Audio Volume Normalizer

Continue in the tool

More pages for this tool

what is LUFS audio normalization

LUFS, RMS, and peak: understand audio volume measurements

Audio volume can be measured in three main ways: peak (the maximum instantaneous amplitude value), RMS (the quadratic mean representing perceived power), and LUFS (integrated loudness weighted by human perception, per ITU-R BS.1770). Each measure serves a different purpose in the production chain.

When to use peak, RMS, or LUFS as a reference

Peak in dBFS is the maximum instantaneous amplitude value, and its only job is telling you whether the signal will distort: keep the sample peak below 0 dBFS and, ideally, the true peak (which accounts for reconstruction between samples) below -1 dBTP before exporting. RMS in dBFS is the quadratic mean of the signal energy, and it is useful to compare average levels between tracks or check consistency in narration, but it does not weight frequencies the way the human ear does.
LUFS is defined by the ITU-R BS.1770-4 standard, the same technical basis behind the EBU R128 used in professional broadcast: the signal runs through a filter that boosts treble above roughly 1682 Hz and cuts bass below about 38 Hz, modelling how the human head accentuates certain frequencies, then the energy is summed in 400 ms windows with 75% overlap between them. Two gates discard sections that do not represent the content's real volume: an absolute one at -70 LUFS ignores pure silence, and a relative one, 10 LU below the already-filtered mean, ignores sparse speech passages that would drag the average down without representing perceived loudness.
The practical difference between RMS and LUFS shows up in content with uneven dynamics: two tracks with the same RMS can sound noticeably different in volume if one has short, frequent peaks and the other is more constant, because RMS does not weight by frequency nor discard silence before averaging. That is why every streaming platform normalizes by integrated LUFS, not by peak or RMS, to decide how much to turn a delivered track down, or leave it untouched.

Typical reference values

Professional podcast narration

Input: Gravação em estúdio home
Expected output: Pico -3 dBFS, RMS -18 dBFS, LUFS ~-16

Typical values before dynamic compression.

Music for Spotify

Input: Mix finalizado antes de mastering
Expected output: Pico -1 dBFS, LUFS integrado -14

Spotify normalizes automatically to -14 LUFS; deliveries above are not amplified.

Same track, broadcast delivery

Input: Faixa já em -14 LUFS (padrão streaming)
Expected output: Precisa de -9 dB para atender -23 LUFS do EBU R128

Broadcast is the only common target that is a contractual requirement, not a platform recommendation.

Full tool FAQ

What is peak normalization?

Peak normalization increases or reduces the overall gain of an audio file so that the maximum peak amplitude reaches a defined target level, usually -1 dBFS to avoid clipping on playback systems. It is the simplest and most direct normalization method.

What is dBFS?

What is RMS and why does it matter?

What is LUFS and how does it differ from RMS?

What is the difference between peak and loudness (LUFS) normalization?

What is audio clipping?

Why is the exported file WAV and not MP3?

What target peak value is recommended?

Frequently asked questions

Why is the LUFS shown in the tool an estimate?

It is not an estimate: it is the full ITU-R BS.1770-4 algorithm, with a weighting filter, 400 ms blocks at 75% overlap, and the two gating stages (absolute at -70 LUFS, relative 10 LU below the mean). Peak is also measured as a true inter-sample peak, in dBTP, not as an isolated sample peak. What the tool is not is a homologated meter for broadcast delivery, where the contractual requirement is a certified analyser.

Do RMS and LUFS measure the same thing?

No. RMS is a pure quadratic mean of amplitude, with no perceptual filter and no silence discarding. LUFS applies a filter that weights frequencies the way the ear does and discards silence and sections far below the mean before integrating, so two tracks with identical RMS can end up with quite different LUFS.

Which measure should I check first when preparing a track for delivery?

LUFS, if the destination is a streaming platform or broadcast, because that is what it uses to decide playback gain. True peak in dBTP comes right after, as a safety ceiling against clipping during signal reconstruction. RMS is useful internally, to compare tracks within the same project, but no platform normalizes by it.

Next steps

When to use peak, RMS, or LUFS as a reference

Peak in dBFS is the maximum instantaneous amplitude value, and its only job is telling you whether the signal will distort: keep the sample peak below 0 dBFS and, ideally, the true peak (which accounts for reconstruction between samples) below -1 dBTP before exporting. RMS in dBFS is the quadratic mean of the signal energy, and it is useful to compare average levels between tracks or check consistency in narration, but it does not weight frequencies the way the human ear does.

LUFS is defined by the ITU-R BS.1770-4 standard, the same technical basis behind the EBU R128 used in professional broadcast: the signal runs through a filter that boosts treble above roughly 1682 Hz and cuts bass below about 38 Hz, modelling how the human head accentuates certain frequencies, then the energy is summed in 400 ms windows with 75% overlap between them. Two gates discard sections that do not represent the content's real volume: an absolute one at -70 LUFS ignores pure silence, and a relative one, 10 LU below the already-filtered mean, ignores sparse speech passages that would drag the average down without representing perceived loudness.

The practical difference between RMS and LUFS shows up in content with uneven dynamics: two tracks with the same RMS can sound noticeably different in volume if one has short, frequent peaks and the other is more constant, because RMS does not weight by frequency nor discard silence before averaging. That is why every streaming platform normalizes by integrated LUFS, not by peak or RMS, to decide how much to turn a delivered track down, or leave it untouched.

Typical reference values

Professional podcast narration

Input: Gravação em estúdio home
Expected output: Pico -3 dBFS, RMS -18 dBFS, LUFS ~-16

Typical values before dynamic compression.

Music for Spotify

Input: Mix finalizado antes de mastering
Expected output: Pico -1 dBFS, LUFS integrado -14

Spotify normalizes automatically to -14 LUFS; deliveries above are not amplified.

Same track, broadcast delivery

Input: Faixa já em -14 LUFS (padrão streaming)
Expected output: Precisa de -9 dB para atender -23 LUFS do EBU R128

Broadcast is the only common target that is a contractual requirement, not a platform recommendation.

Full tool FAQ

What is peak normalization?

What is dBFS?

What is RMS and why does it matter?

What is LUFS and how does it differ from RMS?

What is the difference between peak and loudness (LUFS) normalization?

What is audio clipping?

Why is the exported file WAV and not MP3?

What target peak value is recommended?

Frequently asked questions