What mastering is, and what it is not
Mastering is the last step before a track is released. It takes a finished stereo mix and prepares it for the places people will hear it: streaming platforms, a club system, a car, a phone speaker.
It is a small set of decisions applied to the whole mix at once — broad tonal shaping, dynamic control, and setting the final level so the track sits correctly next to commercial releases.
What mastering cannot do is fix a mix. If the vocal is buried, the kick and bass are fighting, or the snare is too quiet, those are mix problems. A mastering chain works on the sum of everything; it cannot reach inside and move one element. A muddy mix mastered louder is a louder muddy mix.
The honest test: if you can name a single instrument that is wrong, go back to the mix. If the whole thing sounds right but flat, quiet, or dull compared to references, mastering is the right tool.
Before you start: prepare the mix
Leave headroom. Bounce with peaks around −6 to −3 dBFS. This is not because processing needs the room mathematically — it is because a mix that already clips gives the mastering chain damaged material to work with.
Bounce at the project sample rate, 24-bit or 32-bit float. Do not bounce to MP3 and master that. Lossy encoding discards information that the chain will then amplify.
Turn off any limiter on the master bus. A mix bus limiter left running is mastering applied twice, and the second pass has less to work with.
Listen on more than one system. If a mix only sounds right on your monitors, mastering will not make it translate. Check on headphones and a phone speaker before you go further.
Step 1 — Listen against a reference
Pick one or two commercial tracks in the same genre that sound the way you want yours to sound. Load them next to your mix.
Match levels before comparing. This matters more than anything else in this guide. A track that is 3 dB louder sounds better to almost everyone, regardless of quality. If you compare without matching, you will chase loudness and end up crushing your master.
Listen for three things: where the low end sits, how forward the vocal is, and how bright the top is. Those three account for most of the difference between an amateur and a commercial master.
Read more: using reference tracks properly.
Step 2 — Broad tonal shaping
This is EQ, but not the kind used in mixing. In mastering the moves are wide and small: a shelf, a broad bell, rarely more than 2 or 3 dB.
If a specific frequency is a problem, that is a mix issue. A narrow cut at 340 Hz on the whole mix removes 340 Hz from everything — the vocal, the guitar, the room. It fixes one instrument and dulls the rest.
Typical moves that do work: a gentle low shelf if the bottom is thin, a wide cut around 200-400 Hz if the mix is congested, a high shelf if the top is closed. Make one move, listen, then decide whether it needs another.
If your mix sounds thick and unclear rather than wrong in one place, the guide on why mixes sound muddy covers what actually causes it.
Step 3 — Dynamic control
Compression on a master serves a different purpose than in mixing. It is not there to control a loud snare — it is there to make the whole track feel like one piece rather than several elements playing together.
Ratios stay low. Between 1.5:1 and 2.5:1. Gain reduction stays modest: 1 to 3 dB on the loudest sections. If the meter is showing 6 dB of reduction, the compressor is doing mixing work it should not be doing.
Attack slow enough to let transients through — usually 10 to 30 ms. A fast attack on a master flattens drum hits and the track loses its punch, which is the opposite of what most people are trying to achieve.
Many masters need no compression at all. A well-mixed track that already has consistent level is finished after tonal shaping.
Step 4 — Set the final level
This is where most self-taught masters go wrong, because the instinct is to make it as loud as possible.
Streaming platforms normalise playback. Spotify, Apple Music and YouTube all turn loud tracks down to a reference level. A master pushed to −6 LUFS is played back at the same volume as one at −14 — but the loud one arrives with its transients already crushed. You did the damage and got nothing for it.
Sensible targets: around −14 LUFS integrated for streaming, −16 for Apple Music, −8 to −6 for club playback where nothing normalises. See how loud a master should be for the full picture.
Watch true peak, not sample peak. A file that reads −0.1 dBFS on a normal meter can exceed 0 dB once it is converted to MP3 or AAC, because the reconstructed waveform passes between the samples. Leave the ceiling at −1 dBTP and the encoder has room. This is explained in what true peak means.
Step 5 — Check before you release
Listen to the whole track, start to finish, at a normal volume. Not sections. Not loud. The problems that survive to this stage are usually structural — a chorus that got quieter than the verse, a fade that cuts early.
Check in mono. Phone speakers, most club systems and many venues sum to mono. If the vocal disappears or the low end thins out, something in the mix is out of phase and mastering will not fix it.
Compare to your reference again, level-matched. If yours now sounds close in tone and weight, you are finished. If it sounds worse than when you started, undo everything and do less.
Sleep on it. Ears adapt within minutes. A master that sounded perfect at 2am frequently sounds harsh the next morning, and that is the version listeners will hear.
The chain, in order
Mastering is not a fixed recipe, but the order matters because each stage changes what the next one sees:
1. Corrective EQ — remove what should not be there before adding anything.
2. Dynamic control — glue, applied to a mix that is already tonally correct.
3. Tonal EQ — the character moves, made after compression has settled the level.
4. Saturation — optional, adds density and perceived loudness without gain.
5. Stereo adjustment — width, and low-end mono where needed.
6. Limiting — the final ceiling, and nothing after it.
Reversing any two of these changes the result. Compression before corrective EQ, for example, means the compressor reacts to a resonance you were about to remove.
Doing it in Quantara
The studio runs this chain in your browser, with the numbers visible at every stage: integrated and short-term LUFS, true peak, and a spectral view of where the energy sits.
Genre presets give a starting point calibrated from measured tracks rather than guessed by ear — but every control stays adjustable, because a preset is a starting point and not an answer.
The free plan includes three masters a month, with no card required. Open the studio and work through the steps above on your own track.
Frequently asked questions
Yes. The technical part is learnable and the tools are available. The harder part is judgement — knowing when to stop, and hearing your own track objectively after weeks of working on it. Level-matched reference comparison solves most of that, and taking a day away from the track solves the rest.
For a mix that is already in good shape, 20 to 40 minutes of focused work. If it is taking hours, the problem is usually in the mix rather than the master, and going back to fix it there will be faster than continuing.
Mixing balances the individual elements of a song — the relative level of the vocal, drums, bass and everything else. Mastering works on the finished stereo file as one thing, preparing it for release. Mixing decides what the track is; mastering decides how it arrives.
No. Accurate metering matters far more than expensive processing. A basic EQ, a transparent compressor and a limiter, used with a level-matched reference and a true-peak meter, produce better results than a chain of expensive plugins used without measurement.
No. Streaming platforms normalise playback, so a master pushed to -6 LUFS plays at the same volume as one at -14 but arrives with its transients crushed. Master to the target for where the track will be heard, and leave the ceiling at -1 dBTP.
No, and attempting it usually makes things worse. Mastering applies broad moves to everything at once; it cannot reach a single instrument. If you can name what is wrong, fix it in the mix.