Gain Staging

There’s an annoying side-effect from converting analog sound ( = physical audio in the real world) to digital sound ( = on your computer). In the real world? Sound can get as loud as it wants! There’s no limit to how many decibels something can produce!

But digital sound obviously needs a limit. There are only so many bytes of data it’s allowed to store on your computer. It can’t be infinitely loud, as that would crash the system.

What does this mean?

• If you record something too softly, you might miss some detail. Because it only writes a few bytes of data and the rest is silence.
• If you record something too loudly, it might clip. It wants to store more data than it can. So it throws away some data and becomes distorted.

Both situations are undesirable. The first is recoverable, but will just sound dull or noisy. The second is unrecoverable—unless you want that distorted grunge sound. (In which case there are better ways to get it.)

How do you prevent this? How do you record at a volume that is “loud enough”? One that will never clip, but also contain enough detail?

Through the process of gain staging. Which simply means “setting everything in your audio chain to just the right volume”.

Head room

In the old days, it was standard to record everything hot: as loud as you can without clipping.

Nowadays, that’s not needed anymore. We have enough definition—enough bytes—to record everything in high detail and volume.

Instead, it’s recommended to leave enough head room. Enough “air” that you could raise the volume of your recording later on without clipping. Enough “safe zone” in case you suddenly go much louder with a recording. (Or forget to gain stage again after moving the microphone closer :p)

What is that safe zone? What are we aiming it?

Some people simply look at their standard level meters. They check if their sound is generally around -12 dB, and peaks absolutely no higher than -6 dB. This is the “eyeball method”. It’s not recommended if you want to consistently get (professional) work done.

Most software comes with a VU Meter, either built-in or as a plugin. They calculate loudness in a better and more general way. (By looking at intensity and power over longer periods of time.) If you can’t find it, it might be under the “mastering” area, rather than “mixing” or “recording”.

With this, you want your average level around -18 dBFS, with peaks absolutely no higher than -10 dBFS.

How to find my peaks?

Just play the thing you want to record and look at the meters. The peaks will be the hardest hits, the starts of a note or chord—the loudest “transients”. Those are the biggest issue. They will cause clipping, not your average sustained note.

This is a fine test. But most people go louder during an actual performance than while gain staging. Or they move much closer to the microphone.

If you know you do this—or your performer does this—take it into account. Go lower with your volume, ensure the peaks have more than enough space to grow.

Keep gain staging

Obviously, these volumes change as the setup changes. Different instrument, different performer, different microphone placement.

Yes, gain stage once at the start of a session. But keep checking the chain of volumes any time you make any change.

This doesn’t even need to be a big one. Guitar chords are much louder (especially when low) than subtly plucking on that same guitar. Even if everything else stays the same, just playing a different part on the same guitar can cause our your volumes to be completely wrong.

In the Takes chapter, I will tell you to critically evaluate each take after doing it. This is one of the main reasons for that. To check if your volumes are still at proper levels, and modify if not.

Otherwise, you’ll be a train that slowly races towards a cliff of either “recordings that are too soft to hear anything” or “recordings that clip everywhere”. Because you will change your position slightly, you will vary how loud you play, and you will move microphones here and there.

Warning: the right order

When I just started, I encountered many clipping issues. And I did “the naïve fix”: just lower the volume on the master fader. (That’s the big fader that controls volume of the whole track, in every DAW>)

This is wrong. That’s the last thing you should ever touch.

Why? If it clips, that means it’s too loud earlier in the process. Data is thrown away somewhere before that point. You’re not getting that data back by lowering the final volume.

Instead, lower the volume as early as possible: on the individual recordings. If all your recordings have enough “head room”, they can stack and combine without clipping.

It should only rarely be needed to lower the volume of a whole track or group of tracks. If a sound clips when it reaches your master fader, you’re already too late.

In general, I think you should err more on the safe side. When in doubt, give yourself more head room and record at lower volumes. We have the technology and quality to deal with that, even in cheap equipment.

We will never find technology that can magically retrieve bits of data that were never saved :p