How does one actually capture an acoustic IR?

gdgross

gdgross

Experienced
It's easy enough to get IRs of the fishman aura or the D-tar mama bear (which I've done using some of Logic's utilities) as they are completely electronic devices.

It's also straightforward, although a little more work, to get IRs of a guitar cabinet: power amp in/mic out.

However, I've seen other, acoustic IRs floating around like "cello body" and such.

Does anyone know how these are captured? (stick a speaker motor on it and treat it like a guitar cab??) I believe fishman captures some kind of IR of your instrument for their aura system, right? anyone know how they do it?
 
I'm pretty sure I've posted on this elsewhere, possibly in more detail. Fishman simultaneously records the output from the pickup and a mic while the instrument is being played. Then Fishman adjusts for time lag and builds a model based on the two signals. The model attempts to reproduce both linear and nonlinear aspects of the response to being played.

I've seen the report from a university research project where the student simply strummed the instruments (guitar and banjo, I believe) and treated that as an impulse. I suspect it was one hard strum! I know that a starter's pistol has also been used to create an impulse. The ubiquitous gtrbody IR that has been around for a long time came from data collected by a grad student at Stanford.

I've tried a couple of approaches: 1) similar to the way they evaluate the response of a violin, I suspended the instrument in front of a speaker and recorded outputs from a pickup and a contact mic while playing a sweep through the speaker; 2) I attached a motor directly to the instrument and recorded the output via a mic.
 
  • Like
Reactions: jon
gdgross said:
Wow LMO, that's intense. How do you feel those IRs came out?
Click to expand...
I wish I could give you a good answer. To start with, I'm trying to do something weird for an IR, and that's to take the output of my guitar (steel guitar in my case) and make it sound similar to a dobro. Creating an IR gives you a way to replicate the transfer function of a thing, i.e. the changes to the signal that occur from input to output. In the case of a dobro, I may have done a good job capturing how the mechanical motion of the strings translates to audible sound, but I don't intend to use it in quite that way. Instead of plucking strings to create the mechanical input, I've got a signal from a pickup as electrical input.

So, I succeed if I get close enough to the sound of a dobro. Although I haven't tried using it in a band situation, I feel pretty good that I've achieved that goal. But just between you and me, I'd like the results to be better than any of several dobro simulators that are on the market. Even better, it'd be nice to be able to tell which of my dobros is being modeled. I want people to come up to me with tears in their eyes saying, "That sounded just like the dobro my dear departed Uncle Josh used to play..." Well, maybe not that.

I took lots of samples, using different placements and mounting systems for the motor, different mics and placements for each, different signal levels, different post-processing, etc., and that adds up to a lot of files to evaluate. And I'm still thinking up things I could do in post-processing, or things I could do in the preset, or discovering potential shortcomings of one method or another. So at this point it's more of a hobby, something to do when I get the urge to tweak presets and IRs.

But I'll try to give you a more useful answer: of the methods I've tried, I think using the motor is producing the best results. I feel some of the IRs came out pretty well. I took IRs of three dobros, a banjo, and a Martin D28. One thing I'd like to do when I get some time is to compare the D28 IRs with the ubiquitous gtrbody IR and see if that teaches me anything.

5.07 just finished installing. Gotta go.
 
LMO said:
I wish I could give you a good answer. To start with, I'm trying to do something weird for an IR, and that's to take the output of my guitar (steel guitar in my case) and make it sound similar to a dobro. Creating an IR gives you a way to replicate the transfer function of a thing, i.e. the changes to the signal that occur from input to output. In the case of a dobro, I may have done a good job capturing how the mechanical motion of the strings translates to audible sound, but I don't intend to use it in quite that way. Instead of plucking strings to create the mechanical input, I've got a signal from a pickup as electrical input.

So, I succeed if I get close enough to the sound of a dobro. Although I haven't tried using it in a band situation, I feel pretty good that I've achieved that goal. But just between you and me, I'd like the results to be better than any of several dobro simulators that are on the market. Even better, it'd be nice to be able to tell which of my dobros is being modeled. I want people to come up to me with tears in their eyes saying, "That sounded just like the dobro my dear departed Uncle Josh used to play..." Well, maybe not that.

I took lots of samples, using different placements and mounting systems for the motor, different mics and placements for each, different signal levels, different post-processing, etc., and that adds up to a lot of files to evaluate. And I'm still thinking up things I could do in post-processing, or things I could do in the preset, or discovering potential shortcomings of one method or another. So at this point it's more of a hobby, something to do when I get the urge to tweak presets and IRs.

But I'll try to give you a more useful answer: of the methods I've tried, I think using the motor is producing the best results. I feel some of the IRs came out pretty well. I took IRs of three dobros, a banjo, and a Martin D28. One thing I'd like to do when I get some time is to compare the D28 IRs with the ubiquitous gtrbody IR and see if that teaches me anything.

5.07 just finished installing. Gotta go.
Click to expand...

Interesting stuff. Seems to me that not an IR but an "acoustic guitar simulator" should be high on the wish list. Cliff could assign an engineer to study the characteristics of an acoustic guitar vs. an electric (while Cliff works on new firmware of course!) and after all the data is collected Cliff could create a mathematical expression to make an electric sound like an acoustic.

It wouldn't have to be perfect to be useful. Roland has models in their products that do this that sound cool, and while not perfect (far from it) they give live players the chance to put acoustic sounds in their music with no feedback or having to carry and mic the instruments.

Please let us know more about your experiments. I'd love to hear some examples of how they work out.

Gotta go - same reason - 5.07

Thanks.
 
Hmm, if it's just the sound of a dobro you're after, I took IRs of every setting on a d-tar mama bear, which as I recall includes a couple of dobro settings designed to work with USTs. I know I converted then to syx for the ultra. I'd be happy to take a look on my computer at home and see if I still have the wav files. Hopefully I made the IRs long enough to use with the Axe-II.

I hope I still have them...but if not I suspect they are floating around the forum somewhere a couple of years ago.
 
Geoff, thanks for the offer. I'll have to check, but I'm pretty sure I have those, although maybe only in g1 format. I do have the Aura dobro pedal, and have IRs of those images. I posted them a while back.

Not long ago I created a match EQ by comparing the outputs from a dobro with the Fishman UST and a steel. My idea is to take the output from the steel, run it through EQ to make similar to the Fishman pickup, and then run that through the Aura (pedal or IR). That method shows some promise as well.
 
LMO, if i understood right, you said that you instead of sending a predefined signal through a speaker and recording it,
you took an acoustic instrument, played a sound with it and recordet both pickup signal and mic signal.
So to say you used the sound from the instrument as the source to make the IR.
I want to do that! I´m playing my Clarinet with a Pickup through my Axe2 and i want to make it sound more like the mic signal than the pickup signal.
Is there a Software for creating IRs, where you can define the source signal?
So i would record pic and mic signal and tell the Software, the pic signal is the source and the software has to create the IR from the Difference of pic and mic signal.

I already did the ToneMatch this way, but as i understood, the ToneMatch captures only the Freqs, right?
 
You must log in or register to reply here.
Back
Top Bottom