crosstalk and other console emu things

[kensuguro]

I'm still deeply into the console emulation genre of native plugins. One thing I don't quite understand is crosstalk. As a physics phenomenon I get it. Signal from one path bleeds over to another nearby signal through electromagnetic interference. (or similar) And by having copies or facsimiles of a signal routed in random directions, they cause cancellations and other effects when summed. That combined with the EQ variance between each channel seems to account for the widening effect. (3D sound, etc, of spatial perception) That's about as much as I've picked up from plug ins that exist, and reading their sales blurbs.

What I don't get is, how does one go about "designing" such a thing? Because at its very basic, these are not only imperfections, but to me sound like offensively self defeating malfunctions of a system whose job is to route and sum signals as transparently as possible. (or not) Aside from that, because the resulting sound relies on a complex network of cross interference based on input, how is it possible to even anticipate a result and sculpt a certain characteristic while designing or modifying a console?

It seems like with actual consoles, whether a particular instance (not model, but specific console) becomes magical or not. Some consoles seem chosen, and get work done on it by masters who modify it further to make the particular console a unique gem. And since these things happen partially by chance, they are not mass produced or replicated in any predictable way. (hence emulators going back to study specific consoles) So then, the question is, are these things even "designed"? Or is it more the work of mother nature? It's blurbs like "we went back to study channel XX on console YY that Mr. ZZ owns" that make me think that there is no mastermind behind these sounds, a product of pure chance.

Personally, I prefer not to rely on such happenings of chance. I think it's much more meaningful to learn from these natural occurrences and learn the mechanism, the "how", and create something new from that instead. So instead of Audio Acustica Nebula approach, which goes and samples things, I'd prefer the Redline Preamp or LVC-Audio PreAmped approach that are based on a generalization of how the console produces the sound that it does. It is a difference between an literal snapshot versus a parametric model that can be explored. But still, I feel these models are limited in how much of its inner workings is exposed to the user, so I was trying to think of ways to model the console nonlinear behavior so more of it could be revealed and controlled effectively. I dunno if I'll every actually build it, but still fun to mentally explore.

[jksuperstar]

So then, the question is, are these things even "designed"?

I'd say shielding, isolation, and impedance matching are all done specifically to reduce noise...which is defined as anything other than the signal itself. There's no way you are shooting for a 120db S/N spec, only to design in some extra crosstalk. That's not to say there isn't the occasional genius that analyzes and places routes of busses on a board using specific distances to not only cancel, but encourage only specific harmonics. (this statement does NOT include everyone claiming the virtues of the golden ratio instantly makes a design better.)

To me, it sounds a marketing game of "how many bullet points of features can we add below our plug in so we can say we did something above & beyond what the other 50 emulations of Rupert Neve's designs did".

Sometimes, I wish Kurt Vonnegut was still alive and writing reviews of these exceptional human achievements.

[garyb]

yeah, people are playing with computers more than audio these days...
i agree with jksuperstar.

[dawman]

It's in our nature to destroy ourselves.

Arnold in the Terminator
1988 MGM/United Artists

[dante]

I say if the developer plays with the computer, the audio engineers doesn't have to.

For example, playing with the DAS Comp, the emu of a lamp mimics the non linear imperfection that was introduced into the original.

Opto is a way of applying a more natural curve to dynamics based on the way a light bulb dims and brightens.

Someone found out that this phenomenon had a desirable musical effect when built into compressors.

I never used the original or indeed any opto compressor, but now I know the blurb I can explore it without knowing the math of a light going from dim to bright.

So any imperfection that can be measured and expressed mathematically can be modelled, tge more musical it is the more likely to be attempted by a developer.

Ken, check out the 'bleed' switched in Toontracks EZDrummer - theres your crosstalk.

[garyb]

well, that's more mic bleed rather than crosstalk between channels. the same but different.

the optical compressor is about an optical sensor. the signal is changed into light, the bigger the signal the brighter the light. then there is a sensor which measures the amount of light, and which lowers the level based on the state of the sensor. the thing that is neat about this is that when the source material has a lot of peaks, the sensor overloads, it can't respond to rapid changes since the sensor needs a moment to reset. modeling this is different than modeling crosstalk. crosstalk's effect is different in every circuit, and it usually is not a good thing AT ALL(the best consoles limit crosstalk as much as humanly possible). the optical sensor is the same in every circuit since the response of the opto sensor is always pretty much the same.

most of these types of features are snares for the guy who has more money than sense. since he doesn't know how to get the desired result, he figures that just buying more software should make things sound right. the truth is that knowing how the tools work and why they are needed has more effect on getting a good sound than adding crosstalk ever could. native mixes just generally sound bad compared to real hardware, so companies are going out of their way to convince consumers that the situation has suddenly changed. i've heard people using stock Cubase SX1 plugins get better sounding mixes than folks with the latest greatest, just because they knew how to produce music and use the tools. technology cannot replace skill and experience.

emulating crosstalk is a cool concept, but it won't make crap into diamonds. i'm sure it will sell plenty of software.

