Audio Toolkit

21 June 2026 | 5 min read

What is dBFS, and why your meters disagree with your ears

Decibels relative to full scale. One reference point, no positive numbers, and a hard ceiling you cannot negotiate with.

dBFS means decibels relative to full scale. Zero dBFS is the largest number a digital sample can represent at a given bit depth. Everything below it is negative, and in fixed point audio there is nothing above it.

Why there is no plus side

In 16 bit audio a sample is an integer between minus 32768 and 32767. In 24 bit the range is larger, but it is still finite. When a signal tries to exceed the maximum, the sample is clamped, waveform tops flatten, and harmonic distortion appears. That is clipping, and it is destructive.

Sample peak versus true peak

A meter reading sample peak only looks at the sample values. But a digital to analogue converter reconstructs a continuous waveform between samples, and that curve can overshoot above the highest sample. True peak metering, measured in dBTP, oversamples to catch these inter sample peaks. This is why delivery specs ask for minus 1 dBTP rather than 0 dBFS.

dBFS is not dBu or dBSPL

  • dBFS: digital, relative to the maximum representable value, always negative or zero.
  • dBu: analogue voltage reference, 0.775 V, used on preamps and mixing consoles.
  • dBSPL: acoustic pressure in the room, what a sound level meter shows.
  • The alignment between them is a studio convention, commonly minus 18 or minus 20 dBFS equal to 0 dBu.

Why dBFS does not predict loudness

Peak level says nothing about perceived loudness. A dense pop master and a sparse dialogue scene can both peak at minus 1 dBFS and differ by 15 LUFS in loudness. Peak meters protect you from distortion, loudness meters tell you how loud it will sound to a listener. You need both.

The inspector on this site shows peak level, average level, sample rate, bit depth and channel layout for any file you drop in, entirely in your browser.

Frequently asked questions

Is 0 dBFS the loudest possible level?
It is the largest representable sample value in fixed point audio. Going beyond it clips, so it is a ceiling rather than a target.
Why do delivery specs ask for minus 1 dBTP?
To leave room for inter sample peaks and for the distortion that lossy encoders introduce during playback.
Can 32 bit float exceed 0 dBFS?
Yes, floating point can represent values above full scale without clipping, but they must be brought back down before conversion to fixed point.

Keep reading

No more channel headaches

Get your production audio edit-ready

Split poly WAVs, merge mono tracks, map channels, extract audio from video. Everything runs locally in your browser, nothing is uploaded.

Convert my file free