Question about IR's waveform being spikey

J

jimnov

Inspired
IR Guru's:
I played with Cab Lab 3 this weekend. Great product!
I was surprised to see how spikey the frequency response is of all the IR's I tried.
When I look at a Celestion speaker frequency response curve, it doesn't appear that way.
I understand how different mic's and placement will affect the capture curve.
One would think... if the mic and speaker have a reasonably flat response in the 85 Hz to 1200 Hz range, you would get a reasonably flat IR, in that range. The harmonic range 1k to 10 k looks even worse.
Is there a tool available to fix this? The existing tools in Cab Lab do help shape the tone, but I would like to be able to go in and fix/remove/smooth out the spikes. Is that even possible?

My end goal it to have great IR's of my existing cabs that I can send to FOH when I use cabs on stage.
I've gotten pretty close with the blending capability.
Just asking for the final 10 %.

Thx,
JN
 
Last edited:
Are you saying the sound is also overly spikey to your ears? If not, maybe those response curves we see in relation to speakers and headphones and such are just obtained using one method and the impulse responses are intentionally created in a different way; being exposed to energy that leaves a different signature on a graph. But I'm also curious and it will be something to read about further; or hopefully someone more involved in the creation of IR's can shed some light here.
 
G'day,

Have a look at the Y scale in the response graphs you are comparing. Perhaps you are comparing 1dB per Y unit to 10dB!

Thanks
Pauly
 
If you're referring to these Celestion curves:
graph.gif

Then that's a very unscientific curve that is for sure rounded = low resolution. Remember that every curve like this that you see has been captured by something, most likely a microphone and that curve is highly affected by it's position. Therefore those curves Celestion gives us are actually rather unhelpful. I could give you much more usable curves like this for all the speakers I've done.

But yes you are right. If there are too many spikes in an IR it's bad and unusable. You can try using dephase to fix it but IMO you'll get better results by finding a better IR to begin with. :)
 
jimnov said:
I was surprised to see how spikey the frequency response is of all the IR's I tried.
When I look at a Celestion speaker frequency response curve, it doesn't appear that way.
Click to expand...
Manufacturers' published frequency response curves are usually smoothed out. They're also free-field responses. Mount that same speaker on a baffle and put it into a cabinet, and the frequency response changes. "Spikey" doesn't mean inaccurate.
 
You can smooth out the spikes by turning dephase to 10 but that doesn't sound too good.
 
ML SOUND LAB said:
If you're referring to these Celestion curves:
graph.gif

Then that's a very unscientific curve that is for sure rounded = low resolution.
Click to expand...

Agree, resoluton is the key, I have the response graph of my Yamaha msr250 powered monitor in the manual and shows a thick line rounded that cloaks the ups and downs, if I look to the manual of the EV ZLX 12P I can see a graph with a very thin line that draws little spikes. Does this mean that the yamaha is more flat? my ears say no. The EV graph is more accurate and honest.

aziz said:
You can smooth out the spikes by turning dephase to 10 but that doesn't sound too good.
Click to expand...

Agree also, you can smooth to a certain point, but beyond this the IR loose his grace!
This applies also to smoothing parameter of tonematch block, if you go so far the tone loose his identity.
 
Last edited:
What's with all this snow in NY.....
Thanks for all the replies!
I just learned a ton of information.
I'm going to shoot some IR's this weekend with Cab Lab and keep exploring.
Is the IR capture process similar to measuring the frequency response of a home stereo speaker?
Are there any RTA programs you would recommend to shoot IR's. Have you ever used Pink Noise instead of a sweep?

JN
 
Last edited:
ML SOUND LAB said:
If you're referring to these Celestion curves:
graph.gif

Then that's a very unscientific curve that is for sure rounded = low resolution. Remember that every curve like this that you see has been captured by something, most likely a microphone and that curve is highly affected by it's position. Therefore those curves Celestion gives us are actually rather unhelpful. I could give you much more usable curves like this for all the speakers I've done.

But yes you are right. If there are too many spikes in an IR it's bad and unusable. You can try using dephase to fix it but IMO you'll get better results by finding a better IR to begin with. :)
Click to expand...

Hi Mikko.
Thanks for the offer.
I am using a Celestion Lead 80, Vintage 30 and a Warehouse Guitar Speaker ET-65.
JN
 
Morphosis said:
Here`s a visual example, what the guys mean with smoothing:

Both Curves show the identic IR, blue = no smoothing, red = 1/12 smoothing:
public.php
Click to expand...
Wow, this is great.
Having a visual really helps.
What software program are you using.
That big dip at 550 Hz is what I'm concerned about.
Do they sound different?
 
jimnov said:
Wow, this is great.
Having a visual really helps.
What software program are you using.
That big dip at 550 Hz is what I'm concerned about.
Do they sound different?
Click to expand...
Having a visual helps but you also need to know what to look for. F.ex. that dip is a character that exists in most IR's. You will not see that very well in the graph if you're monitoring your guitar tone. Remember that this IR curve is not exactly the same thing. Remember the main rule. Narrow holes are good, wide holes are bad. Wide bumps are good, narrow bumps are bad. That's just the basic rules of EQing altogether. Boost wide, cut narrow.

Spikes are not your enemy. Spikes are "the character" and if you don't have them at all you sound too transparent = fake = digital. Simulate this by turning dephase too high. That's sort of the effect. The whole point of having so many different IR's is to have different types of characters that sound natural. If you went in to boost that 550hz dip it would sound extremely unnatural and honky. So with certain amps a spike at a certain frequency sounds completely different than on other amps. This is why every Cab Pack is extremely powerful and opens new worlds to your Axe-Fx. That's the reason people get equally excited over new Cab Packs as they do over new firmware releases. :) As they should be.
 
I'm fairly sure the larger spikes like the one located around 550 Hz in the cabIR.eu graph are caused by some amounts of phasing. When placing a mic on a speaker you want to be careful to eliminate too large or too many large spikes, but they are still mostly what add different character to the whole process. Even when just using a single mic phasing can still either be friend or foe. You can't really pick out these sort of things by ear, which is why graphs like that are so helpful.
 
jimnov said:
Wow, this is great.
Having a visual really helps.
What software program are you using.
That big dip at 550 Hz is what I'm concerned about.
Do they sound different?
Click to expand...
As Mikko said, a "dip" in especially this are is seen in nearly all IRs i have take alook into. In general, dips like those can result from mechanical reasons of the cabinet/chassis/speaker, particular vibrations for example. Also not (well treated) rooms / cabinet/mic position in the room could result in (unwanted) dips / spikes.
 
NatKingAlford said:
I'm fairly sure the larger spikes like the one located around 550 Hz in the cabIR.eu graph are caused by some amounts of phasing. When placing a mic on a speaker you want to be careful to eliminate too large or too many large spikes,
Click to expand...

Exactly this is, for what we take care. But in this case, the mentioned spike results mostly from the cab itself, IMO.
 
ML SOUND LAB said:
Having a visual helps but you also need to know what to look for. F.ex. that dip is a character that exists in most IR's. You will not see that very well in the graph if you're monitoring your guitar tone. Remember that this IR curve is not exactly the same thing. Remember the main rule. Narrow holes are good, wide holes are bad. Wide bumps are good, narrow bumps are bad. That's just the basic rules of EQing altogether. Boost wide, cut narrow.

Spikes are not your enemy. Spikes are "the character" and if you don't have them at all you sound too transparent = fake = digital. Simulate this by turning dephase too high. That's sort of the effect. The whole point of having so many different IR's is to have different types of characters that sound natural. If you went in to boost that 550hz dip it would sound extremely unnatural and honky. So with certain amps a spike at a certain frequency sounds completely different than on other amps. This is why every Cab Pack is extremely powerful and opens new worlds to your Axe-Fx. That's the reason people get equally excited over new Cab Packs as they do over new firmware releases. :) As they should be.
Click to expand...

Got it.
So everything in the chain affects the end result:
Mic, Mic Preamp, Mic Placement, Room Reflections, Mic Cables, Speaker Cabinet, Type of Speaker Cabinet [Ported, Open Back, Sealed] Amplifier Freq. Response etc...
Was wondering if the IR's I made last night were accurate.
Made a comparison this morning with pink noise running through the same amp, same reference mic, at the same position.
Check out the pics below.
Cab Lab and the RTA are in complete agreement.
You may or may not like the resulting sound, but you can be sure it's accurate!

Cab Lab 3_ET-65_UR.jpg ET-65_Blue Cat RTA.jpg

Cab Lab_RTA Comparison.jpg
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom