Exported Video Quieter Than the Timeline? Loudness Normalisation, Master Fader & Playback Volume Diagnosis Guide

2026-11-21 | GPU Repair | 27 minutes | HK GPU Fix Editorial Team

Exported Video Quieter Than the Timeline? Loudness Normalisation, Master Fader & Playback Volume Diagnosis Guide

Key takeaways: An exported file that sounds quieter than the timeline is usually a gain or loudness-target difference, not a damaged file: a low fader, an export loudness target, or loud monitoring can each create the same gap. Measure the timeline and the file in LUFS, compare the readings, then test the export switch and each fader layer in turn.

A finished export sounds quieter than the timeline did. The dialogue is still there, the music is still there, and the file plays from start to finish without dropping out, but the whole programme sits at a lower level than it did during editing. The first instinct is often to export again, to change the container, or to assume the file is damaged. None of those steps addresses the cause, because the cause is usually a setting that was already in place before the export started: a fader sitting low, an export loudness target sitting below the level the timeline was mixed to, or a monitoring volume that made the timeline sound louder than the file ever was.

The distinction that matters most is between how loud something sounds in the edit room and how loud the file itself is. Timeline playback passes through a monitoring chain — the monitoring volume control, the system volume, and the headphones or speakers — and every one of those controls can make the timeline sound louder without changing a single sample in the mix. Export passes through a different chain: clip gain, track faders, the master fader, and any loudness normalisation applied at export. The exported file records the second chain, not the first.

This guide treats that stable, whole-file gap as a measurement problem first and a settings problem second. A loudness meter reading taken on the timeline and the same reading taken on the exported file turn a vague impression into a number that can be compared and tested against. A gap that matches a fader position points at one layer; a gap that appears only when a particular export switch is on points at another. Each test changes one thing at a time and re-measures, so the layer responsible is identified by the change in the reading, not by guessing.

What shape is this article about?

This article covers only a narrow shape, and five conditions must all hold. First, the subject is a finished exported file: a file that has already been through the export stage. A recording file that is itself quiet is a recording-level matter, not this guide's shape, because the quietness was present before editing began. Second, the timeline and the exported file contain the same material and the same mix: the timeline plays with sound and sounds loud enough, and the exported file plays the same material and sounds steadily quieter. If the timeline itself is quiet, the mix itself is quiet, and there is no export gap to diagnose. Third, the exported file is steadily quieter for its whole duration, from start to finish, by a consistent amount. It is not quiet in some sections and normal in others, it does not grow gradually quieter, and it does not cut in and out. Fourth, the exported file has sound: it is quiet, not silent. A file with no sound at all from start to finish is a different shape, and a file with some sections silent and others audible is a different shape again. Fifth, the comparison must have a stated baseline: which device was used to monitor the timeline, which player was used to play the exported file, and where each volume control was set. Without that baseline, "quieter" is only an impression, and no later comparison can be trusted.

Several neighbouring shapes look similar and must not be confused with this one. A file that is silent from start to finish belongs to the guide on a fully silent export, identified in this library as article 537, where the checks are about mute, zeroed faders, and missing audio. A file in which some sections are silent while others have sound belongs to the guide on partly silent exports, identified as article 522, where the fault is in the playback chain or in particular sections. A file in which dialogue is quiet while effects and music are normal inside a game belongs to the guide on game dialogue balance, identified as article 536, where the fault is in the balance between dialogue and other sounds inside the game mix, not in the gap between a timeline and an export. A file whose pitch and speed have both changed belongs to the sample-rate guide, identified as article 504. A recording in which sound comes from one ear only belongs to the channel guide, identified as article 524. A recording in which the microphone and the game sound drift apart over time belongs to the clock-drift guide, identified as article 523. A recording that is silent on a phone and audible on a computer belongs to the codec-compatibility guide, identified as article 526, where the file has no sound at all on one device rather than being steadily quiet on every device. Each of those guides asks a different question, and a step taken from one of them will not diagnose the steady whole-file loudness gap this guide addresses.

