The analogue-summing '3D' effect: what actually is it?

[Dr Huge Longjohns]

People who espouse the use of external analogue summing mixers often talk about the '3D' effect that these devices create. Somehow the individual instruments or sounds seem to have a space around them that's hard to describe but once you hear it, it's very attractive and hard to miss. Airwindows and other software manufacturers claim to be able to recreate this effect in the box with console emulations and so forth. Here's the Airwindows example which demonstrates it very well if you listen on monitors or phones:

https://www.youtube.com/watch?v=r8eSFNgSpuo

And this piece of software, DrMS, seems to achieve the same with a variety of processing, particularly phase shifting and cross-talk manipulation.

https://www.youtube.com/watch?v=u7TcIVLJN3o&t=677s

I have been able to achieve this effect in Logic Pro X simply by loading Waves Kramer Tape on the stereo bus but it's always hit and miss whether it works or not.

But what is the actual science behind this effect? It's certainly not "digital summing is broken" as some commentators would have us believe. What is going on to make this magical 'space' appear?

[muzines]

If it's "magic", it can't be explained by science!

[Sam Inglis]

I only skimmed through the Airwindows video, but I didn't hear anything magical. I heard mild saturation and a consequent added mid-range content, and small loudness variations. In fact in the very first example where he compared Console 4 to the raw digital mix I preferred the raw mix.

[DanielBeach]

This may be a naive view, but isn’t it possibly just the side effect of each signal having gone through physical connections which can’t help but to add some noise/artefacts/whatever, and therefore colours the sound in some way, boosts some frequencies a little, etc?

[CS70]

Think the common wisdom is that it's about crosstalk between channels. Tough that should _decrease_ separation, not increase it...

No idea however. I just mix it until it sounds good to me, problem solved

[MOF]

I’ve bought the BX Focusrite Console plugin which has TMT in it to emulate the effect of using separate analogue channels, with their random slight variations of eq, phase shift etc, and to be honest if there is a change I can’t hear it so far on the few songs I’ve used it on but maybe you need to get to 24+ channels to hear it; there’s always a slight crossfade/delay when you switch off lots of plugins simultaneously to compare, so without doing several level compensated mixes and cutting between them it’s hard to judge, whatever is happening it’s subtle. Anyway I do like the sound of its eq, compressor etc and that’s the main reason I bought it.
I would say that trying out a Mid Side plugin and experimenting with the width and eq’ing the difference channel will give you more of that effect you want.

[CS70]

One more thing, in Cakewalk you have ProChannel in every track which includes channel and bus emulations. They're supposed to emulate SSL, Neve and API consoles.
I do hear what they do quite clearly and I like it, so I use them quite extensively.

However, having never owned an SLL, Neve or API desk, I have no idea if they actually sound like them

[Dr Huge Longjohns]

I only skimmed through the Airwindows video

At 2 mins in he switches it out. The mix suddenly sounds flat in comparison to me. Is this just the effect of losing the saturation?

[Dr Huge Longjohns]

each signal having gone through physical connections which can’t help but to add some noise/artefacts/whatever, and therefore colours the sound in some way, boosts some frequencies a little, etc?

Yes, I guess it's precisely this sort of thing, plus the cross-talk, phase anomalies etc. But I've never seen this '3d' effect explained properly in terms of what actually happens to the signals. Or as Sam says, it may be just our ears are getting fooled but simple saturation and volume changes.

[resistorman]

I’ve bought the BX Focusrite Console plugin which has TMT in it to emulate the effect of using separate analogue channels, with their random slight variations of eq, phase shift etc, and to be honest if there is a change I can’t hear it so far on the few songs I’ve used it on but maybe you need to get to 24+ channels to hear it; there’s always a slight crossfade/delay when you switch off lots of plugins simultaneously to compare, so without doing several level compensated mixes and cutting between them it’s hard to judge, whatever is happening it’s subtle. Anyway I do like the sound of its eq, compressor etc and that’s the main reason I bought it.

Same here, though I bought the SSL one for $49 when it was on sale. I use it on all my tracks now to start the mix off. And like cs70, never having owned an SSL, I don’t know how it compares, but it sounds great.

[Hugh Robjohns]

Think the common wisdom is that it's about crosstalk between channels. Tough that should _decrease_ separation, not increase it...

Not if the crosstalk signal has a lot of phase shift in some frequency regions and thus makes the image width wider rather than narrower.

