Normalize Audio Levels is the command in DaVinci Resolve that sets a clip’s loudest moment to a level you choose. It sits on the clip’s right-click menu, and it behaves the same in the Edit page and the Fairlight page.
It is narrower than it sounds. Resolve’s manual says the command makes a volume adjustment only, and no dynamics are applied. That is the whole story of the tool: it moves a clip up or down and leaves everything inside the clip alone. The rest of this page is about picking the target, and about the two jobs people expect it to do that it cannot.
What you need
- DaVinci Resolve 21, or 20. The dialog is the same in both.
- The clips on a timeline, in the Edit or the Fairlight page.
- A destination in mind. YouTube, a podcast feed and a broadcast spec each want a different number, and the number is the point of the tool.
1. Select the clips
Click one audio clip, or drag a selection around several and shift-click to add more. Then right-click one of the selected clips. Select first and open the dialog once, rather than running it clip by clip, because the selection is what Relative and Independent act on.
2. Choose Normalize Audio Levels
The right-click menu item opens the Normalize Audio Level dialog. The command also runs in the Fairlight page, so you can do this on whichever page you are already working in.
The change it makes is a gain change on the clip. Your Volume control, the clip’s volume line and Clip > Audio > Increase Audio Level all still work on top of it afterwards, so nothing here is permanent.
3. Pick the mode
Normalization Mode decides what Resolve measures before it works out the gain. The menu offers two kinds of answer.
The loudness modes measure how loud the clip sounds, not how tall its tallest peak is. They are the standards Resolve ships and meters against: BS.1770-4 at the base, with EBU R128, ATSC A/85, OP-59, TR-B32, AGCOM 219, Netflix, YouTube and Disney built on it. Use these when the file has to sit next to other files on a platform.
The peak modes are Sample Peak and True Peak. Sample Peak reads the tallest sample in the clip. True Peak estimates the peaks that fall between samples, and Resolve meters it against BS.1770-4, which is why it catches a peak the sample meter can miss. Use these when the worry is a ceiling rather than loudness.
Which to choose:
- YouTube: the YouTube mode, at -14 LUFS. That figure is Resolve’s own, from the manual.
- A podcast: a loudness mode at -16 LUFS, Apple’s recommendation. The LUFS explainer lists what each platform expects.
- European broadcast: EBU R128, at -23 LUFS.
- Sound effects under dialogue: True Peak, at -1 dBFS.
4. Set the target level
The reference level is the number the loudest part of the clip should end up at. With a peak mode it is a dBFS ceiling, and -1 dBFS is the usual delivery choice for True Peak. With a loudness mode, enter the target the standard is built around: -14 for YouTube, -23 for EBU R128, -16 for a podcast.
5. Choose Relative or Independent
Set Level only matters when you selected more than one clip.
Relative treats the whole selection as one clip. The highest peak in the selection sets the gain change, every selected clip gets the same amount, and the balance between them is kept. For an interview recorded in one sitting, or a single talking head cut into pieces, this is what you want.
Independent measures each clip on its own and gives each its own gain change, so their peaks all land at the same point. That is right for a tray of sound effects captured at very different levels. It is wrong for dialogue: it will pull a quiet room tone up until it is as loud as a shout.
6. Check what you got
- The Loudness meter in Fairlight shows the mix against the standard you picked. Blue is below tolerance, yellow is inside it, red is above it.
- Project Settings > General Options holds the Target Loudness level, -23 LUFS by default, and the Loudness Scale, EBU +9 or EBU +18. Set the target to your destination and the meter’s zero means something.
- To measure a file without playing it, right-click it in the Timeline and choose Analyze Audio Level. Pick the mode, and the panel shows the reading next to that mode’s target after you click Analyze.
When normalize is the wrong tool
Words that move up and down
One gain change cannot make a voice steady. If the loud words are 10 dB above the quiet ones, they stay about 10 dB above them after normalizing.
Dialog Leveler is the tool for that. It rides down the loud parts, lifts the soft ones and holds back background that is not dialogue. The presets are Allow wider dynamics (the default), More lift for low levels, Lift soft whispery sources and Optimize moderate levels. You can add Reduce Loud Dialogue, Lift Soft Dialogue or Background Reduction, and set Output Gain between 0 and +6 dB. It works on mono, stereo and dual-mono tracks, not on tracks with more than two channels.
Resolve’s own tip is to normalize first, to set a baseline, and then run Dialog Leveler on the default preset. That is the order this page uses: normalize the clip, then even out the words inside it.
Noise
A gain change lifts the noise with the voice. If the room, the fan or the keyboard is the problem, normalizing makes all of it louder. Voice Isolation, in the same Inspector, is Resolve’s answer for loud background sound, with Amount around 70 to 80 for a natural result and the same two-channel limit. The manual marks it Studio version only.
Clipping
A clipped peak is already damaged. Turning the clip down makes the distortion quieter, not smaller. Normalize cannot rebuild the top of a wave that was flattened on the way in, so record with headroom instead of trying to recover it later.
Cleaning the audio outside Resolve
When the noise needs more than the Inspector, take the sound out of the video and treat it as a file. Gleam Audio does that on your own computer: every one of its tools takes video files for their sound, mp4, mov, mkv, webm and avi included, and hands back audio. Enhance levels in one step and shows each file’s loudness, like “Now -18.4 LUFS, peaks at -1.1 dBFS”. Remove noise is tuned for speech, which its own known limitations say plainly.
It is free for personal use up to 20 files a week, and a license for paid work is $10, $15 or $20. The export, clean and put-back workflow is in how to reduce background noise in a video, and the leveling side of it is in the audio leveler guide.
Normalizing on the way out
The Deliver page has a third place to normalize. The YouTube presets for 720p, 1080p, 1440p and 2160p carry a Normalize Audio checkbox that normalizes the clip’s audio to YouTube’s standard. It happens during the render, so the clips on your timeline are untouched by it. Treat it as a backstop for a file you already levelled, not as a replacement for setting levels you can hear.
Takeaways
- Normalize Audio Levels sets the loudest part of each selected clip to a target, with one gain change per clip, in the Edit and Fairlight pages.
- Loudness modes match what a platform expects. Peak modes protect a ceiling. Pick by destination.
- Relative keeps the balance inside a selection. Independent levels each clip on its own, which suits effects and wrecks dialogue.
- It cannot even out words, remove noise or repair clipping. Those need Dialog Leveler, Voice Isolation or a separate cleanup pass on the audio file.