Gain Staging Explained: How to Set Levels From Mic to Master

Gain staging is the least glamorous skill in music production and the one that fixes the most problems. If your mixes sound harsh when they get loud, if your plugins never quite behave the way the YouTube videos promise, if the master keeps clipping no matter how much you pull the fader down, the cause is almost never the plugins. It's the levels feeding them.

This guide walks the signal from the microphone to the finished master and tells you what number to aim for at every point, why that number exists, and what goes wrong when you miss it. No mysticism, no "just use your ears" hand-waving. Real targets you can meter, and a repeatable process you can run on every session.

What Gain Staging Actually Is

Gain staging is the practice of setting the level at every point where your signal changes hands, so that each stage receives a signal it can handle comfortably. Your audio passes through a lot of hands: a microphone, a preamp, a converter, a DAW channel, a stack of plugins, a group bus, the master bus, and finally a limiter. Each of those is a stage, and each one has a level it likes best.

The word "gain" is doing specific work here. Gain is level going into a process; volume is level coming out. When you turn a fader down you change how loud something sounds. When you change gain you change how hard the next device gets hit, which changes how it behaves. That distinction is the entire point. A compressor with a threshold at -20 dBFS does something completely different to a signal peaking at -3 than to the same performance peaking at -18, and no amount of fader riding afterward will undo it.

THE CHAIN, STAGE BY STAGE

1. Mic or instrument Perform at full volume A quiet take cannot be fixed later.
2. Preamp Peaks around -12 dBFS The one knob that sets everything after it.
3. A/D converter 24-bit, never clipped This is the only stage you cannot undo.
4. DAW track Average around -18 dBFS Use the trim, not the fader.
5. Group bus Peaks around -10 dBFS Summed tracks add up fast.
6. Master bus Peaks around -6 dBFS Leave the mastering engineer some room.

TitanAudio Share titanaudio.io

Notice that the targets get quieter as you move down the chain, not louder. That feels backwards to most beginners, who assume the goal is to get everything as loud as possible as early as possible. It's the opposite. Loudness is applied last, deliberately, by a limiter. Everything before that exists to hand a clean, well-proportioned signal to the next stage.

Why Digital Audio Still Needs Headroom

Here's the objection everyone raises: modern DAWs mix in 32-bit or 64-bit floating point, which has such an enormous internal range that you effectively cannot clip the mix bus by summing tracks. That's true. So why bother?

Three reasons, and all three bite hard.

Your converters are not floating point. The analog-to-digital converter in your interface is fixed point, usually 24-bit. When you record too hot, the waveform hits the ceiling and the tops of the peaks are simply gone. There is no plugin, no AI restoration, no "clip fix" that puts them back. It's the one truly irreversible mistake in the whole chain. The same applies on the way out: your final bounce is a fixed-point file, and anything above 0 dBFS in it is distortion.

Analog-modeled plugins expect analog levels. This is the big one. Every plugin that emulates a console, a tape machine, a tube compressor, or a classic EQ was modeled at a specific reference level. In the analog world, 0 VU on the meter (the point where a device is working nicely and just beginning to add its character) usually corresponds to somewhere around -18 dBFS in the digital world. Feed one of those plugins a signal averaging -3 dBFS and you're driving the modeled circuit far past where anyone drove the real thing. It will not sound like the hardware. It will sound harsh, thin, and strangely lifeless, and you will blame the plugin.

Plugin controls stop making sense. Compressor thresholds, gate thresholds, saturation drive, transient shaper sensitivity: all of these are calibrated against an assumed input level. When your tracks arrive at wildly different levels, the same setting does something different on every channel, and you spend the session chasing your own tail instead of mixing. Consistent gain staging means a preset that works on one vocal is at least in the right neighborhood on the next one.

The Target Numbers: Peak, Average, and the -18 dBFS Rule

Two numbers describe any signal, and confusing them is the single most common source of gain staging errors.

