How everysong works.
You drop an audio file. We return 13 audio traits and 20 ranked Creative-Commons matches. The whole pipeline runs in 10-15 seconds per readout (allow about a minute for the engine to wake up if it's been idle). This page goes through it slowly. It explains each step, which traits you can trust, and where the limits are.
If you got here looking for a way to find royalty-free music that sounds like a specific song, you're in the right place. People also describe everysong as a music similarity search, an audio fingerprint tool, a copyright-safe music finder, a creative-commons alternative finder, a free song-ID-style tool that returns similar tracks instead of identifying your upload, and a BPM / key / LUFS analyzer with a built-in match engine. They're all describing the same loop: upload → 13-trait audio fingerprint → 20 closest CC-licensed neighbours.
What happens to your song, in five hops
MP3, WAV, FLAC, M4A, or OGG up to 30 MB. The file lands on the API just long enough to analyse, then it's deleted. We never keep the audio.
The listening is done by Vera, my own audio engine. She measures tempo, key, mode, loudness in LUFS and the spectral stats straight from the signal, and turns the sound into a 512-dimensional fingerprint (built on the open CLAP model) that captures the overall feel. Give her the same file twice and you get the same answer.
From that fingerprint Vera reads five named traits: energy, valence (happy/sad), danceability, acousticness, instrumentalness. Together with the eight signal traits from hop 02, that's the 13-trait readout. Her genre call is learned from about 900 labelled songs rather than guessed from word prompts.
Your fingerprint is compared against 12,751 tracks released by their artists for commercial use on Jamendo, every one analysed by Vera the same way as your upload. The closest by sound go through, then a compatible tempo and key move a track up the list, so a swap lands on your edit's beat.
You land on a readout page: 13 trait values up top, 20 ranked matches below, license badge on every match (CC BY, CC BY-SA or CC0: never BY-NC, never restricted), a preview that streams from Jamendo, and a link to get the track on its Jamendo page. There's a permalink so you can share the readout. That's everything, and none of it is behind a paywall.
How we know she's any good
Before Vera was let near your uploads she sat a blind exam: 1,137 songs across 18 genres and six decades, heard only as 30 second clips with no catalogue help, marked against the real catalogue facts. Then torture tests with answers we know for certain: drum loops at set tempos, chords in all 24 keys, songs pitched, sped up, squashed and turned down.
- Tempo: right 75% of the time exactly, 92% counting half or double time (510 songs with a known BPM).
- Key: 24 out of 24 on test chords, and she follows a pitch shift 84% of the time on real songs.
- Genre: in the right family 68% of the time, from sound alone.
- Tough: answers held on 38 of 40 songs turned down 12 dB or squashed into a 64 kbps MP3.
Published as measured, misses included. The full method lives on Vera's own page.
What changed
Updated 17 November 2026. Before and after, straight from the records. Method: same test, old engine against new.
- Catalogue: 5,079 tracks when everysong relaunched on 13 November 2026, 12,751 by 15 November. Same licence filter (CC BY, CC BY-SA and CC0, never BY-NC), and every track measured by Vera.
- Tempo: on the same 30 songs, within 5 BPM with half or double time allowed, Vera went from 71% in July to 86% with her new tempo picker on 12 November. Her average miss dropped from 6.98 BPM to 3.64.
- Genre: in the right family 63% of the time when she guessed from word prompts, 68% once she learned it from 898 labelled songs, checked on songs she never trained on. Her favourite label used to be "latin" on 27% of songs. Now no label tops 7%.
- Matching: since 12 November, matches are shortlisted by sound, then a compatible tempo and key move a track up. There's no published before and after score for that yet, so we're not quoting one.
- Free limit: the daily upload limit had never actually switched on, live included. Fixed on 13 November.
The 13 things Vera measures
v1 ships 13 trait readouts, colour-labelled by how much you should trust them. The boring stuff (signal processing) is solid. The mood-ish stuff is a useful second opinion.
What "GREEN" and "AMBER" actually mean
Signal-processing traits. BPM, key, LUFS, spectral stats. Decades of published research underneath. Well-validated on standard MIR benchmarks. These are the numbers you can quote in a mix note or a brief without hedging.
Mid-level descriptors. Mood, energy, danceability, acousticness. CLAP zero-shot makes educated calls based on millions of hours of training audio: but they're statistical guesses, so treat them as a useful second opinion. The label is right next to the reading so you never have to guess.
No red-tier traits ship in v1. We considered genre-classification and "instrument detection beyond instrumentalness": both either leak training-set bias or aren't reliable enough to publish a number for. We'll add them once they are.
Who it's for, really
Find the song that sounds like the one stuck in your head.
Pick a track and get 20 others that actually sound like it. They're matched on the sound itself, rather than genre tags or what other people listened to.
Find music for your video, podcast or stream that you're allowed to use, for free.
YouTubers, podcasters, streamers, indie game devs: upload a reference you love, get CC-licensed matches with the license shown up front. Always verify the license at the source before publishing: Creative Commons is freely usable, but not strike-proof if you ignore the attribution rules.
Compare your mix with the reference.
