What mastering will not fix

You bounce the track. You upload it, or you send it to an engineer. A day later, or ninety seconds later, you get something back that is louder, a little brighter, a little wider.

And the thing that was bothering you is still there.

This happens constantly, and it is almost never the engineer's fault or the algorithm's fault. It is a category error. Mastering is a finishing process. Most of what producers hope it will do is fixing, and those are different jobs with different tools.

Browse artists and get feedback on your demo →

What a mastering engineer is actually working with

A stereo file. Two channels, already summed, with every balance decision you made baked in permanently.

Their tools are broad by definition: overall level, tonal balance across the whole track, dynamics applied to everything at once, stereo image, and consistency across a release. Good mastering is a set of small, confident moves on a finished thing.

What none of those tools can do is reach inside the file and move one element relative to another. The kick and the bass arrived as one signal. The reverb and the dry synth arrived as one signal. Anything that requires separating them is out of scope.

Stem mastering gives back a little of that control, usually four to eight groups, and it is worth considering if you know the problem is a balance issue. But it is a partial fix at a higher price, and it still will not save a mix where the underlying decisions are wrong.

This applies with more force to AI mastering, not less. An algorithm has no ability to decide that your low end is a mistake. It analyses what you gave it and optimises it, which means it will render your problem faithfully and at competitive loudness.

Here is what survives.

1. Kick and bass fighting each other

The most common one, and the one that decides whether a track works on a large system.

If your kick and your sub occupy the same frequency range at the same moments, they are adding and cancelling against each other in ways that change with the room. In your studio it reads as a slightly soft low end you have stopped noticing. On a club rig it reads as a track with no weight, at the exact moment weight is the only thing anyone is listening for.

Mastering presents this louder and more clearly. Compression at the mastering stage can make it worse, because the combined low-end energy is what triggers the compressor, so the whole track ducks every time the conflict happens.

What resolves it: sidechain, carving a pocket in the sub for the kick fundamental, tuning the sub to the key, or changing one of the two sounds entirely. All mix-stage decisions.

2. A stereo sub

Width below roughly 100 to 120 Hz is a self-inflicted wound. It sounds enormous on headphones and disappears in mono, which is what a lot of club systems and every phone speaker effectively give you. It also makes a track difficult or impossible to cut to vinyl.

An engineer can collapse the low end to mono for you. What they cannot do is recover the energy that was cancelling before they got there. You get back a clean, thin low end instead of a wide, thin one.

3. Crowded low mids

Three or four layers, each of them fine on its own, adding up to a mass of energy between about 200 and 500 Hz. It reads as warmth in your room and as mud everywhere else.

The mastering response is broad EQ, which takes that region out of the whole track, including from the parts where it was doing something useful. The body goes out of the drums along with the mud. You end up with a track that is cleaner and weaker.

What resolves it: deciding which single element owns that range and cutting it out of the others.

4. A mix with nothing left to give

Compression on every channel, then on every bus, then a limiter on the master pushing everything to the ceiling before you exported.

There is no movement left to work with. The track will come back loud and flat and tiring after ninety seconds, and no amount of processing will put the transients back. This is the most permanent item on the list.

It is also the least necessary, because the loudness argument is largely over. Streaming platforms normalise playback level, so a crushed master does not play back louder than a dynamic one, it just plays back with less life. Club playback is a separate conversation, and it is one to have with your engineer rather than to pre-empt with a maximiser.

5. Harshness baked into the sound

Aggressive saturation, cheap distortion, over-driven clipping, exciters. All of these generate high-frequency content that is now inseparable from the content you wanted.

The mastering response is de-essing and top-end correction, which dulls the entire track to get at it. You trade harshness for a lifeless top end.

6. The balance itself

The lead is too quiet. The hats are too loud. The vocal sits behind the pad. Whatever the relationship is between your elements, that is the relationship the record will have.

Broad tonal moves shift the impression slightly. A brightness lift will bring hats forward and a lead forward at the same time, which is not the same as fixing the balance between them.

7. Arrangement

This is the category producers are least willing to hear, and the one that most often decides whether a label replies.

A track that does not develop. A breakdown that arrives before the groove has been established. Sixteen bars too many before the drop. An outro that stops dead, or an intro too short to mix out of. Eight minutes of material that should have been five.

None of this is a processing question. Nobody is going to master your way out of an arrangement problem, and no AI service is going to flag it, because it is not measuring that.

8. Sound selection

If the kick is weak, mastering gives you a louder weak kick. If the lead is thin, you get a brighter thin lead. Processing is an amplifier of decisions, not a substitute for them.

Producers often spend weeks at the mix stage trying to process a sound into being right when the faster path was to replace it in an afternoon.

What mastering does fix

Worth saying, because the list above is long and the point is not that mastering is pointless.

A good master gives you level and translation: your track sitting correctly against everything else in a DJ set or a playlist, tonal balance that holds up across a phone, a car, headphones and a rig, controlled peaks, and coherence across an EP so the tracks sound like they belong together. On a release with multiple tracks, that consistency alone justifies the cost.

It is a real craft. It is just a narrow one, applied at the end, to something that is already working.

The actual problem

Every item above is identifiable by someone with a trained ear in about ninety seconds of listening. That is the frustrating part. These are not subtle or contested problems. They are obvious to anyone who is not you.

You cannot hear them because you have been inside the track for three weeks. Your ears have adapted to the low end. Your sense of the arrangement is built on knowing what happens next, which nobody else has. The specific thing that is wrong has become the thing you expect to hear.

The usual substitutes do not solve it. Friends are kind. Forums produce contradictory advice from people who have not finished anything. Sending to labels and waiting is slow and low-information: silence tells you nothing about what to change, and by the time you conclude something is wrong, six months have gone.

The cheapest useful move, and the one most producers skip, is to get one person who works in your genre and whose ear you trust to listen once and tell you specifically what they hear. Before mastering, while the session is still open and every one of these is still a fixable decision.

After the master comes back, it is not a mix problem any more. It is a next-track problem.

One last thing

Getting signed to a label is one milestone on a longer road. It matters — but it's not the only way forward, and it's not a shortcut to anything. The producers who build lasting careers in underground music do it through consistency: finishing tracks, putting them in front of the right ears, developing their sound over years.

The feedback loop matters as much as the release itself. Every honest critique you receive from someone who knows this music is a data point that makes the next track better than the last.

That's what this is really about.