Peak is the highest instantaneous sample value. It's what a normal DAW meter shows you, and it's what determines whether you clip. Average (measured as RMS, or as LUFS when it's loudness-weighted) is how loud the signal actually feels over time. The gap between them is called crest factor, and it varies enormously by source. A close-miked snare might peak 20 dB above its average. A heavily compressed bass might peak only 6 dB above.

This is why "record to -18" is ambiguous advice and why so many people get it wrong. The classic -18 dBFS reference is an average target, not a peak target. If you set your snare's peaks to -18 you have recorded a very quiet snare. Aim for an average around -18 dBFS and let the peaks land wherever the source's natural crest factor puts them, typically somewhere between -12 and -6.

THE SAME TAKE, RECORDED THREE WAYS

• Too hot peaks at -1 dBFS
• Target peaks at -12 dBFS
• Too quiet peaks at -40 dBFS

TitanAudio Share titanaudio.io

Here are the working targets for each stage of a typical session. Treat them as a starting grid, not a set of laws.

Stage	Aim For	Why
Recording (preamp)	Peaks -12 to -6 dBFS	Safe from clipping on a louder-than-expected take
Individual track	Average around -18 dBFS	The level analog-modeled plugins were built around
Instrument bus	Peaks -12 to -8 dBFS	Summed tracks add level; bus processors need room
Mix bus, before limiting	Peaks -6 to -3 dBFS	Headroom for the mastering chain
Final master	True peak -1 dBTP	Survives lossy encoding without inter-sample clipping

That last row deserves a note. When a streaming service encodes your track to AAC or Ogg, the decoded waveform can overshoot the sample values in your file. Leaving 1 dB of true-peak headroom is what stops that overshoot from turning into audible distortion on someone's phone. Our guide to how loud your master should be covers the loudness side of the same question, and understanding limiters explains the tool that enforces the ceiling.

Stage One: Setting Levels While You Record

This is the stage that matters most, because it's the only one you genuinely cannot redo without recalling the singer.

The process is short. Have the performer run the loudest part of the song, not a polite test phrase. Watch your peak meter. Turn the preamp gain up until the loudest moments are landing around -12 dBFS, then stop. Record a take, look at the waveform, and if the peaks are consistently under -18 or consistently over -6, adjust and go again.

The reason to target -12 rather than -6 is simple insurance. Singers get louder when the red light comes on. Guitarists dig in on the take that counts. That extra 6 dB of margin is what stands between you and a clipped chorus, and at 24-bit resolution it costs you nothing audible. A signal peaking at -12 dBFS in a 24-bit file still has roughly 132 dB of range below it, which is far more than the noise floor of any room you're likely to record in. The old advice to "record as hot as possible" is a leftover from 16-bit digital and from analog tape, where the noise floor was a genuine problem. It is not your problem.

Watch for the gain that isn't on the preamp. Many interfaces have a pad switch, an instrument/line switch, and a separate software gain in the control panel. Some microphones have their own pad. Condenser mics are far more sensitive than dynamics, so a gain setting that works for an SM58 will slam a large-diaphragm condenser. Check the whole path before you decide the interface is broken. If you're still choosing hardware, our roundups of the best audio interfaces and microphone preamps for home studios cover which units give you usable clean gain.

Stage Two: Gain Staging Inside Your DAW

You've got clean recordings. Now the goal is to get every track sitting at a sensible, comparable level before you open a single plugin.

Do this with a trim or gain control, not with the channel fader. The distinction matters because of where each one sits in the signal path. A trim (a gain plugin at the very top of the insert chain, or your DAW's built-in input gain) changes the level going into your plugins. The fader sits after them. If you use the fader to tame a hot track, the plugins are still being hammered; you've only turned down the result.

The practical routine looks like this. Solo a track, play the busiest section, and watch a meter that shows average level rather than just peaks. Most DAWs have one, and free loudness meters like Youlean Loudness Meter or MeldaProduction's MLoudnessAnalyzer will do it if yours doesn't. Adjust the trim until the average sits around -18 dBFS. Move to the next track. When you've been through all of them, your faders should all still be at unity, and the rough balance should already be surprisingly close.

