>
Uncategorized

Sonorium: music discovery that trusts humans, not algorithms

I used to find new music by waiting for friends to text me links, then forgetting about them three days later. Algorithmic playlists on the major streaming services kept pushing the same twelve artists at me, dressed up as “discovery” but really just the same producer roster on repeat. I joined a community-driven music discovery platform called Sonorium in March, and after four months of using it most days, I have stopped opening the algorithmic playlists entirely.

Sonorium is built around a simple idea: real people sharing music they actually love, with context. Every track you add to the service comes with a short note about why you liked it. The note is the point. The note is what other people read before they decide to listen. Algorithmic discovery treats you like a vector in a high-dimensional space (a mathematical way of representing your tastes as a long list of numbers, where similar vectors get similar recommendations). Sonorium treats you like a person with reasons.

What it actually is

The product is a web app with a clean two-pane layout. On the left, a feed of tracks other users have shared. On the right, a player and a comment thread. You can scroll the feed, click something that catches your ear, and read what the person who shared it said about the track. The notes are short – usually one or two sentences – and they cover things algorithms cannot see. “Heard this in a coffee shop in Lisbon and the barista told me the band was from a town 40 minutes away.” “My dad used to play this on Sunday mornings when I was a kid.” “Sounds like a David Lynch score if Lynch had been happier.”

The site also runs a forum-style discussion area where you can ask for music by mood, decade, instrumentation, or just a feeling. Someone posts a thread titled “Something to cook pasta to on a Tuesday night” and the responses come back with context, not just track links. The whole product is built around the assumption that recommendations are a conversation, not a computation.

There is a free tier and a paid tier at $4 per month. The free tier covers everything I actually use. The paid tier adds lossless audio (uncompressed music files, similar to the quality on a CD) and a few discovery filters the free version does not expose. I am on the free tier and have not felt the need to upgrade.

What works better than I expected

The notes are the part I underestimated. I joined mostly because I was tired of Spotify’s algorithmic playlists, which is a low bar. What I did not expect was how much the notes changed my listening behavior. When I see a track is recommended by someone who says “I have been a jazz drummer for 20 years and this is the most honest solo I have heard all year,” I listen to it differently than I would a random algorithmic pick. I pay more attention. I am more likely to revisit.

There is also a social feature I was skeptical about: you can follow other users, and their shares show up in your feed. Following a small number of users with good taste – maybe 15 or 20 – turns the feed into a steady stream of music from people whose taste I trust. It is the inverse of the algorithm model. The algorithm picks for you based on a model. The community model lets you pick the people, then the people pick the music.

The discovery forums are where the product shines. I posted “Looking for ambient electronic that does not sound like a film soundtrack” and got 47 responses in two days, each with a track and a sentence of context. I added 11 of them to my library. None of them would have shown up in my Spotify Discover Weekly, because none of them are in the same audio-feature neighborhood (a group of tracks the algorithm considers similar based on tempo, key, and instrumentation) as anything else I have listened to.

The parts I do not love

The mobile experience is rough. There is a responsive web version that works on a phone, but the layout does not reflow (rearrange gracefully) cleanly, and the player controls are too small to use comfortably while walking. I have started a feature request thread for this and other users have piled on, so the team knows. The official iOS app is in beta and shows a “submit your email” landing page when you try to install it. The Android app does not exist yet.

The catalog is small compared to the major streaming services. Sonorium works as a discovery layer, not a full music library. Tracks you find on Sonorium usually link out to Spotify, Apple Music, or Bandcamp, and you listen to them in those apps. I treat Sonorium as the front door and the streaming services as the back. That works for me, but it is an extra step the algorithm-first services do not require.

Moderation on the platform is light. I have seen one or two threads go off-topic into arguments about genre boundaries, and the moderators were slow to step in. The community is small enough that this has not been a real problem, but a larger community would need more active moderation.

Trade-offs

Sonorium is not free in time. The notes that make it valuable also make it slow. I spend 10 to 15 minutes a day reading the feed and the comment threads, which is more time than I spent scrolling algorithmic playlists. In exchange I find music I actually want to listen to, which is the point. The 10 to 15 minutes a day feels like time well spent now, but it would not have felt that way six months ago.

The community is also the bottleneck. If the people I follow stop posting, the feed goes quiet. I have learned to follow a wider pool of users than I would on a typical social network, specifically to keep the feed full. This is the opposite of the algorithm model, where the algorithm does the curation and you do not have to think about who you follow.

Migration from a streaming-service-only workflow is real. For the first two weeks I found myself going to Spotify out of habit, scrolling for 30 seconds, and remembering I had stopped doing that. The habit took longer to break than I expected.

If you are tired of algorithmic discovery and willing to spend 10 minutes a day reading notes from real people, this is a clear win. If your only goal is to press play and have music in the background with zero effort, the streaming-service algorithmic playlists are still easier. Sonorium is for people who want the discovery part of listening to music to feel like a conversation.

If I could send a message back to the version of me that was about to delete my Spotify algorithmic playlists, I would say four things.

  • Follow 15 to 20 people, not 100. A small pool of users with good taste is the entire point. Following too many people dilutes the feed back into noise.
  • Read the notes before you press play. A 30 second note changes how you listen to the track. Skipping the note is the same as using the algorithm.
  • Treat Sonorium as the front door, not the back. Tracks link out to streaming services. That extra step is the design, not a bug.
  • Give it a month. The first two weeks feel slow. The notes start to feel essential around week three or four.

Bottom line

Sonorium is a real product with a real community and a real point of view about what music discovery should feel like. It is not trying to replace your streaming service. It is trying to be the front door, and it does that well. The mobile experience needs work, the catalog is small, and you will spend 10 minutes a day reading notes. If those trade-offs sound acceptable, give it a month. The notes will start to feel like the most valuable part of your listening routine.

If I only do one thing from this article, I delete my Spotify algorithmic playlists and replace them with a Sonorium feed followed by a small group of users whose taste I trust. The rest can wait.

Leave a comment