What is gain staging and is it important?

[Nick Batzdorf]

Hello Akarin,
I bought VUMeter this summer, thanks for the tip in another thread..
Do I place VUmeter on every track?

Cheers,
W

If you want, but your tracks already have meters. (They're not actually VU meters when you're using digital audio, but that's a discussion for mastering engineer cocktail parties.)

 

[José Herring]

I'm reading through every thoughtful response. It is taking me time. But, I just wanted to post to say that I am reading and taking in a lot of information.

I did find a course online by Denis Sands one of my favorite all time film music mixers. I'm considering taking it but it's a bit prices for 7 hours worth of material. But, I may just do it.

Keep the responses coming. Thx

 

[jcrosby]

From my basic understanding, it’s especially important when using plugins that are modelled on analogue gear, to get the right volume level going in, as this type of equipment would usually have had a limited sweet spot in which it could perform it’s processing magic.

Yes indeed. This is where gain staging makes all the difference when mixing ITB. Modeled plugins are calibrated to their real world, non-0 dBFS big brothers. So while in theory you can run a mix as hot as you want in a 32 bit float engine, the attitude that gain staging doesn't matter goes out the window as soon as you start to work with plugins that are calibrated to mimic 0VU calibration. There's about as much wisdom in that belief as saying if you want to write a melodic piece it doesn't matter if you play in a completely unrelated key. Sadly that misconception is a lot more pervasive than you'd think...

Does that mean you shouldn't deliberately run plugins hot if that's a particular sound you're after? Absolutely not. People have been deliberately hitting tape hard and over-saturating analog gear for ages... That's what decapitator's punish button does, it adds 20 dB of gain... That said if you do this to every single track in a mix you're just going to wind up with mud, and no transient definition.

If you have a hard time sticking to the rules the hornet VU meter's the best 5 euros you'll spend all month. Not only is it useful for getting over bad gain staging habits, it lets you gain stage an entire template in a single play-through.

 

[tc9000]

A great question and great replies here! Gain staging is something I keep coming back to to figure out a bit more...

Thinking about levels going into plugins, and hunting for that sweet spot, I like this section of a review where the reviewer uses an ITB signal generator and VU meter to assess how the PA Lindell 254E is calibrated. Its not rocket science but I never thought of doing this until I saw this:



Also, this:

 

[el-bo]

If you have a hard time sticking to the rules the hornet VU meter's the best 5 euros you'll spend all month. Not only is it useful for getting over bad gain staging habits, it lets you gain stage an entire template in a single play-through.

Yes indeed. This is where gain staging makes all the difference when mixing ITB. Modeled plugins are calibrated to their real world, non-0 dBFS big brothers. So while in theory you can run a mix as hot as you want in a 32 bit float engine, the attitude that gain staging doesn't matter goes out the window as soon as you start to work with plugins that are calibrated to mimic 0VU calibration. There's about as much wisdom in that belief as saying if you want to write a melodic piece it doesn't matter if you play in a completely unrelated key. Sadly that misconception is a lot more pervasive than you'd think...

Does that mean you shouldn't deliberately run plugins hot if that's a particular sound you're after? Absolutely not. People have been deliberately hitting tape hard and over-saturating analog gear for ages... That's what decapitator's punish button does, it adds 20 dB of gain... That said if you do this to every single track in a mix you're just going to wind up with mud, and no transient definition.

If you have a hard time sticking to the rules the hornet VU meter's the best 5 euros you'll spend all month. Not only is it useful for getting over bad gain staging habits, it lets you gain stage an entire template in a single play-through.

There are a couple of plugins that do this GS thing right, though I’m sure there may be more. Both Ohmforce’s Ohmicide and NI’s Supercharger GT have a lit indicator that works hand-in-hand with the input trim. Like this, it guides the user towards the best level to be hitting the processor.

It’s a shame that more plugins don’t follow suit.

