Convolution IR's and emulations...

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

iomegaman

Well-known member
GDIY Supporter
Joined
Dec 22, 2008
Messages
867
Location
Tucson, Az
So pretty much all of your audio plugins are doing some kind of emulation...some use SPICE to emulate the analog gear and more recently we have companies like Townsend Labs and Slate offering devices that are supposed to use some sort of emulation to replace very expensive mics and now even "rooms"...

I recently updated my Liquid Sonics to reverberate 3 and it can handle imported IR's and frankly the bricasti stuff sounds really good to me (having never owned an original I don't really have a lot to go on here)...

Down that road I recently found a guy that does wav* IR's of some very nice acoustic guitars, vintage Martin D28's and Gibson 45's/Hummingbirds etc..and discovered I can load them into Logics "Space Designer" and send them to a bus to get some very interesting boost to frequencies a tracked acoustic might be missing...so the Taylor gets a little
Gibsony boost where needed...without a ton of surgical eq, just cut back the wet signal to the IR host...

Sonicially we already know that UAD and a host of others offers some "Close enough in handgrenades and horses" types of Vintage gear and there are a LOT of fairly famous mix engineers that default to ITB and plugins these days...

But my wondering is if an IR can ever truly replace source material?

The Townsend Lab Sphere mics seemed specifically attuned to capture as best as possible and then run through an emulation of a mic 10x the cost...and many people are suggesting the difference sonically is not worth the difference in price point, seems like a no brainer if you've already agreed to the idea that plugins in general are worthy replacements of vintage gear...

The Slate VMR?X? headphones suggest that a specially tuned set of cans can sonically replace a tuned room and speakers at 1000x the cost...I find it hard to believe especially since Beryllium headphone drivers are not all that and a bag of chips...you can buy the exact same set of headphones from china for 1/4 the price and minus the Slate sticker...

I'm wondering are we "squinting" our ears to hear what we want to hear, because we've purchased the virtual thing?

I know the UAD version of my DIY G-SSL  sounds closer than any other plugin but there is something in terms of "headroom" (for lack of a better word) that the real thing offers that a virtual sheen just does not replace...

I wonder how true this holds across the board, especially with electronic Reverbs and IR's and also with capture devices like microphones...

On one hand we are treating an already captured signal, in another we are capturing signal and treating it as we go...

Apples and Orangutans?

Reverbs seems like you can get away with some of this...even so far as to emulate actual rooms (Altiverb sounds pretty damn good to me)...and maybe some EQ's in digital domain don't really care what GUI you're using because its all code now anyway...

I wonder if I'm too harsh on Slate because I feel like you're inserting something fake in the audio path if you use his plugin with the headphones instead of a real set of speakers...I mean I don't really hesitate to reach for Dave Derrs Arouser (or BIgFreQ now, what a great EQ!) instead of worrying about sonic degradation...

I have yet to find any plugins I actually feel comfortable TRACKING with though...I still find something missing and what I feel is a "synthetic sheen"over the best of them///the UAD 610 sounds nothing like a real tube pre to me...but some of these IR's are actually sonically quite usable...

Thoughts?
 
Well, convolution can only do LTI systems. So no saturation, hysteresis, coupled resonances, .. it is a useful tool though and also has some creative use.
 
Interesting post. I think you can look at it from a couple viewpoints. If the final product, being a compelling song with a compelling sound is all your shooting for, I don't think it matters if the "model" sounds just as "good" as the real thing, since its a subjective decision. If you A/B'ed the SSL model against the real thing with someone who has never heard either, but knows sound and music, you can't tell me they will automatically pick the real thing. I think someone's taste has a lot more to do with the final product than any of the tools do.

Being a mastering engineer and cutting engineer I listen to a wide range of music from throughout history everyday. I'm never drawn to songs recorded in the 60's-70's-80's because of the recording quality. I'm drawn to the talent. I often think to myself, "boy they don't play it like that anymore". I never think, "boy they don't record it like that anymore".
I also master amazing sounding music recorded all in the box pretty regularly. I think people have to free themselves from even thinking about it if you want to maximize the creative output. Use what you got.

Another point of view would be the love of a particular period of music and the way it sounds. If it brings you joy to try and emulate that sound the best you can, then by all means, get the real gear. If the challenge of collecting the real deal and using it, is fulfilling to you, its nobodies business. But its not going to get you a "better" sounding recording in a general way.

Now to me, the mic emulation thing is confusing. I feel its like converting an mp3 to a wav, and calling it a wav. I would think modeling a vintage Neumann capsule and transformer with a Chinese capsule, a couple transistors, and some software would be a hard thing to do. A complete stripping away of the non linearities of the original equipment would not get you the Chinese hardware. There's really no way to get to the base sound of a microphone to be able to add to it with software. There will always be the faults of the base hardware. 
This is true for any emulation but I would think it would be most apparent with a mic emulation.

