Hi Craig, Question about "Mixing in the Box" Keyboard Mag Article

Hi Craig,

I just read your article in the latest issue of Keyboard Magazine (November, 2006) "Mixing In the Box".

You write a section about "Practice proper gain-staging." I never heard that it's better to keep the Master at 0 and adjust the individual channels lower.

Normally I start with the Master at Zero then the Kick drum hiting Zero but as I add other sounds or maybe add some EQ on the kick then the Master starts peaking red so I turn that down...

What your saying is I should probably start lowering all the channels instead?

This could make sense and I will try it because I've found that allthough I'm recording at 24bits and using a 32bit floating engine I do feel there is something funky going on. I'm battling too hard with my levels and the sound isn't as good as I want.

I'm noticing Im more unhappy with the sound quality I am getting using Ableton that when I used Cubase SX3. I heard it has to do with the internal summing engine... maybe this all has to do with what you are talking about to?

I can't go back to Cubase though... Ableton's workflow is a dream.

Any comments or suggestions are greatly appreciated.

Thank you.

Number-crunchers will tell you that the idea of keeping the master at 0 and reducing faders to compensate is no longer necessary in today's hi-res digital age, but others will swear they can hear a difference. So, why not be on the safe side?

As to Live, an important subtlety is that quality goes down as soon as its stretching engine kicks in. In other words, being in a loop recorded at 120 BPM with Live set to 120 BPM, and it will sound just fine. Change Live to 124 BPM, and the stretch engine has to kick in.

For "mission-critical" Live projects where fidelity has to the best possible, I stretch the original loops in a different program to Live's project tempo using the highest possible stretch quality (e.g., the iZotope algorithm in Sonar, or use Rex or Acidization), then bring those loops into Live. Because Live doesn't have to do any stretching, the sound quality is much better.

I wrote an article about this for Sound on Sound, can't remember the issue but it was about Live techniques and appeared within the last six months or so.

Yeah I've read a few places and on forums how the stretching engine in Live will lower the sound quality... that makes sense. However, does that mean if just 1 loop is being stretched that the entire engine will react and sound worse? Or just if say the master volume for everything is changed?

Since your kind enough to put your .2 in....

What do you think about all these new analog summing devices? Honestly the extra work to use seems like a big yawn.

And I know this is an entirely different subject but since it has to do with mix quality... what about harmonic exchiter plug-ins like the BBE Sonic Maximizer?

Say I have a kick drum and I love its sound but it needs more bass... would you eq some in or use something like the BBE? Or are these both b.s. fixes that are just making mud and I need a different kick drum?

I mean I know in the end its just the sound that matters and its more about the song and arrangement...

No, it should affect only that loop.

Theoretically, it shouldn't. 32-bit floating point is a lot of resolution. But you never know...

>

I think they are a sophisticated form of analog dithering. Analog devices are processors, no doubt about that, and some people like the sound of that processing while others don't. Just add some dithering to your digital projects and save yourself a few thousand dollars

>

They were much more relevant in the days of analog tape, where progressive erasure of high frequencies tended to remove "sparkle" from the overall sound.

I'd get the kick sound as close as possible without doing anything, then add EQ.

>

100% correct, but a great song and arrangement sounds even better when the recording's great!

Well, it's one thing for a floating point DAW to gracefully handle "overs" which are corrected at the buss before output to unforgiving real world 24 bit converters...

But that doesn't necessarily necessarily apply to all of the plugs you may be overdriving either in an individual track or across the master -- a scenario where it's easy to imagine overdriving. The plug may be minimized and any metering or clip lights on the plug interface invisible.

And even if a plug-in uses floating point, there may still be negative consequences, performance/effect-wise, since it seems likely many or most plugs are optimized for sub 0dB levels.

I'll admit I hadn't really though much about any of this (odd to think) until recently when the topic of proper gainstaging came up here or maybe at another board I also frequent.

I think if you are starting with the kick hitting zero, you are pulverizing the mix bus. You should be able to look at the waveform of one of your mixes and see if they are clipping.

I used to do the same thing in Acid. I could mix a tune that was like 5 times as loud as radio commercials. Thing is, the distortion and "funkyness" you refer to is subtle.

Lower everything and try to spit out mixes that top around -6. Leaves room for mastering, dynamics, and I think (playback) converters respond better to moderate level material than the hyper-compressed stuff we hear today. Sometimes it's cool, but not all the time.

