally play 3-4 Metrik songs in a row, and then start “exploring” outside of my library on Tidal. Which means it will then play about 90% Royalston songs. Rolanoid only has EP’s and singles and this seems to screw up plex search so it doesn’t think there’s an artist out there with the name Rolanoid, so you don’t even get that in your list. Nevermind the fact that I’ve never heard of Royalston or Rolanoid before, but I have about 80% of Tidal’s similar artists in my library, so it doesn’t seem that accurate.

This is an artist that has 690,350 monthly listeners on Spotify so it’s not exactly an issue of the data not being there due to their popularity.

What are my options here?

  1. Why does the plex metadata not even have the 2 more artists that allmusic has? Is there some way to trigger that update?
  2. Is there some way to update allmusic.com so it sucks less? I’ve submitted data for a ton of local bands on musicbrainz so I’m all for spreading the love on a service where everyone can reap the rewards, but it seems like allmusic.com does not allow community submissions?
  3. Is there a way to do a single metadata fetch from another service (last.fm) for an artist? or mark that artist to always use last.fm?
  4. Is manual entry the only solution? I manually added about 20 related artists and suddenly my mixes and radios were amazing

I’m hoping option #4 isn’t the only solution as it wouldn’t automatically update as new artists emerge, and any work I put into this only benefits me and not the whole plex/music community.

Thanks!
Rory

2

Same question, thanks for raising a topic

3

This definitely could use some work. Not sure if it’s more on the end of allmusic, muscbrainz, last.fm or plex though.

It’d be nice if there was an easy way to have a central user submitted/user voted database of similar artists. I feel like rateyourmusic/sonemic would be a nice integration for this type of stuff but we’d have to wait for their api to be released. Rateyourmusic/sonemic integration would be nixe for a lot of things, to be honest (genres, descriptors, ratings, reviews, related artists, etc.).

4

Just found out, that I was losing tons of similar artists (and artist bios) after the switch to the new scanner and I was wondering why. It boils down to allmusic’s lack of data. It’s really pathetic, when there’s 10+ pages on last.fm as described above and allmusic has NONE!! I mean where do they get their data even from? Could we have another option in the library settings or agents to get at least the similar artists from last.fm again?

5

It would be great to be able to manually link similar artists.

6

Well, I suppose you could manually go through and add them. :stuck_out_tongue:

That’s not really a great solution when you have 1,000+ music artists, though. Would just be nice to be able to combine all music, the first 2-3 pages of last.fm similar artists, and some of tidal’s as well. I feel like adding 20+ pages of similar artists maybe isn’t the solution, but maybe some how cross referencing all three and seeing what artists show up on all three and just adding those would help to have better similiar artists at the very least. Now, i’m not sure how that would be done while also including a decent selection of similiar artists so mixes and radios would still have good variation and not only just 5 or 6 artists. I feel like it’d be more complicated to do something like that, and so I’d be fine with a simple fix of just combining the data from each site (while truncating last fm to maybe just the top 10 or 25 similiar artists). Then maybe just doing a simple algorithm that removes the duplicates.