Studio Report

R.D. said:
Are you sure that's accurate ? AFAIK the actual "length" of the IR isn't any different between 1024 & 2048 ? I thought it had to do with the # of data points sampled ? If that's the case, there wouldn't be any more or any less room artifacts in either .... As a matter of fact, Jay Mitchell has said many times that a properly taken IR WON'T have any room in it at all ?

What's your source ?
Click to expand...

The reason a properly taken IR won't have room is because it will be captured in a big enough room that the reflections arrive after the capture is done.
 
Last edited:
The necessary length of an IR has nothing to do with the room. The necessary length is defined by the Q of the various formants. If the speaker has a particularly high-Q formant then the length of the IR necessary to accurately reproduce that speaker will be longer. Think of it this way: how long of an IR do you need to reproduce a bell or a tuning fork? The length of the IR is defined by the time it takes for all formants to decay to inaudibility.

One cannot make a sweeping generalization about the length of the IR required, however, 2048 samples is long enough to reproduce any speaker I've ever encountered.
 
I don't think any one stated that the room anything to do with the length of the IR. Only that when capturing the cab it is generally good not to have reflections from the room. I just stated the room ought to be big enough to capture the cab with out capturing the room reflections. My comments were that the IRs currently in the axefx do not seem to have much, if any, energy beyond 1024 samples. Is this not the case?
 
Last edited:
javajunkie said:
Exactly. The energy from the cab (low and high) has already dissipated with in the first 1024 point (I believe even for bass - need to pull out the calculator ;)).
Click to expand...
And for 8 strings guitar players and 13 strings bass players, hmmmmmm ?






:D
 
FractalAudio said:
The necessary length of an IR has nothing to do with the room. The necessary length is defined by the Q of the various formants. If the speaker has a particularly high-Q formant then the length of the IR necessary to accurately reproduce that speaker will be longer. Think of it this way: how long of an IR do you need to reproduce a bell or a tuning fork? The length of the IR is defined by the time it takes for all formants to decay to inaudibility.

One cannot make a sweeping generalization about the length of the IR required, however, 2048 samples is long enough to reproduce any speaker I've ever encountered.
Click to expand...

This is interesting, so what your saying is if you were to hit the string o the guitar and let it ring out this is the total length or time of the IR? Also how much time does 2048 samples take? Is it stretched out over the length of time of the IR or is it reoccurring as many times as it takes until inaudibility?... sorry just trying to get my head around the procedure.
 
Sixstring said:
This is interesting, so what your saying is if you were to hit the string o the guitar and let it ring out this is the total length or time of the IR?
Click to expand...
Nope. It's not that.

IR stands for Impulse Response. It's the response of a system (in this case, a speaker cab) to an impulse (a brief burst of energy). If you tap a speaker cone, how long does the speaker take to ring out and then be silent? Not very long. A few milliseconds. That's how long the IR needs to be for a speaker cab.

But if you tap a bell, it's a different story. A bell is a high-Q device (it's very resonant). If you tap a bell, it will take a long time to ring out and then be silent. That's how long an IR needs to be if you're modeling a bell.
 
fremen said:
And for 8 strings guitar players and 13 strings bass players, hmmmmmm ?
:D
Click to expand...

fremen said:
And for 8 strings guitar players and 13 strings bass players, hmmmmmm ?

Point is????

Reading this thread, it's clear there's a thirst for trying to find some flaw in the IR process. I think this thread shows that there is fantastic progress in the IR field from Fractal. As far as the 'pushing the envelope' area, the discussion is all good stuff for the most part, but not sure you can call '8 strings guitar players and 13 string bassists' in the category of reasonable research. Maybe it's something for the future, but feeling like this is a *REALLY?* moment when I read this.

Even in jest...

IMO.

R
Click to expand...
 
Relax, it was just a (stupid) joke (saw the smiley?). I've read Cliff and Java posts and for the first time I understand what the length of an IR is related to. I'm not "trying to find flaws" (if that is directed at me), and btw, I'm happy with only six strings on my guitar ;)
 
After reading many of the latest comments in this thread I don't think anyone is trying to find flaws with anything.
I believe that people only try to understand what an IR actually is and how it is "made". Most people don't have a masters in engineering/acoustics and even then it can be hard to understand what some processess actually do and how to interpret the results. At least if you are newly graduated with not that much experience in the field.
Regards/

And some jokes are not always interpreted as jokes. It would have been easier if everyone could sit down in room somewhere and just talk! A great fun!
 
Rex said:
Nope. It's not that.

IR stands for Impulse Response. It's the response of a system (in this case, a speaker cab) to an impulse (a brief burst of energy). If you tap a speaker cone, how long does the speaker take to ring out and then be silent? Not very long. A few milliseconds. That's how long the IR needs to be for a speaker cab.

But if you tap a bell, it's a different story. A bell is a high-Q device (it's very resonant). If you tap a bell, it will take a long time to ring out and then be silent. That's how long an IR needs to be if you're modeling a bell.
Click to expand...

Ok, I knew what IR stood for I was just a bit confused with the example that Cliff gave. I thought the process was a bit longer but was confusing this with a sign sweep for a response measurement and thought the two were done in the same manor.
 
AdamCook said:
Here are a few pics from the session in no particular order or significance.

Super%20Reverb.jpg


Marshall.jpg


Twin%20Reverb.jpg


Bad%20Cat.jpg


cliff%20in%20control%20room.jpg
Click to expand...

pardon me, Sir, what are those wood boxes with strange writings and funny knobs ???
:-D
 
Sixstring said:
Ok, I knew what IR stood for I was just a bit confused with the example that Cliff gave. I thought the process was a bit longer but was confusing this with a sign sweep for a response measurement and thought the two were done in the same manor.
Click to expand...
Actually you're right, the process is a bit longer. It's not practical to hit a speaker with an actual impulse. A perfect impulse means hitting the speaker for an infinitely small amount of time with what amounts to infinite power, so other methods (like sine-wave sweeps) are used to determine what the frequency response of the cab is. If you know a system's frequency response, you can figure out its impulse response, and vice-versa.
 
Rex said:
Actually you're right, the process is a bit longer. It's not practical to hit a speaker with an actual impulse. A perfect impulse means hitting the speaker for an infinitely small amount of time with what amounts to infinite power, so other methods (like sine-wave sweeps) are used to determine what the frequency response of the cab is. If you know a system's frequency response, you can figure out its impulse response, and vice-versa.
Click to expand...

I think that is a bit off.

You can create a .wav file with one sample at full scale and use it.

IIRC the issue with impulses vs. sine waves is just the volume needed to excite the system. BTW, I believe you can use a pistol shot too as the impulse if you could figure out where it could be done lol

Also, some algorithms take advantage of using the sine wav and filter out extraneous frequencies during the capture. This can help if the capture is being done in an environment with background noise.
 
You can use an actual impulse. This is generally not done though since the SNR achieved with this method is poor. Modern techniques employ sine sweeps, MLS sequences or other analytic signals with low crest factors.
 
FractalAudio said:
....Think of it this way: how long of an IR do you need to reproduce a bell or a tuning fork? The length of the IR is defined by the time it takes for all formants to decay to inaudibility..
Click to expand...


Oh my God, Cliff explained something in layman's terms! There's hope yet that someday I may fully understand what the Hell he's talking about. :mrgreen
 
You must log in or register to reply here.
Back
Top Bottom