Audio Levels and LUFS in DaVinci Resolve

I wouldn't choose between peak and loudness meters, because they answer different questions. A mix can peak safely and still play much louder or quieter than the programs around it. It can also land on the right integrated LUFS while one short transient clips after encoding.

LUFS, dBFS, and True Peak in Plain English

LUFS means Loudness Units relative to Full Scale. LKFS is the equivalent K-weighted term used by some standards; for the same BS.1770 measurement, the numerical values are treated as equivalent. dBFS is plain sample level below digital full scale, and true peak estimates the peaks between samples that a sample meter can miss. The targets and Resolve controls below come from the Resolve 21.1 manual and each standard's own page.

Blackmagic's official Fairlight page confirms the pieces this guide leans on: selectable standards, tolerance color coding, a main-bus loudness graph, and offline clip analysis. The current Resolve 21 Reference Manual documents the meter controls and readouts on pp. 4054–4058.

Which LUFS Target Should You Use?

My first reference is a signed delivery specification whenever one exists. The examples below show where common numbers come from; they don't override a network, distributor, or client document.

The EBU’s loudness guidance specifies -23 LUFS, ±0.2 LU (±1 LU for live programs), and names the momentary, short-term, and integrated meters behind that measurement. ATSC A/85 uses -24 LKFS when loudness metadata is absent or unspecified and recommends keeping true peaks below -2 dBTP in that context. Apple Podcasts’ current requirements recommend about -16 LKFS ±1 dB with true peak no higher than -1 dBFS, both measured to ITU-R BS.1770-5.

You'll see “-14 LUFS for YouTube” repeated everywhere as if it were an upload contract. It can be a handy community mastering reference, but YouTube's current public video and audio formatting guidance doesn't state a universal creator-upload LUFS requirement. Resolve's meter even lists a YouTube profile; that's a meter preset, not a rule YouTube publishes for uploads.

Its separate audio-enhancements guidance says playback audio may get automatic volume and mixing adjustments, and viewers can turn Stable volume or Voice boost on or off. Channel owners can switch these enhancements off for future uploads in YouTube Studio on a computer. Either way, I'd never crush a good mix just to chase a screenshot from “Stats for nerds.”

Fix Levels in Signal-Flow Order

1. Set repeatable monitoring. Pick a sensible speaker or headphone level and leave it alone from scene to scene.
2. Repair clip-level outliers. Use clip gain so unusually quiet or loud recordings reach the track chain in a workable range.
3. Build dialogue first. Balance representative quiet, normal, and loud lines before you touch music and effects.
4. Shape tracks. Reach for EQ, compression, automation, and cleanup only where they fix a problem you can hear.
5. Place music and effects around speech. Lower or automate the competing material rather than forcing dialogue through it.
6. Watch buses and Main. Several safe tracks can still sum into an overloaded subgroup or output.
7. Measure after all processing. The loudness meter has to see the same final bus that the deliverable uses.

Each gain stage has one job. Clip gain is pre-track correction, a track fader balances that track in the scene, a bus fader balances a group, and Main controls the program output. If your routing is unclear, sort that out before metering anything, starting with our Fairlight bus-routing guide.

Build a Dialogue Anchor Before Chasing LUFS

Pick a few passages: a normal line, the quietest intentional line, an excited line, and a section under music. Set clip gain so the track processor receives consistent material without flattening the performance. Then automate the phrases that still wander.

Where the peaks land doesn't tell you whether dialogue is good. Mic distance, noise, spectral balance, accent, room reflections, music masking, and dynamic range all affect intelligibility.

Work in order: clean the obvious problems, shape tone using our Fairlight dialogue EQ guide, control excessive variation with the dialogue compressor workflow, then rebalance in context.

Resolve 21's Voice Isolation (Studio only) and Dialogue Leveler can rescue specific recordings, but neither one is a master-loudness control. Our Voice Isolation vs Dialogue Leveler comparison explains which problem each tool is designed to solve.

