CutScore
Join the waitlist
AUDIO LOUDNESS BLOG / 8 MIN READ

Why is my video audio so quiet?

Almost always for one reason: the mix was never normalised to a loudness target, so the platform left it where it was. Here is why quiet audio happens, and the two numbers that fix it for good.

Check your loudness automatically Jump to the fix
−14 LUFSYouTube target
−1 dBTPpeak ceiling
1cause, most of the time
2 minto fix it

By Thomas, founder of CutScore · Updated June 2026

LOUDNESS CHECK · quiet_take.mp4
A studio mixing console covered in faders and level meters, the kind of desk where a video's loudness gets set before it goes out into the world.
CRAFT SCORE
FIXES ADVISED
your mix, measured against the target
Loudness too low · −22 LUFS, lift +8whole
True peak safe · −6.2 dBTP✓
Noise floor low · clean room tone✓
The 30-second answer Your video audio is too quiet because the mix was never normalised to a loudness target. You almost certainly recorded at a safe, low level to avoid clipping, then exported without bringing the average loudness back up. YouTube measures every upload and normalises toward about −14 LUFS, but it only turns loud videos down, never quiet ones up. So a soft export stays soft next to everyone else. The fix is one pass: normalise the whole mix to roughly −14 LUFS with a limiter holding the true peak at or below −1 dBTP. Not sure where your file sits? That is exactly what CutScore measures in one pass.
WHY THIS KEEPS HAPPENING

I have shipped quiet videos. More than I would like to admit. The take sounded perfect in my headphones at 11pm, then I watched it on my phone the next morning, on the train, and I could barely hear myself over the rattle of the carriage. The audio was not broken. It was just timid. And timid is its own kind of broken once your video has to survive in a feed full of louder ones.

Here is the chain that gets almost everyone. You record carefully, keeping the level low so a sudden laugh or a door slam does not clip and ruin the take. Smart move. Then you edit, you cut, you add music, and you export. Nowhere in that chain did anyone tell the software, "bring the whole thing up to a normal loudness." So the file leaves your machine at the cautious level you recorded it, which is roughly where the problem starts.

Then the platform finishes the job of making you quiet. YouTube, TikTok and the rest measure the loudness of every upload and pull the loud ones down to a shared target. They do not push the quiet ones up. Think about what that means. If your export is below the line, nothing rescues it, and the next creator who did normalise sounds twice as confident as you for free. Loud is not the goal. Consistent is.

THE USUAL SUSPECTS

Why is my video audio so quiet? The five real causes.

Quiet audio is rarely one big mistake. It is usually one of these, and most of them have nothing to do with how much you spent on a microphone.

CauseWhat the meter showsThe fix
No loudness target set≈ −22 LUFSNormalise the full mix toward −14 LUFS before you export. This is the big one.
Recorded too conservativelypeaks near −20 dBFine to record low; just raise the loudness in the edit instead of leaving it.
Mic too far from youlow signal, high noiseMove it closer. A near mic gives you level to spare without amplifying the room.
Platform turned you downloud peaks, low averageA few hot moments made the platform duck the whole track. Even it out, then normalise.
Monitoring too loudmeter, not earsIf you edit at high volume, everything sounds fine. Trust the LUFS meter, not the dial.
The one that surprises peopleYour audio can be quiet because parts of it were too loud. If a single clap or laugh spikes near the ceiling, the platform reads that peak, decides the file is hot, and turns the whole video down to compensate. Tame the spikes first, then the average can come up.
DON'T GUESS YOUR LOUDNESS

Ears adapt and headphones lie. CutScore puts a real EBU R128 meter on your export and tells you the exact gain change to hit the target, before you upload.

Join the waitlist
HOW TO FIX QUIET AUDIO

The two-number fix, step by step.

1. Measure where you actually are

You cannot fix a number you have not read. Drop a loudness meter on your finished timeline and look at the integrated, or program, loudness in LUFS. If it reads around −20 to −24 LUFS, there is your answer: you are roughly six to ten decibels below the −14 LUFS target YouTube normalises toward. That gap is the whole reason your video sounds quiet next to the one that played before it. Write the number down before you touch anything.

2. Tame the peaks before you lift

Before raising the average, deal with any spikes. A single loud laugh or a hand clap near the ceiling will clip the moment you push the whole mix up. Put a limiter on the master with the ceiling set so the true peak stays at or below −1 dBTP. That one decibel of headroom matters, because platforms re-encode your audio and lossy compression can nudge peaks slightly higher than your file shows. The limiter catches the rogue moments so the rest of the mix can come up cleanly.

An audio engineer adjusting faders on a mixing desk while watching the meters, the moment where a quiet mix gets normalised up to a sensible loudness target.
Tame the peaks, then bring the average up to target. Photo: Los Muertos Crew / Pexels.

3. Normalise the whole mix to −14 LUFS

Now raise the average. Most editors have a loudness normalisation step, sometimes called "match loudness" or "normalise to LUFS", and you point it at −14 LUFS for a talking video. It lifts the entire mix evenly so the quiet parts and the loud parts keep their relationship, instead of crushing everything into a wall of sound. If your tool only offers peak normalisation, that is not the same thing; peak normalisation looks at one loud sample, while loudness normalisation looks at how loud the whole thing actually feels.

