Oh lawd, we’re past the half-way mark, aren’t we? After my last post, I’ve been knee-deep in tracking “keeper” takes, mostly of guitars so far. I’m trying to hold myself to a higher standard than for the scratch takes, which means it’s slow going at times, but the project is shaping up nicely. A pleasant surprise has been that, as I dig into the nitty-gritty of capturing decent performances, I’m finding what I recorded earlier is decent enough to still be useful – more often than not, as one voice of a doubled part.
On Sunday, Mike (at Leadfoot) was decent enough to lend me a few microphones that are much nicer and versatile than what I was working with previously. I don’t really have the time to obsess about mic techniques, but it’s opened up a lot of possibilities. More on that later, maybe.
For this post, I’m going to delve into something more on the technical side: digital gain staging, and how I use it to keep my DAW sessions organized.
At first blush, gain staging doesn’t seem like it should be a big issue in the digital realm. In the analog world, tape hiss and other signal degradation is a real problem, so it’s essential to make sure inputs are hot enough to keep above a fairly unforgiving noise floor.
With digital, that’s not really the case: a 24-bit recording has a 144dB signal-to-noise ratio – just a touch more than the range of human hearing. Provided you’re getting any signal at all, and not clipping, any level should work fine for tracking, since you can make drastic adjustments on the faders without sacrificing much by way of sound quality. That works, sure, but it’s an easy way to drive yourself crazy.
A consistent input level
Since there’s not much risk of drowning in noise, job one for digital recording is to avoid clipping. For the longest time, with digital recording, that’s all I did, with decent enough results. Lots of interfaces (like my Focusrite) have lights instead of meters, where green means “signal” and red means “clipping” – the idea is that that adjusting the gain until the light is steadily green is good enough.
However, it’s worthwhile to be more systematic about your input levels, for a few reasons:
- Headroom is your friend. If your level is just barely below clipping, a passage that’s louder than expected might blow an otherwise great take.
- If you want to splice together takes tracked over a few sessions, instead of one right after the other, it’s a lot easier since the levels are similar.
- Having everything recorded at comparable levels makes reading waveforms inside your DAW a lot easier.
After some trial and error, I’ve settled on -18dB as my target level for recording. I don’t obsess about hitting that number exactly, but when I first set up for tracking a new part, I’ll take some time to run through a bit of the song to make sure sound is coming in at that level during typical passages.
Thanks to the REAPER theme that I use, this is dead simple: I’ve got a gradient on track meters that goes yellow right at -18dB. If the tracks I’m recording are just tickling that gradient, I’m in the right place – no fuss, and I’m ready to lay down a great performance. (Or, just as likely, a dozen mediocre performances. But that’s just me.)
As an aside, I’ve heard that -18dB is an ideal level for running signals through outboard gear. I’m doing this project entirely “in the box” so that’s not relevant to me, but I’m curious to learn more about that.
Level matching with pink noise
When I’m tracking, I like to get a simple “level matched” mix put together as quickly as possible. Getting consistent input levels will get you part of the way there, but anyone who’s mixed a track will know that the numbers coming off a meter don’t always reflect how loud a part actually sounds.
This is where pink noise comes to the rescue. “Pink” noise has constant energy per octave – it’s got more bass than white noise, but to our ear (or a frequency analyzer intended for audio) it’s approximately flat from bottom to top.
By crowding the entire audio spectrum evenly, pink noise can stand in for “everything else” when you’re listening to a single track in isolation. While it’s not practical to compare a few dozen – or hundred – tracks all against each other, you can do almost as well by comparing each of them to the same pink noise reference.
The process works like this:
- Find a pink noise generator plugin. REAPER comes with one built-in, and I suspect most other DAWs do as well. There’s also plenty of perfectly cromulent free VSTs around that will also work.
- Choose your reference level. I use -14dB RMS, and wouldn’t suggest going much higher. You want some headroom to work with during mixing and mastering, but you could certainly go lower.
- Put the noise generator in your signal chain – either on your master track, or it’s own channel, and find that reference level. If your DAW has an RMS metering mode, you can use that. Otherwise, something like Voxengo SPAN (pictured above) will do the trick. For REAPER’s noise generator plugin, a noise level of -3dB comes out to exactly -14dB RMS.
- Play your tune alongside with the pink noise, and solo the first track – if you’ve done it right, you’ll hear just that one track, probably drowned in noise.
- Adjust the level of that track until it’s just barely audible over the pink noise. Start low, and slowly sweep the volume up.
