Loudness Normalizer

Knowledge Base

How This Tool Works

The Normalizer measures your file's integrated loudness using the same K-weighted browser analysis as the LUFS File Checker, calculates the gain adjustment needed to hit your target, and applies it as a straightforward, transparent volume change. Use a validated meter for formal delivery acceptance.

Why Peak Safety Sometimes Limits the Result

If pushing your file all the way up to the target LUFS would cause the loudest moments to exceed -1 dBTP (a common safe ceiling used by broadcast and streaming platforms), this tool automatically reduces the applied gain to keep true peak safe, and shows you the resulting loudness instead of forcing potential clipping. This is the same tradeoff a mastering engineer makes manually.

Gain Normalization vs. Compression

This tool only changes overall volume. It doesn't compress or otherwise reshape your dynamics. If your file's loudness problem is really about inconsistent levels between sections rather than overall level, see the Ducking and Sidechain Compression or Compression Basics guides for dynamics-focused solutions instead.

FAQ

Q: Will this fix a file that's already clipped?
A: No, clipping is permanent distortion in the original recording. Gain adjustment (up or down) can't undo it. This tool only corrects overall loudness on otherwise clean audio.

Q: What format does it export?
A: 16-bit PCM WAV at your original file's sample rate and channel count, universally compatible with every podcast host and video editor.

What normalising does and does not fix

Normalising applies one gain change across the whole file so the measured loudness lands on your target. It moves everything by the same amount, so the relationship between the quiet parts and the loud parts is unchanged. This makes it exactly the right tool for a file that is simply too quiet, and the wrong tool for one where levels wander.

Wandering levels are a dynamics problem and need a dynamics solution. If a guest drifts off microphone halfway through, gain applied to the whole file moves the good part and the quiet part together and the difference between them survives. Compression, or manual level rides, is what closes that gap, and normalising afterwards sets the result to target.

Leave true peak headroom. Setting a file to its loudness target can push peaks close to maximum, and lossy encoding reconstructs peaks slightly higher than the samples suggest. Delivering with a small amount of headroom below full scale, commonly one decibel, avoids distortion appearing only after the platform re-encodes, which is the most frustrating version of the problem because your own file sounds fine.

Normalise last. Any processing applied afterwards, including equalisation, compression or adding music, changes the loudness and invalidates the measurement. Doing it as the final step before export is the only way the number you measured is the number you deliver.

Check the result rather than trusting the figure. Hitting a target exactly while sounding wrong is entirely possible, particularly on material with a lot of low frequency energy, which contributes to the measurement differently from how it contributes to perceived level. The meter is a check on your ears, not a replacement for them.

Vocabulary

Normalisation. Applying a single gain change so a file reaches a target level. Loudness normalisation targets perceived loudness in LUFS. Peak normalisation targets the highest sample instead and says nothing about how loud the result sounds.

Gain. A change in level applied equally to everything. Distinct from compression, which changes loud and quiet parts by different amounts and therefore alters the relationship between them.

Headroom. The gap left between your highest peak and the maximum a format can hold. Leaves room for peaks that lossy encoding reconstructs slightly higher than the original samples.

dBTP. Decibels true peak, the unit for measuring peaks between samples rather than at them. Delivery specifications state a true peak ceiling because the reconstructed waveform is what a listener's converter actually produces.

Worked examples: gain, and when the peak limit stops it

The tool sets the gain that moves the measured loudness to the target. It then checks the true peak. If that gain would push the true peak above -1 dBTP, it applies less, so the file ends short of the target. The table runs four files through the tool's logic. Three use the -16 LUFS Apple Podcasts preset and one uses the -23 LUFS broadcast preset.

The second file shows the limit at work. It needs 6.0 dB, but its peak sits only 3.0 dB below the -1 dBTP ceiling, so the tool applies 3.0 dB and the file ends 3.0 LU quieter than the target. Turning a file down never triggers the limit. To bring the second file to -16 LUFS, reduce its peaks first with a limiter, then measure again. The two targets come from Apple's podcast recommendation and EBU R 128, both listed with their sources on the Standards page.

Vocabulary

Integrated loudness. Loudness averaged over the whole file and given in LUFS. It is one figure for the complete programme, so a short loud passage moves it less than it moves a momentary reading.

Peak limited. The state where the tool applies less gain than the target needs, because the full gain would push the true peak above -1 dBTP. The result then ends quieter than the target you chose.