How Tempo & Key Detection Works

1. 01

   Your browser decodes the file

   The Web Audio API reads the selected file into temporary audio data. Stereo channels are mixed for analysis. The file is not uploaded to an audio-processing server.

2. 02

   Analysis moves off the main thread

   A Web Worker resamples the signal to 16 kHz and runs the Essentia.js tempo and key estimators. Loading this analysis code is deferred until you choose audio.

3. 03

   Tempo candidates are calculated

   The BPM estimate is shown as both a rounded and decimal value. Half-time, double-time and milliseconds per beat are derived to make rhythmic ambiguity visible instead of hiding it.

4. 04

   Harmonic profiles estimate the key

   The primary key profile returns a tonic and major or minor mode. A second profile is shown as a possible alternative when it disagrees with the primary estimate.

5. 05

   Music theory adds useful context

   Local, dependency-free functions map the result to its Camelot code, relative key and compatible keys. They do not modify the source audio.

The output is a best-effort estimate, not a verified fact about the recording. Half/double-time rhythm, key changes, sparse harmony, percussion, speech and noise can all produce a plausible but incorrect answer.

Page assets still come from the web host, but selected audio, filenames, decoded PCM and analysis results remain in the browser. There is no upload API, account, cloud audio store or analytics integration in the current application.