The shape also excludes two situations that are easy to mistake for an export fault. If the file only sounds quiet after it has been uploaded to a platform, the file itself may be at a normal loudness and the platform may have applied its own loudness handling after upload; that is a platform layer, and the diagnosis must start by measuring the local file before blaming the export. If the recording file itself was quiet because the recording gain was set too low, the timeline will also sound quiet once the monitoring volume is set to a known level, and there is no export gap at all; that is a recording-gain matter, and it belongs on the recording side of the chain.

How it works: the chain from the timeline to the ear

Sound travels from the timeline to the ear through a series of gates, and each gate multiplies what came before it. If one gate is set low, everything after it is low as well, even if every later gate is left untouched. Timeline playback sounding loud proves only that the monitoring path was loud enough; it proves nothing about the export path, because the two paths share only part of the chain and diverge at exactly the points where this guide's faults live.

1. Clip gain: the gain of each individual clip

Every clip on the timeline carries its own gain. Clip gain sets the starting level for that clip before it reaches the track fader. When one clip has been turned down, that clip alone sounds quiet while the clips around it stay normal, so clip gain is usually not the main cause when the whole file is steadily quiet. It still has to be excluded, because a project in which many clips were turned down together can look like a whole-file fault, and because a single representative clip with a lowered gain setting is the quickest way to confirm that this layer is not involved. The check is to select a representative clip and read its gain or normalisation setting, and to note whether it has been lowered from its original value.

2. Track faders: the fader of each audio track

Each audio track has its own fader. A track fader pulled down makes the whole of that track quieter for the entire programme. This layer is different from mute: mute is zero, and a low fader is quiet but still audible. The distinction matters because the fully silent export guide checks for zero, while this guide checks for low. A track fader at a low setting makes the exported file quieter by the same amount in decibels: a fader at minus 20 dB makes the file quieter by 20 dB than the same mix with the fader at 0 dB, and there is no need to look at the encoder to explain the difference. When several tracks are involved, each fader must be read separately, because one low track can drag the whole mix down while the others sit at their normal positions.

3. The master fader and the main bus: the last fader before export

After all the tracks are mixed together, the combined signal passes through one last fader before export: the master fader, or main bus. In Premiere Pro this is the main output of the audio track mixer; in DaVinci Resolve it is Bus 1, the main bus in the Fairlight page. A low master fader makes the whole programme quieter together, which is why it is the most common shape in this guide. It is also the easiest to miss, because it does not change how the timeline feels during editing: the monitoring volume can be turned up to compensate, so the timeline still sounds loud in the room while the export faithfully records the low master level. The check is to read the master fader position in the mixer and to write the value down before changing anything.

4. Monitoring volume: the amplifier that is not in the file

How loud the timeline sounds during editing is decided by the monitoring volume control, the system volume, and the volume of the headphones or speakers, and all of those sit outside the exported file. A timeline can sound loud simply because the monitoring volume was turned up, while the file itself is at a much lower loudness. This is the single most important sentence in the whole guide: how loud the timeline sounds is not how loud the file is. Monitoring volume is an amplifier outside the measurement, and it does not exist at export time. Any comparison that does not record the monitoring level is comparing a feeling with a file, and the feeling will always win while the file is blamed.

5. Loudness normalisation on export: the gate that only the export passes through

At export, a switch can be turned on that pulls the whole mix toward a target loudness. This gate exists only on the export path; timeline monitoring never passes through it. It is not a "make it louder" switch. It is a "pull it toward the target" switch: if the mix is louder than the target, normalisation pulls it down; if the mix is quieter than the target, it pulls it up; and if the peaks hit the ceiling, it may not be able to reach the target at all. The target value must be written down as part of the diagnosis, because a file exported to a target of minus 23 LUFS will sound quiet when played in a place that expects minus 14 LUFS, and that is a target chosen too low, not a broken export. Normalisation changes only the level, not the dynamic range: the Resolve manual states that its normalise command applies no dynamics, and the same point is made in the Resolve normalisation guides, so a normalised file is the same mix at a different overall level, not a compressed mix.

