← All posts

New music discovery: tools and sites worth using in 2026

The best music discovery apps, tools, and sites in 2026. Community charts, genre finders, credit-based tracking, and more.

New music discovery: tools and sites worth using in 2026

New music discovery: tools and sites worth using in 2026

There are dozens of music discovery tools online. Most do the same thing: you type in an artist you like, they suggest similar ones. That's useful exactly once. After that, you get diminishing returns fast, because similar-artist algorithms all draw from the same well: listening data, genre tags, collaborative filtering.

The new music discovery tools worth keeping around are the ones that show you something the algorithm can't reach. Here's what's actually good right now.

The similar-artist tools

Gnoosic and Music-Map are siblings, both part of the Gnod project, both built on crowdsourced listener data. Gnoosic asks for three bands, then feeds you recommendations one at a time while learning from your yes/no reactions. Music-Map shows the same data as a visual cloud: type a name, see related artists clustered by proximity. They look like they were designed in 2004 and they still work. One-person projects by Marek Gibney in Hamburg, running since 2005. No accounts, no Spotify integration, nothing to sell you.

Chosic is the most feature-rich of the bunch: similar-artist finder, playlist generator that connects to Spotify, mood-based recommendations, and a visual artist map. It also works as a genre finder: paste a track and it'll break down what genres it falls under, which is handy when you've heard something you like but don't know what to search for next. Nearly two million monthly visitors, which makes it the most popular dedicated discovery tool online.

Spotify's AI Playlist lets you type a vibe in plain language and generates a mix. Good for mood-based listening. Less useful for finding specific artists, labels, or scenes, because it draws from the same collaborative filtering data as Discover Weekly.

These tools are good for a cold start. They're less useful once you already know a scene. If you're deep into UK garage and someone recommends Disclosure, the tool has run out of things to tell you.

Where diggers actually hang out

Rate Your Music has 6.6 million releases cataloged and 147 million ratings submitted. That data powers the most detailed genre taxonomy on the internet. You can filter charts by subgenre, decade, country, and mood descriptors. Highest-rated dub techno albums from Germany in the last five years? RYM gives you a ranked list with user reviews attached. No streaming platform comes close to that kind of specificity. It's also the place where people will spend 400 words arguing about whether something is post-punk or darkwave, which is either maddening or exactly your thing.

Album of the Year aggregates critic scores alongside user ratings for current releases. Less useful for catalog diving, better for answering "what's actually good this week" without reading ten review sites. The release calendar alone makes it worth bookmarking.

Hype Machine has been running for 21 years, aggregating posts from around 800 music blogs and surfacing what's trending. It nearly shut down in 2017 when ad revenue dried up, then switched to listener funding. A few thousand supporters keep it alive. It's the closest thing left to the old music blog era: human curation at scale, no algorithm in sight. For indie, electronic, or anything between genres, it still finds things before anyone else.

Then there's Bandcamp, which I keep coming back to. It's the one platform where following a label actually works. Hit follow and new releases show up in your feed. Bandcamp Fridays, where the platform waives its revenue share eight times a year, reliably surface deep-catalog and small-press releases you'd never see on Spotify. Fans have paid artists over $1.6 billion through the platform, most of it going to people who'd never last on a major distributor. The Songtradr acquisition and staff cuts hurt the editorial side, but Bandcamp Daily is still publishing and the marketplace is intact. For underground electronic, ambient, experimental, and independent hip-hop, nothing else is close.

Every Noise at Once (RIP)

Glenn McDonald built it while at Spotify, through The Echo Nest acquisition. An interactive map of over 6,000 music genres. Click any of them and it played a sample. You could spend hours bouncing between genres you'd never heard of, clicking a name like "deep deep house" or "Finnish tango" and hearing what it actually sounded like.

Spotify laid off 1,500 people in December 2023. McDonald was one of them. He lost access to the internal data that powered the site. It still loads as a frozen snapshot, but it hasn't been updated since. The best music discovery tool anyone ever built, killed by a layoff spreadsheet.

Any discovery tool that depends entirely on one platform's internal data can disappear the moment that platform decides to cut costs. Every Noise is the case study.

Credits: the approach nobody else takes

Every track on Spotify has credits: producers, songwriters, engineers, session musicians. Most discovery tools ignore them entirely. Which is odd, because credits catch the connections listening data never will.

When the same producer keeps showing up on records across different genres, that's not a coincidence. It's a map. A.G. Cook executive-produced Charli XCX's Brat, with production credits on 13 of its 15 tracks, and founded PC Music, the label behind SOPHIE and a wave of deconstructed pop. Follow his credit trail and you end up in hyperpop and avant-club, scenes listening data wouldn't have pointed you toward. Or take Four Tet: he arranged Madlib's Sound Ancestors, connecting UK electronic to the LA beat scene in a way no similar-artist algorithm would ever surface. Listening data can't see that link. Credits can.

Spotify seems to agree. In March 2026 they launched SongDNA, a Premium-only feature that maps connections between songs through shared producers, samples, and covers, powered by their WhoSampled acquisition. It's a step toward making credits browseable. But it's manual exploration - you tap around and see what's connected. It doesn't monitor anything, doesn't alert you to new releases, doesn't build a playlist.

I built Tracknack to do that part. You tell it which artists and labels you want to follow. It keeps a playlist updated on Spotify or Apple Music with their new releases and sends you email alerts. Because it pulls credits from both Spotify and Discogs, it also catches releases where someone you're tracking shows up as a co-producer or engineer, not just under their main name. SongDNA lets you explore what already exists. Tracknack watches for what comes next.

Most people pick one discovery method and stick with it until it stops working. A better move is running a few in parallel. RYM surfaces charts the algorithm won't build. Bandcamp holds music that never reaches streaming. And credits expose the wiring between records - producers, labels, collaborators - that shared-listening data can't see. None of these alone gets you everything. All three together get close.

Sources and notes

Checked on April 2, 2026.

Stop missing releases.

Tracknack digs through album credits — producers, engineers, labels — so you don't have to. One Spotify or Apple Music playlist, always up to date.