Convolution inverses to take out mic character?

DrNick

DrNick

Inspired
I read somewhere that the way the Axe-FX adds mic character to a cab block is via convolution with something that encodes the EQ curve of the mic. (I'm a mathematician, and convolution comes up in many places in math - I'm just assuming the term is being used here in the same manner, which would indeed make sense.)

If this is so, I wonder if one could do just the opposite to approximate removing the mic character from a IR pulled from a cab. For example, I might capture an IR with my SM57 but want to remove some of the coloration coming from the 57. It would be amusing to have an "inverse SM57" option from the mic menu.

Mathematically, it seems that this would amount to computing an inverse (with respect to convolution) of the SM57 data. Is this feasible?

Thanks,

-DrNick
 
I've been exploring this exact topic, but in regards to speakers, not mics. Trying to make some guitar speakers more flat in order to use custom cabs with them. I just wrote up a huge post about it in the cabs section :)

I'm not sure how you would do this with a mic, you'd have to somehow be able to measure the response of the mic.. It can be done for sure, I just can't think up he exact mechanics of doing it at the moment. Gotta have a known source of some kind. Something that produces pink noise totally flat would work. Lots of variables, though.
 
Refering to the MIC IRs in the Axe and assuming, they are "correct" ... all the data you`ll need to do inversion IRs are there ;)
 
Morphosis said:
Refering to the MIC IRs in the Axe and assuming, they are "correct" ... all the data you`ll need to do inversion IRs are there ;)
Click to expand...

My thoughts exactly. I just don't know whether these inverses have been computed, and whether there is some way a user can implement them...
 
DrNick said:
My thoughts exactly. I just don't know whether these inverses have been computed, and whether there is some way a user can implement them...
Click to expand...

Measurement with e.g. roomeqwizard, inverse result, export as .wav ir, convert wit cablab to UR - ready.
 
Hmm. If we only had a NULL cab. But, how about using a cab with a 57 and a cab with a NULL mic? Tone match the difference to get the inverse 57?


Sent from my iPhone using Tapatalk
 
fractalz said:
Hmm. If we only had a NULL cab. But, how about using a cab with a 57 and a cab with a NULL mic? Tone match the difference to get the inverse 57?

Sent from my iPhone using Tapatalk
Click to expand...


That's an interesting idea! If I understand all this correctly, then perhaps a good test of the result would be to compare a CAB block with a null mic with the same CAB block with an SM57 followed by this TMA block.
 
DrNick said:
That's an interesting idea! If I understand all this correctly, then perhaps a good test of the result would be to compare a CAB block with a null mic with the same CAB block with an SM57 followed by this TMA block.
Click to expand...

This would be one possibility ..., personally i would measure the frequency response of the MIC IRs via DIGI in/out in roomEQwizard, invert the result, export it as .wav and convert to UR with cablab.

Test would be:
CAB1: "cabir z" + "micIR i" from Block + CAB2 "inverse "micIR i" from Block

vs.

CAB1: "cabir z"


but

sine-sweep -> "Nullcab"+micIR -> tma (local: live,peak hold, highres)
sine-sweep -> tma (ref) should give same inversion of the micIR
 
Last edited:
fractalz said:
Hmm. If we only had a NULL cab.
Click to expand...
What is the difference between a 'NULL' cab and the existing "No Cab"? I understand what the OP is talking about wrt to inverting a specific mic, but am not sure what a null cab would be without some relative reference.

Also, with regard to the OP, isn't a mic inverse sort of meaningless unless it is very specifically tied to a single mic position, I thought the frequency response of a mic varied dramatically with position relative to the speaker? I guess you just want one specific to the positions modeled by the built-ins?
 
I thought no cab muted the output? Maybe not? I'll have to try it again.

Sent from my iPhone using Tapatalk
 
zenaxe said:
What is the difference between a 'NULL' cab and the existing "No Cab"? I understand what the OP is talking about wrt to inverting a specific mic, but am not sure what a null cab would be without some relative reference.

Also, with regard to the OP, isn't a mic inverse sort of meaningless unless it is very specifically tied to a single mic position, I thought the frequency response of a mic varied dramatically with position relative to the speaker? I guess you just want one specific to the positions modeled by the built-ins?
Click to expand...

I don't think that the frequency response depends on its placement (the mic doesn't know where it is placed...). Certainly the sound of a mic'ed cab depends on placement, but this is because speakers sound different at different parts of the cone not because the mic responds differently.

So, in effect, what I would be trying to reproduce with this experiment is what the speaker "really sounds like" at a certain place (like center dust cap) in a manner uncolored by the mic used to record the speaker at that position.
 
DrNick said:
I don't think that the frequency response depends on its placement (the mic doesn't know where it is placed...). Certainly the sound of a mic'ed cab depends on placement, but this is because speakers sound different at different parts of the cone not because the mic responds differently.

So, in effect, what I would be trying to reproduce with this experiment is what the speaker "really sounds like" at a certain place (like center dust cap) in a manner uncolored by the mic used to record the speaker at that position.
Click to expand...

I get what you're saying, I was thinking in terms of the combined effect of placement and the mic's frequency response. You are willing to tolerate the placement effect but want the mics contribution nullified. Got it. Thanks.
 
Smilzo said:
I think when you got some zeros your inverse goes to infinite.
Click to expand...

You're thinking of inverses with respect to multiplication (so effectively division). Convolution is a different animal.

That said, it does have its own troublesome issues vis-a-vis inverses.
 
Yeh, I know. If you invert the matrix H you could have some zeros. H*H'=null matrix. H is a monodimensional matrix (vector). I may be completly wrong... :D
 
Smilzo said:
Yeh, I know. If you invert the matrix H you could have some zeros. H*H'=null matrix. H is a monodimensional matrix (vector). I may be completly wrong... :D
Click to expand...

Well, convolution also isn't a priori matrix multiplication either. Also, once cannot multiply vectors and arrive at vectors in general. That said, any linear transformation from vectors to vectors, such as convolution with a fixed function, can be given by multiplication by a genuinely 2-dimensional matrix... But I think we're getting too far afield here :)

I'm going to try fracalz's idea first and will report back to the thread with my findings.
 
Invertability typically won't be an issue for a mic IR as their is plenty of response at all points on the unit circle. So H*H^-1 is never singular.

The problem is that a mic is typically used in the near-field. There isn't any way to know a priori what the IR of the mic was at the time the IR was acquired. IOW, the IR of a mic varies with it's position relative to the speaker until you get into BOTH the speaker's far field and the mic's far field.

You can find the far field IR of a mic by putting it in the far field of a speaker in very close proximity to a known mic (i.e. something like an Earthworks TC30). The difference between the two IRs is the mic IR.

The mic IRs in the Axe-Fx II assume a nominal position and distance from a typical speaker.
 
You must log in or register to reply here.
Back
Top Bottom