Gain Staging

SeanHurley

Hey guys I was just wondering what your approach to gain staging is, particularly ITB.

Like what level do you record to your DAW at?

Does the level you record at effect plugin performance? - obviously it has an impact on compression but are there optimum levels for plugins (I think this is true modelling plugs like liquid mix)?

What level should tracks be peaking at post inserts?

How loud should the final mix pre-mastering be?

Sorry for all the questions it is just that I feel this is something I really need to work on.

Thanks in advance.

S

Obi-Jim

Its not really important if your working ITB. Once your recordings have good signal to noise, your pretty much good to go. (this is one of the "arguments" that pops up about digital vs. analogue, etc etc.)

I suppose this is one method:

Keep signal peaks to -18dBFS. That's it pretty much. I like this because it sticks to convention a little and it helps with sessions that have a larger amount of tracks. You don't find yourself having to pull back faders, it allows you to keep a lot of faders in and around the zero area.

Mastering, i guess its best to ask the mastering engineer you are giving stuff to. But anything with a bit of headroom should be good. Peaking at: -6dBFS - -3dBFS

PaulBrewer

I recently had a 'virtual' conversation with Brady Barnett a Nashville engineer/DAW operator on the same subject.

One thing he often sees on tracks he works on is stuff recorded 'too loud'.

This isn't 'too loud' in the ' actually distorted' sense but if you, for example, add 4dB at 5k (as could easily happen) the warning lights start flashing on plugins.

What he does is reduce the gain by inserting the Trim plug-in in Protools.

He will also used that Plug-In to RAISE level too.

This is particularly relevant for ITB mixing as it allows you to have maximum control by using the fader in it's most sensitive region i.e. around the 0dB point.

Obvious when you say it out loud innit !

MilanPan!c

The best advice I ever got on levels related to mixing was to mix down, i.e. if you are try to get something louder in a mix, turn other things down.

That's the easiest way to avoid the age old problem of overly loud mixes, which usually means less dynamic mixes.

i57dwun4yb1pt8

i record all my tracks at -16 digital on the converter in my setup

but im an ammie , so the pros here my diverge on this .


sonalksis have a free trimmer / meter

http://www.sonalksis.com/index.php?section_id=99

PaulBrewer

DaDumTish wrote: »

i record all my tracks at -16 digital on the converter in my setup

but im an ammie , so the pros here my diverge on this .

What? Peaks at? Averages at ? Wha ? -16dBFS ?

madtheory

+1 for turning other stuff down! Very easy to do in Pro Tools, slightly less easy if you've done some automation already.

Most plugins don't change sound according to input level, all else being equal (eg threshold in a compressor or WHY) they either clip or they don't. SNR is not an issue because you're running at a very high bit depth. For example, if you're overloading the main outs, you can just turn the master fader down til it stops clipping, which of course doesn't work in the analogue domain. Some plugins have kind of crappy gain controls though, like the Massey tape head. It works better at unity gain and "driving" it with an eq. That's not really driving in the analogue sense though!

The trim plugin trick is cool, because you can set up a rough balance with the faders, then copy the gain/ attenuation to the trim plugin, and then have all faders at zero. Then when your mix goes awry after a few hours, you can actually see on the faders how far out you are. In the tape days, you would zero faders then do a quick rough balance using the input gain controls. They used to do that on the SSL too with the fader trim thingomebob.

But if you're good at keeping peak level at -18dBFS, then the above exercise is kind of already done.

PaulBrewer

madtheory wrote: »

The trim plugin trick is cool, because you can set up a rough balance with the faders, then copy the gain/ attenuation to the trim plugin, and then have all faders at zero.

That is indeed cool !

Maybe we should introduce a few Stickies for some Tips n Tricks for each of the Popular Daws ?

SeanHurley

I like that trim plug idea, I have done something similar a few times with the Analog Channel 1 plug from McDSP.

