Every major streaming platform uses some version of the same approach to recommend music. They analyze audio features (tempo, key, energy, instrumentation, mood) and match you with tracks sharing those properties. It's called sonic matching, content-based filtering, or some variation of both.
It works. Up to a point.
The problem is the ceiling. Sonic matching can only recommend within the boundaries of what sounds similar. It will never connect a metal fan to a classical composer, even if the overlap in their listener bases is massive. It will never suggest a Brazilian bossa nova artist to someone who listens to UK garage, even if thousands of people move between those worlds every day.
Taste is wider than sound. And the best music discovery tools need to reflect how people listen, not how songs sound.
Taste affinity is a different approach. Instead of asking "what does this song sound like?" it asks "what do the people who love this artist also love?"
The distinction matters because human listening behavior is messy, unpredictable, and far more interesting than any audio analysis. A person who listens to Radiohead might also listen to Miles Davis, Aphex Twin, and Fela Kuti. No sonic matching algorithm would connect those four artists. But the listener data says the overlap is real.
Taste affinity follows the listener, not the song. It maps connections between artists based on shared audiences using a proprietary recommendation engine, then surfaces the patterns appearing consistently. If the same listeners keep showing up across two seemingly unrelated artists, the connection is real, even if no algorithm analyzing audio waveforms would ever find it.
Sonic matching is good at finding more of the same. If you love a specific subgenre and want to go deeper into it, audio-based recommendations will serve you well. Spotify's Discover Weekly does this competently.
But most people don't want more of the same forever. They want to be surprised. They want the feeling of hearing something they never would have found on their own and realizing they love it. The feeling almost never comes from a song sounding like what you already play. It comes from a song living in a completely different sonic world but loved by the same kind of listener.
Taste affinity is designed for those moments. At lower depths, the results stay close to the seed artist's core identity. Familiar territory, well-curated. At higher depths, the engine follows the listener patterns further out, crossing genre boundaries, connecting scenes and eras sharing audiences but not sounds.
This is where the results get interesting. Seed a post-punk band and watch electronic producers surface. Seed a jazz vocalist and find connections to ambient music. The engine isn't guessing. It's following real people's listening habits to their logical conclusions.
Most platforms offer some version of a "similar artists" feature. Last.fm, Spotify, Apple Music, and YouTube Music all have them. The problem is how they define "similar."
For most platforms, "similar" means a combination of genre tags, audio features, and collaborative filtering weighted toward popularity. The result is a list of artists who occupy the same commercial space. If you look up a popular indie rock band, you'll see five other popular indie rock bands. The list is accurate in a narrow sense and useless for genuine discovery.
Taste affinity cuts through this because it doesn't care about genre tags or commercial positioning. It cares about what real listeners do. Two artists can share zero genre tags and still have a significant listener overlap. Those are the connections worth finding, because they represent something the algorithm can't manufacture: real human taste crossing boundaries the industry draws between genres.
Sonic Oracle's proprietary recommendation engine is built entirely on taste affinity. You seed an artist. The engine analyzes listener behavior to find artists connected through shared audiences. It then filters and selects representative tracks from each recommended artist, and builds a permanent playlist directly in your Tidal, Qobuz, or streaming library.
The Adventure Dial controls how far the engine travels from the seed. Essential stays tight and genre-pure. Balanced mixes familiar and surprising. Adventurous follows the listener data across genre lines and surfaces connections most people never expect.
The playlist lands in your library. You don't need a separate app. You don't need to copy or import anything. Open Roon, Audirvana, or your player of choice, and it's there. Permanent. Ready to play.
Your streaming app says "this sounds like that." Sonic Oracle says "people who love this also love that."
One approach keeps you in a loop. The other breaks you out of it.
Three playlists free. No credit card needed.
Try Sonic Oracle