That last point is the payoff people don't expect. When every track arrives at a consistent level, your faders start at zero and the mix is roughly in balance before you've done anything. You then move faders to make musical decisions rather than to correct for the fact that the bass was recorded 15 dB hotter than the guitars.

One caveat: -18 dBFS average is a reference point, not a requirement. A pad that sits quietly under everything does not need to be normalized up to -18. What you're really after is that nothing arrives absurdly hot and nothing arrives so quiet that you have to add 20 dB of plugin gain to hear it.

Stage Three: Stopping Gain Creep in a Plugin Chain

This is where most in-the-box mixes quietly fall apart. Every processor you add tends to make things louder. EQ boosts add gain. Compressor makeup gain adds gain. Saturation adds gain. Each one sounds like an improvement in isolation, partly because louder always sounds better for the first few seconds. Stack four of them and the channel is 15 dB hotter than it was, every plugin after the first is being fed a level it wasn't designed for, and the mix bus is drowning.

GAIN CREEP THROUGH A PLUGIN CHAIN

Without trim With trim after each plugin

• Input -18 dBFS -18 dBFS
• EQ boost -13 dBFS -18 dBFS, trim -5
• Compressor -7 dBFS -18 dBFS, trim -6
• Saturator -2 dBFS -18 dBFS, trim -5
• Mix bus clipping -18 dBFS

TitanAudio Share titanaudio.io

The fix is a habit, and it takes about four seconds per plugin. After you set any processor, use its output or makeup control to bring the level back to roughly where it was at the input. Almost every plugin has one. If a plugin genuinely doesn't, drop a free gain utility after it.

The test for whether you've done it right is the bypass button. Toggle the plugin in and out. If the bypassed version sounds noticeably quieter, you are not comparing the processing, you are comparing loudness, and loudness always wins that comparison. Match the levels and toggle again. Sometimes the plugin you were sure was helping turns out to be doing nothing but making things louder, and you get to delete it.

This is doubly true for anything that adds harmonics. Saturation, tape emulation, tube preamp plugins and console emulations all get more aggressive the harder you hit them, and their output climbs as a side effect. Drive them deliberately by turning up their input, then pull their output back down. Never let the drive be an accident of what happened to be feeding them.

Stage Four: Buses, the Master, and Room for Mastering

Buses are where levels sneak up on you, because summing adds level. Two identical signals summed together produce a result 6 dB louder. Eight drum tracks routed to a drum bus can arrive well above what any of them looked like individually, and if you've put a compressor on that bus, it's now working far harder than the settings suggest.

So gain stage your buses the same way you gain staged your channels. Put a trim at the top of the bus chain, get the incoming level into a sensible range (peaks somewhere around -10 dBFS is a good working figure), then process. Bus compressors in particular are threshold-dependent, and a bus that arrives 8 dB hotter than you planned will get squashed in a way you did not intend. Our guide to how to use compression goes deeper on why threshold and input level are really the same conversation.

The master bus is the last checkpoint before your mix leaves the building. Aim for peaks around -6 dBFS with nothing on the master but a meter. That sounds quiet compared to a commercial reference, and it should. You are not competing on loudness yet. The mastering stage is where loudness happens, and it needs something to work with.

If you are mastering your own material, this still applies: finish the mix at -6 dBFS peaks, bounce it, and open the bounce in a fresh session to master. That separation stops you from mixing into a limiter and mistaking "louder" for "better" for three hours. Our walkthrough on how to master a track at home picks up exactly where this guide leaves off.

One more thing about the master: take the limiter off while you mix. A limiter on the master during mixing hides every gain staging problem you're creating, right up until you remove it and discover the mix only worked because the limiter was doing 8 dB of gain reduction. If you want a rough loudness reference while you work, use a loudness meter to look at the number instead of a limiter to squash the signal.