I think I have been recording everything too hot. It always leaves me with a hellish mix of trying to get everything to fit. It makes complete sense tho to leave a little headroom for plugs to work in.

madtheory

Plugins don't really need headroom. Most DAWs are 32 bit float, that's approaching a dynamic range of 15,000dB. That's why if the main out is overloading, you just lower the master fader and the clipping stops. The headroom is for the analogue electronics in the converter: -18dBFS= +4dBu= 0VU on most systems.

i57dwun4yb1pt8

PaulBrewer wrote: »

What? Peaks at? Averages at ? Wha ? -16dBFS ?

slim shady, yes, no , slime shady , yes

trackmixstudio

I track peaks at -12 then mix at -6 peak.

madtheory

Can we please state which scale we're using! dBFS, dBu etc. etc. otherwise we won't know what you're talking about...

For ITB mixing to 24 bit, peak level can be -0.3dBFS

PaulBrewer

madtheory wrote: »

Can we please state which scale we're using! dBFS, dBu etc. etc. otherwise we won't know what you're talking about...

For ITB mixing to 24 bit, peak level can be -0.3dBFS

I like to take it to -.2 for the choruses , give it a bit of ooomph, like.

SeanHurley

madtheory wrote: »

Plugins don't really need headroom. Most DAWs are 32 bit float, that's approaching a dynamic range of 15,000dB. That's why if the main out is overloading, you just lower the master fader and the clipping stops. The headroom is for the analogue electronics in the converter: -18dBFS= +4dBu= 0VU on most systems.

Makes sense now that you say that.

madtheory

PaulBrewer wrote: »

I like to take it to -.2 for the choruses , give it a bit of ooomph, like.

Hey, get your own football if it upsets you that much!

But seriously, -0.3dB is a useable margin to prevent distorition in poor DA converters due to intersample peaks.

trackmixstudio

madtheory wrote: »

Can we please state which scale we're using! dBFS, dBu etc. etc. otherwise we won't know what you're talking about...

For ITB mixing to 24 bit, peak level can be -0.3dBFS

We are talking about ITB so that would be dbfs or digital zero.
Regardless of what you think, DAWs perform better with lower overall levels through the system. This has been widely acknowledged by many industry professionals.
It is also very important to set your meters to pre fade display because if your fader is below zero and the channel is peaking somewhere in the chain your meter will not show peaks if set to post fader metering.
I am a big fan of pres with input and output gains like my API A2D or DRS so you can saturate the input but lower the output to get the level you want hitting the converters.
The only problem with running at lower levels is, as stated earlier compressor/limiter plugins where you have to push the input gain then lower the output gain.

trackmixstudio

madtheory wrote: »

Hey, get your own football if it upsets you that much!

But seriously, -0.3dB is a useable margin to prevent distorition in poor DA converters due to intersample peaks.

If you track at -.3dbfs you have no headroom for any processing. What if you want to boost with eq. Also many eq plugins actually slightly increase overall level even when you cut due to phase distortion.
24bit has a very low signal to noise ratio so there is no need to try to get close to zero.

Here is a very interesting (but VERY long) read on the subject.
Look out especially for Paul Frindle's input on the subject. He was the designer of Sonnox plugins which are widely regarded as some of the best available, especially the EQ plugin.
http://www.gearslutz.com/board/so-much-gear-so-little-time/463010-reason-most-itb-mixes-don-t-sound-good-analog-mixes-restored.html

madtheory

Guys, you've misread my post(s). I already said -18dBFS for tracking, and how it affects the analogue bits. The -0.3dB is the peak for mixing down ITB. Also, you can go a lot higher than -18dB if you're mixing via some analogue bits. You don't need the headroom because you will know exactly what your peaks will be.

PS it's dB, capital B for Mr. Bell.

PaulBrewer

madtheory wrote: »

But seriously, -0.3dB is a useable margin to prevent distorition in poor DA converters due to intersample peaks.

No , you're right ok

Just as the CD player market hit critical mass point all sorts of cheap stuff was produced and the -0.3dB FS point was considered safe.

I remember I had a Country CD that used to peak some cheapo players at the time but not others.

