Mixing vs mastering

At a glance

What is mixing?

Mixing is the stage where all the separate recordings and instruments in a song become one piece of music. Each one sits on its own channel in your DAW, with its own fader, pan control and plug-ins, and you adjust them until every part can be heard and the song feels the way you want it to.

Most of a mix is balance: which part is loudest, where each one sits from left to right and which frequencies each one gets to own. A bass and a kick fighting over 60 Hz is a mixing problem, and so is a vocal that disappears under the guitars. Both come down to one part covering another, which engineers call masking, and you fix it on the channels involved with a fader, an EQ cut or a sidechain compressor.

Mixing also covers the creative decisions about each sound: compression to control a vocal, reverb and delay to give things space, automation to bring a part up in the chorus. The result is a single stereo file, usually called the mix or the bounce. The signal flow lesson shows how the channels and buses inside a DAW lead to the master.

What is mastering?

Mastering is the last stage before release. It starts from the finished mix, a single stereo file, and makes small changes to the whole song at once: a little EQ to adjust the overall tone, sometimes a touch of compression or saturation, and a limiter at the end to set the loudness and keep the peaks under a ceiling.

Because everything goes through every plug-in, mastering moves are small. A 1 dB shelf on a master is a clear change, since it brightens or darkens every instrument at the same time. The mastering chain lesson goes through the plug-ins in order, with settings for each.

The other half of mastering is release work. Every track on an album is brought to a similar loudness and tone so none of them jumps out, the songs are put in order with the gaps and fades decided, and the release files are made: WAV files for a distributor or a DDP image for CD manufacturing, along with codes such as the ISRC that identify each recording. Mastering is also the last careful listen, where clicks and digital overs get caught before anyone else hears them.

Hearing the difference in the lab

The lab runs the same three loops, drums, bass and chords, through both stages. In Mixing, each loop has its own fader and the bass has its own low shelf, so you can change one part without touching the others. In Mastering, the three are added into one bounce, the faders lock, and you get a low shelf and a limiter that work on the bounce as a whole.

The change-per-part screen is the one to watch. It has a column for each loop and shows how far that loop's level has moved since step 1: the teal bar for moves made on its own channel, the pink bar for what the master EQ does to it. The lab works the pink bars out by running the master EQ on each loop alone, which gives the same result as running it on the bounce, because an EQ treats every part of a signal the same way. A mix move lights up one column, and a master move lights up all three.

The loudness screen compares the before version with what you're hearing, in LUFS, with the true peak marked above it. Match loudness is on to start with, so every A/B plays both versions at the same loudness and only the sound changes.

Can mastering fix a bad mix?

Only a little. Say the bass in your mix is 4 dB too loud and it's covering the kick. In the mix, the fix takes a second: pull the bass fader down 4 dB. In the lab, that moves the bass by exactly −4.0 dB and leaves the drums and the chords at 0.0.

Once the mix is bounced, the bass fader is gone. The closest thing a master has is an EQ cut where the bass lives, so step 3 uses a low shelf, 6 dB down at 150 Hz. It takes the bass down 2.2 dB, which helps, but the chords lose 1.3 dB and the drums 1.1 dB, because the shelf can't tell them apart.

Cutting harder makes it worse. A 12 dB shelf still only moves the bass 3.6 dB, short of what the fader did, and the drums lose 5.8 dB below 150 Hz. Most of what the drums have down there is the kick, so the master has thinned out the very part the fix was meant to uncover.

Step 5 moves the same 6 dB shelf at 150 Hz onto the bass channel. The bass drops by the same 2.2 dB as it did on the master, and the drums and chords don't move at all, because nothing but the bass passes through that channel's EQ.

The same goes for every other kind of fix. A de-esser on a master catches the hi-hats along with the vocal's s sounds, and a compressor on a master turns the whole song down every time the kick hits. A master can lean a balanced mix a little brighter or warmer, or calm a mix that's slightly harsh everywhere.

Loudness is the master's job

Turning a song up to release level happens in mastering, with a limiter. The limiter turns the whole mix up into a ceiling the peaks aren't allowed to pass, so the quieter parts come up and the loudest hits stay put. In step 6 the fixed mix goes into the limiter with 8 dB of gain and a −1 dB ceiling, and the bounce goes from −20.2 to −12.7 LUFS.

The master comes out 7.5 dB louder rather than the full 8, because the limiter turns the loudest moments down to keep them under the ceiling. How far it turns them down is the gain reduction, up to 3.5 dB here. The true peak lands at −0.7 dBTP, a little over the ceiling, because the lab's limiter watches the samples and the wave can rise slightly between them. A limiter with true peak detection, which most mastering limiters have, catches that.