6. Player volume and platform normalisation after export: the second amplifier

Once the file exists, the player volume control, the system volume, and the headphones or speakers form a second chain of amplifiers, and after upload a platform may apply its own loudness handling as well. This layer explains why the same file can sound very different in different places while the file itself has not changed at all. It is a comparison layer, not a repair layer: turning the player volume up makes the file sound louder in that player, and it changes nothing about the file.

LayerWhere it sitsWhat it does to the fileHow it shows up
Clip gainEach clip, before the track faderLowers one clip onlyOne clip quiet, the rest normal
Track faderEach audio trackLowers one whole trackThat track quiet for the whole programme
Master fader and main busAfter the mix, before exportLowers the whole programme togetherWhole file steadily quiet
Monitoring volumeOutside the file, during editingChanges nothing in the fileTimeline sounds loud, file measures quiet
Export loudness normalisationExport onlyPulls the whole mix toward a targetFile lands near the target value
Player volume and platform handlingAfter export, at playbackChanges nothing in the fileSame file sounds different in different players

Loudness in LUFS, and how it differs from peak level in dBFS

Loudness is measured in LUFS, Loudness Units Full Scale, which measures how loud a programme sounds to a listener, using the measurement method of ITU-R BS.1770 with K-weighting to model human hearing. Integrated loudness is the average loudness of a whole programme from start to finish. The broadcast standard EBU R 128 recommends normalising broadcast programmes to minus 23 LUFS, with a true peak ceiling of minus 1 dBTP. Peak level, measured in dBFS or in dBTP for true peak, measures the highest instantaneous point of the signal, not how loud it sounds. Two clips can have the same peak level and differ in integrated loudness by more than 10 loudness units: one is sustained loud dialogue, the other is a single loud burst with long quiet passages around it. A peak meter sitting at the top proves nothing about loudness, which is why this guide uses a loudness meter and not a peak meter. One loudness unit is close to one decibel for this diagnosis: a timeline at minus 14 LUFS and an exported file at minus 23 LUFS differ by 9 LU, close to 9 dB, and the fader or target that accounts for 9 dB is the layer to look for.

Loudness tools in each editor, kept separate

Premiere Pro and DaVinci Resolve handle loudness in different places, and the two must not be mixed up. In Premiere Pro, loudness normalisation is found in the export settings, under the Effects tab, where Loudness Normalization offers standards including ATSC A/85, EBU R 128, and ITU BS.1770-3, of which only ITU BS.1770-3 allows a custom target loudness; measurement uses the Loudness Meter added to the first effect slot of the Mix track in the Audio Track Mixer. In DaVinci Resolve, measurement is done on the Fairlight page, under Meters and then Loudness, by playing the whole timeline and reading the integrated value; the target loudness in Project Settings, under Fairlight and then Audio Metering, defaults to minus 23 LUFS, and can be changed to minus 14 to match the reference value measured from YouTube player data. A clip in Resolve can be normalised by right-clicking it and choosing Normalize Audio Levels, with Set Level offering Relative, which moves a group of clips together, and Independent, which moves each clip on its own.

Premiere Pro also has an Audio Gain dialogue, opened with the G key, and it must not be confused with loudness normalisation. Its Normalize Max Peak and Normalize All Peaks options pull the loudest peak to a chosen dBFS value, and they measure peaks, not LUFS. After peak normalisation, two clips can have the same peak level and still sound different in loudness. A peak-normalised clip is not a loudness-normalised clip, and a peak-normalised export can still measure quiet in LUFS even though its peaks reach the chosen dBFS value.