I don't think it's the issue it once was these days though.

madtheory

PaulBrewer wrote: »

I don't think it's the issue it once was these days though.

true enough! converters are very good these days.

cqd2

Funnily enough, ive been looking at this side of my mixes for the last couple of months..i read through as much as i could on the epic itb v otb thread over on gearslutz(started by the dude that designed the sonnox plugs), and i really must say, it does make an awful difference...(tracking low, and keeping the volume down itb, trying to hit plugins at optimal levels ~-10dB)..apparently once you engage any filtering in a plugin, you can cause intersample peaks, which dont show up on the clip indicators, but do actually distort..you can get plugs that'll spot it but you've to pretty much check after each plug...some dither, some dont..there's quite a bit to it, but my mixes have just way more headroom..im getting clearer, more open mixes..nowhere near as much eq needed..just sounding noticably better...you can bang up the volume then once you've bounced..My mix buss is hitting maybe -8 to -6..



just saw trakmix linked to said thread..as they say...fools rarely differ..

PaulBrewer

cqd2 wrote: »

Funnily enough, ive been looking at this side of my mixes for the last couple of months..i read through as much as i could on the epic itb v otb thread over on gearslutz(started by the dude that designed the sonnox plugs), and i really must say, it does make an awful difference...(tracking low, and keeping the volume down itb, trying to hit plugins at optimal levels ~-10dB)..apparently once you engage any filtering in a plugin, you can cause intersample peaks, which dont show up on the clip indicators, but do actually distort..you can get plugs that'll spot it but you've to pretty much check after each plug...some dither, some dont..there's quite a bit to it, but my mixes have just way more headroom..im getting clearer, more open mixes..nowhere near as much eq needed..just sounding noticably better...you can bang up the volume then once you've bounced..My mix buss is hitting maybe -8 to -6..



just saw trakmix linked to said thread..as they say...fools rarely differ..

That's interesting - so are you saying that if you had a track that was at 'ok' level i.e. not distorting and reduced the input on every chan to the plugins with the Trim plug in PTools you'd have a nicer sounding mix ?

Obi-Jim

PaulBrewer wrote: »

That's interesting - so are you saying that if you had a track that was at 'ok' level i.e. not distorting and reduced the input on every chan to the plugins with the Trim plug in PTools you'd have a nicer sounding mix ?

ha! imagine that! if I get all my levels right, from start to finish, my mix will sound better!

But, for the record, yes. I think the theory more so applies to tracking levels. if you track everything to peak at say -3dBFS and ensure that nothing clips along the way, it won't sound as good as tracking it to peak at -18dBFS and ensuring everything is ok.

However, the only real way to test this, i suppose is to record the same thing both ways.
But there's no harm in tracking at -18dBFS anyway. It's 0VU. It's grand and it allows you keep those faders closer to 0 and you don't really have to worry about plugins clipping or anything (unless you add a lot of gain somewhere along the way). Most importantly, i find, is that it eliminates any worry about this subject while you mix. You do your thing and nothing clips.

And yea most DAWS have a 32bit floating mix bus, so clipping can be dealt with. PTHD is 48bit fixed though, a little different.

PaulBrewer

Obi-Jim wrote: »

.

And yea most DAWS have a 32bit floating mix bus, so clipping can be dealt with. PTHD is 48bit fixed though, a little different.

Should a floating point mix bus sound better, theoretically at least ? I know nothing of the science here.

I recall reading somewhere about Logic's Summing sounding better (or was it worse? ) than PT's HD ?

Anyone done an A/B ?

Aridstarling

Consensus in Windmill the other week was Logic "just sounded better" than PT. I don't know for sure as I'm a Cubase user, but it was interesting to hear opinions on the matter. Very definite.

On the subject of tracking, yeah, -18dBfs, peak the mix somewhere between -3 and -6. I don't know if it inherently sounds better, I think it just allows you to work without any worrying about clipping. Works for me anyway.

Obi-Jim

PaulBrewer wrote: »

Should a floating point mix bus sound better, theoretically at least ? I know nothing of the science here.

I recall reading somewhere about Logic's Summing sounding better (or was it worse? ) than PT's HD ?

Anyone done an A/B ?

