Methodology
BPM, key, loudness and peaks answer different questions. Find out how the tools calculate them and when to check the result by listening.
A starting point for tempo
The index contains editorial tempo references by genre, with production notes. Use it to choose a starting point; the analyzer estimates the tempo of a particular audio file.
Dataset version: 2026-07-10-slashfix
Explore tempo ranges →From your own audio
Your browser decodes and measures the file locally. The audio file is not uploaded to a server.
01 · LUFS / LRA
Integrated LUFS describes overall loudness. LRA describes how much the measured loudness changes.
The implementation uses K-weighting, 400 ms blocks and absolute / relative gating. LRA uses loudness percentiles. These are local calculations; this page does not claim certified compliance.
02 · TRUE PEAK
The dBTP result estimates peaks in the reconstructed signal, which can be higher than the stored samples.
The engine uses 4× oversampling with a windowed-sinc FIR interpolator. Lossy encoding can change the peaks: measure the encoded file again.
03 · BPM
A tempo estimate to check against the rhythm. Half-time and double-time readings can both fit the same music.
Spectral-flux autocorrelation searches from 50 to 200 BPM, with peak refinement and half / double-time heuristics.
04 · KEY
An estimated major or minor key. Confirm the result with the melody and bass before choosing a harmonic transition.
A 12-bin chromagram is compared with major / minor profiles. The Camelot position is derived from the estimated key.
The interface compares your measured LUFS with reference values configured in the tool. It does not simulate the full playback system or certify acceptance by a service. Use the comparison alongside peaks and listening, not as a compulsory mastering target.
Measure your file →Reproducible checks · 30 October 2026
Generate a 1 kHz sine at 48 kHz, with a peak amplitude of 0.1 (−20 dBFS), for 8 seconds. Measure it as one channel, then as two identical channels. Our local calculation was compared with the independent FFmpeg ebur128 filter.
| Signal | Reference | Observed agreement |
|---|---|---|
| 1 kHz · mono | −23,0 LUFS | < 0,1 LU |
| 1 kHz · dual mono | −20,0 LUFS | < 0,1 LU |
| 4 s silence + 8 s mono tone | −23,1 LUFS | < 0,15 LU |
| Four EBU loudness-step examples | LRA: 5 / 10 / 15 / 20 LU | ±1 LU |
The LRA checks use the level-step signals described in EBU Tech 3342. Passing these examples does not certify every audio format, sample rate, true-peak case or musical programme. These tolerances describe the tests above, not a universal accuracy promise.
Processing scope: mono or stereo; tempo and key inspect the first 20 seconds; integrated loudness and LRA inspect up to the first 60 seconds. Longer files are explicitly marked as excerpts. Peaks inspect the decoded file. Silence and insufficient windows produce an unavailable reading rather than a default tempo, key or invented dynamic range.
Keep listening
ITU-R BS.1770 · EBU Tech 3342 · LRA
References explain the measurement methods. They do not certify this implementation. Reviewed 6 October 2026.