Guide
Play your own file first. That is the entire diagnosis
In most cases nothing took your volume away, the file was quiet when it left. Platforms turn loud videos down far more readily than they turn quiet ones up, so a quiet master stays quiet while everything around it sits at a comfortable level. This page shows how to tell that apart from real platform processing, in about two minutes.
- To prove where the volume went
- 2 min
- A typical gap between a phone read and the feed
- 8 dB
- Fader that will not fix it
- 1
Open your export in a plain player and compare it to anything else
Do not compare inside the app. Play the file you uploaded in a normal video player, at a volume you do not touch, then play any other video from your library at the same volume.
If your file is already the quiet one, nothing was taken. It left home like that and you have a mix problem, not a platform problem.
If your file is level and only the uploaded version is quiet, that is real platform handling and it is worth chasing. In our experience that is the rarer case by a wide margin.
The reason people get this wrong is that they judge their own file inside their editor, where the monitoring level is set for working rather than for listening.
The two minute test
Two plays and one comparison. It separates a mix problem from a platform problem before you change anything.
Play your export
Plain player, fixed volume
Play any other video
Same volume, do not touch it
Quiet already?
Your mix. Fixable today.
Only quiet online?
Real processing. Rare.
Turning the fader up is not the fix, because the fader was never the problem
The instinct is to raise the master until the meter looks healthy. That raises the peaks first and the perceived loudness barely at all.
A read with big consonants hits the ceiling long before the average gets anywhere. You run out of headroom at -20 LUFS and the file starts clipping while still sounding thin.
What actually moves perceived loudness is reducing the distance between the loud and quiet parts, then lifting the whole thing. Gentle compression, then gain, then a limiter to catch what is left.
Doing it in the other order is why so much amateur audio is simultaneously quiet and harsh. The peaks got loud, the voice did not.
This is the whole editor
Highlight a phrase and a clip lands on those exact words. No timeline, no keyframes, no layers.
Three things make a read quiet, and two of them happen before you press record
Microphone distance is the biggest one. Doubling the distance from your mouth to the mic costs roughly six decibels, and a phone on a tripod across a room is several doublings.
The second is a peak-normalized export. Plenty of tools offer normalize as a single checkbox and it normalizes peaks, which does nothing at all for loudness on a dynamic recording.
The third is untreated dynamics. One loud laugh in an otherwise level read sets the ceiling for the whole file, and every other second sits far below it.
None of the three is fixed by the platform, an upload setting, or a different app. They are fixed at the microphone and in the mix.
What distance costs you, before any processing
Roughly six decibels per doubling. This is why a phone at arm's length beats an expensive microphone across the room.
- Phone at 20 cmreference
- Phone at 40 cm-6 dB
- Phone at 80 cm-12 dB
- Tripod across the room-20 dB or worse
The ninety second fix, in whatever you already use
Find the loudness meter in your editor and read the integrated figure on the finished cut. If it says something in the low twenties, that is your answer in one number.
Apply a gentle compressor to the voice track first. Three or four decibels of reduction on the loudest syllables is plenty and is not audible as an effect.
Then raise the whole mix until the integrated figure sits near -14, and put a limiter at the end to hold the peaks under -1.
If you would rather not learn any of that, an automated pipeline is a reasonable answer. Every Cutroom export runs the measure, gain and limit chain without asking, which is the entire benefit and also the entire limitation: there is no dial to turn if you disagree with the target.
Questions people ask
- Does TikTok turn quiet videos up?
- Not dependably, and it does not say. What is consistent across platforms is that loud material gets pulled down more readily than quiet material gets pushed up, which is why a quiet master stays the quiet video in the feed.
- Will exporting at a higher bitrate make it louder?
- No. Bitrate is how much detail survives the encode and has nothing to do with level. A 50 Mbps file at -22 LUFS is exactly as quiet as an 8 Mbps one.
- Should I just record with the mic gain higher?
- Up to a point. Higher gain raises the noise floor with the signal, and a clipped recording cannot be repaired at all. Getting the microphone closer beats turning it up every time.
- My music is fine but my voice is buried. Same problem?
- Different one. That is a balance issue between two tracks rather than an overall level issue, and the fix is ducking the music under the speech rather than raising the master.
Prove where the volume went before you change anything, because the fix for a quiet mix and the fix for platform processing have nothing in common. If you would rather not think about it again, the chain that measures, gains and limits runs on every export we produce.