Gain staging in plain English
Every sound in a mix passes through a chain of stages before you hear it. A vocal goes through a microphone, a preamp and a converter, then into your DAW, through a few plug-ins on its channel, up its fader, into the master bus and out through your audio interface or into an exported file. Each of those stages has a level going in and a level coming out, and gain staging means keeping that level sensible at every one of them.
The level should never be so high that a stage runs out of room and clips, and it should never be so low that the sound sinks towards the noise of the equipment. In between is a range where each stage works the way its designer intended, and the job is to land in that range everywhere, which you can’t do by setting one fader at the end.
Gain is amplification, measured in decibels. A stage with +6 dB of gain doubles the height of the waveform, and one with −6 dB halves it, so gain staging is really a series of small decisions about how far to turn things up or down and where. If decibels are new to you, the decibels lesson explains the scale and why digital levels are negative numbers.
Reading the lab
The lab is one mixer channel. The loop comes in on the left, goes through a trim control, one plug-in slot and a fader, and ends up in an exported 24-bit file. Like a real DAW it works in 32-bit floating point between those stages, so levels above 0 dBFS are stored without damage until they reach the export.
The left screen has a meter for every stage, in the order the sound travels. The amber block is the average level of the whole loop and the pink line is its highest peak, with the figures for both underneath. While the loop plays, a white bar shows the level right now, and it turns pink above 0 dBFS. Two reference lines run across all five meters: the dashed pink one at 0 dBFS and the dotted teal one at −18 dBFS.
The right screen is the plug-in’s transfer curve, with level in along the bottom and level out up the side. The shaded band shows where your sound lands on that curve after the trim, from its average level up to its peaks. Turn the trim and the band slides along the curve while the curve itself stays put.
Underneath, the waveform shows the loop as it reaches the export, on a scale that runs 6 dB past 0 dBFS. Anything that crosses the dashed lines turns red, because that’s the part the export slices off. The panel next to it describes what your current settings are doing, using numbers measured from the loop, and the A/B button bypasses the plug-in with the loudness matched, so you can hear exactly what it’s adding.
Does gain staging matter in a DAW?
On an analogue mixing desk, gain staging was about survival. Every channel and every piece of outboard gear had a level where it distorted and a level where its hiss took over, and an engineer had to keep every stage between the two by hand.
A DAW works differently. Ableton Live, Logic Pro, Pro Tools, Cubase, FL Studio and Reaper all process audio as floating-point numbers, which can hold levels hundreds of decibels above 0 dBFS. A channel can go over 0 dBFS, the meter can turn red, and as long as the level comes back down before it leaves the computer, nothing has been lost. You can check that in the lab before anything else.
With the trim at +12 dB the loop reaches +11.1 dBFS and the export clips. Now pull the fader down by the same 12 dB. The overs are gone, and the coach compares the exported file with the original loop sample by sample: they match to within −120 dB, so as far as anyone could hear, they’re identical.
Channel overs inside the DAW are mostly harmless, but some places still clip for real. Your audio interface can’t play anything past 0 dBFS, so a hot master crackles through your speakers. A file exported at 16 or 24 bits can’t store anything past 0 dBFS either, so the tops of those waves are cut flat for good. And some plug-ins distort or clip inside when the level going in is too high. That last one is where most gain staging problems in a modern mix come from.
Why plug-ins care about the level going in
A clean digital EQ or a plain volume plug-in does the same thing whatever level you send it. Plenty of plug-ins don’t work like that. Anything with a threshold, a drive control or a model of analogue circuitry behaves differently depending on how loud the input is, and the trim decides that input level.
Saturators and analogue models
The saturator in the lab uses a smooth tanh curve, a common way to model analogue saturation: a straight line for quiet signals, a gentle bend from around −12 dBFS and a hard squash as the level gets near 0 dBFS. Hover over its curve and you can read the numbers. At −18 dBFS it takes off about 0.2 dB; at 0 dBFS it takes off just over 6 dB.
Feed it the loop with +12 dB of trim and it flattens every hit. Turning the fader down afterwards makes the result quieter and leaves the distortion exactly where it was, which you can check with the coach: the distortion figure doesn’t move when you drag the fader, and it changes as soon as you touch the trim.
Bring the trim down to −6 dB and the same plug-in, with the same settings, now rounds off only the loudest hits. Neither result is wrong, because heavy saturation is a sound people choose on purpose. Gain staging makes it your choice, so a sample that arrived at full scale doesn’t make it for you.
