Mastering for musicians: what it actually does and when to book it

Mastering is the most misunderstood stage in music production. Plenty of self-releasing musicians skip it, confuse it with mixing, or reduce it to « making the track louder ». That’s a shame, because it is also the cheapest of the stages that separate a finished project from a releasable one. Here is what it actually covers, what it changes for your music, and where it belongs in your project.

What exactly is mastering?

Mastering is the last check before your music goes out into the world. Concretely, three things happen at once: listening in a reference environment, a fresh ear that hasn’t spent three weeks on the track, and the technical preparation of the delivery format — today, a file destined for the platforms.

Note the order. What you primarily buy from a mastering engineer is not a chain of compressors: it is a listening position and critical distance. The rest follows from that.

It also explains why mastering is a more technical stage than an artistic one. By the time the file reaches the engineer’s desk, the aesthetic decisions have already been taken: the balance between instruments, the placement of the voice, the timbres, the motifs that make up your sonic signature. Mastering does not rewrite those choices; it validates them, corrects them at the margins, and makes them portable.

Mixing and mastering: the difference in one sentence

Mixing works inside the track, one channel at a time. Mastering works on the track, as a stereo block, and on the coherence of a set of tracks with each other.

Where the trade comes from: from the groove to streaming

Understanding the history of mastering settles just about every misunderstanding around the trade, because its mission has changed three times in a century.

Originally, the mastering engineer is a format technician. Wax cylinder, then disc: somebody has to know how to fit an audio signal into a physical object. On vinyl that constraint is brutal and perfectly measurable — too much bass, badly distributed between the two channels, and the cutting head loses the groove. This is not aesthetics, it is physics. The lab coat in the studio is not a cliché, it is a job description.

The second age arrives with multitrack recording, and above all with compilations and multi-CD changers. The day two artists find themselves back to back in the same playlist, a new question appears: why does my track sound quieter than the one before? The comparison becomes direct, permanent, and the general public spontaneously equates « quieter » with « worse ». The loudness war was born there — not from some engineers’ whim, but from a listening context.

The third age is ours. The physical format has disappeared in favour of a file played by platforms that decide the playback level themselves. It is a complete reversal, and not everyone has absorbed it yet.

Why your mix doesn’t sound the same at different volumes

Here is the point worth understanding before everything else, because it governs how you work at home.

The ear is not a linear microphone. Its sensitivity varies with frequency and with level. That is what the equal-loudness contours describe — historically the Fletcher–Munson curves, today standardised under ISO 226. What they say in practice: at low volume you perceive less bass and less treble; the higher the level, the flatter the curve and the more evenly you hear the spectrum.

This is exactly what the « Loudness » button on old hi-fi amplifiers was for: it compensated for that perceived loss at low volume. And it is also why it was so massively misused — left engaged at high volume, it compensated for nothing and simply caricatured.

Direct consequence for you: two mixing sessions in the same place but at two different volumes will not produce the same decisions. If you mix loud, you will tend to under-dose the bass; if you mix quiet, to overload it.

The end of the loudness war: what normalisation changes for you

Streaming platforms apply loudness normalisation: they measure the perceived level of your track and adjust the playback volume to align it with a common target. The aim is simple: a listener moving from one track to the next shouldn’t have to touch the volume knob.

That table contains one very good piece of news and one trap.

The good news: crushing your master with a limiter no longer buys you anything in perceived volume. You will be brought back to the target anyway, but you will have lost the dynamics on the way. In other words, the argument that justified twenty years of over-compression has disappeared.

The trap: platforms do not all behave the same way with a master that is too quiet. Spotify turns it up; YouTube and Apple’s Sound Check do not. A master at −20 LUFS will therefore play discreetly on some services and normally on others — the inconsistency is on your side, not theirs.

In practice, one master is enough for everyone. Spotify recommends aiming for −14 LUFS integrated and staying below −1 dBTP true peak, with the headroom taken to −2 dBTP if your master exceeds the loudness target, to absorb the overshoots introduced by lossy encoding. Those values are a reasonable starting point even if you are targeting other services.

One nuance, so as not to replace one dogma with another: −14 LUFS is not a grade to achieve. A folk track recorded with three microphones has no reason to reach the same loudness as a club record, and that is entirely as it should be. The target exists to frame your headroom, not to standardise music.

Do you have to pay a mastering engineer?

Let’s put the question differently: what exactly are you buying?

• An outside ear. This is the central point. Music production has become a very solitary activity, and by the third week the relationship you have with your own music is no longer remotely objective.
• Reference monitoring and acoustics. A mastering studio invests there first — monitoring system and acoustic treatment — because it is the one tool that governs all the others.
• End-of-chain experience. In mastering, every move is global. Retouching a snare in a mix is trivial; retouching it on a stereo file also moves the voice and the guitars. That gymnastics has to be learned.

From that list comes the rule worth keeping even if you never pay anyone: never master under the same listening conditions you mixed in. Otherwise the biases your room and your speakers introduced during the mix will never be detected — you will carry them forward, amplified.

Does that mean treating gear as sacred? No. Some well-known mastering engineers work on headphones, including on the move, relying on references they know by heart. Gear is a tool, never an excuse: what counts is the consistency of the setup, knowing its flaws, and the workflow. It is in fact something it shares with the very personal relationship every musician has with their equipment: working habits make the result, not the shopping list.

If you master yourself: the minimum checklist

• Let time pass. Never master straight after mixing. Two or three days of separation is enough to recover a usable ear.
• Change listening context. Another room, another system, headphones you know well. Nothing is worse than doing everything in the same place.
• Work with references. Two or three tracks in the same register, level-matched to yours by perceived loudness — otherwise the comparison is worthless.
• Check on real-world systems. Car stereo, laptop speaker, earbuds. A resonance that goes unnoticed on your monitors shows up in three seconds in a car.
• Measure. A loudness and true-peak meter costs less than one session and saves you from gross errors.
• Doubt. If you are torn between two versions, the difference probably isn’t where you think it is.

How to prepare a mix for mastering

Whether you hand the work to someone else or do it yourself, preparation follows the same principles.

Add one element that gets systematically forgotten: say what you want. A sonic reference, two sentences on the intention of the track and on what bothers you in the mix, are worth more than a ten-line technical brief.

Frequently asked questions

Can mastering rescue a bad mix?

No, and it is better to know that early. Mastering acts on a global signal: it can correct a slight spectral imbalance, tighten a dynamic range, make an album coherent. It cannot move a buried vocal or repair a badly treated snare. If the problem is identifiable in one listen, it belongs in the mix.

How long should I wait between mixing and mastering?

Long enough to lose sight of the track. A few days is usually enough. The goal is not to observe a delay, it is to recover the ability to hear the track the way your listener will: without knowing every detail in advance.

What about automatic AI mastering?

It is a calibration tool, not an interlocutor. It will bring your track into a loudness range and roughly correct a curve, which is already useful for a demo or a working listen. What it will never tell you is that your bass is an artefact of your room, or that one track on the album clashes with the others. And that is precisely what mastering adds.

Do I need a different master per platform?

In the vast majority of cases, no. A coherent master with sufficient peak headroom behaves correctly everywhere. Dedicated files are justified for formats with real constraints — vinyl, a mix for picture, a cinema release — not for moving from one streaming service to another.

What if my budget is zero?

Then apply the free principles: the rest period, the change of listening context, level-matched references, the low-volume test, checking on ordinary systems. They account for most of the benefit. You keep the budget for the day a project genuinely deserves it — the same logic as only going to a studio when you need to.