Tool 01
Spectrum Analyser
Load a loop or a mix and read it: level, crest factor, spectral tilt and the balance between nine frequency bands. Nothing is uploaded, and analysis runs the moment a file is decoded.
Analyser Workspace
or drop a file here. WAV, MP3, OGG, FLAC and M4A are decoded by the browser.
Load a file or a demo loop to begin.
- Peak
- — dBFS
- RMS
- — dBFS
- Crest Factor
- — dB
- Crest Reading
- —
- Spectral Centre
- — Hz
- Tilt
- — dB/oct
- Duration
- —
- Integrated Loudness
- — LUFS
- Loudness Range
- — LU
- Clipped Samples
- —
Tone Report
Load audio to generate a report. The figures below come from an offline pass over the whole file, not the live display, so short transients are not missed.
—
- Tilt
- Dynamics
- Balance
| Band | Centre | Move | Q | Why |
|---|---|---|---|---|
| No analysis yet. | ||||
How to Read the Numbers
Peak is the single loudest sample, in dBFS, where 0 dBFS is the point at which the file would clip in a fixed-point converter. Anything at or above −0.1 dBFS is effectively at the ceiling. RMS is the average level and is the better guide to perceived loudness. Crest factor is the difference between the two: the amount of transient information the material still has.
Spectral centre is the power-weighted mean frequency. It is not a note value; it is a rough indicator of where the file's energy sits. A kick-and-bass loop will centre well below 500 Hz; a hi-hat-heavy loop will centre far higher. Tilt is a least-squares slope across 200 Hz to 8 kHz, in dB per octave. A flat tilt means the midrange is even; a negative tilt means the top end falls away.
What the Band Table Is For
The nine bands are contiguous, so their shares add up to 100 per cent. A band that sits more than about 3 dB above its neighbours is a candidate for a cut; one well below them is usually better left alone than boosted, because a boost raises whatever noise is already there.
Treat the suggested moves as a first pass, not a verdict. Half a decibel at the right frequency is often enough, and it is always worth comparing bypassed against processed at matched loudness before committing.
Limits
- The analysis is mono-summed, so phase differences between channels are not reported.
- Very long files cost memory; keep test material to a few minutes.
- Loudness normalisation (LUFS) is not yet computed on this page.
Questions
What Does Crest Factor Tell Me?
Crest factor is the gap between the loudest peak and the average level. A wide gap (over about 16 dB) means transients are intact and there is room to compress before anything sounds squashed. A narrow gap (under 8 dB) means the material is already dense.
Is My Audio Uploaded Anywhere?
No. The file is read and decoded in the browser tab using the Web Audio API. No network request containing audio is made.
Why Does the Spectrum Look Different from My DAW Analyser?
Analysers differ in FFT size, averaging and the smoothing applied between frames. This one uses a 4096-sample transform with a 0.75 smoothing constant and a logarithmic frequency axis, which favours readable low-end detail over speed.
What Is Spectral Tilt?
Spectral tilt is the slope of the spectrum across 200 Hz to 8 kHz, measured in decibels per octave. A negative value means the top end falls away from the bottom, a positive value means the opposite, and a value near zero means the midrange is even.