How do I normalize audio to -14 LUFS for YouTube?

The steps at a glance

Normalizing is one gain change for the whole video, so it comes last, after everything that changes the balance:

1. Finish the mix before you normalize.
2. Measure the integrated loudness of the whole video.
3. Normalize the full mix, not each clip.
4. Catch the peaks with a limiter.
5. Or do steps 2 to 4 in one go with my free normalizer.
6. Export and measure the file you will upload.
7. Check what YouTube does with it.

What YouTube does with loud and quiet videos

YouTube Help publishes no loudness number, but the player reports what it does. Right-click any video and choose Stats for nerds: the Volume / Normalized line shows the content loudness of its audio and the target. I read it on five videos from my loudness study on September 27, 2026:

The three loud files played at the -14 target, their extra loudness taken away. The two quiet ones played about 6 dB under it and were not raised. Five videos in one desktop browser is a small check, and during it the player was playing its DRC version of the audio (Stats for nerds read “100%/100% DRC”), so phones, TVs and other apps may handle loudness differently. The pattern held on all five, though, and Blackmagic’s DaVinci Resolve manual states the target directly: “the YouTube target LUFS specification is -14 LUFS”.

Big channels do not share a level. The 50 videos had a median of -16.3 LUFS, from -24.3 to -7.4, with 12 of them louder than -14 and 9 under -20, and by my count from the study’s spreadsheet only 11 sat within 1 dB of -14. Mastering quieter is a valid house sound if every upload matches; it simply plays softer than the videos around it, because YouTube does not turn it up.

YouTube Help adds that Stable volume, which “balances the range between quiet and loud”, is on by default for viewers, though not on every video, and creators can opt future uploads out of audio enhancements in Studio. Your file’s loudness is what you control.

Step 1: Finish the mix before you normalize

Normalization changes the volume of the whole file and nothing else; DaVinci Resolve’s manual says of its own Normalize command that “no dynamics are applied”. Uneven takes, music over the words and noise in the pauses come out exactly as they went in. So clean the voice first (removing background noise), set the music under it (background music under a voiceover) and even out the dialogue clips:

• Premiere Pro: select the clip, open Window, Essential Sound, expand Loudness and click Auto-Match. Adobe says it keeps loudness consistent across clips and shows the clip’s measured LUFS under the button.
• DaVinci Resolve: right-click the dialogue clips, choose Normalize Audio Levels and pick a loudness mode, with Set Level on Relative to move a consistent recording together or Independent to match clips recorded at very different levels. The Dialogue Leveler in the Inspector rides loud and soft passages.

Step 2: Measure the integrated loudness of the whole video

LUFS (LKFS in the ITU’s naming) is loudness as ITU-R BS.1770 defines it; its fifth version has been in force since November 2023. What counts is the integrated value, the average over the whole video. Short-term readings move with every sentence.

• Premiere Pro: in the Audio Track Mixer, open the first effect slot of the Mix track and choose Special, Loudness Meter; Adobe says its presets include YouTube. Play the sequence from start to finish and read Integrated. The meter never changes the audio.
• DaVinci Resolve: in Project Settings, Fairlight, set Target Loudness level to -14 (the default is -23 LUFS) so the meters’ 0 mark sits at YouTube’s level. The meters’ options menu chooses the standard, and YouTube is on the list. Right-click a clip and choose Analyze Audio Level for a reading without playing it all.
• Any editor: open the exported mix in my free LUFS meter, which reads the integrated loudness and true peak of the whole file in your browser.

