Workflow guide

Peak normalization vs loudness normalization

Understand what peak and loudness normalization change, why neither replaces mixing, and how to check the result.

Concept illustration of audio recording and a waveform
Concept illustration · Creator Edit Lab

Peak normalization sets gain from the largest measured peak. Loudness normalization sets gain from a loudness measurement over the program. They answer different questions, and neither automatically repairs distorted audio or makes an uneven dialogue mix consistent.

On this page
  1. Choose the measurement for the job
  2. An arithmetic example
  3. A practical order of work
  4. Know when a limiter is changing more than gain
  5. Sources and scope

Choose the measurement for the job

Two normalization goals
MethodWhat controls the gainWhat it does not establish
Sample-peak normalizationThe largest sample amplitudePerceived program loudness or true-peak behavior after encoding
Loudness normalizationA chosen loudness measurement and targetA consistently balanced mix or recovery of clipped source audio

A quiet interview with one loud click can behave very differently under these methods. A sample peak can restrict how much gain is added even when most dialogue remains difficult to hear. Removing the click or adjusting the mix may be more useful than choosing a more aggressive normalization target.

An arithmetic example

Suppose the highest sample peak is −10 dBFS and you choose a sample-peak target of −3 dBFS. A gain-only operation adds 7 dB to the entire signal. The relationship between the quiet and loud sections remains unchanged. This is a numerical example, not a measured recording.

A practical order of work

  1. Listen for clipping, clicks, noise and inconsistent dialogue before changing global gain.
  2. Correct distracting events and balance the useful sections.
  3. Measure the full program with the meter and integration method your delivery requires.
  4. Apply gain or suitable processing to reach the selected target without exceeding the relevant peak limit.
  5. Export, reopen and measure the delivered file rather than relying only on the timeline meter.

Know when a limiter is changing more than gain

If the selected loudness target would push peaks too high, a processor may limit or otherwise change the signal. That is not merely turning the whole file up. Listen to transients and dialogue intelligibility, and document whether the workflow was gain-only or included dynamics processing.

A compressed delivery codec can change reconstructed peaks. Keep a suitable margin for the actual delivery and verify the encoded result. Continue with LUFS and true peak; for storage planning, use the audio size calculator.

Sources and scope

This is an explanatory workflow, not a measured hardware or software test. Examples are illustrative. Check the current documentation for your version. Sources checked October 6, 2026.