Skip to the article

Looking for SpinZ CZ remixes? Click here

Troubleshooting

Why your master sounds distorted

Crackle, fizz, a kick that sounds broken. It is almost always one of five things, and they need different fixes. Work out which one you have before you touch anything.

Do not reach for a plugin yet.

Distortion in a master has five common causes. They sound similar through a laptop speaker and they need completely different fixes. Guess wrong and you will make the song worse while feeling like you did something.

So the first job is finding where it starts.

  1. Play the mix on its own, before any mastering

    If the distortion is already there, stop. It is a mix problem and nothing downstream will fix it.

  2. Bypass the limiter

    If it goes away, the limiter is working too hard. That is cause two.

  3. Solo the low end

    Filter out everything above about 200 Hz. If the ugliness is down there, it is a low end problem.

  4. Check the true peak

    Set a meter to true peak. If it reads above 0 dBTP, you have found it.

  5. Convert to MP3 and listen again

    If it only appears here, the encoder is the trigger and the true peak is the cause.

Worth knowing: more than one of these can be true at the same time. Fix them in this order, listening after each one, or you will not know what helped.

Cause one: the mix was already clipped

This is the most common one, and the most annoying, because by the time you hear it in the master the damage is permanent.

A digital file has a hard ceiling. Push a wave past it and the top of the wave is not squashed, it is cut off. Where there was a rounded peak there is now a flat edge, and a flat edge is a burst of harmonics that were never in the music.

The same 12 millisecond passage of real audio shown clean and then clipped flat against the ceiling
Twelve milliseconds of a real track, before and after. Same music, same ceiling. The top panel has room and every peak keeps its shape. The bottom panel is the same audio pushed 8 dB harder, so the peaks hit the ceiling and go flat. Those flat sections are the distortion. Turning the file down afterwards does not bring the curve back.

How to tell: play the mix on its own with no mastering. If it is already there, it is here. Look at the mix in your editor and zoom right in on the loudest part. Flat tops mean clipping.

The fix is in the mix session, not the master. Pull the master fader down until nothing hits the ceiling, then export again. If the mix bus has a clipper or limiter on it purely for loudness, take it off. If the clipping is on one instrument, fix that instrument.

A trap to avoid: exporting at 32-bit float hides clipping rather than preventing it. The numbers survive above the ceiling in the file, then clip the moment anything converts them to a normal format. Turn the level down, do not rely on the format.

Cause two: the limiter is working too hard

A limiter holds the peaks down so you can turn the whole song up. Ask it for a decibel or two and you do not hear it working. Ask it for six or eight and you hear everything.

What it sounds like: the kick loses its front edge and turns into a thud. The vocal ducks down every time the kick lands. Cymbals get a fizzy edge. The whole song breathes in and out in a way that gets tiring after a minute.

How to tell: bypass it. If the song opens up and the distortion goes, that is your answer. Also watch the gain reduction meter. If it is pinned down and rarely comes back up, the limiter is not limiting peaks any more, it is compressing the whole song.

What that looks like on a limiter.

FabFilter Pro-L 2 limiter interface showing gain reduction, true peak and loudness meters
Limiter: watch the gain reduction, not the output.The output meter can look perfectly safe while the gain reduction lane shows the limiter catching six or more decibels on every hit. That lane is the honest one. If it sits pinned through the loud sections, you have found the distortion. Official FabFilter product page

These are editorial examples, not endorsements or a buying list. Product names and interface images belong to their respective owners.

The fix is to ask it for less. Get the loudness from earlier in the chain by controlling the low end and gluing the loud moments together, so the limiter only has to catch the last decibel or two. If you still need the loudness, accept a quieter master. Streaming turns everything down anyway, which is covered in the guide on LUFS targets.

Cause three: the low end is overloading everything

Low frequencies carry most of the energy in a song. A sub note at 40 Hz moves the meter far more than a hi-hat does, even though you barely hear it on a laptop.

So a rumbling, uncontrolled low end eats all the headroom, the limiter clamps down on every bass note, and the whole song pumps. The distortion sounds like it is in the midrange, but it starts underneath.