Step 3: Normalize the full mix, not each clip

• Premiere Pro and Media Encoder: Adobe documents this in Media Encoder: open the Export Settings of the queued file, go to the Effects tab and check Loudness Normalization. Of the three standards (ATSC A/85, EBU R128, ITU BS.1770-3) only ITU BS.1770-3 lets you change the target, so pick it and enter -14. Configure True Peak Limiter sets the limiter; Write Loudness Report saves loudness and true peak to an XML file. Premiere’s Export screen has its own Effects section; if your version lists Loudness Normalization there, use the same standard and target.
• DaVinci Resolve: the quickest route is the YouTube render preset on the Deliver page, whose Normalize Audio checkbox, in Blackmagic’s words, “will normalize the clip’s audio level to YouTube’s standard”. Check it, render, and measure the file anyway (Step 6). For control over the standard and the level, turn on Audio Normalization in the Audio tab of the render settings instead: Normalize to Standard scales loudness or peak, whichever crosses its threshold first; Optimize to Standard brings loudness into range and keeps peaks under the threshold. Choose the Standard and the Target Level. In Resolve Studio, the AI Audio Assistant (Timeline, AI Tools, Audio Assistant) builds a mix for the delivery standard you pick and adds mastering plugins for it.
• CapCut: CapCut’s own article describes a Loudness normalization box in the Audio tab that sets the clip to “a standard value” and shows the level used. It describes no field for -14, so measure the export and use the free normalizer in Step 5 if it is off.

Peak normalization is a different thing. Premiere’s Audio Gain dialog (Normalize Max Peak to, Normalize All Peaks to) and the Sample Peak and True Peak modes of Resolve’s Normalize command set the highest peak, not how loud the video sounds. Normalizing each clip or track to -14 is a mistake too: voice, music and effects playing together add up to more than -14.

Step 4: Catch the peaks with a limiter

Turning a quiet mix up to -14 pushes its loudest moments up as well, often past 0. Limiters come in two kinds: a sample-peak limiter caps the samples themselves, while a true-peak limiter also catches inter-sample peaks, where the waveform between two samples rises above both (Resolve’s manual notes its meter catches these). EBU R 128 caps broadcast true peak at -1 dBTP, a sensible ceiling for YouTube: in my study, 29 of the 50 videos had true peaks above 0 dBTP on the audio YouTube served, after its own encoding.

• Premiere Pro: the Hard Limiter effect in Amplitude and Compression (Maximum Amplitude, Look Ahead, Release) is a sample-peak limiter: Adobe describes Maximum Amplitude as “the maximum sample amplitude allowed”. Set it below your -1 dBTP ceiling, for example -2 dB, and check the true peak of the export in Step 6. For a true-peak limit, use Configure True Peak Limiter in Media Encoder’s Loudness Normalization (Step 3).
• DaVinci Resolve: the Limiter Fairlight FX on the main bus, which Blackmagic’s manual describes as a true peak limiter that looks 64 samples ahead, or Optimize to Standard on the Deliver page, which keeps peaks under the standard’s threshold.

If the limiter has to work hard, a gentle compressor before it sounds more natural.

Step 5: Or do steps 2 to 4 in one go with my free normalizer

If your editor has no loudness tools, or you want to check its work, open my free normalize audio to -14 LUFS tool. Load the audio or the whole video file (up to an hour, in common formats such as WAV, MP3 or MP4), keep the target at -14 LUFS and the ceiling on -1 dBFS (or -2 dBFS if it will be encoded to MP3 or AAC) and press Normalize. It measures with the BS.1770 method, sets the gain, limits only the peaks that need it, measures again and gives you a WAV, without uploading anything. Its ceiling caps sample peaks in dBFS, not true peaks, so the true peak can still land above it: check it in Step 6, and if it reads above -1 dBTP, run the file again with the -2 dBFS ceiling. Swap that WAV in for the original audio on your timeline, then export the video normally.

Step 6: Export and measure the file you will upload

YouTube Help recommends AAC-LC (or Opus) audio at 48 kHz, stereo at 384 kbps. Encoders can move the true peak a little, so measure the exported file in the LUFS meter: the integrated loudness should read close to -14 and the true peak at or below -1 dBTP. If the true peak reads higher, lower the limiter or the normalizer’s ceiling to -2 and export again. If the voice sounds harsh once turned up, the original probably clipped: the audio clipping detector finds every clipped moment, and how to fix clipped audio covers the repair.