Platform target values: only what has a source

YouTube has not published a loudness target figure. The value of minus 14 LUFS that is commonly quoted for YouTube is a reference value measured from YouTube player data, in the Volume and Normalized column of the Stats for nerds display, and the DaVinci Resolve manual states the YouTube target loudness specification as minus 14 LUFS. This guide therefore never presents minus 14 LUFS as a figure published by YouTube; it is a measured reference value and a manual statement, and should be treated as such wherever it appears. YouTube playback only turns down material that is louder than that reference value; material quieter than it is not turned up, so a file below minus 14 LUFS simply sounds quieter next to other videos. Target loudness values for other platforms were not verified from primary sources for this guide, and they are not given here: each platform has its own target, and the creator documentation current at the time must be checked before any figure is quoted.

The EBU R 128 target of minus 23 LUFS carries a tolerance that changed between versions of the standard, so this guide quotes the target value only and does not state a tolerance figure.

Diagnostic steps: six tests in order

The order below matters, because each step narrows the fault to one layer before the next step is attempted. Only one thing is changed at a time, and the exported file is re-measured after each change, by ear alone. A short test export is used for every comparison, so that each change can be checked quickly without re-exporting the whole programme for every test.

Step 1 — Confirm the shape and write down the baseline

Confirm that the exported file is steadily quieter for its whole duration, not quiet in some sections, not growing gradually quieter, and not cutting in and out. Then write down the baseline before anything is changed: the editor and its version, the export preset, the device used to monitor the timeline and where its volume was set, and the player used for the exported file and where its volume was set. Also note whether the timeline itself sounds loud at a normal monitoring level, and whether the exported file sounds quiet at the same player volume used for other files. Without this written baseline, later comparisons have nothing to compare against, and a change in player volume halfway through the diagnosis can look like a change in the file.

Step 2 — Measure the timeline against the exported file with a loudness meter

This is the first decisive test, and it is the one that turns the impression into a number. Play the whole timeline and read its integrated loudness in LUFS. Then measure the exported file with a loudness meter — a free loudness meter plug-in, the loudnorm filter in ffmpeg, or the same editor by importing the exported file and measuring it there — and read the same integrated figure. Place the two numbers side by side; the difference between them is the gap itself.

If the timeline reads minus 14 LUFS and the file reads minus 23 LUFS, the gap of 9 LU sits in the export layer, and the diagnosis goes to Step 3. If both read near minus 23 LUFS, the gap is not in the export and the mix itself is quiet, so the diagnosis goes to Step 4. If both read near minus 14 LUFS, the file itself has no gap and the fault is in playback, so the diagnosis goes to Step 5. The example readings here are illustrations of the logic, using the two reference values named in this guide; the reader's own readings are the evidence, and they must be written down exactly as measured. A gap that matches a fader position is a strong clue: if the master fader sits 9 dB low and the gap is 9 LU, the two numbers are describing the same fault.

Step 3 — Compare the export with loudness normalisation on and off

Export the same timeline twice, once with export loudness normalisation switched on and once with it switched off, using a short test export for each, and measure the integrated loudness of both files. The difference between the two readings is what the normalisation layer added or removed. If the normalisation is on with a target of minus 23 LUFS, the exported file will land near minus 23 LUFS, and that is not a fault: it is the setting doing what it was set to do. If the two exports measure the same and both are quiet, normalisation is not the cause, and the diagnosis moves to the fader layers in Step 4. If the target was set to minus 23 LUFS and the file is played in a place that expects minus 14 LUFS, the file will sound quiet there, and the handling is to choose the target that matches where the file will be played, not to blame the export. Note the target value in writing each time, because a target that is never recorded cannot be compared with the next export.

Step 4 — Check the fader layers one at a time, from clip to master

Work through the layers in order, from the smallest to the largest, and write down each value before changing it. First, clip gain: select a representative clip and read its gain or normalisation setting, and note whether it has been lowered. Second, track faders: read the fader position of every audio track in decibels, and note whether any track sits well below 0 dB. Third, the master fader or main bus: read the position of the main output in the audio track mixer in Premiere Pro, or of Bus 1 in Fairlight in DaVinci Resolve, and write the value down.

