Mic characteristics plugins?

[audiodef]

Are there any LV2 plugins or Linux software that emulate characteristics of vintage and popular mics? I'm looking for plugins or presets that emulate specific microphones, not EQ's. This isn't to cover up a poor mic's recording, but to emulate the sound of vintage and/or popular microphones that have specific characteristics. It could be a plugin, or it could be a web page somewhere with a chart detailing EQ settings that approach the sound of specific brand/model mics. (Although I would think EQ alone is not 100% responsible for the tone of many a vintage mic.) My intention is to use these on top of good-quality recordings, so that a good recording done with a mic with a flat response for the way it was used can be made to sound a little more like it was done with a U47, for example. Maybe such a thing doesn't really exist, or simply isn't done, but I won't know until I ask around. Or maybe this hasn't really been done, at least for Linux, but might be desired.

[mclstr]

I'm not sure there would be much of a demand and could you trust that it could really accurately make a certain dynamic mic sound like a certain condenser, ribbon, ... mic if there was such a thing. Transient response, microphonics, frequency response, non-linearity and more come into play.

From my experience classic microphones of the fifties and sixties of the same model do not sound identical.
I worked in a studio that had no less than 20 classic Neuman and Telefunken condenser microphones as well as many of the classic ribbon and dynamic types.
At one point we compared 4 similar era well maintained U47s and found none could be used as a stereo pair because none matched.

The technology at the time didn't allow the precision that we have now and some components can change tolerance over time.

Over a long time I came to the conclusion that classic microphones couldn't be trusted in serious situations. I had many moments when they would change sound as they warmed up, would become microphonic, crackle, frequency drift, ... or just fail.

My most memorable experience was working with a very well known celebrity using a favorite microphone from the 50's that ended up changing sound during the recording session to the point that we switched to a modern microphone and re-tracked all the parts because matching the sound of the various punch-ins became too tedious.

I also found that modern microphones often sounded better and/or be more accurate than the classics and were more consistent from studio to studio.
Many times a lesser expensive dynamic mic would work better on certain instruments and voices than expensive classic or modern microphones.

[tramp]

Not a plugin, but here you'll find a couple of IR files from over 40 different Mic's, to use with your favourite convolver:
https://micirp.blogspot.de/

[Frank Carvalho]

The characteristics of a microphone means the way it captures - and not captures - sound. With an bidirectional shape you capture sound behind the microphone as well. With a cardioid and supercardioid shape you capture mainly sound in front of the microphone. WIth an omnidirectional mic you capture in all directions. So when you pick a mic and record, you have already made your choice, and no plugin can recreate the sound that was *not* captured by the original mic.
Second, the other characteristics of a mic are all described relative to a completely neutral sound. But anything you capture with a mic will already be coloured by that particular mic, and so is not neutral to begin with. So applying any sort of IR or EQ or whatever to such a signal, will not reproduce the sound of the modelled mic anyway, but at best be a strange combination of the sound of two different mics.
So I dont think it would be of much use.

/Frank

[ufug]

Interesting idea. For this to work, wouldn't you need to first need to neutralize the characteristics of the mic you are using before applying the characteristics of the mic you are attempting to emulate? Seems like a slippery slope into voodoo valley.

[CrocoDuck]

Seems like a slippery slope into voodoo valley.

It is slippery indeed, but not very voodoo I think. In fact, most of the stuff discussed above can be modelled/measured and native response compensation can be done (it is common practice in psychoacoustics experiments). It is all pretty scientific actually, and mathematically elegant... I like to call it "voodoo" anyway tho, but that's not the common meaning of the thing.

What you capture with a microphone depends upon the microphone, of course, but also the environment in which the microphone and sources are, which dictates how the sounds gets transmitted from the sources to the microphones. What Frank mentioned is known in academia as source/receiver directivity and I cannot think of any easy way to undo/do directivity effects for microphones: one should have the mic directivity as a function of 3D polar coordinates measured in an anechoic chamber, then know for each source the law transmitting sound from them to the mic, which depends on their coordinates in the said reference frame and the environment the systems are into...

It is common practice to do these measurements (or simulate the whole thing algorithmically) to conduct experiments and research, I don't think it is very practical for plugins tho (for sure not in this particular formulation), although similar things are done realtime with HRTF for spatial audio.

