Short

Let your computer scan new music releases for good dynamics - automatically!

Motivation

Overcompressed music annoys me.

There are albums out there that are not that much compressed. They are rare and therefore hard to find. It is a manual process by searching for them.Once you find a list somewhere you need to check for genre that one likes, then one needs to search and listen to it.

Another approach is to use the dynamic range database. There is no genre filter and no quick listening. One finds more music there than searching lists, but it is still manual.

You can use tools like the de-limiter.

All that is not satisfying.

Idea

What about a program that

  • uses freely available online sources like
    • Musicbrainz (Listenbrainz - basically youtube)
    • youtube
    • qobuz previews
    • spotify or
    • bandcamp to
  • scan the dynamics of new releases and
  • pushes the results (maybe only with dynamics greater than X) in a small package (just text - json or so - maybe reuse existing services like MusicBrainz) to an online website that lists
    • Band, Album, Year, Album cover (linked)
    • Genre
    • Dynamics (to be discussed - I would love to see integrated LUFS and (true) peak of some songs or the entire album) and
    • a freely available preview (linked) and
  • which can be filtered for example for a genre

I think this is possible, though it might be tough. Maybe something like this is already existing.

I would benefit from it.

It could be decentralized. A scanning application that could be run on demand. It checks for missing dynamics data at the online service and then fills it if the one runs the app.
Like torrents - where everyone helps to provide the solution.

I wanna discuss this. Maybe this is a stupid idea. What do you think?

    • topas184@alien.topOPB
      link
      fedilink
      English
      arrow-up
      1
      ·
      10 months ago

      I (or better the community - it’s too much for me alone) would like to do this with music I don’t own yet. So grab freely available online music content (like YouTube videos) and scan THEM for loudness characteristics continuously.

      Then upload that data somewhere to let others find music based also based on loudness data easily.

      I’ll add this more clear in my first post.

  • aerozol@alien.topB
    link
    fedilink
    English
    arrow-up
    1
    ·
    10 months ago

    You can’t scan music on MusicBrainz - it doesn’t host any audio.

    But MetaBrainz did work on something like this - AcousticBrainz (RIP). Long story short, the algorithms turned out not to be as accurate as was hoped, especially with some genres, and better technology (e.g. machine learning) started emerging soon after the project was released, which meant that starting over would have been the best choice.

    MusicBrainz would work as a base for pretty much everything else though. You could submit most of this data to MusicBrainz as tags, for tracks, releases or artists. Links to streaming services/previews can also be added. You could start adding tags to MusicBrainz already, e.g. [genre], lossless