Shfl is a music recommendation tool — one that depends only on the very simplest algorithm of them all, random sampling. It is built from a massive number of specific recommendations from music critics, musicians, and a few carefully chosen lists from around the web. There’s no robots peeking over your shoulder here, we’re trying to optimize for surprise, which is why you can go straight from Danny Brown to Scarlatti.

Shfl is a big, densely-connected graph — there’s lots of ways to customize your path through it. Tapping on a tag or the icons next to the date, label, or album recommender will restrict the sampler to albums that share that attribute. Filters can also be added manually, and they can be combined — you can filter on African music from the 70’s, for instance, or norwegian black metal, or jazz guitar. If you’re not sure where to start, we also publish guides to various musical subjects, collections of albums grouped by topic, and there’s a best-of section with jumping-off points for many of the most popular genres on the site.

  • Lime Buzz (fae/she)@beehaw.org
    link
    fedilink
    English
    arrow-up
    5
    ·
    edit-2
    22 days ago

    It looks neat but it’s very limited at the moment, both in genres and artists. We did some testing and it only got one of the artists we searched for and none of the genres so far, sadly.

  • Allero
    link
    fedilink
    arrow-up
    2
    ·
    7 days ago

    ListenBrainz is the most mature open-source solution. It may connect to your player of choice (kinda like last.fm) and save what you’ve listened to to give you recommendations later.

    • I know about it, I use it myself. But I don’t “scrobble” directly to ListenBrainz, it’s connected to my Last.fm.

      I’m not a big fan of the recommendations it gives me, not obscure enough for me. Maybe they will improve as more people use it, because AFAIK the algorithm is only based on listening data of users. I feel like the recommendations could already be better if MusicBrainz data like tags, labels, compilations albums, etc. were used additionally. (Example: You listen to artist X. Songs by artist Y are included in multiple compilation albums which also feature songs by artist X. Maybe you also like artist Y!)

      For now I prefer tools like the one I posted here: There’s a collection of albums tagged with music genres, styles, etc. You want to discover albums with a certain tag. The tool provides you with a random album with that certain tag.

      (You can also do this manually with Rate Your Music charts by using a random number generator. I kinda wish they added a feature to do this automatically.)