BPM & Key Finder
Tempo, key with Camelot code, tuning reference, and loudness measurements
100% In-Browser · Private & Instant · Free Forever
Your file is decoded, processed, and re-encoded in this browser tab. Nothing is uploaded to an OfflineTTS processing API.
Interactive Workspace
BPM & Key Finder
Measure a track before you work with it. This tool reports the tempo in BPM, the musical key with its Camelot code, the tuning reference the recording was made at, and three loudness figures — peak, average RMS, and crest factor.
How tempo is detected: the audio is reduced to a low-frequency envelope, energy peaks are picked out with a refractory gap that caps the detection at 300 BPM, and the intervals between peaks are collected into a histogram folded into a 70–190 BPM range. The most heavily weighted interval wins. Looking at several peak intervals rather than only the neighbouring one keeps the estimate stable when individual beats are missed.
How key is detected: a chromagram is built from an FFT pass, then correlated against the Krumhansl-Schmuckler major and minor profiles for all twelve tonics. The best match becomes the reported key, and the Camelot code is derived from it so you can check harmonic compatibility with another track.
How tuning is detected: the strongest low-frequency peak in the spectrum is located with parabolic interpolation, and its deviation from equal temperament is reported as a reference frequency. A result near 440 Hz means standard tuning; 432 Hz or 415 Hz indicates a different reference.
What this tool does not do: it does not identify genre, detect instruments, separate stems, transcribe notes, or measure loudness to a broadcast standard. The tempo and key estimates are statistical — they are a good starting point, and they can be wrong on unusual material. Files are decoded, measured, and re-encoded inside this browser tab. Nothing is uploaded and no result is retained after you close the page.
How It Works
Load a track
Add any audio file, or the audio track of a video file.
Measurement runs
Tempo, key, tuning, and loudness are computed in this tab.
Read the confidence
Tempo reports a confidence value; low confidence means check by ear.
Act on the numbers
Use the gain figure with the volume booster, or the key for harmonic mixing.
Features
Tempo Detection
BPM from an onset-interval histogram with a confidence score.
Key & Camelot
Krumhansl-Schmuckler key estimation with the DJ-friendly Camelot code.
Tuning Reference
Reports whether a recording sits at 440 Hz, 432 Hz, or another reference.
Loudness Facts
Peak, RMS, crest factor, and the gain needed to reach -1 dBFS.
Reading Tempo and Key Results Critically
Both tempo and key are statistical estimates, and both can be confidently wrong. A tempo histogram will report a clean peak even when the peak corresponds to a subdivision rather than the beat. A key correlation will report its best match even when the best match is weak. The confidence score exists to communicate that uncertainty, and it should be read as part of the answer rather than as a decoration.
The practical response is to verify against your own ears. Tap along to the reported tempo — if your taps line up with the beat for eight bars without drifting, the number is right. Play the reported scale against the track — if the tonic sounds like home and the third matches the chord quality, the key is right. Ten seconds of checking beats trusting an uncertain figure.
Material that consistently defeats both estimates includes free-tempo recordings, solo piano without a steady pulse, spoken word, and heavily compressed dance music where the kick dominates the spectrum. In those cases the loudness measurements are still reliable even when tempo and key are not.
Loudness Numbers and What to Do With Them
Peak level is the loudest single sample; average RMS is the overall level; crest factor is the distance between them. A crest factor near 6 dB indicates a heavily limited master with very little dynamic range left. A crest factor above 14 dB suggests either a natural, uncompressed recording or a track with one loud transient among quiet material.
The reported normalize gain is the amount needed to bring the file's peak to -1 dBFS. That leaves a decibel of headroom, which is a reasonable precaution against the small peak increases that lossy encoding can introduce. If the file already peaks at 0 dBFS, the reported gain is zero and the peak measurement will flag clipping — in that case the recording is already at its limit and no gain will improve it.
Related Tools
Volume Booster
Make a quiet recording louder and normalise inconsistent levels
Tempo & Speed Changer
Slow down or speed up audio while holding the original pitch
Audio Converter
Change sample rate and channel count, normalize, and re-export
Tone Generator
Test tones, tuning references, and frequency sweeps generated locally
BPM & Key Finder — FAQ
How accurate is the BPM detection?
On electronic and pop music with a clear kick drum, it is usually exact or within one BPM. On live recordings, classical music, free-tempo material, and tracks with heavy syncopation it can lock onto a related tempo such as double or half the true value. The confidence score tells you how concentrated the evidence was.
Why does it report double the BPM I expected?
Tempo detection cannot distinguish between a beat and its subdivision. A 70 BPM track with a busy hi-hat pattern contains strong energy at 140 BPM, and either answer is defensible. The histogram is folded into a 70–190 range specifically to reduce this, but it cannot eliminate it.
What is a Camelot code?
It is a notation for musical keys used by DJs to find harmonically compatible tracks. Keys are numbered 1–12 with an A or B suffix; adjacent numbers and matching letters mix smoothly. The code is derived directly from the detected key and scale.
What does the tuning reference tell me?
It reports what frequency the recording treats as A4. Standard modern tuning is 440 Hz; some orchestras use 442 or 443 Hz, Baroque performances often use 415 Hz, and some producers deliberately work at 432 Hz. Knowing the reference matters when you layer or retune material.
Can I analyse a file I have not downloaded?
No. Analysis needs the audio data, so you must have the file locally. This tool does not fetch from a URL.
More Local Audio Tools
Every audio utility here runs in the browser tab. Browse the full set of editing, conversion, and measurement tools.