And I suspect even-order harmonic distortions probably play a part too.

[Dr Huge Longjohns]

If you get a chance to listen to that Airwindows clip, Hugh, I'd love to hear your view on what's causing the sonic differences.

[CS70]

maybe you need to get to 24+ channels to hear it

It does certainly need to "sum up" a bit. Maybe not 24 channels, but the effect should be very subtle (after all, the original desks were built as usual to be as transparent as possible, or at least not with the current "let's emulate analogue" fixation) so it should get noticeable only with more than a handful of channels.

There's also the much more immediately noticeable "driving" of a channel's emulated preamp, which pops out right away with even a single channel (because you're distorting).

[CS70]

Think the common wisdom is that it's about crosstalk between channels. Tough that should _decrease_ separation, not increase it...

Not if the crosstalk signal has a lot of phase shift in some frequency regions and thus makes the image width wider rather than narrower.

Good point, hadn't thought of that.

[Martin Walker]

I only skimmed through the Airwindows video

At 2 mins in he switches it out. The mix suddenly sounds flat in comparison to me. Is this just the effect of losing the saturation?

JUMPING ONTO ONE OF MY FAVOURITE HOBBY HORSES

Yes - this is exactly the hardware 'analogue summing' experience Chris has attempted to mimic, and is exactly what I hear too, having spent a long time working with his Console plug-ins.

Like you I could hear the switch from 'space round each individual sound' to flat, as heard in that mix (and I've since heard exactly the same differences between ITB and hardware analogue summing demos - once you lock onto how the sound is changing you can invariably hear it in future summing examples, whatever the gear involved).

However, at first I found the Airwindows Console benefits annoyingly elusive in my own mixes, and spent a long time exploring why, eventually writing up the findings on my web site:

http://www.yewtreemagic.co.uk/soundworlds7.php

The reasons I had difficulties? Chris' example of himself singing a single note around 20 times proved to be the perfect subject for revealing the difference, because:

a) Console 5 summing only became audible to me once there were at least a dozen tracks being mixed.

b) Console 5 provided vanishingly small audible summing improvements until the combined peak level of the tracks came close to 0dBFS, and since his singing was of a sustained note, he could optimise this drive level for maximum effect.

My initial demos were of more percussive music and around eight tracks, neither of which gave significant improvements, only occasionally being almost vanishingly audible under headphones, but as I explored more I began to get the idea, and finally began to reap the benefits, essentially by mixing together more channels with combined transient peaks around -2dBFS (going much higher ran into nasty distortion, so with Console 5 it proved to be a fine dividing line between maximum effect and overload).

I published my results, Chris himself was impressed by my efforts and confirmed all my findings, but just a couple of days later he released Console 6, essentially the same technique but using a different algorithm that kicked in at significantly lower audio levels, making it less necessary to push levels so close to 0dBFS to hear the improvements.

You can hear my audio demos via the link above - I offer the same short musical section three times; the first being standard ITB digital summing, the second using Console 5 carefully set to peak at -2dBFS, and the third being Console 6 with -6dBFS peaks.

Now to the audible differences (which I can still hear quite easily - see if you agree )

The digital summing does sound flat, while Console 5 and Console 6 open this up - listen in particular to the voices in each version. To my ears the percussive voices sound more focused, giving the effect of being to discern each 'singer' more clearly, while the sustained voices sound significantly further away in the Console versions, giving the effect of more 3D 'depth'.

Finally, to the reasons for these changes. The Airwindows Console plug-ins add slight 'pre-distortion' to each channel via its Channel plug-in instances, and then the combined mixed signal is passed through a single Console Buss plug-in, which provides the reciprocal 'undistort' curve:

Now, with a few channels, the 'difference' signal is inaudible, but with a dozen or more channels peaking a few dBS below 0dBFS, some of the initial distortion applied to the peaks of the individual channels gets left in, mimicking the effect of a non-perfect mixing buss, where ambience gets pushed into the far distance. The actual sounds don't seem particularly altered, but the space between the sounds gets hammered with the full 'undistort' treatment, which sounds very distant and thin all by itself. This (to my ears anyway) gives slightly more 3D 'detail' to them i.e. more depth and a sense of enhanced "foreground", which is the effect commonly claimed for high-end analogue mixing desks.

