I think we're losing most potential contributors before they make their first contribution

Development
6

no player (that I know of) capable of tapping into both sources and exposing a seamless combined collection.

I think this is part of the gimmick with the audiophile service Roon. What is Roon? : Everything You Need to Be Roon Ready | World Wide Stereo

the beets DB could include entries for tracks in people’s cloud libraries?

I wonder if there’s room for collab with listenbrainz or musicbrainz?

curate cloud collections

In audio communities, people make lists like “Rolling Stone top 500 albums”, “my favorite electronic of 2022”, “whimsicore essentials”. This could be really cool. Musicbrainz supports collections which may be similar.

or every track and album, know its ID on every metadata platform (MusicBrainz, Discogs, etc.) and every streaming service (Spotify et al.)

Musicbrainz has been trying to do this for ages, but it’s not clear if most users care. I think too much wheel-reinventing happens in music communities today.

(1) and (2), but for all media I care about—movies, TV shows, podcasts, etc.—not only music.

I’d have to see more of the pitch here. Podcasts maybe, but movies and TV have very little commonalities with the beets data model, typical consumption (repeat listen vs. less frequent rewatching), data formats, and consumption model (watch and delete? vs. keep forever).

My personal wishlist for expanding the beets appeal is taking the manual tagging and throwing it into the sun. The return on time investment is not there unless you are really particular about knowing whether you have the 1984 or 1992 master of The Wall. We should focus on matching release groups, not releases. The false matches happen more on popular tracks, which is what normal people are matching most of the time.

But beets does not have to be about only local music files.

This has been a small conflict for me over the last ~3 years - understanding where beets ends and other projects should come into play.

Back in 2017, I got into beets with simple goals:

  1. Make all my scrobbles of the same track (recording, in Musicbrainz terms) map back to one track. So make all my tags “musicbrainz grade”, i.e. no “feat.” in title, no “remaster” in release title.
  2. Make my playlist display (in Foobar2000 then and now, but this is not a requirement) show a uniform view. (Same as #1, everything should match.)
  3. Make my file folders and files uniform.
  4. Automate or semi-automate the import process so all new music conforms to 1-3.

Since 2017, my already large music collection has exploded. My new goals, which replace the old ones, are:

  1. Basic smart playlist creation. I want to tag my top 50 albums and always have those with me. I want playlists for my favorite genres so I don’t get stuck in feedback loops playing what I remember.
  2. Curate my library, slicing it in interesting ways. Basically, make playlists, but this can be more advanced than playlisting. This is something that streamers have started doing and have only gotten better at over time. The difference is there isn’t much of a “library” on the unlimited streamers (Spotify, Deezer etc.). I.e. there’s little difference between “a track you’ve heard” and “a track in the streamer’s library”. I don’t think an open source $0 project can compete with the machine learning people at the streamers, but I do think it can provide interesting (good) results by focusing more on the human angle (tags, ratings, user’s scrobble data). This ML vs. human curation debate comes up now and then, I’ve read a few non-technical articles on the topic.
  3. Have a way to play my library on my Android phone. I thought I would use convert for this, but the case for streaming + seamless local caching on the phone is getting better thanks to people on these forums.
  4. Have a way to use those curated slices on Android.
  5. automate the import process. luckily I’m much handier in python/automation than I was.

As for my original goals:

  1. Make all my scrobbles of the same track (recording, in Musicbrainz terms) map back to one track. I hope someday that ML will be good enough to do NER to fix my historical scrobbles. It’s too much of a pain and the time to fix vs. value to me isn’t worth it.
  2. Don’t care enough Make my playlist display (in Foobar2000 then and now, but this is not a requirement) show a uniform view. (Same as #1, everything should match.)
  3. Symlinks and playlists should replace the need to look at raw files Make my file folders and files uniform.

Aside: I see Musicbrainz got Google summer of code contributors.

As a Google Summer of Code’22 contributor, I worked for MetaBrainz, on the MusicBrainz Android app and added a music playback feature to the app, which we call BrainzPlayer.

Is SoC worth exploring for 23? Google Summer of Code - Wikipedia

Does anyone know anyone at MetaBrainz? I wonder what allows MetaBrainz/MusicBrainz to achieve a critical mass of people while beets doesn’t have many contributors. Or is this even true? the activity graphs look similar for the last month.

1 Like