I hope this is okay for me to post in here. It’s about music :)

So I’m currently digitizing my records. This is easy to do for CDs. I got myself a dvd drive and just rip them to flac files on my computer. Tagging is easy too pulling the data from music brainz. Vinyl records are a little bit of a different story though because for each side of a record I get one long audio file. Splitting up the files to individual tracks is a very tedious process. I’m currently doing that manually with audacity and then exporting them to flac files.

Does anyone have an idea to do that more easily? Maybe even automate parts of the process. After all music brainz has all the track lenghts available. So in theory it should be possible to automate this to some degree.

Any ideas are welcome. Thank you.

  • sanataseva@kbin.social
    link
    fedilink
    arrow-up
    2
    ·
    1 year ago

    MusicBrainz won’t help, because every rip will have enough variance to thwart digital identification. Also, the song lengths will be off because your beginning and end times will differ from CD rips, and this will compound until you’re significantly off course by the end. I’ll be happy if someone posts a better method, but I’m still using Audacity and setting manual track labels.

  • Illecors@lemmy.cafe
    link
    fedilink
    arrow-up
    2
    ·
    1 year ago

    Technically - doable.

    As someone who’s never had a vinyl - this would be my angle of attack:

    • Slice the first 10 (or however many are necessary) seconds of the track
    • Hit some Shazam or Google API with the slice
    • Get the title
    • Hit Music Brainz API (or some other) to get the legth of the track
    • Cut the single track into title+remainder
    • Rinse and repeat.

    ffmpeg could deal with cutting the file

    cUrl can hit the APIs

    some shell script (bash would be my choice) would tie it all together

    • honk@feddit.deOP
      link
      fedilink
      arrow-up
      1
      ·
      1 year ago

      Labeling the parts of the song and then exporting multiple is how I currently do it.

      The auto labelling actually somewhat works. The error rate is relatively high though. Set the threshold too low and the background noise of the record will make it detect nothing. Setting it too high and even small quieter parts get detected. It’s far from being “automated” but it actually helps a bit. Thank you.