More thoughts on IRs

[Doug B]

FWIW: I remember Thomas Blug of Amp1/AmpX talking about their development of some kind of dynamic IRs for their soon to be released AmpX.

Isn't the UA Ox Stomp supposed to use dynamic IRs?

[jzucker]

Isn't the UA Ox Stomp supposed to use dynamic IRs?

i don't know specifically what Ox is doing but the typical dynamic IR as used by fractal and line6 are simply multiple IRs captured at different distances and speaker placements and with different mics. HOWEVER, they are only dynamic during the configuration phase. In configuration, you dynamically move the mic around the speaker and vary the distance but once you press OK, you're back to a single IR. The dynamic-ness is simply that you don't have to deal with a folder of 100 IRs with different mics and mic positions.

Maybe OX is actually varying the IR dynamically but I've seen nothing from them saying they are doing this.

[skykomishone]

I bought a Fishman Aura Imaging DI, which is an IR (I think), to use with my Taylor and Martin non-archtop guitars.
It sits on the shelf. I currently have virtually no use for it with my archtop guitars.
For what it's worth ($0) it was expensive. Someday, I may drag it out and try to use it with my Barbera Soloist transducer equipped classical. It's just a shelf decoration at this point, and a painful reminder of all of the technological distractions that I've allowed myself that interfere with my actual guitar playing.

[rpjazzguitar]

IR's are available in large numbers on line, both for free and for $$.

I'm not entirely clear on how you would go about finding the best one for your application.

I guess you'd look at the gear (cabs and speakers, not amps, if I understand it) being modeled to get in the ballpark. Some on line sources provide that info, some don't

But to get, say, Joe Pass' Polytone sound, you'd need an amp model of the Polytone and then run it through the IR of a Polytone cab (which you can find on line). And, then, you'd have to do some careful listening tests to figure out if it's going to sound the way you want in the rooms you gig in.

So, is that much correct? And, if so, has anybody tried it? Did it work?

[jzucker]

IR's are available in large numbers on line, both for free and for $$.

I'm not entirely clear on how you would go about finding the best one for your application.

I guess you'd look at the gear (cabs and speakers, not amps, if I understand it) being modeled to get in the ballpark. Some on line sources provide that info, some don't

But to get, say, Joe Pass' Polytone sound, you'd need an amp model of the Polytone and then run it through the IR of a Polytone cab (which you can find on line). And, then, you'd have to do some careful listening tests to figure out if it's going to sound the way you want in the rooms you gig in.

So, is that much correct? And, if so, has anybody tried it? Did it work?

the amp and its settings also come into play. if you generate an IR with all the tone controls set on 10 or the gain control all the way up, you will end up with a different IR than if the gain is on 2 and tone controls on 5. The same thing is true of sample player modelers like kemper, tonex, quad cortex.

I've told this story a few times but I once bought a fender twin sample for kemper from a very well known seller of amp profiles and IRs (I bought the amp sample, not a speaker IR). When I got it, it sounded like !#@$. I reached out to the company and asked them what settings they sampled the twin at TREBLE, MID and BASS on 10, they said.

WHO USES A FENDER TWIN WITH THOSE SETTINGS???

With celestion IRs, they have variations such as thin, fat, bright, dark, balanced. Other IRs I've bought were labeled high gain, low gain, clean.

In an ideal world, the technology would support a generic, clean signal to generate the IR but because speakers are motors, they react differently with different content and different levels of gain.

Sadly, the growth area of the market are djent/hardcore players. Not old guys with thinning hair playing '50s archtops...

[rpjazzguitar]

the amp and its settings also come into play. if you generate an IR with all the tone controls set on 10 or the gain control all the way up, you will end up with a different IR than if the gain is on 2 and tone controls on 5. The same thing is true of sample player modelers like kemper, tonex, quad cortex.

I've told this story a few times but I once bought a fender twin sample for kemper from a very well known seller of amp profiles and IRs (I bought the amp sample, not a speaker IR). When I got it, it sounded like !#@$. I reached out to the company and asked them what settings they sampled the twin at TREBLE, MID and BASS on 10, they said.

WHO USES A FENDER TWIN WITH THOSE SETTINGS???