Well, which sounds better, i don't know, i've not tested it. I don't think it really ever comes into the equation too much (unless you really use master faders or auxes a lot, to stop your clipping to your converters).

An "important" difference is to do with the noise floor. 32 bit floating has a floating noise floor, which essentially means a pumping noise floor. 48 bit, obviously you don't get this. But, come one, it's far too low to ever hear, probably. I'm sure there are far more important things to worry about in any setup than whether it's 32 bit floating or 48 bit fixed.

32 bit floating = 4,294,967,296 points
48 bit fixed = 281,474,976,710,656 points

24 bit = 16,777,216 (which is what your really working at anyway)
16 bit = 65,536

There are other differences when you read the maths of it, but it's not worth the brain power unless you're really interested in that stuff. (I don't claim to know the whole story)

I dunno if that answered anything, but sure gwan.


(just on PTHD: Basically in this mix bus, you have to ensure no auxes/masters clip while going to the main master output. All aux and masters input at 24 bit and output at 48 bit. This is not the same with a 32 bit floating system)

But anyway, just don't be clipping stuff and we'll all be grand :-)

cqd2

PaulBrewer wrote: »

That's interesting - so are you saying that if you had a track that was at 'ok' level i.e. not distorting and reduced the input on every chan to the plugins with the Trim plug in PTools you'd have a nicer sounding mix ?

Yeah, pretty much, ive was putting the trim on first (or just adjusting the input gain or the eq)..and then, say if you've sansamp on something, it can kind of get out of hand...maybe put a trim after the plugs..so you're hitting the mixbuss at roughly the same level...you also have more room for manouvre(spelling?...brainfart) with the faders in PT this way too..i dont know if you'l understand the analogy now, but its like i had almost hit a noise ceiling before..didn't have that much control over the overall dynamic.., just a few dB under said ceiling...i dunno...im definately noticing an improvement though...(although, this could well have something to do with a certain acusticaudio product as well, which kind of started me looking at the whole thing)

cqd2

Aridstarling wrote: »

Consensus in Windmill the other week was Logic "just sounded better" than PT. I don't know for sure as I'm a Cubase user, but it was interesting to hear opinions on the matter. Very definite.

Pan laws or gain structure maybe...its one's and zero's..

jtsuited

cqd2 wrote: »

Pan laws or gain structure maybe...its one's and zero's..

this always confuses me when people make this implication that it's all ones and zeroes.

summing two audio signals is a very complex process and summing algorithms is where so much DAW development has happened in the past 5-10 years. DAWs' summing algos are not created equal.

I remember trying to mix in ableton 6 and jesus it was a nightmare because of their summing algorithm. of course a lot of people said 'sure it's all just ones and zeroes etc.' but then ableton addressed it in version 7, and lo and behold the distinctive ableton mix buss sh1tness had dissappeared.

cqd2

Well, if you put 15 audio tracks in each with no processing they'd null..anything that doesn't might need to ba panned a bit differently..

The difference between the two is the panners. In Logic, a stereo track does not have a separate left and right pan control, only a single rotary control that you can turn left or right. Then if you look at the bus sends, they are all stereo and have no pan controls.

Aridstarling

cqd2 wrote: »

Pan laws or gain structure maybe...its one's and zero's..

I'd imagine there's more to it but I think Logic's pan laws were changed recently and I'm sure this does indeed have something to do with it. But jtuited makes a good point about Ableton, that was a very noticable difference.

kfoltman

jtsuited wrote: »

this always confuses me when people make this implication that it's all ones and zeroes.

summing two audio signals is a very complex process and summing algorithms is where so much DAW development has happened in the past 5-10 years.

:O

Palm, meet face...

jtsuited

cqd2 wrote: »

Well, if you put 15 audio tracks in each with no processing they'd null..anything that doesn't might need to ba panned a bit differently..

Where are you getting that from?

Years ago I moved entirely to logic from ableton because i did a null test (and yes pan law was the same) and it turned out there was a huge difference between logic 8 and ableton 6.

madtheory

This is old. Check out Lynn Fuston's Awsome DAWsum CD.