[dante]

Agreed bleed is not exactly the same as crosstalk, but thats what I mean by developer will only work on whats worth doing, so Toontrack took up a model of bleed thats not bad. I always turn it off, since I prefer to control the mix myself in STM and having bleed turned on means loosing seperation at the mixing console.

Crosstalk, you probably wont find modelled, because its too varied and inprecise for a developer to deal/be bothered with, and musical value less qualified.

So, bleed is probably as close an example we will see - unless someone else has a crosstalk one.

[jksuperstar]

I doubt there was very much research done, other than to take a quick sample from a particular desk to find the crosstalk, delays, and the filtering characteristics (you can pull the same thing off with one of those headphone simulators, there's a free scope device called CAN CONTROL, and turn all the knobs down to almost 0.) That'd be about it.

[kensuguro]

That's not to say there isn't the occasional genius that analyzes and places routes of busses on a board using specific distances to not only cancel, but encourage only specific harmonics.

So at least some people do design this into the circuitry then. In my mind I'm likening it to the amp maestros that tune guitar/bass amps through specific material and circuitry, but maybe it's technically very different. Still, what puzzles me is that even with the big name consoles, only specific ones are labeled as "magical", others not. Or perhaps they were actually individually unique since they were hand crafted and designs were improved through consecutive builds.

I did mess with the crosstalk settings for sknote's stripbus, and realized it was just "bleeding" in other tracks.. not sure if that adds to the overall sound in a positive way at all since everything was bleeding into all channels, whereas in physical circuitry, the electromagnetic bleeding should have a proximity effect. Anyway, I guess I care less about what really happens or how the actual bleeding occurs. I'm curious why people deem it as something that makes consoles sound good. (or perhaps it's just a marketing blurb after all) I do feel that the nonlinear saturation characteristic is largely responsible for the sound character, more than crosstalk.

[garyb]

yes, those legendary consoles were all hand made for the studio or tour that they were purchased to operate in.

later consoles were factory made, but in small batches, so they are very close to custom. in addition, the designer usually upgraded the design every time a new solution or need was recognized. it was a production line, but not mass production. the real hardware is still made this way. then there's the garbage that everyone thinks is good quality....and it is, for what it costs....

native=bullsh^t.
it might be VERY usable bullsh^t, it might be wonderful for growing your audio garden, but it's still bullsh^t. a crosstalk control is also bullsh^t.

[jhulk]

ex neve employee and now retired makes and designs tube only mixers with passive eq and tube make up gain with summing bus and master bus

most things were point to point wiring and they used lots of metal shielding and input and out transformers for noise cancellation









its easy to build a diy neve products because they came with schematics so that engineers could fault find on site when engineers were that they were electronics specialist and sound specialist they learnt about acoustic theory and about sounds and how to lay them the passive inductance circuits for eq have crosstalk because of the interactions of the magnetic fields but they sounded so good but you do get a lot of volume cut and so this has to be made up to the same level as the input by a make up gain stage

worse cross talk comes from the panning circuits on the fx returns where the left and right channels can bleed into one another

but even on paper in designing when you make the product it never performs as it does on paper due to heat components used and quality of the electrical supply

and to emulate inductance via a native plug in is nigh impossible so its all relative using the same type frequencies

its the same with most emulation unless you are actually doing circuit emulation down to the resistors opamps or discrete circuits as is the case with the older neve products where the opamps where created by hand with match components dbx being the most famous one the blackmer amp vca

so to me as i use analog processing and make the diy pro audio versions and can tell you the plugs dont come close in sound or operation and why there is a big community making new and improved versions of the same 50,60,70,80 pro audio gear

like ian who is making new tube mixer designs but using top grade 1% resistors and wima caps and sowter or mag or other transformers for better crosstalk and noise suppression

as the top drawing shows his new ez mixer channel goes beyond the neve cnannel with pre and post eq and routing options i have built several channels of this channel strip with the red47/rs eq and the meq5 eq and the helios eq and a summing box and the difference in sound to a mass produced opamp based 1 pcb design like mackie is over whelming

[dante]

Awesome pics jhulk ....!!

[dawman]

Now I know where Roland got the idea of scientists in white coats for their 10,000 USD V-Piano video.