Set Up Fairlight Loudness Monitoring

1. Go to the Fairlight page and open the Mixer and the Control Room monitoring controls.
2. In the monitoring menu, confirm that you're listening to the same Main bus you'll use for export.
3. Open the loudness meter options and choose the required standard. The 21.1 list runs from BS.1770-4, ATSC A/85, and EBU R128 to OP-59, TR-B32, AGCOM 219, Netflix, YouTube, and two Disney profiles.
4. Confirm the meter’s target and tolerance against the current delivery document. The reference level itself is Target Loudness Level in the General Options of Project Settings, and it defaults to -23 LUFS.
5. Read maximum true peak at the top of the Control Room meter; Resolve 21 measures it to BS.1770-4, so inter-sample peaks show up there.
6. Reset the integrated measurement right before the program start. With Lock Metering to Transport enabled, moving the playhead resets the analysis, so know how that option is set before a compliance pass.
7. Play the complete delivery range without jumping the playhead. If you have to stop, use Pause and then Resume, not Reset.
8. Write down the metrics the delivery spec asks for: Integrated, the relevant Short/Short Max reading or Short Term history view, Loudness Range, and maximum True Peak.

Resolve’s colored loudness display is a compliance aid, not a mix decision. In the manual's wording, blue means below tolerance, yellow within it, and red above it, always against the selected standard. So a blue reading under EBU R128 isn't necessarily wrong for a podcast, and a yellow reading under a custom -14 preset isn't broadcast approval.

Measure the Correct Range

Integrated loudness only means something for the program that played after the last reset. Start at the first frame delivery requires and stop after the required tail. If the spec puts bars, tone, an ident, or silence outside the measured program, follow its measurement boundaries exactly.

For a bounced or rendered audio clip, right-click it in the timeline and choose Analyze Audio Level for the offline check. The dialog offers the same standards as the live meter and shows each target in parentheses next to the reading.

The three-second Short Term history curve helps you find a shout, a music sting, or a weak scene. The numeric Short/Short Max readout is a different Resolve measurement, which the manual describes over a 30-second window, and Momentary loudness is more local still at 400 ms.

Neither one replaces a complete integrated pass. If you change a bus limiter, music automation, dialogue gain, or the measured duration, reset and measure again.

Use a Limiter for Peaks, Not as a Loudness Bulldozer

A true-peak limiter on the final bus can catch the transients that are left, but it's the guardrail after a balanced mix, not the mix itself. If it's shaving several decibels off every chorus or sentence, I'd rather pull down the buses feeding it or revise compression upstream. Constant limiting flattens punch, ambience, and speech texture.

Take the output ceiling from the delivery spec. Online work often leaves at least 1 dB of true-peak headroom because AAC or other lossy encoders can create inter-sample overs. Broadcast and platform contracts may require a different ceiling. Measure after the limiter. For the mechanics behind inter-sample overs, see true peak explained.

Should You Normalize Audio on Export?

Resolve’s Deliver page offers two different operations. Normalize to Standard scales the level toward the chosen standard until either the loudness or the peak constraint governs. Optimize to Standard brings loudness into the desired range while keeping peaks below the selected threshold.

Neither one replaces mixing. Delivery processing can't tell that one interview is buried under music, repair a clipped bus, or protect the drama you built into each scene.

1. Measure and save the final mix first.
2. On the Deliver page, select the correct bus and channel format.
3. Choose Normalize to Standard or Optimize to Standard deliberately for each output, and select the exact standard only when the delivery requires or benefits from it.
4. Check what happens when the requested gain would push past the true-peak ceiling.
5. Render a short test file.
6. Reimport that file, route it straight to a clean measurement bus, reset, and measure it again.

Our DaVinci Resolve audio export guide covers bus selection, mono/stereo layout, WAV versus AAC, and reimport verification in more detail.

