Skip to main content
SafeStudio

( Free · in the browser · no sign-up )

Normalize audio onlineto a −1 dBFS sample peak

Measure the decoded file's highest sample, calculate one linear gain adjustment, preview it, and export from the same browser-based audio graph.

Or drop an audio or video file anywhere on this page. Compatibility depends on the file codecs and device: MP3, WAV, M4A, AAC, FLAC, OGG, WebM, MP4, MOV, MKV, and more.

The file you choose is processed in this tab; the page still loads site resources and consented analytics. Read the Privacy Policy.

  • 01Sample-peak measurement
  • 02Automatic −1 dBFS target
  • 03Local processing

( What it does )

Peak normalization is a repeatable level adjustment

This tool finds the largest absolute decoded sample and applies the same gain ratio to every channel. It is useful for setting a predictable sample-peak ceiling, but it does not measure perceived loudness.

  • 01

    Before and after

    The editor shows the original peak, calculated gain, and predicted peak before export.

  • 02

    Channel balance

    One multiplier is used for every channel, preserving the relationship between left and right.

  • 03

    Preview equals export

    Playback and export use the same gain value in the shared processing graph.

( How it works )

How to normalize an audio peak

Use the measured values as a check, not just an automatic button. The result is most useful when you know what the target represents.

  1. 1

    Open and measure the file

    Choose the recording and wait for the browser to decode every channel. The original sample peak is measured before any gain is applied.

  2. 2

    Review the calculated gain

    The tool targets −1 dBFS with one multiplier for the whole file. Check the predicted peak and note when the 200% gain cap prevents the target from being reached.

  3. 3

    Preview and export

    Listen to quiet and loud passages, then export. If you need matched perceived loudness across several files, continue with a dedicated LUFS workflow.

( Limits )

Important limits

  • This is sample-peak normalization, not LUFS loudness normalization or true-peak limiting.
  • A file that already reaches 0 dBFS may already be clipped; lowering it cannot reconstruct flattened samples.
  • The automatic gain is limited to 200%, so very quiet files may remain below −1 dBFS.
  • Compressed export re-encodes the result and can introduce codec-dependent changes.

( Uses )

Useful for music, voice, and podcasts when the goal is a peak ceiling

  • Bring a quiet recording closer to a consistent maximum without manual gain calculation.
  • Reduce a hot file to leave a small sample-peak margin.
  • Prepare clips for a later loudness workflow while keeping this step clearly separate from LUFS matching.

( Concepts )

Sample peak, true peak, and LUFS are different measurements

Choosing the right measurement avoids treating peak normalization as a solution for every loudness problem.

Sample peak (dBFS)

The highest decoded digital sample in the file.

This is what the current normalizer measures and targets at −1 dBFS.

True peak (dBTP)

An estimate of peaks that can occur between samples during playback or conversion.

This tool does not calculate it; use a true-peak meter or limiter when delivery rules require dBTP.

Integrated loudness (LUFS)

A weighted measurement of perceived loudness over time.

Use it to match programs or episodes; a shared peak value does not guarantee equal perceived volume.

( Other tools )

Related tools with different intentions

( Where your audio goes )

Normalization runs locally

The selected audio is decoded and processed in browser memory. It is not uploaded to the application server to calculate or export the normalized result. Site resources and consented analytics may still use network requests.

( Questions? Answered )

Peak-normalization questions

Clear answers about what changes, what stays intact, and when another audio process is needed.

What does normalizing audio actually do?

It finds the loudest single sample in the file and multiplies every sample by one number, so that loudest point lands on the target. Nothing inside the recording changes relative to anything else — the whole thing moves up or down together, which is why it is predictable and why it cannot fix a bad balance.

How do I normalize an MP3 online?

Choose the file, wait for the browser to decode it, read the original peak and the gain the tool calculated, then preview and export. Nothing is uploaded: the decoding, the measurement and the export all happen in this tab.

Is this free, and do I need an account?

Free, no account, no sign-up and no watermark. There is no cap on how many files you normalize; the practical limit is how much audio the browser tab can hold.

Why −1 dBFS and not 0?

A sample peak of exactly 0 dBFS leaves nothing for what happens after. Converting to MP3 or AAC can push the reconstructed waveform slightly past the samples it came from, so a file that measured 0 can clip on playback. One decibel of margin costs nothing audible and avoids that.

Will every normalized file sound equally loud?

No. Two files can have the same sample peak and very different average energy, dynamics and perceived loudness. A dense master and a sparse acoustic take can both peak at −1 dBFS and sound nothing alike. Matching perceived loudness is LUFS work, and this is not that.

Normalize or adjust the volume by hand?

Normalize when you want a predictable ceiling and do not want to do the arithmetic. Use the volume tool when you want a specific amount of change, or when the automatic result is not what you wanted.

Can normalization repair clipped audio?

No. Lowering a clipped recording creates headroom but does not reconstruct waveform details already flattened at the recording or encoding stage. The distortion is in the samples; moving them down moves the distortion down with them.

Does it change the stereo balance?

No. The same linear gain is applied to all channels, so the relationship between left and right is preserved exactly as it was.

Why can the predicted peak stay below −1 dBFS?

Automatic amplification is capped at 200% for safety. A very quiet source may need more gain than that cap allows, and the tool stops rather than pushing a noisy recording up until the noise is obvious.

Does normalizing reduce the audio quality?

The gain itself is a multiplication, which is not where quality is lost. The export is: saving back to MP3 runs another encoding pass. Choose 16-bit PCM WAV or FLAC when you want to avoid that.

Which formats can I open and export?

Common audio and video containers are accepted; the editor tries native decoding first and falls back to a local FFmpeg build to pull the first compatible audio track. Exports are MP3 from 64 to 320 kbps, 16-bit PCM WAV, M4A/AAC at 192 kbps, and FLAC.

Is my file uploaded to a server?

No. The audio is decoded, measured and normalized in this browser tab and is not sent to the application server. The page itself still loads site resources and consented analytics.

( Field notes )

Learn the technique behind the tool

Guides written and reviewed by the person who builds the editor, including the real limits of each feature.

See all articles