All that said, I heard a Slate mic setup in a friends studio a while back and was really impressed. I actually liked the base hardware mic with no emulation. He had a real U47 there and demoed an A/B of the Slate model. It didn't sound like the U47. But do any U47's sound alike? Lol. Still sounded pretty good though. And I guess that's the only thing that matters in the end.
 
I'm hearing exceptional things about the Townsend Sphere mics...and they just filed and got awarded another patent of come sort...apparently the mics are really well calibrated (dual capsule) and they are the only ones who allow the user software for free...I've tried it with a stereo mic recording and was really impressed it does not really sound like some sort of EQ masking because it does a few things with the sound field.
 
That's really cool! Seems they are taking it a step further than just copying another sound. Using the technology to make something new is a good direction to move in IMHO.
 
I will add that I made some burst captures in a nice live room last year at the end of a tracking session, and I can 90% recreate the space on new overdubs with the IR's.  I have used them on most things I've mixed since.  Pretty amazing. 
 
EmRR said:
I will add that I made some burst captures in a nice live room last year at the end of a tracking session, and I can 90% recreate the space on new overdubs with the IR's.  I have used them on most things I've mixed since.  Pretty amazing.

I should get off my ass and learn how to do this, I'd like an IR from my living room/dining room. :)
 
I though that Moore's law had broken down...

True it's slowed down, but the application of new discoveries are always around the bend.

Thought this one was bs until it showed up in nature mag:
https://www.nature.com/articles/d41586-020-02895-0



 
boji said:
True it's slowed down, but the application of new discoveries are always around the bend.
The mechanism driving Moore's law was miniaturization. AFAIK this is limited by wavelength of light for minimum photolithography dimensions. I have heard arguments that alternate 3D IC design opens up opportunity for more density. It seems in principle that multilayer ICs could be constructed like multilayer PCBs by stacking die on top of each other, but this may have issues with heat dissipation, yield, and more. 
Thought this one was bs until it showed up in nature mag:
https://www.nature.com/articles/d41586-020-02895-0
Interesting but 15C and extreme pressure is not exactly room temperature. I look forward to being able to transmit power through the WWW.

JR
 
Impulse response or convolution will give you anything of linear nature - but nothing of the nonlinear. Which excludes most of the "sound" that is why we like some analogue processors in the first place.

Try an IR of a tape machine. Or simply a transformer :)

The mic modelling idea is BS, basically no different from the likewise BS'hitty antares mic modeller of yestercentury. Carried by marketing and suggested-perception. There is simply much more to a microphone than can be captured in 1½D - if they did an array like the soundfield, we could start talking about further details..

/Jakob E.
 
cgi-sucks-22-728.jpg
 
gyraf said:
Impulse response or convolution will give you anything of linear nature - but nothing of the nonlinear. Which excludes most of the "sound" that is why we like some analogue processors in the first place.

Try an IR of a tape machine. Or simply a transformer :)

The mic modelling idea is BS, basically no different from the likewise BS'hitty antares mic modeller of yestercentury. Carried by marketing and suggested-perception. There is simply much more to a microphone than can be captured in 1½D - if they did an array like the soundfield, we could start talking about further details..

/Jakob E.

That actually makes sense...no mic modelling mic has the ability to determine what its actual distance is from the source and then to subsequently adjust its frequency response to accommodate that measurement...even though everyone of these plugins allow you yo adjust for "distance" and proximity...

I guess we tend to not investigate the physics side of things when we see names like Paul Wolff attached to products...I know I would probably take any criticism of Rupert Neve products with a grain of salt at least until he stops being involved personally...but then again Rupert doesn't seem to be overly concerned with anything other than what he's good at.
 
Like the Kemper guitar amp " profiler " It doesn't model or profile the amp but only one setting of the amp
everything else is post profile , like you're mastering.  When you change the eq, you're not changing a fender or marshall eq , it's the same eq no matter what the profile. there are positive benefits as a tool , in this case
use the exact recording chain used on your project, tour w/ multiple amps in a light weight box.
Some people maybe fooled listening to them, less actually playing them, but no substitution for pants flapping volume ...........And I don't like vinyl better !
 
But my wondering is if an IR can ever truly replace source material?

I did IR for two plate reverbs (mine and a friends) and a great sounding barn - they were really impressive. Not a full substitute for the real thing, but very usable
 
Makes me wonder about the emulation method used in Nebula plugins.  Are "Vectorial Volterra Kernels" arrayed IR's?
 

Latest posts

Back
Top