- Repeat this for each track. Move quickly: even for a huge project this should only take a few minutes.
- Disable the noise generator, make sure nothing is solo’d, and admire your level-matched mix.
For me, this technique was a revelation. It doesn’t do the creative work of mixing, but it’s amazing how quickly you can turn a huge number of tracks into a decent-sounding rough mix this way.
Zero out your faders
So you’ve tracked everything at a good input level, you’ve got a rough mix put, and it’s sounding great – but now all your faders are all over the place. That’s not really a problem, per se, but I find it’s easy to mess things up by forgetting where the faders are supposed to be.
I like to set things up so my level matched mix has all the faders at 0dB. That way, I can see at a glance the relative change I’ve made to each fader as I work through the mix, and can easily reset a channel back to “neutral” just by popping the fader back to 0.
There’s a lot of ways to set this up:
- Your DAW probably has an “item gain” setting, or something similar, but I’m not a huge fan of the workflow for that in REAPER.
- If you’ve got some effects on a track, like a compressor or an EQ, they should have an output gain setting.
- You can find plenty of free gain-staging plugins that are nothing more than one volume knob or fader. I’m partial to GGain, if only because the whole GVST library is a valuable part of my toolkit.
If you’ve already done level matching on the faders, you can transfer the value from your fader into whatever you’re using for gain control, and zero out the fader, and you’re done. You can also work the other way around: put gain control plugins on your channels before doing level matching. Over the course of a session, I find myself doing both.
This might seem a bit fiddly, but it’s worth it. Effectively, it makes fader values creatively meaningful: -3dB means “a bit into the background” rather than just “a bit quieter than it got recorded.”
Gain staging for plugin chains
A big issue when it comes to processing your tracks – either just roughly during tracking, or with more finesse during a final mix – is that effects can mess up your gain staging. It’s easy for our brains to get fooled into thinking that something that’s louder sounds better. This can be an issue at the best of times, but some plugin developers seem to exploit this by having presets crank up the gain: one delay plugin I have, and otherwise really like, adds about 10dB just on the “default” setting.
To stay objective, it’s helpful to gain stage your plugin chains to keep the impact on levels to a minimum. By using the output gain on each plugin, or instances of something like GGain, you can ensure levels stay the same as the signal goes through each plugin.
You can use pink noise for this as well, in more or less the same way: just solo the one channel, disable all the effects, then bring them in one by one. For each plugin, adjust the output gain to make sure the track is just peeking through the pink noise. That way, you can be confident that the change you hear from the effects isn’t being coloured by loudness changes, and you can tweak an effects chain without messing up your rough mix.
You can also use a pink noise reference to give you some direction on how to apply effects. For example, as you bring a track up through pink noise, the first thing you’ll hear is the part of the frequency spectrum where that track is loudest, and where it will take up the most space in the mix. Sometimes, I find that it’s audibly the wrong part of the sound, like muddy low-end rumble on an acoustic guitar, or fret noise on an electric bass. That can be a good clue for where to find a quick win for your rough mix – slap an EQ on the track, and dial it in until what you hear over the pink noise is what you actually want.
Gain staging is a journey
The proof is in the pudding: here’s a short sample of one of my RPM Challenge tracks, using all these techniques. It’s the one all these screenshots come from, so this is how it sounds with all the faders at 0:
That’s far from a finished mix, or even a finished recording, but it’s great for this stage in the process. Most importantly, there’s 25 tracks in the session (including buses and the reverb send) and, to my ear at least, each one is sitting comfortably in the mix.
It may seem, from how I’ve described this process, like it’s something you do across all your tracks all in one giant pass, but in practice that’s typically not the approach I use. Instead, checking my levels and gain staging against a pink noise reference is something I do pretty much constantly, while I’m tracking or getting ready to start mixing. For example:
- After I’ve finished tracking a part and am thinking of putting some edits on it, I’ll do a quick pink noise level check on it so I have a good perspective on the edit and how it fits into the song.
- If I’m looking to double a part – say, adding a second guitar to fatten up the sound – I’ll track a verse or two, then level match it so I can get a sense for how the tone and performance will complement the existing part.
- Sometimes I’ll just check a few tracks to give myself something more “objective” to do when I’m in the middle of tracking or mixing and want to step back to refocus my perspective.
It might only take 30 seconds here, or a minute there, but in my experience at least that effort pays huge dividends for keeping a big, and growing, project organized and sounding good every step of the way.
But enough about gain staging – I’ve got an album to make, and so do you!