Compressors and presets
A compressor turns a sound down when it goes over the threshold, so the level going in decides how much it works. The lab’s compressor has its threshold fixed at −20 dBFS. Send it the loop at full level and the drum hits go up to 19 dB over; send it 12 dB less and they only go 7 dB over, so the gain reduction drops a long way even though you haven’t touched a single compressor control.
This is why a compressor preset called “Vocal, gentle” can squash one vocal flat and do nothing to another. The preset’s threshold assumes a certain input level, and a vocal recorded 10 dB hotter or quieter than that gets a very different result. Set the level going in first and presets start behaving the way their names say. The compression lesson covers the controls themselves.
The −18 dBFS rule
Analogue studio equipment is designed to run at a nominal level that reads 0 on a VU meter, and studio converters are usually lined up so that level lands somewhere between −18 and −20 dBFS. Many plug-ins that model analogue hardware copy that calibration, so an input averaging −18 dBFS makes them respond the way the hardware did at its normal operating level. Above it they start to saturate, which is also what the hardware did.
The −18 refers to the average level, the way an RMS or VU meter reads it, so a drum loop averaging −18 dBFS can have peaks at −6 dBFS and that’s fine. It’s also only a starting point for analogue-style plug-ins; a clean digital EQ doesn’t care, and plenty of plug-ins have an input control or a calibration setting that moves their sweet spot somewhere else.
Trim, clip gain and the fader
In almost every DAW the plug-ins on a channel come before its fader, so the fader can only change the level after them. To change what the plug-ins hear, you need a control that comes earlier.
| Control | Where it sits | Changes what plug-ins hear? | Use it for |
|---|---|---|---|
| Interface input gain | Before the converter | Yes, and it’s recorded | Setting the recording level |
| Clip gain | On the audio clip itself | Yes | Evening out a recording before any plug-in |
| Trim or utility plug-in | First insert slot | Yes | Setting the level into the plug-ins after it |
| Plug-in output control | Inside each plug-in | Only the ones after it | Undoing the level a plug-in added |
| Channel fader | After the inserts | No | The balance of the mix |
| Master fader | After the master bus inserts | No | Overall level into the interface or export |
Pro Tools calls the level control on an audio clip clip gain. Ableton Live puts a Gain slider on every audio clip, and Logic Pro has a Gain setting in the region inspector. For a trim plug-in, Ableton has Utility, Logic has Gain and most other DAWs have an equivalent. A plug-in’s own output knob is worth using too: if a saturator or an EQ boost makes the sound 4 dB louder, turning its output down 4 dB keeps the next plug-in in the chain at the same level and lets you compare it fairly.
Gain staging when you record
The one stage floating point can’t help with is the converter in your audio interface. It turns the voltage from the preamp into fixed-point numbers, and anything past 0 dBFS gets cut off right there, before the DAW sees it. A clipped recording stays clipped, and no fader or plug-in can put the missing tops of the waves back.
At 24 bits there’s no reason to record hot. A decent interface has a noise floor more than 100 dB below 0 dBFS, so setting the input gain for peaks between about −12 and −6 dBFS leaves plenty of room for a singer who gets louder in the chorus while keeping the noise far below anything you’ll hear. Recording at 16 bits leaves less space under the sound, which is one reason almost everyone records at 24. The sample rate and bit depth lesson explains what the bits do.
Turning the input down also protects the stages after it, because a recording that peaks at −8 dBFS arrives at your plug-ins near the level they were designed for.
How to gain stage a mix, step by step
This order works in any DAW and takes a few minutes at the start of a mix.
- Pull every fader to 0 dBStart from unity so the faders aren’t hiding anything. You’ll set the balance again at the end.
- Set each channel’s level before the plug-insUse clip gain or a trim plug-in in the first slot so the meter averages around −18 dBFS, with peaks no higher than about −6. Loops and synths often need 6 to 12 dB of cut; quiet recordings may need a boost.
- Check the level after each plug-inA saturator or a wide EQ boost can add several decibels. Match its output to its input by ear and with the meters, so the next plug-in still gets the level you set.
- Watch the busesTen channels at −18 dBFS add up to much more than −18 on the drum bus or the master. If a bus has its own compressor or saturator, trim the bus input too.
