Gain staging: setting healthy levels from recording to mix

Gain staging means managing signal level at every stage of the chain so that each device or plugin works in its sweet spot: far enough above the noise floor and safely below clipping. In practice, you record with peaks around -12 to -6 dBFS, bring each track to an average level near -18 dBFS before the plugins, compare every process at matched loudness, and leave a few decibels of headroom on the master bus.

Why does gain staging still matter in digital audio?

Most DAWs now mix in 32-bit or 64-bit floating point. In that format, a signal that goes over 0 dBFS inside the mixer is not clipped: pulling down the master fader gives you a clean signal back. It is tempting to conclude that gain staging is obsolete. That would be a mistake, for four reasons:

• Converters clip. At the input and output of your audio interface, 0 dBFS is a hard ceiling. A clipped recording stays clipped.
• Many plugins are level-dependent. Compressors, saturators and especially analog emulations behave differently depending on what they receive: a threshold, a colour or a distortion changes with input level.
• Some processing runs in fixed point. Certain plugins, outboard gear and 16- or 24-bit export formats clip above 0 dBFS.
• Consistent levels make a mix readable. Faders near their nominal position, meters that mean something, honest comparisons: you make better decisions.

dBFS, dBu, VU: which units do you need to know?

Analog and digital are linked by an alignment level. The EBU R68 recommendation maps 0 dBu to -18 dBFS; the SMPTE RP 155 standard aligns +4 dBu with -20 dBFS. In the plugin world, a widespread convention treats 0 VU as roughly -18 dBFS, but every developer sets its own reference, so check the manual of your emulations.

Above all, remember the difference between peak and average (RMS or VU). A peak meter shows the risk of clipping; an average meter shows what the ear perceives as loudness.

What level should you aim for when recording?

At 24 bits, the theoretical dynamic range is about 144 dB (6 dB per bit). Real converters deliver less, but still plenty to record with headroom and no worries about quantisation noise. There is no reason to push into the red.

1. Set the preamp gain while the performer plays the loudest part of the song, not during the warm-up.
2. Aim for peaks between roughly -12 and -6 dBFS. For a very dynamic vocal or percussion, leave more room.
3. Check the average: a vocal or guitar take often sits around -20 to -16 dBFS RMS, which puts it straight into the plugins' working range.
4. Let the headroom do its job: a slightly quiet take beats a clipped peak that cannot be repaired cleanly.

A well-chosen level at the source makes everything downstream easier. It is also the first thing to check with a noisy recording; our guide to cleaning up a recording covers the other causes of noise.

How should you set up the plugin chain when mixing?

1. Start with clip gain or a gain plugin at the top of the track, bringing each track to an average close to -18 dBFS. Faders then stay near 0 dB, where their travel is most precise.
2. Set each plugin in chain order. After an EQ or a compressor, adjust the output gain so the level leaves as it came in.
3. Compare against bypass at matched loudness. If a process only sounds "better" because it is louder, it is not helping.
4. Keep an eye on virtual instruments. Many synths and samplers output very hot signals, sometimes close to 0 dBFS: turn down their output rather than the fader.
5. Manage the buses. Summing several tracks raises the level: treat a drum or backing-vocal bus like a track, with its own headroom.
6. Let the master breathe. No limiter while mixing; peaks around -6 to -3 dBFS on the master leave comfortable room for mastering.

Why compare at matched loudness?

Our ears do not hear every frequency the same way at every level, as the equal-loudness contours standardised in ISO 226 describe. The louder a sound, the more present its lows and highs seem. As a result, between two versions, the louder one almost always seems better.

That is the main trap in mixing. An EQ adding 2 dB of top end, a compressor lifting the track, a saturator thickening the signal: each seems to improve the sound when it mostly makes it louder. Compensating the output gain before comparing is the only way to judge a process.

The same reflex applies in mastering, where the temptation to push the limiter is strong; our guide to the mastering steps explains how to compare a reference and your master at equal loudness.

Which starting levels should you remember?

These figures are working references, not rules: what matters is that every stage receives a suitable level and none clips by accident.

Which mistakes should you avoid?

• Recording too hot "to get level": all you gain is the risk of clipping.
• Mixing into a limiter on the master: it hides level problems and skews every decision.
• Pushing every fader up to bring one track forward: turn the others down instead.
• Forgetting plugin outputs: a level that creeps up from plugin to plugin ends up overdriving an emulation or a bus.
• Judging a process without matching loudness: you pick the louder one, not the better one.
• Ignoring bus peak meters: a sum of sensible tracks can still clip a fixed-point bus or an output.

The same principles apply before a reverb or delay on an aux send: a return that is too hot tempts you to turn the effect down instead of setting the send properly (see our article on reverb types). And if you are considering a hardware desk, our article on physical vs virtual mixing consoles shows where input level shapes the colour.

Key takeaways

1. Record with peaks around -12 to -6 dBFS: headroom costs nothing at 24 bits.
2. Bring each track to an average near -18 dBFS before the plugins.
3. Match each plugin's output to its input and compare against bypass at equal loudness.
4. Turn down hot sources (virtual instruments, buses) rather than pushing everything else.
5. Keep a few decibels of headroom on the master, with no limiter while mixing.

Frequently asked questions

Can a track clip in a 32-bit float DAW?

Inside the mixer, going over 0 dBFS does not clip as long as the signal stays in floating point. It will clip at the interface output, inside a plugin that works in fixed point, or when exporting to 16 or 24 bits. Healthy levels are still the safer choice.

Why is -18 dBFS mentioned so often?

Because many emulations of analog hardware are calibrated so that around -18 dBFS matches their nominal level (0 VU). Their response, including their saturation, then resembles the original unit's. The exact value varies between developers.

Does gain staging change the sound?

With perfectly linear plugins, no: only the level changes. With compressors, saturators and emulations, yes, because their behaviour depends on the level they receive. Controlling those changes is precisely why levels deserve care.

Should you aim for the same levels when recording at 16 bits?

At 16 bits, the theoretical dynamic range is about 96 dB: you can afford slightly less headroom, but the risk of clipping is unchanged. For recording, 24 bits is now the norm and gives you more freedom.

To go further, explore our mixing tools on the Mixing and mastering page.