And thanks for the VU meter link. Seen it recommended on a few occasions recently, but never sprung for it. However, even though HS is not too difficult to grok, it still remains somewhat of a dark art. At this price, it even makes sense as a tutorial and checker, ti make sure everything is in it’s right place (...in it’s riiiight plaaace)

 

[robgb]

Gain staging is very important and it doesn't apply to analogue gear only, it's an universal thing regarding your mix.

Gain staging is basically to take care of your volume meters and your "volume budget", the purpose of gain staging is to make sure that you never exceed your volume budget to avoid distortion or clipping, and beyond that, give all of your instruments enough space to shine and sound good.

You could peak at 0dB with just a bassline and a lead synth, or with a full orchestra with 45 tracks. To fit the latter into the same space as the first while still sound clear and powerful, gain staging is inevitable.

More in detail it's as well about the signals between your plugin chain.
If you'd have a chain like this: Synth -> Chorus -> Delay - Reverb, you could easily be clipping above 0dB at the chorus plugin already but lower the post-volume afterwards, which would still lower the overall quality of your signal (if not done on purpose, sound design wise). So with gain staging you make sure that your levels between the plugins are all fine for a clear final sound.



That's simply not true, as explained in my example of a single channel where you could easily run into distortion while your master bus and even that channel's output meter both look fine.

The master bus doesn't contain any information about the clarity of the signal, it just shows you the final output volume.

Watch Kenny's video. Proof.

 

[wst3]

I hate to state the obvious, but at some point your signal is going to be analog. If you are recording a live source then gain staging is required in the analog stages, and the first stages in the DAW. The remaining stages will benefit from a proper gain structure, but yeah, you can cheat like crazy and probably get away with it.

Oh wait... if you signal is going to hit a D/A stage you need to worry about gain staging. A D/A converter has an analog half. And any analog equipment that follows your D/A conversion will also care about levels.

 

[JEPA]

I'll leave this here...

 

[Voider]

Watch Kenny's video. Proof.

I've tried it and to my surprise I must admit, it works with real time VSTs, but if you record something, do render in place of your synths/vsts you'd still bake that distortion into the audio file. And I am not sure if it always works with real time VSTs as there are pre- and post gain knobs and different constellations could lead to different results.

Here are 4 beautiful piano notes (I'm kidding, before anyone believes that), I've rendered them in place while they were peaking above 0dB, now I've turned the channel fader of that audio down so that it peaks at -1.4dB on the master fader - you can still hear the distortion from the piano being rendered above 0dB.

 

[robgb]

I've tried it and to my surprise I must admit, it works with real time VSTs, but if you record something, do render in place of your synths/vsts you'd still bake that distortion into the audio file.

If you're rendering in place, the audio is still being drawn from the master bus, so I'm not sure why you would be baking in any distortion unless you're overloading the master.

As Kenny says, we're talking only about mixing, not recording. Recording input levels are still important.

 

[ProfoundSilence]

I turn down a professional mix I like a little and then use that as my reference for what "loud" is supposed to be like. I always warm my ears up a little with that sound - so that when I'm adding things - they make sense volume wise.

With orchestral instruments - gain staging should be done for the desired body/peaks during loud sections... gain staging each individual instrument to a certain level is absolutely silly, given a clarinet is not as loud as 3 trombones.

 

[NoamL]

given a clarinet is not as loud as 3 trombones.

agreed, but they have the same "reference loudness"

 

[SlHarder]

the positions of the faders become useful information. I know that if I have a fader pulled way down low, it's because it's a track that I want to be very quiet in the mix and not a track that is so hot that I have to have to lower the fader to compensate.

This.

The positions of the faders then represent artistic decisions you've made in the mix, not technically driven decisions.

 

[vitocorleone123]

Side note: TBProAudio mvmeter2 vumeter plugin is free - there's no technical reason to pay for a VU meter plugin for a couple years. It's just a matter of preference and money.

 

[Kent]

agreed, but they have the same "reference loudness"

Yep. Set two mf clarinets to one mf horn, two mf horns to one mf trumpet, and you're in the Rimsky-Korsakov ballpark.

(this is an oversimplification obviously please don't hate on it)

 

[rudi]

There are several factors to take into account.