- Balance with the fadersNow that the plug-ins are getting the right levels, use the faders to set how loud each part is in the mix.
- Leave headroom on the masterBefore any limiter, aim for master peaks around −6 to −3 dBFS. If it’s hotter, pull all the channel faders down together; the next section explains why the master fader is the wrong tool for this.
Headroom on the master
Headroom is the gap between your loudest peak and 0 dBFS. A mix with no headroom has nowhere to go: one louder snare hit clips the output, and whoever masters it has to turn it down before they can start. Leaving 3 to 6 dB costs nothing, because the final loudness is set at the end by a limiter, and streaming services turn loud tracks down to their own level anyway, which the LUFS lesson lets you hear.
The master fader is a poor way to make that space. It comes after any plug-ins on the master bus, so pulling it down leaves a bus compressor or a limiter on the master hitting exactly as hard as before. If the master is too hot, select all the channel faders and lower them together, or trim the master bus at its input. The headroom lesson goes further into how much to leave and why.
Common gain staging mistakes
Fixing a hot channel with its fader
The meter goes red, you pull the fader down, the meter goes back to green. The plug-ins on that channel are still being hit just as hard as before, because the fader comes after them. Turn down the clip gain or add a trim at the top of the chain instead.
Recording too hot
Floating point protects everything inside the DAW and nothing before it. If the interface’s clip light flashes while you record, the damage is in the file. Turn the input gain down and record the take again.
Treating −18 dBFS as a law
It’s a useful target for analogue-style plug-ins and a sensible habit everywhere else, but no plug-in breaks at −17. If a saturator sounds better driven a few decibels harder, drive it harder and match its output so the next stage still sees a sensible level.
Comparing at different volumes
Turn a plug-in on and the channel gets 3 dB louder, so it sounds better, and you keep it. Most of that improvement is the volume. Match the plug-in’s output to its input before you decide; the A/B button in the lab does that for you.
Forgetting that channels add up
Every channel can sit at a sensible level while the master still clips, because ten parts playing at once are louder than any one of them. Check the buses and the master each time you add a part.
Training your ears to hear gain staging
The quiz under the lab plays the loop through the same plug-in twice. One version has more trim going in and less fader coming out, the two are matched in loudness, and the screens are covered, so the only clue is how hard the plug-in is working. Saturators are the easier rounds, because the harder one sounds crunchier on the loud hits; compressors are harder, and you’re listening for the hits sounding flatter and the space between them coming up. Your best score is saved in your browser.
Questions people ask
What level should I gain stage to?
Aim for an average level around −18 dBFS on each channel going into your plug-ins, with peaks somewhere between −12 and −6 dBFS. On the master bus, keep peaks below about −6 to −3 dBFS before any limiting. These are starting points for analogue-style plug-ins; clean digital ones such as a plain EQ work the same at any level.
Does gain staging matter in a 32-bit float DAW?
Yes, but for different reasons than on a mixing desk. A channel going over 0 dBFS inside the DAW does no harm, because floating point can hold levels far above it. What still counts is the level going into plug-ins that react to it, the level reaching your audio interface, and the level of any file you export at 16 or 24 bits.
Should I use the fader or clip gain to gain stage?
Clip gain, or a trim or utility plug-in in the first insert slot. In almost every DAW the fader comes after the inserts, so turning it down changes how loud the channel is and does nothing to the level your plug-ins receive. Use the fader for the balance of the mix once the levels going in are right.
Why −18 dBFS?
Analogue studio gear is designed to run at a nominal level that reads 0 on a VU meter, and studio converters are usually lined up so that this level lands somewhere between −18 and −20 dBFS. Plug-ins that model analogue gear often copy that calibration, so −18 dBFS going in makes them behave the way the hardware did at its normal operating level.
Is normalising the same as gain staging?
No. Normalising turns a whole clip up or down so its loudest peak hits a level you pick, which is often 0 dBFS. That is about as hot as a clip can be, so normalising to 0 dBFS and then feeding the clip into analogue-style plug-ins is usually the opposite of good gain staging. If you normalise, pick a lower target such as −6 dBFS.
How do I gain stage synths and samples?
Many synth presets and samples come out close to 0 dBFS. Turn them down at the source: the synth’s own output or volume control, clip gain on the sample, or a utility plug-in before any effects. Then check the meter again after each plug-in that adds level, such as a saturator or an EQ boost.