The mix itself doesn't need to be loud, and it's easier to judge when it isn't. Streaming services measure the integrated loudness of every track and play it at a set level: about −14 LUFS on Spotify and YouTube, and about −16 on Apple Music. The streaming loudness lesson shows what that does to a loud master.

When you compare a mix with its master, match their loudness first. A louder version sounds fuller and brighter for that reason alone, and the master is always the louder one. Turn Match loudness off in step 7 to hear how much the extra level flatters it.

How loud should a mix be before mastering?

A sensible target is peaks somewhere between −6 and −3 dBFS, which leaves headroom for the mastering EQ and keeps the mix clear of clipping. The mix in the lab peaks at −4.0 dBFS, inside that range.

The peak level isn't worth chasing to the decibel. A 24-bit file has so much range below its peaks that a mix at −6 dBFS loses nothing against one at −3. Just make sure it doesn't clip and that there's no limiter on the mix bus squashing it first. A 32-bit float WAV even keeps a stray peak over 0 dBFS intact, though it's still worth fixing. The gain staging lesson covers levels inside the mix.

Processing on the mix bus: mixing or mastering?

Plug-ins on the mix bus while you're still mixing are part of the mix. You can put a bus compressor on the mix bus early, often a VCA design modelled on the bus compressor in SSL's 4000-series consoles, and mix into it, so every fader move is heard through it. That's a mixing choice, and it stays on when you bounce.

The line is whether the balance can still change. While you're moving faders, anything on the bus is mixing. Once the balance is fixed and you're working on the whole file for loudness and overall tone, you're mastering, even with the same plug-in on the same channel. Many DAWs call that channel the master, which is part of why the two jobs get confused.

Should you master your own music?

Yes, if you're releasing it yourself and the budget isn't there. A simple chain of EQ and a limiter set to a sensible loudness, checked with an A/B at matched loudness, gets a well-balanced mix most of the way. The mastering chain lesson walks through it.

A mastering engineer gives you a second pair of ears in a different room. They haven't heard your song two hundred times and their monitoring is usually more accurate than a home studio's, so they'll often catch a mix problem while you can still fix it. If you master yourself, compare the master with reference tracks at matched loudness, and play it on something other than your monitors, such as headphones or a car stereo.

Automatic mastering, such as iZotope Ozone's Master Assistant or online services like LANDR, analyses the mix and sets a chain for you. It's a reasonable starting point. It can't fix the mix for the same reason a human master can't, so the mix still has to be finished first.

From mix to master, step by step

1. Finish the balanceGet every part where you want it in the mix. If something only sounds right with an EQ move on the master, go back and make that move on the channel it belongs to.
2. Check the mix busTake off any limiter or loudness plug-in on the mix bus, unless the sound depends on it. Keep the bus compressor or tape plug-in you mixed into.
3. Bounce with headroomExport the full song as a 24-bit or 32-bit float WAV at the session's sample rate, with peaks somewhere around −6 to −3 dBFS and no clipping.
4. Listen to the bounce as a filePlay it in a new session or a different player, on more than one system. Note anything that bothers you and decide whether it's a mix fix or a master fix.
5. Master in a separate sessionAdd the mastering chain: EQ, any compression or saturation, and a limiter last with its ceiling at −1 dBTP and true peak detection on.
6. Compare at matched loudnessTurn the master down to the bounce's loudness for every A/B, and compare it with a reference track the same way.
7. Export the release filesBounce the master in the format your distributor asks for, with dither as the last step if you're going down to 16-bit, and check the file on a LUFS meter.

Common mistakes

Fixing mix problems on the master

Every master move lands on every part, as steps 3 and 4 in the lab show. If you need more than about 2 dB of EQ on the master, or a narrow cut to deal with one instrument, the fix belongs in the mix.

Mixing with a limiter on the master channel

A limiter that's always on hides how the mix really behaves. Peaks that would clip get squashed instead, and you end up balancing the song at a level it won't have until it's mastered. Mix without one, or switch it off regularly to check.

Judging the master by how loud it is

The master is louder than the mix, so it wins a plain A/B every time. Compare at matched loudness, and remember that streaming services turn very loud masters down anyway.

Mastering straight after mixing

After hours on a mix, your ears have got used to it. Take a break, and if you can, master the next day, when you'll hear the song more like a listener will.

Training your ears to tell them apart

The quiz under the lab plays the step 1 mix with a low shelf cut hidden somewhere: either on the bass channel or on the master. A/B plays the mix without the cut, and the version with the cut is matched to it in loudness. Listen to the kick. If it keeps its weight while the bass thins out, the cut is on the bass channel; if the kick goes soft too, it's on the master.

Questions people ask

Keep going