Step 7: Check what YouTube does with it

Once the upload has processed, read Volume / Normalized in Stats for nerds. Content loudness near the target means it plays at YouTube’s level; well below means it plays quieter than the videos around it.

How I set loudness on client videos

On every video I deliver, loudness gets set last, once the edit and the mix are locked, at the level YouTube plays back. The player check above is why I see no point in a louder master: YouTube simply takes the extra loudness away. To have the mix handled along with the edit, see sound design and mix.

Loudness normalization FAQ

Should I normalize audio to -14 LUFS for YouTube?

For a voice-led video, yes, -14 LUFS integrated is the practical target. In a 2026 check of YouTube’s player, videos mastered louder than -14 played at -14, so the extra loudness was taken away, while videos mastered near -24 played about 6 dB quieter than the rest.

Will YouTube turn up a video that is too quiet?

In my check, no. Two videos mastered near -24 LUFS showed a content loudness around -20 in Stats for nerds and were not raised further, while the louder videos were turned down to -14. That was five videos in one desktop browser, with the player serving its DRC audio version, so other devices and apps may behave differently.

What true peak should I use for YouTube?

Keep true peaks at or under -1 dBTP, the ceiling EBU R 128 sets for broadcast, and leave a little more room, around -2, if the file will be encoded to MP3 or AAC before upload. A sample-peak limiter set to -1 dB can still let inter-sample peaks through, so check the export on a true-peak meter. In my study, 29 of 50 files served by YouTube had true peaks above 0 after YouTube’s own encoding.

What is the difference between peak normalization and loudness normalization?

Peak normalization moves the whole file until its highest sample reaches a value such as -1 dB. Loudness normalization moves it until its average loudness, measured in LUFS with the ITU-R BS.1770 method, reaches a target such as -14. Two files with the same peak can sound very different in loudness, which is why YouTube’s level is a loudness number.

Is it bad to upload louder than -14 LUFS?

It does not make the video louder. In my check, a MrBeast upload mastered at -7.4 LUFS played at the same -14 as the others. The heavier limiting needed to reach that level stays in the sound, so you lose dynamics for nothing.

How do I normalize audio in DaVinci Resolve for YouTube?

The quickest way is the YouTube render preset on the Deliver page: check its Normalize Audio box, which Blackmagic says normalizes the audio to YouTube’s standard. For more control, set Project Settings, Fairlight, Target Loudness level to -14 so the meters show YouTube’s level and mix to it with the Limiter on the main bus, or turn on Audio Normalization in the Audio tab of the render settings with your own standard and target. Either way, measure the rendered file before you upload it.

Sources

• Player behaviour, the 50-video figures and true peaks after encoding: my loudness study (player check September 27, 2026, five videos in one desktop browser with the DRC audio version). The 11 of 50 within 1 dB of -14 is my count from its spreadsheet.
• YouTube Help: Control video volume on your device, Video and audio quality enhancements and recommended upload encoding settings, read October 20, 2026.
• Adobe Help: Automatic loudness correction in Media Encoder (updated July 21, 2026; Configure True Peak Limiter); Loudness Meter overview, Measure track loudness, Apply Loudness Meter to tracks, Auto-match audio loudness, Adjust audio gain in Premiere and Apply audio effects (Hard Limiter, “the maximum sample amplitude allowed”) in Premiere, read October 20, 2026.
• Blackmagic Design: DaVinci Resolve 21 Reference Manual (Fairlight settings, Normalize Audio Levels, audio meters with the -14 LUFS YouTube target, the Limiter Fairlight FX as a true peak limiter, the YouTube render preset’s Normalize Audio checkbox and Deliver page normalization) and DaVinci Resolve 20 New Features Guide (AI Audio Assistant), read October 20, 2026.
• CapCut: How to Use an MP3 Normalizer, read October 20, 2026.
• Standards: ITU-R BS.1770 (BS.1770-5 in force since November 2023) and EBU R 128 (version 5.0, November 2023: -23 LUFS, true peak not above -1 dBTP), read October 20, 2026.