Then verify by changing only the suspected layer. Return that fader to 0 dB, or add back the same number of decibels as the measured gap, and make a short test export. If the exported file's LUFS rises by the same number of loudness units as the gain added, that layer is the cause. A fader at negative infinity is silence, not quietness, and it belongs to the fully silent export guide, not this one. If no fader layer accounts for the gap, return to Step 3 and re-check the export settings.

Step 5 — Compare the player with a known good reference file

Play the problem file and a reference file of known normal loudness — another exported file that sounds normal — in the same player, on the same machine, at the same volume, through the same headphones or speakers. If the reference file sounds normal and the problem file sounds quiet, the fault is in the file layer, and the diagnosis returns to Steps 2 to 4. If both files sound quiet, the fault is in the player, the system volume, or the output device, and the file is not the problem. This is a shared check, not a new test of this guide, and it must not be presented as the decisive step: it separates the file from the playback chain, and the loudness meter readings in Step 2 are what identify the layer inside the file.

Windows loudness equalisation, a system playback enhancement, can also change how loud playback sounds, and it belongs on the playback side of this comparison: it changes what is heard, and it is a playback-side direction to suspect, not an export cause to assert without evidence.

Step 6 — Re-export a short test, changing only one thing

For every change, re-export a short test section only, and change only one thing at a time: the normalisation switch, one fader, or the target value, never two or three together. After each change, measure the new exported file in LUFS again and write the reading down next to the change that produced it. Re-exporting with the same settings will produce the same LUFS reading, so a second export on its own repairs nothing; only a changed setting changes the reading. Once the reading matches the intended level, export the full programme with those settings and measure the full file once more, because a short test confirms the direction of the change and the full export confirms the result.

Boundaries: what this article does not cover

A fully silent export, with no sound at all from start to finish, is article 537. Its checks are about mute, zeroed faders, and missing audio, and they ask whether the signal is present at all, while this guide asks how loud a present signal is. A partly silent export, with some sections silent and others audible, is article 522, where the fault is in the playback chain or in particular sections. Game dialogue that is quiet while effects and music are normal inside a game is article 536, where the fault is the balance between dialogue and other sounds inside the game mix. A file whose pitch and speed have both changed is article 504, where the fault is in the sample rate. A recording with sound in one ear only is article 524, where the fault is in the channels. A recording in which the microphone and the game sound drift apart over time is article 523, where the fault is clock drift. A recording that is silent on a phone and audible on a computer is article 526, where the fault is codec compatibility and the file has no sound at all on one device.

Two further situations sit outside this guide even though they feel close to it. If a file only sounds quiet after upload to a platform, the local file must be measured first, because the platform may have applied its own loudness handling after upload and the export may be correct. If a recording file was itself quiet because the recording gain was set too low, the timeline will also be quiet once the monitoring volume is set to a known level, and the fault belongs on the recording side. The loudness target values of platforms other than the YouTube reference value were not verified from primary sources for this guide, and they are not given here. The loudness normalisation switches of editors other than Premiere Pro and DaVinci Resolve were not verified for this guide, and this guide does not assume whether other editors have such a switch or what it is called. The effect of individual players' built-in loudness handling was not verified either.