Six Gain Staging Mistakes That Wreck Mixes

Recording as hot as possible. The advice made sense in 1996. At 24-bit it buys you nothing and costs you the take when the singer over-delivers. Peaks around -12 dBFS, and stop worrying.

Using the fader as a trim. The fader is after the inserts. Pulling it down does not protect your plugins from a hot signal; it only turns down whatever they did to it. Use a gain plugin at the top of the chain instead.

Normalizing everything to 0 dBFS. Normalize is not gain staging. Peak-normalizing a file to 0 dBFS gives you the least headroom possible and makes every track's average level unpredictable, because peak-to-average ratios differ wildly by source. If your DAW offers loudness normalization, that is far more useful.

Comparing at unmatched levels. Every A/B test you run at mismatched loudness will pick the louder option. This is not a matter of discipline or trained ears; it is how hearing works. Match first, then decide.

Ignoring the low end. Sub frequencies carry enormous energy and barely register on small speakers, so a rumbling kick or an untamed 40 Hz build-up can eat 3 or 4 dB of headroom without you hearing it. A high-pass filter on everything that does not need bottom end is the cheapest headroom you will ever find. Our Frequency Reference Chart shows where each instrument's low end actually starts, so you know how high you can safely filter.

Fixing it at the end. Gain staging is not a mastering step. By the time the mix is finished, a badly staged chain has already baked in the wrong compressor behavior, the wrong saturation character, and the wrong EQ interactions. Turning the master down at the end changes none of that. Set the levels first and the rest of the mix gets easier.

A Complete Gain Staging Walkthrough

Here's the whole process as a checklist you can run on any session.

1. Set the session to 24-bit before you record anything.
2. Set preamp gain on the loudest part of the performance, targeting peaks around -12 dBFS. Leave the margin.
3. Record, then check the take. Look at the waveform, not just the meter. Consistently under -18 or over -6 means adjust and redo.
4. Trim every track to an average near -18 dBFS using a gain plugin at the top of the insert chain. Leave the faders at unity.
5. Build a rough balance with faders only, no plugins yet. If the staging is right, this takes minutes.
6. Add processing one plugin at a time, compensating output gain after each so the channel level stays roughly constant.
7. Bypass-test every plugin at matched level. Delete the ones that were only making things louder.
8. Trim your buses before their processing, aiming for peaks around -10 dBFS.
9. Leave the master clean at peaks around -6 dBFS. No limiter while mixing.
10. Bounce, then master in a fresh session, finishing at -1 dBTP true peak.

None of this is creative work, and that's the point. Gain staging is the plumbing that lets the creative decisions actually land. Get it right once, make it a habit, and you stop fighting your own signal chain.

When you're ready to put the numbers to work, the dB Calculator handles the decibel math for gain compensation, and the BPM Calculator covers the timing side of the same toolkit. Both live in the Titan Audio production tools collection.

Frequently Asked Questions

What is gain staging?

Gain staging is setting the level at every stage of a signal path so each processor receives a signal in the range it was designed for. It matters because a compressor with a threshold at -20 dBFS treats a track peaking at -3 completely differently from the same performance peaking at -18.

What level should I record and mix at?

The -18 dBFS rule is the standard reference: aim for an average around -18 dBFS with peaks well below the ceiling. dBFS is measured relative to full scale, where 0 dBFS is the loudest a digital file can go, so -18 means 18 dB of headroom left.

Does digital audio still need headroom?

Digital audio still needs headroom, because plugins and summing add gain even though the file itself has no noise floor to worry about. Running hot into a chain leaves nothing for a compressor's make-up gain or a saturator to work with, and fader riding afterwards cannot undo it.

Can I fix bad gain staging later with faders?

Fader moves cannot undo bad gain staging, because every processor in the chain has already reacted to the wrong level. Levels have to be right going in; the fader only changes what comes out after those decisions have been made.

Related Tools

Do the level math instead of guessing

📊 dB CalculatorConvert decibels to gain ratios for exact level matching 📈 Frequency ChartSee where the low-end energy eating your headroom lives