4. Keep your voice on top of the music

Quiet voice and present music is a different problem that feels like the same one. If your speech is buried, the answer is not "make everything louder", it is "make the music quieter under the talking." Pull the music down four or five decibels when you are speaking, then let it breathe in the gaps. Loudness normalisation will not save a mix where the balance is wrong, so fix the relationship first, then set the target. The order matters more than people expect.

5. Check it on the worst speaker you own

This is the step I skip when I am tired and the step I always regret. After you export, play the file on a phone speaker at a normal volume, in a room with some background noise. Can you hear every word without leaning in? Good. If you are still reaching for the volume button, your meter and your ears disagree, and the ears win, because that is what your viewers actually have. The meter gets you to the target; the worst-speaker test confirms you got there.

RATHER SEE IT THAN READ IT?

Here is a real CutScore report for an everyday vlog: loudness, true peak, voice-versus-music balance and noise, each scored with timestamps and the exact gain change to make.

See a sample report
SHORT ON TIME

If you only do three things.

Most quiet-audio complaints disappear after these three moves. Do them in order and you will sit at the same volume as everyone else in the feed.

1
2-MIN FIXAUDIO
Normalise the whole mix to about −14 LUFS
This is the single move that fixes quiet audio for most people, and it has nothing to do with your microphone. Match the integrated loudness of your export toward −14 LUFS and your video stops sounding timid the instant it lands in the feed.
How Use your editor's loudness normalisation, or let CutScore measure it and hand you the exact gain change.
2
AUDIOSAFETY
Hold the true peak at or below −1 dBTP
Before you lift the average, put a limiter on the master so nothing clips when you push it up. One decibel of headroom protects you from the slight peak rise that lossy re-encoding adds after upload, so your audio stays clean instead of crackling.
How Set a brickwall limiter ceiling to −1 dBTP. See true peak.
3
QUICKBALANCE
Pull the music down under your voice
If your speech is the quiet part, raising the whole mix just makes loud music with quiet talking. Duck the music four or five decibels whenever you are speaking, then set the loudness target. Balance first, level second, every time.
How Lower the music track under speech, or use a ducking effect that does it automatically.
THE PART NOBODY TELLS YOU

Quiet audio costs you the view before quiet picture does.

Here is the thing I had backwards for a long time. I worried about my picture, my colour, my framing, the soft shot at 0:40. Meanwhile the audio was the first thing pushing people away, and I could not hear it because I was too close to the project. People will forgive a slightly soft shot. They will not lean in to hear you. If they have to reach for the volume in the first three seconds, a chunk of them just leave, and you never find out why.

In my experience, a correctly normalised mix is the cheapest upgrade in all of video. It does not need new gear, it does not need talent, it needs a meter and one export. A clean, on-target voice reads as "this person knows what they are doing" before you have said anything clever. Quiet audio reads as the opposite, fairly or not. That is a lot of judgement riding on one number you can fix in two minutes, which is honestly good news.

How CutScore checks your loudness for you CutScore is an AI video quality coach for pre-publish QC. It runs a real EBU R128 loudness meter over your file, reports the integrated loudness in LUFS and the true peak in dBTP, and compares both to the right target for your genre and platform. You get the exact gain change to make, the spots where peaks risk clipping, and whether your voice is sitting under the music, all before anyone else hears the video. It judges the craft of the audio, not your tags or thumbnail, so it sits next to a growth tool rather than competing with one. More on the method and the standards.
Get your loudness checked automatically See everything we check
KEEP READING
Check your audio levels How loud a YouTube video should be Music louder than voice LUFS for YouTube A full sample report
QUESTIONS

Frequently asked.

Because the mix was never measured against a loudness target. You recorded at a safe, low level to avoid clipping, then exported the file without bringing the average loudness back up. YouTube normalises toward about −14 LUFS and only ever turns loud videos down, never quiet ones up, so a soft export stays soft next to everyone else.
Sometimes, but it is the wrong tool. Raising a clip gain knob lifts the loud parts and the hiss and the room tone all at once, and you can push the peaks past zero and start clipping. The right move is to normalise the whole mix to a loudness target, around −14 LUFS, with a limiter holding the true peak at or below −1 dBTP.
For YouTube, aim for an integrated loudness of about −14 LUFS with a true peak at or below −1 dBTP. That is the level the platform normalises toward, so your video sits at the same volume as the one before it. TikTok and Reels run a little louder in practice, but −14 LUFS is the safe default for a talking video.
Platforms measure your loudness and turn loud videos down to match their target, but they leave quiet videos alone. If your export sat below the target, nothing changes it, so it plays back soft. The file on your desktop and the uploaded version are the same loudness; the difference is that everything around it on the platform is louder.
EARLY ACCESS

Never ship a quiet video again.

CutScore measures your loudness and true peak against the right target and tells you the exact gain change, with the evidence to back it up. Join the waitlist for early access.

Join the waitlist