If you are recording an audio signal, staging is vitally important to ensure the best noise to signal ratio.

However, once the signal has been recorded, it is stored in your DAW as a series of numbers.

In the early days of digital audio for technical reasons (e.g memory, processor speed, maths operations etc.) the resolution was limited. From 8, to 12, to 16 bits. As computers became more powerful audio could be stored and processed at 24bits, and even 32bits or more.

Each increase in bit size gave you a greater range i.e more bits > more steps > more dynamic range. You jump from 255 steps and ~48dB at 8bits, to 4,294,967,296 steps and ~192dB at 32 bits - which is not far off the maximum loudness range ~210dB.

But no matter how much dynamic range you have you still need to be careful not to go over the maximum possible resolution. You also have to be careful when adding gain (multiply), reducing gain (divide), or mixing (adding) to avoid rounding errors and exceeding your maximum 0dbFS ceiling.

But what if you could you increase your audio resolution without the above pitfalls? The good news is that you can - by using some clever binary and maths manipulation.

We'll stick to 32 bits integers. Instead of storing an integer as a single value, you can divide the 32 invidual bits to represent different parts of a number, similar to scientific notation when dealing with very large of very small numbers. There are many ways of doing that, but one of them is by using a floating integer.

In our example we'll divide our 32 individual bits as follows:
- a sign for + or - (1 bit)
- an exponent (8 bits)
- a mantissa or floating point value (23 bits)

This gives you an incredibly large range of possible values. You jump from a range of 0 to 4,294,967,296 for a fixed 32 bits integer, to a range of ~1.2 x 10(-38) to ~3.4 x 10(38) values which is incredibly large! In terms of dB range it works out at ~1528dB!

It is also a signed range so you can go from hugely negative to hugely positive values. Even if your recorded signal hits its maxinum dB range in 32 bits integer ~192dB, you can still increase or decrease its gain by ~770dB of headroom!

Of course, when you output your end signal to an audio file, it will be scaled down to whatever format the file uses, but in the intervening stages in your DAW you won't have to worry about distortion.

Regarding FXs that have a "sweet spot", they are generally modeled after analogue pieces of equipment which distort if you overload them. If the internal resolution of the FX uses fixed integers, they will still distort the signal - but this is usually done by design and for aesthetic choices.

https://en.wikipedia.org/wiki/Audio_bit_depth
https://www.hdvideopro.com/columns/audio-assist/what-makes-32-bit-floating-point-audio-powerful/

 

[NoamL]

Yep. Set two mf clarinets to one mf horn, two mf horns to one mf trumpet, and you're in the Rimsky-Korsakov ballpark.

(this is an oversimplification obviously please don't hate on it)

hmm sort of... I have no idea how to communicate the idea of "reference loudness"... it's like, if instrument X is averaging at X dB at mezzo forte then instruments W, Y, Z will all have a correct loudness at mezzo forte relative to that...

 

[DS_Joost]

I've heard the term a lot and did some research a few years ago on gain staging. I ran across an old engineer trying to explain it and poor dude. It just seemed like he was trying to apply some old analog knowledge and hadn't really caught up to the digital age.

But, then I'm like but what if? What if it's important?

So what is gain staging? Is it worth learning and doing?

José, I know you are a Reason user, but this tip also applies to most other DAWs:

 

[Snoobydoobydoo]

Sonimus has written an Article about Gain staging,
and their Plugin's Britson / Satson are an interesting approach.
Sonimus, what is gain staging.

 

[jaketanner]

Been mixing professionally for well over 15 years...from analog to now ITB...here is what I have found. The plugins that emulate analog units..example: Acustica Audio's Acqua plugins...they are sample modeled...not algorithms but also also plugins that emulate analog like the BX console from Brainworx should also be gain staged to get the most from them...-18 is standard. Many plugins like Slate Digital's "trim" within their rack is calibrated at -18 = ZERO...so if that trim is at the end of a plugin chain, you want to make sure it is hovering around zero, so that whatever you put on after it, is properly calibrated.