Analog and digital recordings...... making them sound better..... (From Gearslutz....)

The Reason Most ITB mixes don’t Sound as good as Analog mixes. This is a repost from another thread. Hope you find it usefull.

Ok, I'm going to try and give you An ITB education, as my over 24 years has taught me. Here is what I try and teach to students. I'll try and keep the math to a minimum.

First, I own a high end analog setup's Via an SSL 4K with 1/2' 2 Track YADA YADA, ICON with Killer OB FX And classic Compressors, YADA YADA/ Hybrid Setup Via AWS 900w/ 24 Channels Of Xlogic Killer OTB FX and Comps YADA YADA. Point is not to impress, or brag in any way, but to let you know everyday I work on a verity of systems. This has led me to The Following conclusion.

To learn to mix ITB coming from an analog world you must revisit what Voltage reference Analog consoles work at, and make appropriate adjustments to translate this to work ITB.

The first thing we must ask is simply what is 0VU. What does it mean to us. Lets use an SSL G+ as our point of reference mainly because I work on those every day. If we put a signal into the line input of the SSL so the channel meter reads 0vu, that also, is referenced as +4 or 1.23 volts. A kick ass SSL will go out to about +24DB, so we have approximately 20 DB of headroom above the 0 VU point on the meter before the signal goes to crap.

Now let take a common situation. A Client hands you a Protools session and you spread it out over the SSL console. Like most people today every track is recorded as hot as hell. Most pro Eng's will use proper gain staging and get the now slammed meters reading around 0VU or 1.23 volts. By lowering the line trim we now have a good level into the desk so we can Compress/Gate/EQ the Signal without it overloading the processing. Sounds simple right? Remember that all outboard equipment was designed to work around the 0VU/+4/ 1.23 Volt reference. So by putting the incoming signal at around this reference, your rack equipment will work better as well.

Why use a +4 reference? Well remember that the 1.23 volt reference came from the tube days where 1.23 volts was enough voltage over the plate noise that you still had a good signal to noise ratio, but still left room above 1.23 volts to allow for normal audio operations.

Now to ITB. Lets pretend we have the same setup as we did on the SSL. Client hands you a session that’s recorded hot as hell. Now most folks mixing ITB don't understand reference levels when relating it to Digital. To have the same amount of "headroom" as we do on the SSL we must create a reference of 0VU or 1.23 volts at -20 from 0DBFS or the top of the Digital scale.

So if you simply place the good old trim plugin as the very first plugin, you now have the ability to adjust your tracks to our Mixing (+4/1.23 volt) reference IE -20. Just like you did on the SSL. You have have the same amount of headroom. Now with your tracks properly gain staged, you can add EQ/dynamic plugins and not run out of headroom. You can also insert hardware and they will operate much better as they are operating at the level they were designed to operate at.

Plugins use the same reference at real equipment. Never try and drive them to the top of the Digital scale. Don't try and make your mix look like a master. You don't do that on an analog console, so why do we do it ITB?

The answer is simple. DAW meters suck Butt. There should be a meter mode in all DAW's that makes the meter at 3/4 scale equal -20 at 1.23 volts. Just like the old VU. This way, novices will quit corn-holeing their levels.

Something to think about. The noise floor of an analog desk is about -75 DB from our +4 reference. Our equivalent "problem level" below our -20 reference in digital is well over 100 DB. So please don't let people tell you analog has more "headroom" than digital. This is simply not true. Headroom is only relative to your noise floor below your reference. Remember if the volume is to low, turn up the darn speaker volume.

Running a Digital mix right to the top of the scale is like running your SSL mix buss where the VU meters are slammed all the way to the right and you are constantly hitting it at +25. No one will get a good sounding running the desk like that. You won’t get a good sounding mix in digital either.


So what does all this mean? Put simply, proper gain staging is essential to both analog and digital mixing. You just need to correlate the references between the two. Once you figure this out, I'll Guarantee your mixes will start to sound open and wide, just like the good old analog days.
__________________
Regards, Skip Burrows, Chief Whatever

Just wondering if the original poster of that info realizes what "ITB" or "IDB" is short for anywhere else on the internet.

and yes, I'm 13 years old

MIght want to read this thread too:  http://recforums.prosoundweb.com/index.php/t/29301/15910/
 
It's not very long and there's some great info about levels and what they mean in relation to one another (RMS vs. dBfs vs. dBVU, etc.).

Gain staging is the most over-looked and yet essential element in any recording chain.
 Improper gain staging will make mixing a nightmare that never ends, and frankly you will know it when you get it right.

Nice post.

The OPs quoted post seems like it is from 1999... is this back again?

But again - it sounds GREAT and does not sound clipped or distorted.

If you re-import the exported file, I wager it will not be clipped. Something  in your mix config attenuated the level before it got exported.

hat the heck does a master LOOK like?

These days it tends to like a solid wall of sound with no dynamics. 

Da=man


I really think a lot of this is waffle. I could be wrong but I still think there is nothing wrong with running hot levels as long as you aren't clipping.

mike_mccue


The OPs quoted post seems like it is from 1999... is this back again?

No waffle, no time travel.
 
Gain staging was typically simpler when all gear was analog, but still mis-understood and done wrong.
 
Even now, with digital recording a mature paradigm, gain staging is still not done right.
 
