What does RMS actually measure?

It is the root mean square of the samples, which for a waveform is a measure of average energy. It responds to how much signal is present rather than to peaks, so a dense mix and a sparse one can share an RMS figure while sounding quite different in loudness. That is why it is used for consistency rather than for perceptual matching.

Why did my file not reach the target?

Because the peak ceiling stopped it. The tool computes the smaller of two gains, the one that reaches the target and the one that keeps the peak under the ceiling, and applies whichever is smaller. When the ceiling wins, the footer says so and names the ceiling that intervened. Bringing the peaks down first lets the level come up further.

Is this the same as peak normalising?

No. Peak normalising targets the loudest sample, so it guarantees no clipping but says nothing about average loudness. RMS normalising targets the average, which is a better match for how loud a file feels, at the cost of needing a ceiling to stay safe.

Why does it measure the channels jointly?

So that a stereo file is treated as one thing rather than as two independent channels. Measuring jointly and applying one gain to both preserves the balance between the sides, and it also prevents a quiet channel from being boosted much further than a loud one, which would distort the image.

Should I use this or the LUFS normaliser?

The LUFS tool if the destination is a streaming service, because that is the measurement those platforms use. This one if the goal is a consistent average across a set of files, or if you specifically want the broadcast-style method. The two can give different results on the same file by a couple of decibels, which is expected.

RMS Normalizer

You Might Also Need

Loudness Normalizer

Normalize to LUFS loudness standards

Audio Normalizer

Normalize audio levels for consistency

Audio Leveler

Automatically level audio volume

Dynamic Range Compressor

Professional-grade audio compression

RMS Normalizer applies one static gain so a file's average level lands on the target you choose, with a peak ceiling that honestly reports when it had to stop short. It is the classic broadcast style of normalising.

What is RMS Normalizer?

Drop a file, set the target average level and a ceiling, then read the report. The footer gives the level before and after in dBFS, the gain applied and a note when the ceiling prevented the target from being reached.

The controls and the readout:

A target RMS from minus 30 to minus 9 dBFS in single decibel steps, defaulting to minus 18.
A peak ceiling from minus 6 to zero dBFS in half decibel steps, defaulting to minus 1.
A footer reporting the measured level before and after the pass, the gain applied and whether the ceiling interfered.
An output WAV named after the source with an rms suffix, limited to two channels.
Guidance in the panel noting that very peaky material may hit the ceiling before it reaches the target.

The measurement is taken jointly across the channels: every channel's squared samples are summed and divided by the total sample count before the square root is taken, so the figure describes the file rather than one side of it, and the worst case peak is tracked at the same time. The gain that follows is a single static gain, chosen as whichever is smaller between the gain needed to reach the target and the gain that would push the peak past the ceiling. That second condition is the honest part: rather than quietly overshooting in order to hit the number, the tool stops at the ceiling and the report says so, naming both the target it wanted and the ceiling that intervened. Because RMS is an average, it responds to how much energy is present rather than to how loud the material sounds, which is why it is brighter and more percussive content that tends to trip the ceiling first. The perceptual alternative is the loudness normaliser, which weights the measurement the way hearing does and applies gating as well.

What RMS normalising does not give you:

RMS is not a perceptual measure, so two files normalised to the same RMS can sound noticeably different in loudness.
A file with sharp transients will hit the ceiling and land below the target, which is common on percussive material.
The gain is constant, so level variation inside the recording is not addressed at all.
It does not change dynamics; a compressed file stays compressed at its new level.
There is no limiter, so the ceiling is respected by reducing the gain, not by controlling the peaks.
Only two channels are accepted, and a mono file is measured on its single channel.

Use it when a consistent average level matters more than a perceived loudness figure, which is the case in a lot of broadcast-adjacent work and when matching a set of files to each other. Watch for the ceiling note in the footer: if it appears often, either the target is too ambitious for the material or the peaks need controlling before the level is raised. A useful habit is to check the result with a loudness meter as well, since two files at the same average level can still feel different in loudness.

How to use RMS Normalizer

Drop an audio file onto the panel; the tool accepts up to two channels.
Set the target RMS level, starting from the default of minus 18 dBFS.
Set a peak ceiling so the static gain cannot push the output into clipping.
Run the pass and read the footer, noting whether the ceiling limited the gain, then download the normalised WAV.

When to use RMS Normalizer vs related tools

RMS Normalizer suits work where a consistent average level is the goal, such as matching a series of files to each other. Loudness Normalizer is the perceptual equivalent and the one to use when a streaming platform is the destination, and Audio Normalizer covers the simple peak based approach. When the reason the target is missed is that the level drifts within the file, Audio Leveler is the tool that fixes that first.

Privacy & Security

This tool runs entirely in your browser — no data ever leaves your device. There is no server round-trip, no upload, no logging, and no account required. Your input is processed locally using client-side JavaScript and is never stored, transmitted, or accessible to anyone else. When you close the tab, everything disappears.