convoluting amps/mics/compressors

[joel Oporto]

I was sufing about convoluting reverbs and came across a site called www.noisevault.com. which was dedicated to convoluting reverbs. Interestingly they had impulses from mixers like api, neve and ssl. they also had pre-amp impulses from those and the manley voxbox. Aside from that, they also had impulses for various compressors.

I can understand sampling acoustic space responses/impulses, but I dont get how they capture compressor/mixer/preamp effects and use it (like a reverb?). I can also understand emulations of mixers/preamps and compressors, but capturing their effects? What gives?

[Alndln2]

They use .wav file of a dirac or sine wave and play it through the hardware. To be honest,most of those attempts ony result in something that sounds out of phase. The correct way to to something like that is with multiple sweeps in frequency's which can be a difficult undrtaking. It is suspected that that was the process used for Focusrite's Liquid Channel and Waves Q-Clone plugin and the Sintefex hardware. You can learn a bit more at the Sintefex site http://www.sintefex.com/

[joel Oporto]

I understand that they put in a reverse phased original signal along with the "wet signal" to cancel it out and be left with the "noise" of the comp/amp/mixer/whatever, but how do you use that "noise" to color your sound to make it like pehaps a "neve preamp effect" It seems understandable with an antares mic modeler effect or amplitube effect, but as a convlouting reverb???

[jonmatifa]

For reverb typically you have a very short playback of pink noise through a speaker in the enviroment you wish to capture, when the microphone(s) pick up the result, what you've got is a snap shot of what that pink noise sounded like in that enviroment and whatever reverb effect it had. The pink noise is used because it generalizes the properties of any sound source. At noise valt, you can litterly play back the convolution wave files on your computer to hear what they actually recorded.

[Kiwiburger]

It's important to know whether the impulse you are using is intended as a 'Pipeline' impulse or not. With a typical reverb impulse, you would normally apply this as Send effect, where the Wet sound is layered over the Dry sound.

 

With impulses taken from eq's and compressors, these are considered 'pipeline' impulses and you must use them 100% wet. Otherwise you will get nasty phase cancellation.

 

I don't have a lot of time for most pipeline impulses (except for Tape and cabinet impulses. Until Dynamic convolution becomes possible, these are just static impulses, and therefore can't model what a compressor does at different dynamic levels. They can capture an eq curve, but it's a pain not being able to adjust it.

[Phil O'Keefe]

Originally posted by Kiwiburger It's important to know whether the impulse you are using is intended as a 'Pipeline' impulse or not. With a typical reverb impulse, you would normally apply this as Send effect, where the Wet sound is layered over the Dry sound. With impulses taken from eq's and compressors, these are considered 'pipeline' impulses and you must use them 100% wet. Otherwise you will get nasty phase cancellation. I don't have a lot of time for most pipeline impulses (except for Tape and cabinet impulses. Until Dynamic convolution becomes possible, these are just static impulses, and therefore can't model what a compressor does at different dynamic levels. They can capture an eq curve, but it's a pain not being able to adjust it.

 

Exactly. And while an EQ curve may be applicable to one instrument in one mix, chances are it's not going to be suitable for the next. And presets (of all kinds) can be useful to give you some idea of what a device can do or to use as a "starting point", the real fun kicks in when you start messing around with adjusting things on those presets, or doing really wacky things to find out what else the device can do.

 

Convo verbs are useful, but a EQ or compressor is something I feel still has significant limitations with the current technology. However, while I find some modeling and convo stuff interesting and even useful at times, for the most part, there's still nothin' like the real thing.

[jonmatifa]

Its still a relatively young technology, rest assured they'll find new and better ways to measure the outputs of this stuff.

[Kiwiburger]

It's a bit like comparing sampled sounds against real instruments, or against synthesisers. While a sampled sound might fool a lot of people, it's never as good as the real thing. You can improve things a little by taking lots and lots of samples. Convolution of effects could be improved by using lots of samples, but think of the CPU load.

 

Samples - even multisamples - are never quite ideal. That's where synths can sometimes be better than samplers (or in this analogy, DSP 'analog' modeling of effects can be better than convolution of impulses.

 

As with digital instruments, sometimes the best is a combination of samples with synthesis. Or in this case, hybrid effects that use algorithmns for the stuff that algorithmns are good at (saturation, eg, compression) and convolution for the things that convolution is good at (time smearing effects). Some of my favorite plugins use convolution this way - like the Voxengo Analog Flux suite.

 

In my ideal world, as processor speeds increase, we should be able to see some good additive re-synthesis instruments. They exist now (like Camel Audio Cameleon) but to my ears they are lacking a little. But the concept of a synth that creates extremely realistic patches from a few samples and can interpolate all the in-between sounds seems like the perfect solution.

 

So as far as the ideal effect technology, I would like to see something that can analyse samples from a hardware compressor (for example) and create a algorithmn that perfectly models it. In my view that would be way better than mutli-sample convolution.

[joel Oporto]

okay, I get it now. I just thought that I was way behind in that there was a new app/plugin that can really adjust your audio to make it seem like its going through a neve or ssl, or a manley etc.

much like amplitube, antares and q clone are doing.

 

I really agree that nothing beats the real thing though, but then I have to confess, being based in cebu, I have never seen an SSL, a Neve, or an API except in pictures and videos.

[Alndln2]

Originally posted by joel Oporto okay, I get it now. I just thought that I was way behind in that there was a new app/plugin that can really adjust your audio to make it seem like its going through a neve or ssl, or a manley etc. much like amplitube, antares and q clone are doing. (

I wouldn't write Q-Clone off completly and they are not using standard convolution techniques,it does come very close but's it's a very awkward way to work,I mean every time you want to go back and tweak something you have to resample your hardware. Also the Tritone EQ plugin is supposedly based on hardware sampling and does sound very close as well. Another option is Tritones Colortone that loads some type of IR's and Waves new SSL channel strip and EQ and Comp seems to be getting really good feedback from those who heard it at Namm. None of the plugs Iv'e mentioned thus far use standard Diracs or sine waves used in standard convolution methods and have come the closest to hardware Iv'e heard,but those are all EQ's and Iv'e yet to hear any pre-amp/compressor emulation,or at least any that sound like anything worthwhile. I guess I'll give Tritones Colortone free a shot and see how it fares.

[joel Oporto]

as I understand it , q clone basically gets the eq curve of a sample and tries to adjust your mix/file to approximate the sample. I heard it works really well and even george massenburg is reportedly using it on occassion. I think i'm going to try colortone too as with the waves.