With celestion IRs, they have variations such as thin, fat, bright, dark, balanced. Other IRs I've bought were labeled high gain, low gain, clean.

In an ideal world, the technology would support a generic, clean signal to generate the IR but because speakers are motors, they react differently with different content and different levels of gain.

Sadly, the growth area of the market are djent/hardcore players. Not old guys with thinning hair playing '50s archtops...

Thanks for this.

I read, someplace, that the "proper" technique is to use a clean, linear amp to create the impulse and that the IR is then a model of the cabinet, the speaker within and the mic placement, all of which are supposed to be close enough to linear (not that I could give a succinct definition of "linear" in this context).

But, if the creator of the IR does it some other way (liked dimed settings on a non-linear amp) which can, say, introduce clipping or an odd EQ curve, apparently the resulting IR won't fully reflect the cab/speaker/mic. And, I don't get what it would reflect later, when you plug it into your modeler.

Does the Fourier transform work with clipping? Usually, elegant math requires some assumptions of things acting "nicely".

[BreckerFan]

Thanks for this.

I read, someplace, that the "proper" technique is to use a clean, linear amp to create the impulse and that the IR is then a model of the cabinet, the speaker within and the mic placement, all of which are supposed to be close enough to linear (not that I could give a succinct definition of "linear" in this context).

But, if the creator of the IR does it some other way (liked dimed settings on a non-linear amp) which can, say, introduce clipping or an odd EQ curve, apparently the resulting IR won't fully reflect the cab/speaker/mic. And, I don't get what it would reflect later, when you plug it into your modeler.

Does the Fourier transform work with clipping? Usually, elegant math requires some assumptions of things acting "nicely".

Linear just means the response can be described using a linear equation of the form y=mx+b. Anything other than this is non-linear. For example, compression is non-linear, as you have a linear response up to the threshold value, after which the slope of the response line changes. You can't describe the response with y=mx+b because m is not constant. There are varying degrees and types of non-linearity.

An impulse response can not reproduce any non-linearity. Depending on the level on non-linearity you try to capture, your impulse may sound a little weird (maybe if the source has some light compression) or completely awful (heavy clipping).

[jzucker]

I still think we are due for a better version of speaker emulation that utilizes frequency and amplitude dynamically. We do not have that currently. The dyna-cabs implementation of cabs in fractal and a similar concept in helix are an improvement but we need cabinets that vary with frequency and amplitude dynamically. Current IRs are incapable of that.

[rpjazzguitar]

So far, my limited foray into the world of modeling (which I've recounted in unfortunate detail in some other threads) has been frustrating.

The idea was to find a way to use a powered speaker and sound good.

But, I've had multiple experiences of going into PA systems - without modeling - which sounded just fine.

All had Guitar> ME80, then into:

Little Jazz > JBL line array (the one with 7 inputs). Sounded as good as the guitar has ever sounded. Not cheap and not small or light. I wish I had a chance to try bypassing the LJ to see if it really mattered.

Little Jazz > Yamaha MGU10 mixer > Mackie SRM350. Sounded great. Sounds good as loud as I'll ever play. A lot of pieces to carry and setup.

Then there was Guitar>Yamaha MGU10 > SRM350. Different night. By memory, it sounded about the same without the LJ. This test seems to suggest that the LJ doesn't add much. I thought the Yamaha's EQ helped and it may have warmed up the sound somehow, although I didn't think a mixer was supposed to do that.

The Rig Du Jour (earlier today in the practice room) was Guitar>ME70 (with cab modeling activated by putting in the dummy plug) > Bose S1 sounded good. The ME70 cab modeling seemed to help a bit on chords - my biggest modeling success thus far. But, the same rig sounded a little sterile with the Big Band earlier this week (huge room with tile floor and few soft surfaces), so some adjustment may be needed. I'll use this same rig on Friday's gig, adjust it differently and see how it does.

Putting the Korg PX5D in the system helped a little with chords but not high single notes.

The point is that modeling doesn't seem to be necessary in every situation to get the sort of tone I use, which is clean, a little reverb and tending towards the dark end of things. I'm finicky about how how notes ring and sustain. And, a lot of the modeling elements (mostly in the ME90) I tried were not helpful.