Normalize Audio Online

Make loudness more consistent

Recordings often come from different microphones, rooms, apps and devices. One clip may be quiet, another may jump loudly, and a video soundtrack may be hard to hear next to other media. Loudness normalization helps create a more consistent listening level so speech, interviews, lectures, voice notes and rough podcast material are easier to review or share. This tool exports a normalized MP3 from an audio file or from the audio track inside a video.

The workflow is 100% client-side. Your file is not uploaded to a server for analysis or processing. The page loads ffmpeg.wasm when needed, writes the selected media into the browser engine, applies ffmpeg's loudnorm audio filter and downloads the result locally. That keeps personal recordings, internal meetings, client drafts and unreleased material on your device.

How loudnorm helps

The loudnorm filter targets perceived loudness rather than simply multiplying every sample by the same amount. It also uses a true-peak limit and loudness range target, which can reduce the risk of clipping while making quiet material easier to hear. It is a practical one-click pass before sharing a voice recording, preparing audio for transcription or collecting clips from multiple sources into a more even listening set.

Review before publishing

Automatic normalization is helpful, but it is not a full mastering process. Very noisy recordings, distorted peaks or sudden background sounds may still need manual editing. The first run may pause while the WebAssembly engine downloads; later normalizations in the same tab can reuse it. Keep the original file until you have listened to the MP3 and confirmed that speech, music and quiet sections sound natural enough for your use.

Target levels used

The filter is set to an integrated loudness of −16 LUFS, a true-peak ceiling of −1.5 dBTP and a loudness range of 11 LU. Those are common targets for spoken-word content and stereo podcasts, and they sit close to what many streaming and video platforms play back comfortably. Music intended for commercial release is usually mastered louder, so this tool is better suited to speech, lectures and casual listening than to final music masters.

The output is always an MP3 at 192 kbps, and any video picture is removed. The loudnorm filter analyzes the file as it processes, so very short clips of a few seconds may not be measured reliably and can end up at a slightly different level than expected. If the result is still too quiet, the source probably has a large dynamic range or long silences. Trimming the silence first often makes the measurement more representative of the speech you care about.