Put the two readouts side by side and see how close your mix is on BPM, key, LUFS, spectral balance and stereo width. It's the boring stuff that matters at mastering time.
Just geek out over a song you love.
Upload a song you love and have a look at what's going on in it: its tempo, key, loudness and mood.
Honest answers to the obvious questions
How it stacks up against other tools
People often ask how everysong sits next to other audio tools. The short answer: it's purpose-built for indie creators who need a copyright-safe replacement for a copyrighted reference track, and it's free. Here's the longer answer:
Song-ID apps identify the recording. everysong finds tracks that sound similar.
If you upload "Bohemian Rhapsody" to a song-ID app, it tells you it's "Bohemian Rhapsody". If you upload it to everysong, you get 20 Creative-Commons tracks that share its BPM, key, energy and overall feel: songs you can actually use in your video, podcast or game without paying licensing fees. They do different jobs.
Industry audio-AI platforms aimed at labels and supervisors. Subscription, hundreds-to-thousands per month.
These are excellent products and substantially more comprehensive than everysong: they handle catalogues of millions of tracks for music supervision, sync licensing, and label A&R. But they're not for the indie YouTuber who just needs a free track that sounds like the demo they put on their rough cut. everysong is the free indie-tier version: smaller catalogue, narrower features, no subscription.
AI music generators. They invent new tracks. everysong matches existing ones.
Generators are great when you want exactly-fitting bespoke audio. The trade-offs: AI-generated music has unsettled copyright status, often sounds generic, and rarely matches the specific feel of a reference. everysong returns real recordings by real humans, with clean Creative-Commons licenses and verifiable provenance.
Royalty-free subscription libraries. Big catalogue, monthly fee.
Subscription libraries are the right choice if you publish a lot and want a constant tap. everysong is for the case where you have this specific reference track and want a free, license-safe match: not a curated mood-based playlist.
Free and hand-picked, but no similarity search.
The free music libraries built into video platforms are great for browsing by mood and genre. everysong starts from your reference track and goes outward by audio similarity, so it's a different way of finding music.
What people search for to find this
People stumble onto everysong via dozens of different phrasings. If any of these sound like you, you're in the right place:
- "Free music for YouTube videos": yes, the matches are CC-licensed and YouTube-monetisation-safe (with attribution where the license requires it).
- "Royalty-free music similar to [song]": that's literally the core feature: upload the reference, get 20 CC matches.
- "No copyright music that sounds like [song]": same thing, different phrasing. The catalogue is filtered to CC0, CC BY, CC BY-SA only: no BY-NC, no commercial restrictions beyond attribution.
- "DMCA-safe background music for Twitch / podcast / video essay": Creative Commons tracks won't trigger DMCA takedowns when used per the license terms. Always verify on the source page before publishing.
- "Find a creative commons alternative to [popular song]": that's the use case the whole tool was built around.
- "Audio similarity tool" / "music similarity search" / "song matcher": same engine, technical phrasing.
- "BPM analyzer online" / "key detector" / "LUFS meter for free": yes, the readout includes all three plus 10 other audio traits, free, no account needed.
- "Audio fingerprint tool" / "music DNA" / "audio embedding similarity": under the hood, that's exactly what's running: a CLAP audio embedding decoded into 13 named traits, then matched by cosine distance.
- "Cheap audio-AI alternative" / "indie audio AI" / "audio similarity API for one person": this is the small-budget version.
- "Free music for indie game development": entire use case. Drop a reference from your favourite game soundtrack, ship CC-licensed alternatives with your build.
- "Free music for podcasts" / "podcast intro music free": works for intros, beds, transitions, ad-spot music. License shown on every match.
- "Free music for streamers" / "Twitch DMCA safe music": yes, Creative Commons tracks are stream-safe under their license terms.
- "Free song-ID alternative": close but not quite. Song-ID apps identify; everysong finds similar. If you want the original identified, use a song-ID app. If you want a license-clean lookalike, use everysong.
- "How do I find royalty-free music that sounds like a specific song": exactly this. Upload, wait 15 seconds, browse 20 matches.
What's in the Creative-Commons catalogue
The catalogue is 12,751 tracks: CC BY and CC BY-SA tracks from Jamendo, one of the largest CC music platforms (broad genre coverage, lots of EDM, rock, acoustic and world), plus 5 CC0 tracks. everysong doesn't host the audio: previews stream from Jamendo, and you get each track on its Jamendo page.
License coverage: CC BY (attribution required, commercial use allowed), CC BY-SA (attribution + share-alike) and CC0 (public-domain dedication, no attribution needed). We exclude BY-NC tracks because they prohibit commercial use: most creators end up needing commercial-safe music. There's also a "strictest only" filter that hides BY-SA tracks if you want zero share-alike obligations on a client deliverable.
Other names for this kind of tool
For posterity and Google: this category gets called music information retrieval, audio similarity search, music recommendation by audio, audio fingerprinting and matching, content-based music retrieval, audio embedding similarity, CLAP-based music search, music feature extraction, audio MIR (Music Information Retrieval), and informally music DNA or audio genome. Academic papers tend to say "MIR"; vendors tend to say "audio AI"; creators tend to say "find similar songs".