Peak normalisation scales a file until its loudest single sample reaches the top. That is not what anyone hears: one click or one snare sets the level for the whole recording, and two files treated that way still play back at different volumes. This measures loudness instead, using the gated LUFS method broadcasters and streaming platforms use, and turns the file up or down by the difference. The ceiling is checked as true peak — the level the waveform reaches between samples, which is what a converter has to reproduce — so the result does not clip on playback even when it reads under 0 dBFS.
Upload
Adjust & Preview
Pixlane is loading on your device. Supported file contents are not uploaded for processing. Site analytics and the assistant are separate; tool-specific limits may apply.
Match audio to a loudness target the way broadcasters and streaming platforms measure it, with a true-peak ceiling.
Peak normalisation scales a file until its loudest single sample reaches the top. That is not what anyone hears: one click or one snare sets the level for the whole recording, and two files treated that way still play back at different volumes. This measures loudness instead, using the gated LUFS method broadcasters and streaming platforms use, and turns the file up or down by the difference. The ceiling is checked as true peak — the level the waveform reaches between samples, which is what a converter has to reproduce — so the result does not clip on playback even when it reads under 0 dBFS.
LUFS measures loudness the way hearing does — frequency-weighted and averaged over time — while a peak meter reports one sample. A quiet podcast with a single door slam has a high peak and a low loudness, so peak normalisation leaves it quiet. Loudness normalisation raises it to match everything else.
-14 LUFS is around what music streaming services play back at, -16 is common for podcasts, and -23 is the European broadcast standard. If your file is going somewhere with its own specification, use that number.
Between two samples the reconstructed waveform can rise above both of them, so a file that reads -0.1 dBFS can still ask a converter for more than full scale and distort. Measuring 4x between samples finds that overshoot. A -1 dBTP ceiling leaves room for it, which is why delivery specifications ask for it.
Yes, and it is the part cheap implementations skip. Blocks below -70 LUFS are dropped, then blocks more than 10 LU below the rest are dropped too, so pauses and silence do not pull the measurement down and cause the file to be over-boosted.
Because writing MP3 again would put a second generation of lossy compression on the audio. WAV keeps the corrected signal exactly as measured; it is simply larger.
Supported audio content is not uploaded for processing. Reading, measuring, and export run in your browser; site analytics and the assistant are separate.