Normalize YouTube Audio: Stable Volume and Uneven Videos

A quiet video and a video that swings between whispers and loud effects are different problems. Identify which one you have before adding processing. This guide concerns listening to YouTube; an uploader needs to edit and measure the source file before publication.

Which problem are you hearing?

Playback problem	First comparison
Everything on the device is quiet	Check output device and hardware volume with another recording
Only YouTube is quiet	Check player volume and any separate browser volume control
One whole video is quiet	Compare another video on the same output
Quiet speech alternates with loud effects	Compare Stable volume if offered
Music covers the voice	Try Voice boost if offered, or captions; more overall gain raises both

Use a repeatable passage and keep your device volume comfortable. Note whether you changed the source, player, browser or hardware so you can restore the previous setting.

How do I enable Stable volume on desktop?

Open the video’s Settings and toggle Stable volume if present. YouTube describes it as a control that reduces the range between quiet and loud parts. It is unavailable on some videos and disabled for YouTube Music and official music videos. Supported videos may also offer Voice boost; enabling that switches Stable volume off.

Compare the same quiet sentence and the following loud passage with the setting on and off. Keep whichever result works for that recording. If the option is absent, refreshing repeatedly or reinstalling an extension is not evidence that the feature should be available for that video.

Is YouTube normalization always −14 LUFS?

The linked YouTube help page does not publish a universal LUFS target for every upload and playback mode. Do not confuse a measured result for one video with a delivery requirement for the whole service.

LUFS describes program loudness. A player control can change listening levels without guaranteeing a particular integrated measurement. Our downloadable equal-peak example shows why a peak reading alone cannot establish loudness consistency.

Can a volume booster normalize a playlist?

A gain control multiplies the signal. If a quiet sentence and a loud effect are both present, gain alone raises both. A compressor changes dynamics according to its settings, but that still does not prove that successive videos reach the same integrated LUFS.

For uniformly quiet browser playback, follow the YouTube volume-booster setup. Hearably Free includes up to 120% gain and three-band EQ; Pro includes up to 800% and ten-band EQ. Test the actual supported desktop tab before upgrading. The extension does not change YouTube’s uploaded file or process the mobile YouTube app.

If added gain makes speech rough or effects uncomfortable, reduce it and compare with other processing disabled. A limiter can control new peaks in its processing path; it does not repair distortion already present in the recording.

What if different speakers have different levels?

A mixed video may contain one very quiet guest and one loud host. Raising their combined mix raises both. Listener controls have limited access to the original tracks. Captions or a modest dynamics adjustment may help you follow it; the publisher can make a more targeted correction using the separate recordings.

If you own the source, first balance speakers with clip or track gain, then check transitions and use compression only where needed. See the compression guide. Avoid using heavy master processing to compensate for one badly balanced track.

How can an uploader check the source?

Keep an unprocessed copy, edit a separate version and measure the entire intended upload. FFmpeg’s ebur128 filter can report integrated loudness and estimated true peak; the LUFS example shows a reproducible measurement command. Listen to the encoded result as well as the editor timeline.

Publish a private or unlisted test only when appropriate for your material and compare playback on your actual devices. Record the player settings you used. A successful listening check is evidence about that setup, not every future playback route.