Gain Structure

[Robtoons]

I'm sure some of you wizards can answer this one for me, (if you're willing)
I was eager to try out some techniques I heard in Ronan's class at the rally (or as he refers to himself- the hairy guy) great class! That dude is cool. He was saying he never records to track any hotter than about -10 to -15 db. So I'm trying this out and I'm wondering where I'm supposed to get the gain from to get to the master. I can push the 2mix fader all the way up to the top (and that's not a good idea is it?) but it's still not enough. If I put the brick wall on the master buss I can shove that thing up to about +15 and then there's plenty of gain, but I get quite a bit of distortion...
I usually normalise tracks that are low, but he says no no. So- where does the gain come from?
Thanks in advance! -Rob
(Oh I use cubase btw if that makes any difference)

[andygabrys]

my 2 cents.

recording that low (its not really that low) just leaves headroom. so you can leave your faders up high around unity gain. Its easier to make automated fader moves when the faders are close to unity cause moving them an inch doesn't change the gain much. if you recorded really hot, then you have to pull everything way down, and moving a fader an inch at -25 db makes a huge difference in gain.

if you had 20 tracks that were all peaking at -15 db, then the additive volume on the master fader would sure be a lot hotter than -15db.

IMO you mix with headroom. when you go to master, then sure, normalize the whole MIX to peak at -3.0 db or similar if its way low. then apply your eq and limiting.

so you don't normalize the individual tracks, you use that headroom to get a clean mix.

[Kolstad]

If you start your mixes at say -10, you can have additional gain from an insert compressor, a bus compressor as well as a compressor at the master bus (and even mastering). The -10 allows you headroom to add compression in small amounts at several stages, and when you turn the track up after all that compression, it really can add up in the final mix. So, you'd want that headroom for mixing, because it gives you more control over your sound and levels.

That's why you want to start that low (or even lower). For example, an analog console at 0db (unity gain) is usually set so there's 20db of headroom while you are recording, so if you want more of that analog depth when recording to digital, you can consider leaving more headroom. That will also leave you with more opportunity to use the best of software compressors ect. That way you really get to hear what good software can do.

But also remember that just turning the audio up and down in your daw is not normalizing in a bad way. The bad normalization is when you pull up the peaks to 0, leaving no headroom. When you adjust the gain on individual tracks in your daw, you just want the peaks to reach -10db (or the level you prefer as a starting point) for the start of your mix.

It can be helpful to differentiate between recording and mixing, because you can allign the levels again after recording, before you start mixing, as part of the gain staging process. So you can still record at say -6db, and get a sound signal to noise ratio, and then afterwards adjust your start levels for the mix down to -10 to get the desired headroom.

Lots of ways to do this, though, so it's just my2c..

This may be a helpful article http://www.learndigitalaudio.com/blog/normalize-audio

[Russell Landwehr]

The main idea behind gain staging is to introduce the lowest amount of noise and distortion at each point in the path (unless you want it then there is a VST to add that in). I'm guessing that he is talking about achieving that -15 to -10 level at the last point in the signal chain when tracking. It certainly does give wiggle room when mixing. If you need to change the level of those tracks you can do it with faders or make-up gain in compressors. Whatever sounds the most musical.

The problem I see with normalizing, is that it changes the dynamic texture of the track since zero is still zero, but the peak of the track is now higher.

The problem I see with recording at too low of a level, is that if the track needs to be made louder during the mix, you could also increase noise since your noise margin is less.

And I wonder if he was talking about RMS or Peak.

I've always figured that if I need to move the master fader any more than about 3db from unity, then I've got a problem up the chain a bit that I need to fix.


Additionally, when mixing, once I've got everything pretty close, I put a Multiband Compressor and/or an L3 on the master channel. I tweak them in so that they sound musical. Then I finish tweaking the mix since the compressor and limiter will tend to change the mix a bit. But then before I slap the compressor and limiter on the master channel, I've already mixed so that I'm getting some overs on the master channel. Plus, I've always figured that if I need to move the master fader any more than about 3db from unity, then I've got a problem up the chain a bit that I need to fix.

Russell

[Robtoons]

Thanks Andy, Magne and Russell,
I think this is making some sense to me, I'll do some experimentation with this.
So, Andy are you saying mix down the tracks to a new track and then normalize that track before doing the mastering comp and limit?

Lots of food for thought here, thanks you guys! -Rob

[Russell Landwehr]

mix down the tracks to a new track and then normalize that track before doing the mastering comp and limit?

I'm not speaking for Andy here. But I am putting my 2c in on this. I would say never normalize on anything ever... unless it's an MP3 you got from someone else and you don't care how it sounds. Normalizing is just not musical.

Instead of normalizing, use the mastering comp and limiter to get the desired loudness, peaks, RMS, etc.

"Mastering" to me has two different meanings.

1. The process of compiling an album from individual tracks where EQing and Compressing and Limiting them individually makes them fit each other as a cohesive and naturally flowing performance.