One thing to remember (which gets changed a lot since computers serve all these functions) is to try and remember to set your volume level consistently when you mix.

Hopefully in the future, software will display the various signal levels throughout the software (on one screen) so that proper gain staging can be achieved (what with all the plugs, etc).

Hi "Stranger",

You think I should start with the kick at -6 ?

Even if I'm making a dance track?

I guess the thinking here is the overall level will go up and also it's ok to do because Im in 32bit more so everything is sitting above 16bits anyway?

Thoughts?

Thanks!!!

Even if you're recording a polka band.

You can always make things louder after you're happy with the mix, but you can't tell much about what's happening with your levels if the meter stays at full scale all the time. With a decent 24-bit system there is never a reason to worry about levels being too low unless it's excessively low (like never gets above -20 dBFS). But bad things can happen if the level is too high, and you actualy have to LISTEN in order to hear that.

Admittedly, a waveform that peaks at -6 dBFS doesn't fill up much of the screen, so you might not think it's loud enough. Don't let that bother you. You can always turn up your monitor volume if you want to hear it louder, and you can always "master" the mix to bring up the peak level without clipping once you have everything in the mix.

OK OK!!!!

So I just got back into my studio since I posted the 1st question about all this and I loaded up a song I am working on...

With this song I had the Master about 12db lower than 0 and some individual channels approaching zero.....

I now changed it... I lowered the Channels, the kick being the loudest to about -10db and placed the master at 0db....

Guess what???

MAJOR IMPROVEMENT!!!

Amazing... Im really suprised... in fact its a very large improvement in clarity!

So what the hell does this mean? None of the individual faders come to zero but they accumulate to it...

Its very noticable in Ableton... maybe Steinberg is going some tricks (even maybe just visual) so people can mix precievable hotter.

Some of my channels are not only hitting 50% up on the fader... I guess this is ok?

I think part of my thinking was too old school! I thought if I lowered the faders Id be loosing bits... time to reprogram my old self!

I use Steinbergs Nuendo.

Peaks on individual channels are between -15 and -12.

Masterfader is allways on 0.

That's how Phil and Craig taught me and it allways works.

It's really just basic summing. Whenever you combine several signals together, the result is going to be a louder signal going thru the next stage. On an analog console you'd adjust the channel gain so that the fader could sit at 0, mainly to minimize the audible effect of the fader itself, i.e. it would be more or less a 1:1 passing of the signal. With digital gain the effect of each stage is less obvious, but the basic principal remains.

I've taken a different approach with Live, and now I'm curious if there's any difference in output quality. I use the clip's gain adjustment to drop the signal, much like one would adjust the preamp & then mix with the fader. This has a good & a bad side: the problem with changing the fader drastically is that you now have less room to mix with; but the problem with using the clip gain is that you have to set it on individual clips, even in the same track. Not necessarily a bad thing - especially if you are using different source material on the same track, then it's a must - but it can be more work. Last night I was working with some drum fills from a loop CD, and after deciding where I wanted them to sit & doind a little slice & dice, I had to adjust the gain of each clip to match (I should have set the gain first, I guess... )

Craig, any idea as to whether adjusting the clip level or the channel fader is better, or if it matters at all?

"Whenever you combine several signals together, the result is going to be a louder signal going thru the next stage."

Whats funny is for some reason I didnt think that would be the case but then again if you put a bunch of ingredients in a bowl to mix them you need a big bowl to hold all the little parts...

I'm actually finishing songs for a new album and man I'm so happy to have realized this... hah.

As one is sorting through this it is CRITICAL to understand how your DAW handles 0 dBfs...

On my DAW, Sonar, per my highly informal and limited tests, the DAW itself gracefully handles overs at one stage that are subsequently corrected... presumably because it uses a floating point schema, as opposed to fixed point math. Presumably.

In my test, I took a signal that peaked at -1 dBfs and jacked the channel trim 6 dB. When I bounced that track through the master with the master bus fader at unity gain, that produced the squared off, distorted overs you would expect.

I then lowered the master fader a compensatory 6 dB and bounced again.

The resulting wave form was visually identical to (down to sample level) the original.

Now, I don't know how a fixed point math DAW would handle that. Perhaps some fixed point DAWs have a way of handling minor overs; perhaps not.