Believe your ears or not as the case may be, but I can reliably hear the difference, and have been using Console 6 on every track I've written since exploring the possibilities.

Martin

[Drew Stephenson]

I did try Console 5 on three of my mixes but ended up removing it from all of them. I could hear a difference but to me it didn't add any magical space and separation, just additional mud (as if I'd stuck a saturation plugin on every channel funnily enough). It also seemed to make the lead vocal narrower in the mix.
I preferred the way Gullfoss enabled you to get that kind of additional separation and space but not enough to spend £150 on it (and it also did that vocal narrowing thing).
Perhaps my recordings are too muddy to begin with and it's just highlighting that fact, but I've never run anything through an analogue summing device of any kind so maybe I'm just expecting the wrong thing.

[Dr Huge Longjohns]

Yes - this is exactly the hardware 'analogue summing' experience Chris has attempted to mimic, and is exactly what I hear too, having spent a long time working with his Console plug-ins.

Fantastic, Martin, I remember you saying you were a big convert to Console 5 in another thread. Thanks for such a detailed reply, I will do my homework on your research tomorrow morning! (Something to get up for, for a change.)

[Dr Huge Longjohns]

Thanks for your article Martin, most interesting. Will do some C5 experiments today. I downloaded it a while back but, like you, could never get the 3D thing going and saw it simply as adding a bit of saturation mojo etc.

Question, though. I run a pretty weedy Mac and have to freeze (render to audio) a lot of tracks on even fairly small projects. Does C5 etc still work if you've frozen a track or does it rely on live interaction with the buss plugin?

[Dr Huge Longjohns]

so maybe I'm just expecting the wrong thing

Once you've heard it there's no going back! As I say, I've achieved a similar effect with a tape emulation but it doesn't always work.

[ManFromGlass]

Sometimes I’m a little slow -
Martin a few questions
- Console 5 or 6 should be the last plugin on the Master bus?
- The plug works best when the track isn’t swimming in reverb?
- something like a jazz trio may not benefit as there may not be enough tracks for the plug to trigger?

[Dr Huge Longjohns]

There are two parts to the plugin, the channel and the buss. They suggest you put the Channel plugin as the final plug on each channel, then the buss plugin is first on the stereo out.

[Martin Walker]

Question, though. I run a pretty weedy Mac and have to freeze (render to audio) a lot of tracks on even fairly small projects. Does C5 etc still work if you've frozen a track or does it rely on live interaction with the buss plugin?

You do need the interaction of channel and buss plug-ins, but you should be able to freeze your tracks without the Console Channel plug-ins in place, and then add them afterwards as the one and only plug-in on each channel, with the channel faders all at 0dB.

Martin

[Martin Walker]

Sometimes I’m a little slow -
Martin a few questions
- Console 5 or 6 should be the last plugin on the Master bus?
- The plug works best when the track isn’t swimming in reverb?
- something like a jazz trio may not benefit as there may not be enough tracks for the plug to trigger?

1. Huge has already answered this correctly - console channel plug-in last on each channel, and the console buss plug-in first in the master channel.

2. Not necessarily, although since its aim is to provide extra detail you'll probably hear it more easily if your mix is swimming in reverb

3. Exactly - I'd say a minimum of a dozen tracks to make it worth all the effort.

Martin

[Dr Huge Longjohns]

Martin, I think I'm confusing you with Logic terminology, sorry. In Logic a Freeze is a temporary rendering to audio which saves CPU. You can freeze pre plugins but it's normally the plugins that are using the CPU of course. So what I would have to do is fully render the tracks to audio using what Logic calls Bounce in Place then add the C5 plug on the audio produced.

But if all the C5 channel plug is doing is adding some additional signal, shouldn't this be incorporated into the 'frozen' track like any other effect ie if it's not actually 'talking' to the buss effect in some way? Rather like good old Dolby!

This is fairly trivial to test, actually, I'll have a play around tomorrow.

[Drew Stephenson]

Question, though. I run a pretty weedy Mac and have to freeze (render to audio) a lot of tracks on even fairly small projects. Does C5 etc still work if you've frozen a track or does it rely on live interaction with the buss plugin?

You do need the interaction of channel and buss plug-ins, but you should be able to freeze your tracks without the Console Channel plug-ins in place, and then add them afterwards as the one and only plug-in on each channel, with the channel faders all at 0dB.

Martin

Hmmm. Maybe that's where i was going wrong.