Resolve Free vs Studio

The core Fairlight mixer, bus routing, standard loudness monitoring, and audio rendering workflow all work in Resolve Free. Studio adds advanced features such as immersive 3D audio mixing, but buying Studio doesn't choose the correct loudness specification or mix the program for you.

Exact standards, presets, effects, and labels can change between Resolve releases. This guide follows Resolve 21.1; verify your installed point release and the current client document before final QC.

Troubleshooting Audio Levels and LUFS

Integrated LUFS never settles

It keeps updating as new material plays; that's what integration means. Reset at the true program start and play the complete range once. Looping a loud section won't give you a program value.

Dialogue meets LUFS but is hard to understand

LUFS doesn't diagnose masking, noise, reverb, poor articulation, or spectral imbalance. Lower the competing music, automate phrases, and repair the dialogue before you touch the master target.

The Main bus clips although tracks look safe

Tracks sum at buses. Check subgroup gain, parallel sends, makeup gain, and any processing on Main. Fix the earliest overloaded stage; don't park a limiter after accidental clipping.

The exported file measures differently

Confirm the Deliver page rendered the intended bus, channel count, timeline range, and Normalize/Optimize setting. A different bus or per-output normalization choice changes the measurement, and lossy encoding can move true peak. Reimport the rendered file and measure it through a clean path rather than trusting only the live timeline meter.

Music is loud enough but masks every word

Volume is only part of masking. Automate the music under speech and pull back competing midrange selectively. Don't keep boosting dialogue until the Main bus runs out of headroom.

Final Audio QC Checklist

• Listen through the complete program at a repeatable monitor level.
• Check the dialogue on speakers, headphones, and a small mono-compatible device.
• Verify that clip, track, subgroup, and Main stages don't clip.
• Reset and measure the exact delivery range.
• Note integrated loudness and maximum true peak.
• Compare those numbers with the current destination specification.
• Render the actual bus and channel layout.
• Reimport and remeasure the file after encoding.
• Check sync, start, tail, fades, channel order, and any unexpected silence.

The meter proves compliance with a measurement. Your ears prove that the dialogue is clear, the music supports the scene, and the dynamics still make sense. I don't call a mix finished until it has both.

The Fairlight Audio in DaVinci Resolve hub puts loudness measurement after routing, dialogue processing, and mix decisions, not before them.

DaVinci Resolve Audio Levels and LUFS FAQ

What is LUFS?

LUFS is a K-weighted loudness measurement relative to digital full scale. Integrated LUFS sums up the whole measured program; short-term and momentary views show what's happening locally.

Is LUFS the same as dBFS peak level?

No. Peak readings show how much headroom the signal has; LUFS estimates perceived loudness across time. A mix can pass one limit and fail the other.

What LUFS should I use for YouTube?

YouTube doesn't publish a universal -14 LUFS creator-upload requirement in its public formatting guidance. Deliver a sensible web master with true-peak headroom, skip destructive loudness chasing, upload privately or unlisted, and check playback.

What is the EBU R128 loudness target?

EBU R128 specifies -23 LUFS program loudness, ±0.2 LU for most programmes. Current delivery documents can also define their own tolerance, true peak, and Loudness Range requirements.

Should I normalize audio on export?

Only after the mix is balanced and you understand the selected standard. Render normalization can shift overall level, but it can't repair dialogue masking, bad routing, or excessive dynamics.

Why does LUFS change after export?

The rendered bus, channel layout, range, normalization, and codec may differ from your monitoring path. Reimport the actual file and measure it through a clean bus.

Can DaVinci Resolve Free measure LUFS?

Yes. The core Fairlight loudness-monitoring workflow is part of Resolve Free. Confirm the exact standards and controls in your installed Resolve 21 build.

Jason Miller I run DaVinci Resolve Club, an independent publication. My guides explain editing, color, Fairlight, Fusion and delivery, with clear Free vs Studio limits.