As for the other aspects (distortion, transient response, hysteresis...), it would be nice to study those in detail. Perhaps, the best approach would be to decompose a vintage mic in a blocks model, with linear and nonlinear elements. Then, transform the most important (and practical to use) of these blocks into DSP modules that can be used on normal microphones signals, so to "vintage up" their sound. It will work to a limited extent as these blocks are gonna be mutually coupled for sure... But probably this could be a cool and practical idea, which could convey in few cases a lot of tonal resemblance to vintage captured sound I reckon. Or just something cool and nice to experiment with.

I am not aware of anything like this existing at the moment. I will add it to the list of things to look at "when I grow up".

[tramp]

Second, the other characteristics of a mic are all described relative to a completely neutral sound. But anything you capture with a mic will already be coloured by that particular mic, and so is not neutral to begin with. So applying any sort of IR or EQ or whatever to such a signal, will not reproduce the sound of the modelled mic anyway, but at best be a strange combination of the sound of two different mics.
So I dont think it would be of much use.

If you use a Mic with a relative flat response curve, like audiodef pointed out, this effect will be minimal. Otherwise, you could as well do a de-convolution with the response of the used Mic to reduce this effect even more.
As mclstr pointed out, old vintage MIc's has anyway all different responses due to u-precision in the used parts.
So I guess the IR-files will help a lot to come close to the wonted Mic sound. At least you may wont to add a saturator for the non-lin parts and be done.

[CrocoDuck]

So I guess the IR-files will help a lot to come close to the wonted Mic sound. At least you may wont to add a saturator for the non-lin parts and be done.

That's how I would do it if I had to try it myself with what's available right now. Perhaps after some fiddling about it will get pretty close to the real thing. it would be an "artisanal" version of this:

Perhaps, the best approach would be to decompose a vintage mic in a blocks model, with linear and nonlinear elements. Then, transform the most important (and practical to use) of these blocks into DSP modules that can be used on normal microphones signals, so to "vintage up" their sound.

Incoming compulsive system analysis

When taking impulse responses of nonlinear systems the quantity that we get is not actually a proper representation of the input/output relationship of the system. If we take the Fourier transform of it we end up with a sort of representation of energy density. So, using such quantities to model nonlinear systems will fail unless the nonlinearity is perturbative, i.e. the system can be described as linear plus a small nonlinear perturbation (not sure whether vintage mics fall in this category). In this case convolver with vintage mic "IR" + nonlinear saturator (or other nonlinear block) will work very nicely I reckon.

Not that it has to be exact to be cool... Or even perceptually similar to the target mic. Just wanted to nerd about it

[Frank Carvalho]

If you use a Mic with a relative flat response curve, like audiodef pointed out, this effect will be minimal.

Yes, but most modern quality microphones claim to have a flat response, which of course they don't in reality.

Otherwise, you could as well do a de-convolution with the response of the used Mic to reduce this effect even more.

I agree, but if this process is to be applied to any mic you choose to record with, then it would require you to be able to produce a proper inverse IR file of that particular mic as well to reverse to neutral, and that would requires you to define "neutral". What is neutral input to a microphone, and how does one separate the microphone characteristics from the room characteristics? Its not a simple process for a plugin at least.

But perhaps the generic technical specs of the microphones could be used to produce standard reverse IR's for standard mic types?

/Frank

[CrocoDuck]

Yes, but most modern quality microphones claim to have a flat response, which of course they don't in reality.

Usually the low end is pretty flat, high frequency could be plagued by resonance and modal behaviour, it depends on mic construction. For example, condenser mics can be made very flat as the mechanical properties of the diaphragm can be tuned to get rid of most resonances in the audible range. Microphones constructed for scientific measurements (we have a couple at work) are dead flat in the audible range (but also ludicrously expensive).

that would requires you to define "neutral"

Neutral is defined as ideal Dirac delta response, which means that the frequency response is flat and the phase response linear.

how does one separate the microphone characteristics from the room characteristics?

That's the trickiest part. The classic method is to measure the mic in an room with massive absorption (anechoic) which means that it has the same response of free field (a neutral delta response, a part for a gain factor). There are other methods to try to do that in situ, but I am not very familiar with those, a part for the referenced technique, which requires another microphone to sense the field in input to the microphone under test.

Its not a simple process for a plugin at least.

A part for the above, responses of linear systems are straightforward to measure in a plugin friendly way. I am working on that.

But perhaps the generic technical specs of the microphones could be used to produce standard reverse IR's for standard mic types?

Yes, that can be done. It is the modelling approach.