SonoVault is in open beta. Signups are live. Get your free API key →
For startups & builders

Add music recognition
to whatever you're building

Send a fingerprint or an audio file and get the matching track back with ISRC, artist, release, and cross-platform IDs, across 90M+ recordings. One key, no OAuth, no acoustic index to build yourself. A free tier to start, and audio recognition on paid plans from 29 euro a month.

By Christiaan Maks, Founder & DeveloperUpdated

What you could build

Recognition is one endpoint. Here is what you can build on it.

Every one of these ships on top of a single call to POST /v1/tracks/identify. Identify the recording, then use the id to pull in whatever metadata the product needs.

Setlist & now-playing logger

For a venue screen, DJ software, or a personal listening-history app: fingerprint whatever is playing and log the exact recording, timestamped, with its ISRC.

Powered byPOST /v1/tracks/identify
Radio & podcast airplay tracker

Identify tracks inside recorded episodes, or point stream monitoring at a live broadcast and build a continuous spin log with the ISRC and label a report of use needs.

Powered byIdentify + stream monitoring
DJ 'what's this track?' tool

Let users identify a track from a short recording, then resolve it to a clean title, its ISRC, and buy/listen links on Beatport, Spotify, and Apple Music.

Powered byIdentify + cross-platform links
Auto-tag music in user uploads

On a UGC or creator platform, detect the recording behind a video or voice memo and attach its ISRC so rights and reporting are handled from the moment of upload.

Powered byIdentify then GET /v1/tracks/:id
Sample & clearance helper

Identify the recording in a sample, then cross to its ISRC and the ISWC composition behind it to start the clearance conversation with the right rights holders.

Powered byIdentify + ISWC lookup
Second-screen companion

Recognise what a user is listening to or watching and surface metadata, context, or related links in real time, without asking them to type anything.

Powered byPOST /v1/tracks/identify
How it works

From audio to a track id in three steps

You bring the audio; we bring the catalog and do the matching.

  1. 1

    Fingerprint on the device, or upload

    Run Chromaprint's fpcalc -raw client-side and send a ~1 KB integer array, so no audio ever leaves the device. Or POST the raw file and we fingerprint it for you. Same endpoint, same response.

  2. 2

    We match 90M+ recordings

    Your fingerprint is matched against an acoustic index spanning the whole catalog, robust to bitrate, format, and encoding differences.

  3. 3

    Get an id + confidence, then hydrate

    You get ranked matches, each slim: id, title, artists, and a 0 to 1 confidence score. Look the winner up by id for its ISRC, genre, label, release date, and cross-platform IDs.

The call

One POST, in your language

The recommended path: fingerprint with Chromaprint's fpcalc client-side and send the integer array. Nothing but a ~1 KB fingerprint leaves the device.

curl
# Fingerprint on the device, no audio leaves it
fpcalc -raw -length 30 track.mp3     # -> FINGERPRINT=1684564321,-1109695284,...


curl -X POST https://api.sonovault.now/v1/tracks/identify \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"fingerprint":[1684564321,-1109695284,1085316]}'

Prefer not to run fpcalc? Send the raw file and we fingerprint it server-side (up to 100 MB, any ffmpeg-decodable format).

bash
# No fpcalc? POST the raw file and we fingerprint it server-side.
# Whole file is the reliable choice; 128 kbps is plenty (320/lossless just costs more).
curl -X POST https://api.sonovault.now/v1/tracks/identify \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/octet-stream" \
  --data-binary @track.mp3

The response is deliberately small: a ranked results array, best match first, each with a confidence score. Hydrate the winner by id for ISRC, genre, and cross-platform links.

json
{
  "matched": true,
  "results": [
    {
      "id": 123,
      "title": "One More Time",
      "artists": [{ "id": 9001, "name": "Daft Punk", "is_primary": true }],
      "confidence": 0.98
    }
  ],
  "credits_charged": 10
}
Why build on SonoVault

The recognition, without the infrastructure

Rolling your own means building an acoustic index over tens of millions of tracks, then wiring up every metadata source separately. This is that, minus the months.

No reference database to build

The hard part of recognition is the 90M+ acoustic index and keeping it current. That is hosted for you. You send a fingerprint and get a match, with nothing to train, store, or scale.

One key, no OAuth

A single x-api-key. No token refresh, no per-platform integration, and no commercial-use approval queue to clear before you can ship.

Fingerprint or upload

Match client-side with fpcalc (private, ~1 KB) or hand us the file and skip the tooling. Both hit the same endpoint and return the same shape.

A match, plus the metadata

Every hit resolves to ISRC, genre, label, release date, and IDs on Spotify, Apple Music, Tidal, Beatport, Discogs, and MusicBrainz, so recognition feeds straight into the rest of your product.

Roll your ownWith SonoVault
Reference catalogAssemble and fingerprint tens of millions of tracks yourself.90M+ recordings already indexed and updated daily.
FingerprintingStand up Chromaprint / an index and tune matching.One POST; fingerprint client-side or let us do it.
Metadata after a matchWire up Spotify, Apple, Discogs, MusicBrainz separately.ISRC, label, genre, and every platform ID in one lookup.
AccessPer-platform OAuth, tokens, and approval reviews.One x-api-key, free tier, self-serve.
Time to first matchWeeks of infrastructure before a single result.Minutes: sign up, send a fingerprint, read the id.

Want the full technical rundown of the endpoint, from fingerprint formats to retry behaviour? The audio recognition reference has it, and the API docs cover every route you would hydrate a match with.

Audio recognition is a paid-plan feature. A fingerprint lookup is 10 credits; a raw-audio upload is 10 + 1 credit per MB uploaded, so the fingerprint path is cheapest. Matches are factual catalog data aggregated from third parties: they can be incomplete or wrong, so verify before any rights, royalty, or licensing use. See pricing and the full API reference.
FAQ

Frequently asked questions

No. That is the part SonoVault hosts for you. We maintain the acoustic fingerprint index over 90M+ recordings; you send a Chromaprint fingerprint (or raw audio) and get the matching track back. There is nothing to train, store, or keep in sync on your side.
Yes. Run Chromaprint's fpcalc -raw on the device and send the integer array (about 1 KB) as JSON. The raw audio stays local and you save the bandwidth. If you would rather not bundle fpcalc, POST the file as application/octet-stream and we fingerprint it server-side. Both paths hit POST /v1/tracks/identify and return the same shape.
You're billed per megabyte you upload, so file size sets the cost, not audio quality. 128 kbps is plenty for recognition; 320 kbps or lossless just costs more for no better match. What improves accuracy is duration: the matchable section can sit anywhere in a track, so a short clip is cheapest but can miss the match, while the whole file is the reliable choice. About 10 seconds is the practical minimum. Cheapest is a short clip at ~128 kbps; best value is the whole file at 128 kbps.
A matched boolean and a ranked results array. Each result is slim on purpose: id, title, artists, and a confidence score from 0 to 1. When you need the ISRC, genre, label, release date, or cross-platform links, look the track up by its id with GET /v1/tracks/:id.
No. Uploaded audio is fingerprinted and discarded within the same request. It is not written to disk or retained; only the fingerprint is used for matching.
You can create a free account and build against the catalog API (search, ISRC lookup, platform links) at 1,000 requests a month with no credit card. Audio recognition runs on paid plans, from Starter at 29 euro a month. A fingerprint lookup is 10 credits; a raw-audio upload is 10 credits plus 1 per megabyte uploaded, so the fingerprint path is cheapest. There is no contact-sales wall and no per-platform approval.
Shazam has no open API to build products on, and dedicated recognition vendors return a match and little else. SonoVault gives you recognition plus the full metadata graph behind one key: the ISRC, the ISWC composition, and the track's IDs on every major platform. It is a fit when identifying the track is step one and you actually need the data around it. Names are used factually; SonoVault is not affiliated with Shazam or ACRCloud.

Start building with audio recognition

Create a free account, grab a key, and send your first fingerprint straight away. No approval queue, no index to maintain.

Get a free API keyRead the technical guide →