It's simple if you remember that when recording to tape, anything above 0 VU was clipping to some extent, and with digital recording anything above your reference level - where analog 0 VU is displayed as -18 (or-so) VU - is in danger of clipping.
 
Even with 24-bit recording, if you record (or sum multiple tracks) up against -0 dBFS (over which everything is digitally clipped), you are not going to get a recording that sounds better than if it was recorded with peaks no higher than -20 dBFS. Nyquist theory. OK, if you peak at -60 dBFS, room noise, cumulative analog electronic equipment noise, etc., might mask some important stuff and low-leve stuff might run into the actual 24-bit noise floor (you'd have to try real hard to make this mistake, though).
 
With 16-bit recording, it's a lot easier to hit -0 dBFS, create intersample peaks that go over, and hit the noise floor on the bottom. So, it's important to set your peaks a little lower, and watch your dynamic range so important low-level stuff does not get masked by the noise floor.
 
Remember that even if you record each track with lots of headroom, add up eight of them and you have to lower each enough so that you are again hitting your target peaks - add sixteen or twenty four tracks, or plugins, or analog inserts with D-A/A-D conversion, etc., and you have to watch them even more carefully.
 
There are of course times when, once levels are under control, that gain stages can be manipulated for effect - some plugins and analog gear sound different/better when driven hard or soft, tape recorders can clip in a euphonic way, some mastering engineers clip their D-A instead of dynamic limiting for effect, etc. - but you can't do this stuff until everything else is at the right level.

Oh boy, it's the old intersample peak story again.

:-)

I'm just now realizing all that analog gain staging I used to do, still do, and look forward to doing in the future is "mis-understood and done wrong".

:-)

this post reminded me of something i read online circa 2001. it was a post from a "known good" source and referred to analog mixing but applies to digital as well. the post was simple. i paraphrase:
"do this. make a mix like you want it, record that. now turn every channel down 6 db and record that. listen to both and pick which ones sounds better. hint: it will be the one thats quieter"

i did it. granted, at the time i was recording and mixing mostly rock music. and it freaked me out.

brundlefly

Even with 24-bit recording, if you record (or sum multiple tracks) up against -0 dBFS (over which everything is digitally clipped),

Bzzzzt! Wrong Answer! The only places anything clips in a DAW are at the A/D input, and the D/A output.  You can sum to your heart's delight, and so long as you pull down the Master bus I/O levels to get the output to the DAC below 0dBfs, you will not get clipping.

you are not going to get a recording that sounds better than if it was recorded with peaks no higher than -20 dBFS. Nyquist theory.

What does the Nyquist "theory" [sic] have to do with levels?

Correct about the DAW, especially SONAR's 64-bit processing, not clipping, but with enough hot tracks summed, even in SONAR, audible distortion can be heard if you don't pull down the master or bus fader (or address the gain staging in some other manner). And this mistake is made by many, experienced or newbie. I'm saying this to support pulling down each track, in SONAR with Trims, so that the summed tracks hit the master fader at something like -18 average and no higher than -6 peak.
 
Nyquist theorem (pre-first cup of coffe, shut up :-) ) Recording at 44.1, for example, stipulates a 22.05 Hz upper frequency limit, imposed with anti-aliasing filters. If you record high frequencies hot, these filters can sound bad - better to record high frequencies at a reasonable level, or not to approach 0 dBFS. And the noise floor is caused by quantization error, and this error is much lower - theoretically (theoremically?) -144 dBFS in 24-bit as opposed to -96 dBFS in 16-bit. This noise is not heard the same as the noise floor of tape or analog gear, but it is heard, influencing the sound of low level stuff. So, recording and summing levels have a lot to do with basic digital audio recording math.

Well, as I recall from the Izhaki book, he contends that with 32 or 64 bit floats being fairly common as an internal representtion of audio with modern DAWs, the only place the any of this matters very much is at the output bus. I assume what he's saying is that even though your meters might be hitting 0dB fairly often, inside the DAW the internal numerical representations are still tracking the actual magnitude of the waveform, since they are not numerically limited to 24 bit integers. These large values only become a serious issue when they get fed into a D/A where they can be summarily truncated into clipping. I haven't thought through what this might do to the dithering process, but usually that would follow a master bus limiter so it may have no impact.
 
I should emphasize though that this discussion is confined to mixing and not recording. Keeping levels conservative during recording makes all the differnece, and 24 bit recording gives us a ton of headroom to record at moderate levels and still capture transients accurately.

I'm not sure I totally agree with this "damn the meters, full speed ahead" theory, since there may be some plugins that don't follow the same resolution as the host DAW, also there are some plugins that definitely behave differently when you drive them harder (ie amp sims). In general, though I think he makes a valid point.

 

feedback50


I'm not sure I totally agree with this "damn the meters, full speed ahead" theory, since there may be some plugins that don't follow the same resolution as the host DAW, also there are some plugins that definitely behave differently when you drive them harder (ie amp sims).
 

Bingo!
 
This is why I will never clip anywhere inside the DAW no matter how much "headroom" 64 bit processing has.
 
Once you start bringing plugins into the mix, all bets are off.  So it's better to keep levels safe at ALL times. 
 
Like they say, "better safe than sorry".

I don't know what "ITB" and "IDB" is short for on the rest of the internet. Tell me.

ITB = In the Box
TDB = In Da Box
Cj

Most of this is above me. I'm just trying to learn how to record and mix using my ears instead of my eyes.