cross-posted from: https://lemmy.ml/post/6528233

I’m working on building a personal alternative to Spotify and YouTube Music and I’ve hit two roadblocks.

  1. Genre labeling in my library is inconsistent and manually updating 2500+ MP3s isn’t feasible. I’ve tried using beets with the LastFM plugin in quiet mode but no luck. Any ideas?

  2. Where can I bulk download diverse music catalogs? I’ve snagged some ‘top 90s/80s/hiphop/etc’ collections, but 2500 songs don’t go far.

What I miss about Spotify and YouTube Music is their ability to auto-play similar tunes based on my current selection. Any advice would be appreciated.

  • For bulk diverse, pre-tagged and sorted collections you really have to go to p2p like soulseek.

    For tagging picard is really the only option that can handle bulk tagging with some level of trust and authority, but definitely have a backup or work in chunks just in case it goes awry.

    For a spotify replacement, Plexamp is really doing well as long as you have decently tagged media. It can do all the mood radio and playlist suggestion stuff that Spotify and others have been doing.

  •  Petter1   ( @Petter1@lemm.ee ) 
    link
    fedilink
    English
    48 months ago

    I use lidarr together with plex and jellyfin. For downloading, I have chosen the usenet and it works pretty good full automatic. You just need the right indexers.

  • Beets will work, but you have to put in the effort to get all your music imported correctly (you have a backup, right!), then lastfm plugin will work fine (and metadata, and album art, and lyrics, and …). Once you get started, it’s pretty easy to add new stuff in. It does tend to work better with albums, I’ve mostly moved to getting an album when I’ve wanted a single song, because space is cheap, and I’ve found some interesting stuff that way.

  •  Appoxo   ( @Appoxo@lemmy.dbzer0.com ) 
    link
    fedilink
    English
    2
    edit-2
    8 months ago

    Identifying: Picard (I think) by musicbrainz
    Organizing and keeping track: Lidarr

    There are alternatives to lidarr but that’s what I did for my ~300 files.
    Picard matches by current metadata and to some extend by acoustic ID (similar to shazam).

    Still better to review files before commiting the metadata (album, year, version, miss matched releases similar releases in name by cover bands or release name etc.)

    For downloading more: I had great success with scarce releases (rare touhou songs) on soulseek.

  • I use mpd-sima for autoplay functionality. It would in turn require using mpd or mopidy.

    It queries last.fm for recommended artists and tries to find a match in your playlist. It works, but honestly it’s not great. I don’t think it’s any fault of the software. Last.FM doesn’t know your library and seems to return limited results, so it’s like a game of battleship actually getting a hit. (Not quite that bad, but the analogy holds water, overall)

  • Genre tagging is a huge pain in the ass!

    Picard is probably the most popular tool for this, but a problem with any auto-tagger is that you may disagree with the genre selection. Because the data is all crowdsourced you also see inconsistencies… Like “Hip-Hop” and “Hip Hop” both in use. Either you live with that stuff… Or, you do it yourself.

    2500 tracks isn’t actually that much to do manually. Chip away at it with a great tool like mp3tag, and once you are caught up, do new additions as you get them.

    Someone here said:

    It does tend to work better with albums, I’ve mostly moved to getting an album when I’ve wanted a single song…

    This is so true. Loose single tracks are a pain in the ass too. Try to get full releases.

  •  DjMeas   ( @DjMeas@lemm.ee ) 
    link
    fedilink
    English
    1
    edit-2
    8 months ago

    For downloading music in bulk, I personally subscribe to Deezer and then run Deemix in Docker which allows me to download offline audio in FLAC. I then serve those FLAC using Plex and Navidrome.