2. The process of making an individual track sound the best it can after mixing.

With #2 in mind, the "master" compressor, EQ and/or limiter can be applied to your mix-down in a different piece of editing software like WaveLab or SoundForge. Editors like that have a "normalize" function... that I NEVER USE. If your mix isn't hitting -.5 or 0 or whatever your target is (mine is always -.5db) then use the fader or the makeup gain in any of your master plug-ins.

OR (and here is my preferred method)

Also with #2 in mind. Put the master compressor, EQ and/or limiter on the master channel of your DAW, so that when you use them (at the end of the mix process), you are free to go back and tweak any of your mix that may become imbalanced due to the inserted master channel stuff. Then when you render your mix, you have a file that is already "normalized" (gawd I even hate using the term) and your peaks and RMS are right where you want them... then all you need to use your expensive Editor for is editing off the pre-roll and post-roll and making sure your fade or tail is right.

When "mastering" in the DAW this way, I will occasionally bypass my master inserts just to make sure that they are still making the mix sound better. Overusing compressors and limiters can really suck the life out of a mix if you aren't careful.

Now I'm going to say something about "overs" during the mixing process. Most DAWs are processing internally at 32bit which can handle channels (specially the submix (group) channels) occasionally going into the red a bit without any bad distortion. Some VSTs can handle the red internally somewhat also, but I try not to let them because they can add some ugly distortion. The master channel can also handle a bit of red, but of course when it hits your soundcard's DA chip and heads to your monitors, then it starts sounding nasty. That's why we go for a 0db or less mix.
With this in mind, if I'm going for a hot (and clean) sounding mix, I'll usually mix so that I'm getting peaks around +1db to +2.5db on the master channel before kicking in my master channel processing such as Multiband Compressor and UltraMaximizer or Limiter. (and one of these days I'm going to have to have Andy teach me how to use the MultiMaximizer... )

The preceeding statements are strictly the beliefs and opinions of... blah blah blah.

Just remember... use your ears. Whatever you do, make sure it sounds musical.

Russell

[andygabrys]

you gotta be real about what you are doing when you are normalizing or adding gain to a track.

as I understand it:

1. normalizing is increasing the gain of a sound.

2. adding a trim plugin is increasing the gain of a sound.

3. pushing a fader up is increasing the gain of a sound.

4. even adding makeup gain (or input trim) on a plugin is increasing the gain of a sound.

all these are not destructive processes as long as they don't push the signal above 0.0db.

not definitive - but this is how I understand Normalization to work - see Peak Normalization. Its a gain change event, not a dynamic expansion technique.

http://en.wikipedia.org/wiki/Audio_normalization

and

https://www.custcenter.com/app/answers/ ... dio-tracks

but to test for yourself, make a short track with some medium loud hits and really quiet hits. Then normalize it so the loud hits are just below digital zero. did the really quiet hits get louder? did it seem like it was dynamic range expansion where the quietest hits stayed as quiet? or mor like an even application of a certain amound of gain to all points of the track.

[andygabrys]

Interesting!

With #2 in mind, the "master" compressor, EQ and/or limiter can be applied to your mix-down in a different piece of editing software like WaveLab or SoundForge. Editors like that have a "normalize" function... that I NEVER USE. If your mix isn't hitting -.5 or 0 or whatever your target is (mine is always -.5db) then use the fader or the makeup gain in any of your master plug-ins.

some people like to mix with much more headroom than -0.5db cause meters on DAW are kinda sketchy depending on the program material, and there is not point to have it that hot straight off - if you are mixing a track intending to send it to someone to master. cause just adding a little eq could put the track over 0.0db.

......something about "overs" during the mixing process. Most DAWs are processing internally at 32bit which can handle channels (specially the submix (group) channels) occasionally going into the red a bit without any bad distortion. Some VSTs can handle the red internally somewhat also, but I try not to let them because they can add some ugly distortion.

The master channel can also handle a bit of red, but of course when it hits your soundcard's DA chip and heads to your monitors, then it starts sounding nasty. That's why we go for a 0db or less mix.
With this in mind, if I'm going for a hot (and clean) sounding mix, I'll usually mix so that I'm getting peaks around +1db to +2.5db on the master channel before kicking in my master channel

it might be generally assumed here that you are overdriving the mix buss of your DAW, thinking of a it like a console. an electronic track (especially a really hard hitting one) is going to have a lot of tolerance for this, it will sound loud. the distortion if any will be masked.

but an acoustic mix, or a delicate guitar / vocal is going to have less tolerance for this.

anyways..........

[Robtoons]

Wow, you guys are a veritable encyclopedia.
I'm reeling with info here. Thanks so much for your time! I will keep reading and re-reading this, hoping some of it will soak in.
Cheers! -Rob

[cassmcentee]

I'm soaking in it too Rob, thanks for asking the question.
And thanks A & R ( ) for the advice.