But -- crucially -- EVEN if your DAW handles 'corrected' overs gracefully, that doesn't mean that your bus plug-ins will... if you hit them "too hard" it may well produce undesirable effects -- even if the plug uses floating point math, since it is presumably optimized for dealing with sub 0dB levels.

Another satisfied customer.

(I can't get on to the boards as UstadKhanAli anymore for some reason, so I am now reinvented, with no posts and no avatar).

Originally posted by AfroKen Another satisfied customer. (I can't get on to the boards as UstadKhanAli anymore for some reason, so I am now reinvented, with no posts and no avatar).

Weird yeah,.. I had the same thing today,... I had to get a new password and got two emails back,... one for boosh and one for boosh2 ,..... Are we in a parallel Universe??

I got confused when I read your post here because I've been calling Ustankahnali ken for a while and thought I'd been talking to the wrong guy when U showed up,...

Actually -- just to be irritating, since we hopefully all already know this -- that quote in the post above is NOT actually true.

But your analogy DOES have some parallels...

What you will get, in principle, is THE SUM of the two signals.

But, as we all know, signals have what we define as "postitive" AND "negative" dynamic values. (Vaguely, somewhat indirectly analogous t your mixing bowl analogy, we may presume that SOME of your ingredients have an "air component" which may collapse with the addition of water, or simply with crushing/mashing/mixing.)

In the extreme -- if you combine two identical signals you'll get double the volume (dynamic value), assuming you have the headroom to hold it. But if you combine a signal with an exact inverse of itself, you will get dead silence.

Hmmm makes sense... and good actually because there lies some of the magic.

You're right, blue - you're being irritating.

Actually that's true, yes - but for sake of simplicity & the point of this discussion I opted to avoid that. It's quite relevant for mixing several tracks of the same source together, but somewhat less so for combining several different sources, i.e. loops/instruments recorded at different times in different places, etc. Not completely irrelevant, of course - cancellation & buildup of frequencies is a definite factor, yeah. But that's covered later, in Mixing 102.

BTW, I posted my question over on the Ableton forum, and someone has told me that the clip gain is first in the chain, which does basically agree with my view that it is akin to channel gain. (One consideration is that the fader is a post-FX adjustment, so adjusting the clip level will have an effect on devices inserted into that channel.)

Maybe I can test it a bit later & see if there's any appreciable difference between radical gain adjustments in one stage or the other.

Originally posted by franknputer You're right, blue - you're being irritating. ...[snip]

My daddy just said, whatever you decide to do, son, just make sure you do the best you can at it...

I don't know if other DAWs have the same type of documentation (you'd like to think) but Sonar has some handy signal flow diagrams that show potential routings through the DAW, insert points, signalflow switch points, meter points, etc. (Don't forget many contemporary DAWs have somewhat flexible routings, so your particular settings may well be a factor.)

Originally posted by Anderton > I think they are a sophisticated form of analog dithering. Analog devices are processors, no doubt about that, and some people like the sound of that processing while others don't. Just add some dithering to your digital projects and save yourself a few thousand dollars

Craig (and anyone else who wants to chime in), do you add dithering to your ITB mixes? If so, how do you apply it exactly? Are you doing something different from simply applying dithering when bouncing to disk? Thanks.

I usually bounce down from Ableton 24bit then load the file into Cubase SX3 and sometimes use Izotope Ozone for minor mastering then using the Apogee plug-in to dither. I also insert the Apogee on the master at the end of the inserts (I forgot off hand if its pre or post whatever... )

If anyone thinks there is a better way I can do things let me know.

I know you know Craig's being whimsical, there, about analog board noise as a form of dither [ *] , but some of our less experienced or less-English-familiar friends may start getting confused in here, somewhere...

* OK... I got dibs on the idea of a dithering plug 'modeled' on analog board noise... it's mine... I'm claiming a People's Patent on it, right here... can't you just read the marketing copy... Based on the noise profiles of famous high end mixing desks... blue2blueDither imparts the warm, analog...

Well, I was actually being serious as to whether he was adding some sort of noise or distortion in his ITB recordings! The reason I asked was because quite a few years ago when I was using SAW+ as my DAW I was doing a mix and, as a complete joke, added tape hiss. The band thought it was funny, but oddly, people thought it sounded great and we kept it in!!!

And yes, I know he was being whimsical about the analog board analog dither...