Gain staging, or why you should record quietly

The volumes of your recordings are of utmost importance while you are recording, mixing, and mastering. In other words, it really matters how loud your stuff is. This affects the quality of your finished product and the ease by which you can manage multiple tracks and get the best sonic results.

The trouble is, it can be confusing at times to know how loud is appropriate. How loud should we record? How loud should we mix? What about the ever elusive mastering stage? That’s what I’m here to explain.

Two important facts

I was convinced that recording to a level of 0 dB was the “sweet spot” for recording when I first starting digital recording. The problem with this statement is lack of context. 0 dB is a sweet spot only if you’re recording to an analog system and you’re measuring average volumes. The first fact to realize is that analog and digital measurements for volume are very different.

A typical analog meter measures volume from silence to a maximum much higher than 0 dB. A digital meter measures volume to a maximum of 0 dB. If you line up the two meters, you can see that the “sweet spot” of 0 db in analog is about the same level as -18 db in digital.

digital vs analog meters
digital vs analog meters

But the nature of recording hasn’t changed from analog to digital. You still need a reasonably loud source to record at a conservative level. The difference lies in how the measurement systems work. In analog, volume is measured in dbVU (decibel volume units). In digital, it’s measured in dbFS (decibels full-scale).

Analog meters show average levels (also called RMS levels). Digital meters show peak levels. In the digital realm, peak levels are critically important for one simple reason. If your signal exceeds 0 dbFS in digital, digital distortion is introduced. This is not the warm, fuzzy distortion of analog tape or tube saturation. Digital distortion is nasty and not musical in any way. In other words, you want to stay way clear of peak levels reaching that 0 dbFS mark.

Here’s a sample vocal recording I did where the peak levels reached about -12 dbFS.

And here’s another sample where the peak levels exceeded 0 dbFS.

This brings us to our second important fact. A finished, mastered song peaks very close to 0 dbFS. This is how we hear songs on CD, on the radio, and on streaming services. They are loud. But this loudness is the result of mastering, not recording. All professional recordings are captured and mixed at much lower levels. Part of the mastering process boosts the volume to get close to 0 dbFS without going over.

Recording

So you’ve armed a track for recording in your DAW (digital audio workstation) software. Note that in most DAWs, when a track is armed for recording, the meter shows the level of the incoming / recorded signal. When the track is not armed, the meter shows the playback level; this can be lower or higher than the recorded level depending on fader position and/or plug-ins. There also may be a setting for meters to show average levels, peak levels, or both. I prefer to have all my meters showing both.

When you’re performing (guitar, vocal, whatever) pay close attention to the peak levels. Most meters will “sticky” the peak level, meaning it shows a little mark where the highest peak occurred and keeps it there until there’s a higher signal. This is important because you want to make sure the highest peak doesn’t get anywhere close to 0 dbFS. I usually shoot for peaks around -12 dbFS to give myself enough room, in case something does peak at -10 dbFS or -6 dbFS.

Only things in the analog world determine the recorded level: the volume of your performance (or output level of your synth or amp), the distance from the microphone, and the gain setting of your preamp (i.e. the gain knob on your audio interface). That’s it. If you exceed 0 dbFS in your recording (referred to as “clipping”) then there’s nothing in the digital realm that can fix the distortion. You have to re-do the take at a lower level by reducing the level of the audio before it hits the digital converters.

If the signal sounds too quiet in your headphones, increase the level of the headphones. If the waveform looks small on the screen, zoom the view to see it better. You can always monitor at whatever level you like, just don’t record too loud.

Mixing and Summation

Avoiding digital distortion isn’t the only reason to record quietly. As you overdub track after track, each signal adds gain to the mix. Compare the peak levels of the individual tracks against the summed master channel below; it peaks much higher.

mixing summation
mixing summation

If the level in your master channel is getting too high (sometimes referred to as “hot”) then reduce the levels of your individual tracks. With projects having several or many tracks, they can be playing back at much lower levels than the recorded audio.

Plug-ins like compressors and EQs are optimized to work at conservative levels, usually around -18 dbFS or -12 dbFS. The specifics here don’t matter as much as just making sure your tracks are playing back at a reasonable level.

Sonic Clarity

Recording and mixing at conservative levels can result in mixes that are more dynamic, open, and detailed.  Plug-ins and your DAW have room to breathe, and perform all their complex algorithms in their optimized zones.

In the final mastering phase, a limiter is applied to achieve the final volume. A limiter is a plug-in that limits the volume to a prescribed level (usually just shy of 0 dbFS) without letting anything clip. For most mixes, the limiter can apply several decibels of volume, making the mastered song much louder than the mixed version. This is the only time in the production any level gets close to 0 dbFS; the final glossy coat that finishes the recording and prepares it for release to the world.