How to tell: filter out everything above roughly 200 Hz and listen to what is left. If it is muddy, wandering or has notes at wildly different volumes, that is your problem. Also check whether the kick and bass are fighting in the same place.

The fix is in the mix. High-pass the things that do not need low end. Get the kick and bass to take turns rather than share. Check for rumble below 30 Hz that you cannot hear on your speakers but that is still using up your headroom.

Cause four: the peak between the samples

This one is invisible on a normal meter, which is why it catches so many people.

Your file stores a series of numbers. Your speaker does not play those numbers as steps, it plays a smooth curve drawn through them. That curve can rise higher than any of the stored numbers. A normal peak meter only reads the numbers, so it tells you everything is fine while the actual sound goes over the ceiling.

A measured inter-sample peak where the reconstructed curve rises above the highest stored sample
Measured in one of my own releases. The blue dots are the numbers stored in the file. The pink line is the curve your converter draws through them, which is what you actually hear. Here the curve reaches -1.62 dBFS while the highest stored sample is only -2.56 dBFS. A sample peak meter shows none of that.

How to tell: switch your meter from sample peak to true peak. If it reads above 0 dBTP, or even close to it, you have this.

A meter that reads the curve, not just the numbers.

Youlean Loudness Meter 2 showing integrated loudness and a true peak readout
True peak readout: the number your sample meter hides.The true peak value sits beside the loudness number. Anything above -1 dBTP is worth a second look, and anything above 0 dBTP is already going over on playback. The free version shows this. Official Youlean product page

These are editorial examples, not endorsements or a buying list. Product names and interface images belong to their respective owners.

The fix is simple. Leave room. Set your ceiling to -1 dBTP, and to -2 dBTP if the master is loud. You lose a decibel of loudness that streaming was going to take off you anyway.

Cause five: it only appears after conversion

You check the WAV and it is clean. You upload it, play it back from the platform, and there is a crackle that was not there before.

This is cause four showing up later. MP3 and AAC do not store your waveform, they store a description of it and rebuild it on playback. The rebuilt wave is close but not identical, and its peaks can land higher than the original. If your master had no headroom, those new peaks have nowhere to go.

How to tell: export an MP3 yourself at 320 kbps, load it back in and listen. If the crackle appears, this is it.

The fix is the same as cause four. Give the master a true peak ceiling of -1 dBTP, and check your master as an MP3 before you upload rather than after. There is more on what conversion does to a file in the WAV versus MP3 guide.

When distortion is the point

Worth saying, because this guide is all about removing it.

Plenty of records are built on distortion. Saturated drums, a driven bass, a vocal pushed through something dirty. That is a sound, and it is a good one.

The difference is where it happens. Distortion you chose, printed into a track in the mix, is part of the music. Distortion that appears at the very end because the file ran out of room is damage, and it lands on everything at once rather than on the thing you wanted to colour.

If you want the grit, put it on the instrument in the mix and leave the master clean.

Common questions

Why does my master sound fine loud but bad quiet?
Usually limiting. Heavy limiting flattens the differences that make a song feel alive. At high volume you do not notice, at low volume the song sounds flat and lifeless because there is nothing left moving.
I exported at 32-bit float, so I cannot clip. Right?
The float file itself will not clip, but the moment it is converted to 24-bit, 16-bit, MP3 or AAC, anything above the ceiling clips then. Float postpones the problem, it does not remove it.
What is the difference between sample peak and true peak?
Sample peak reads the highest number stored in your file. True peak estimates the highest point of the smooth curve your converter draws between those numbers, which is what actually reaches the speaker. True peak is the one that matters.
My meter says -0.1 dBFS and it still distorts. Why?
That will be a sample peak reading. Switch the meter to true peak and it will almost certainly read above 0 dBTP.
Can mastering remove distortion that is already in my mix?
No. Clipping removes information, and it cannot be put back. The only fix is to go back to the mix, lower the level and export again.
Should I use a clipper instead of a limiter to get loud?
Clippers are a legitimate tool and a lot of loud records use them deliberately. The difference is control. Used gently on a controlled low end it can sound great. Used to force loudness onto an uncontrolled mix it just cuts the tops off everything.