Common misconceptions: eight claims that mislead

  1. "Turning the player volume up fixes it." Wrong. Player volume amplifies the whole file together, and the file's own loudness does not change. Upload the file to a platform or play it on another machine, and the problem returns at once. Player volume is a comparison, not a repair.
  2. "Exporting again fixes it." Wrong. Exporting again with the same settings produces the same loudness target, the same fader positions, and the same LUFS reading. Only a changed setting changes the result.
  3. "The file is damaged." Wrong. The data is complete, the file plays to the end, and the whole programme is evenly quiet. Damage does not make a whole programme evenly quieter by a steady amount.
  4. "A full peak meter means it is loud enough." Wrong. A peak meter can sit at minus 1 dB while the integrated loudness is still minus 23 LUFS. Peak and loudness are two different measures, and loudness must be measured with a loudness meter.
  5. "Switching loudness normalisation on always makes it louder." Wrong. Normalisation pulls the mix toward its target, and a target below the mix level makes the file quieter. A file normalised to minus 23 LUFS will sound quiet in a place that expects minus 14 LUFS.
  6. "If the timeline sounds loud, the file must be just as loud." Wrong. Monitoring volume is not written into the file. A loud timeline may only mean the monitoring volume was turned up.
  7. "A low fader is the same as mute." Wrong. A low fader is quiet; a fader at negative infinity, or a mute switch, is zero. The two have different diagnostic paths: this guide checks for low, and the fully silent export guide checks for zero.
  8. "Changing the encoder or the container makes it louder." Wrong. The encoder and the container do not change loudness. Loudness is decided in the mix and in the normalisation layer, and changing the container leaves both untouched.

FAQ

My exported file sounds quieter than the timeline. Is the file damaged?

No. If the file plays from start to finish, the sound is present throughout, and the whole programme is evenly quieter, the data is complete and the file structure is not the problem. Damage does not produce a steady, even loudness gap across a whole programme. Measure the timeline and the exported file in LUFS first: the difference between the two readings is the gap, and the fader or normalisation layer that accounts for that difference is the cause.

The timeline and the exported file measure the same in LUFS, and both are quiet. What does that mean?

It means the gap is not in the export. The mix itself is quiet, and the timeline only sounded loud because the monitoring volume was turned up. Go to the fader layers: check clip gain, each track fader, and the master fader, return the suspected layer to 0 dB, and make a short test export to see whether the LUFS reading rises by the same amount.

The timeline and the exported file measure the same in LUFS, and both are loud enough. Why does the file still sound quiet?

It means the file itself has no gap, and the fault is in playback. Compare the problem file with a reference file of known normal loudness in the same player, at the same volume, through the same headphones or speakers. If both sound quiet, the player volume, the system volume, or the output device is the layer to check. If the reference file sounds normal and the problem file sounds quiet, re-check the LUFS readings, because the two results cannot both be right.

I switched loudness normalisation on and the file got quieter. Is the switch broken?

No. Normalisation pulls the mix toward its target, and a target below the mix level makes the file quieter. Check the target value: a file normalised to minus 23 LUFS will sound quiet in a place that expects the minus 14 LUFS reference value measured from YouTube player data. The handling is to choose the target that matches where the file will be played, and to export a short test with normalisation off to confirm how much the switch itself changed the reading.

My peak meter reaches the top, but the file still sounds quiet. How is that possible?

Because peak level and loudness are different measures. Peak level measures the highest instantaneous point, while integrated loudness in LUFS measures how loud the whole programme sounds on average. A programme with one loud burst and long quiet passages can have a full peak meter and a low integrated loudness. Use a loudness meter, not a peak meter, and judge the file by its integrated LUFS reading.

I exported the same timeline twice and got the same quiet file. What should I change?

Change a setting, not the export count. Re-exporting with the same settings produces the same result. Change one thing at a time — the normalisation switch, one fader, or the target value — make a short test export, and measure the new file in LUFS. When the reading rises by the amount expected from the change, that layer is the cause, and the full programme can then be exported with those settings.

If the fault still cannot be divided

Write down the timeline integrated LUFS reading, the exported file integrated LUFS reading, the export loudness normalisation setting and its target value, the position of each track fader and of the master fader, and the player and volume used for each comparison, and bring that written record with the file. Bring the written record described above to Room 1209, Dali Plaza, Shek Mun for inspection. The telephone number is 2186 8010 and WhatsApp is 6558 6806. There is no Mong Kok shop, and there is no on-site service. The site's published three-tier fee is HK$400–780, HK$480–1,380, and HK$680–2,800, an overall range of HK$400–2,800, with the actual fee quoted after inspection.

Related articles

Related services

If you need Chip-level GPU repair, see these service pages: