Free Song Key Finder & BPM Finder

1. Upload an AAC, AIFF, FLAC, M4A, MP3, OGG, or WAV file.
2. Analysis runs automatically — no settings to configure.
3. Get the detected key, BPM, and Camelot code in a few seconds.

Every piece of tonal music sits in a key — a home note and scale the melody and chords are built around. BPMis how fast the track pulses. Together, they're the two numbers DJs and producers need before mixing, remixing, or layering two tracks.

Harmonic mixing — blending tracks with compatible keys — is what separates a set that flows from one that clashes.

A detected key is always either major or minor. Major keys generally read as brighter or more resolved; minor keys read as darker or more emotional. Every major key shares its exact notes with a relative minor key — which is exactly why they sit at the same Camelot number with a different letter (8A and 8B, for example), and why that pairing is always a safe harmonic move.

The Camelot Wheelrenames the 24 musical keys as numbers 1–12 followed by "A" (minor) or "B" (major). From any key, you can safely mix into the same number, the next number up, or the next number down.

Want the full breakdown of how to use this for building a set? Read The Camelot Wheel Explained: Harmonic Mixing for DJs.

Once you've got key and BPM tagged, the next step is grouping tracks by Camelot compatibility and ordering them for energy before you play. Read the 6-Step DJ Set Prep Checklist.

Only the first three minutes are analysed. The server writes a trimmed copy, reads that, and deletes it — which is why a nine-minute mix comes back as fast as a three-minute single, and why a longer file buys nothing in accuracy.

That makes the opening the whole story. A track whose first three minutes represent the song reads well. A track that opens with a long ambient pad or a drum-only intro can read badly no matter how clear the rest of it is, because the analyser never reaches the rest. If that's your track, trim to a section with the harmony in it using the Audio Trimmer and analyse that instead.

Beyond the intro: consistent tempo and clear harmonic content help, while live recordings, heavy distortion, mid-track tempo changes and spoken-word audio all give the analysis less to lock onto. If your recording has significant background noise, running it through the Noise Remover first can improve detection.

How accurate is it, measured? BPM detection scores 85% exact on the GiantSteps tempo set after moving to a pretrained model; key detection is around 50% and is the harder problem. The full write-up — what the detectors are, what failed, and how the numbers were measured — is in BPM Detection: From 42% to 85% Accuracy With a Pretrained Model. For the plain-language version of why a reading can come back half or double, see how key and BPM detection works.