Properties of Reflection Free IRs

csweldon

Member
Messages
886
Here are some more charts. The first pair contains the response of a far field IR that I could use. The second pair contains the close-mic'ed response of the same cab with the mic position chosen so as to produce a similar spectral balance. The third pair contains the response of a far field IR of an open 1x12 cab (not the same cab as the first two) taken by someone else under conditions unknown to me.

Below is the log IR^2 of the same cab as before (1x12 open back), taken with the same mic at the same distance, but at 20 degrees off axis.



This is its far field magnitude response.


Here's close-mic'ed data for the same cab with the mic placed between the center and edge of the cone.




Here is data on the far field IR of an open 1x12, taken under undisclosed conditions.​




I've got all three loaded in my modeler and have auditioned them. Because the third IR is not from the same cab/speaker as the first two, some of the sonic differences are due to that. However, the apparent deep - and very audible - suckouts in its response from ca. 200-500Hz and 1200-2kHz are definitely not due to the intrinsic behavior of the cab and speaker. They are caused by early room reflections that were captured in the IR.
If you'd like to audition these IRs, shoot me a PM.
I am unable to to PM you due to forum privacy settings, but I would very much like the opportunity to audition these if at all possible!

Regardless, I have been following this thread with great interest and appreciate you taking the time to explain things to us all.
 

TheSchwartz

Member
Messages
286
If the structural/acoustic integrity of a cab is a factor in its sound, then that's its own problem. Generally speaking, I'd say an IR of such a cab would be of little value, except possibly as a negative example.

OTOH, if, by "construction of the cab," you mean the dimensions, aspect ratios, type of cab (open, sealed, ported), presence/absence of internal damping material, etc., then a close-mic'ed IR will be much less affected by those parameters as compared to an IR taken at a realistic distance.
Thanks, Jay. It was the latter I was wondering about. I really have to say that coming to understand IRs and miking techniques has been extremely valuable, especially since almost all of my music these days is made using modelers. In my opinion, people really can’t be successful with modelers unless they have at least a rudimentary understanding of guitar recording techniques.
 

gigsup

Supporting Member
Messages
1,314
We would put the sub in the middle of an empty parking lot and lay a calibrated measurement mic on the ground 2 meters away.
Essentially a boundary microphone. Another half space option is to sink the speaker into the ground so that the speaker is parallel and flush with the ground surface and then change the distance of the mic from the speaker, versus putting the speaker high enough in free space to eliminate all reflections below a specific frequency.
But if you have a lift or a boom and some rigging, it's not a big deal. Probably a better solution for open back and vented cabinets.

If the goal is to eliminate reflections in the measurements of the speaker, as it appears to be, what do you have to work with?
Distance from the nearest possible reflective surface outside of the bandwidth you want to measure is your friend.

What else? Sound waves propagate at about 1100 ft./sec in air. Air is a gas.
A smaller room filled with a different gas where sound moves at a different velocity?
 

Chocol8

Member
Messages
314
2. That a 200ms (or 500ms) IR will add useful ambience.
This is another thing I had not previously thought about or realized. I assumed a close mic’ed IR captured and would allow the recreation of a close mic’ed recording.

Now that I realize the IR captures some but not all of the room effect I am starting to understand some of the issues with IR’s better.

I am also starting to see that neither guitarist nor the guys recording IRs seem to have a sufficient level of understanding of how they work. That in itself explains a lot!

My takeaways so far are:

IR’s should be reflection free to accurately model a cab. (Accurate, not necessarily better sounding which is subjective)

Standard 200ms IR’s can’t fully reproduce room ambiance.

With an FRFR, the reflections and room ambiance is added back naturally (so they really shouldn’t be in the IR).

For recording or headphone use ambiance needs to be added separately.

Thanks again for the thread Jay. You have definitely helped me understand better, and I hope to learn more as it progresses.
 

hippietim

Silver Supporting Member
Messages
6,462
IR’s should be reflection free to accurately model a cab. (Accurate, not necessarily better sounding which is subjective)
This is the most important takeaway IMO. It never really occurred to me before. But duh, there it is smacking me in the head. You just have to stop and think about it critically for a second. When you use any IR that's not reflection free, you are running your amp sim into a cab sim with reflections from some other room and via your playback system you're adding reflections from the room you're in.
 

Jay Mitchell

Member
Messages
5,648
Here's a set of short comparison clips:

Dry

Early reflection -6dB

Early reflection-3dB

I created the dry clip several years ago for a now-forgotten tone-demonstration purpose. The first clip was made with a reflection-free IR in an Amplifire 3 pedal. I may or may not have had the reverb block active; I don't recall. In the second clip, a delayed (2ms), lowpassed (2kHz first order) copy of the dry clip was mixed in at -6dB. This is a pretty good simulation of a floor reflection when a 4x12 is mic'ed on one of the bottom speakers. In the third clip, the level of the delayed track is mixed in at a higher level (-3dB).

The point of this exercise is twofold: to establish that, yes, you can record with a reflection-free IR and that, if you get a guilty conscience that your dearly-beloved floor reflection is missing (I never miss it myself, but that's my personal taste), you can easily add one in as a postprocess.
 
Last edited:

Watt McCo

Member
Messages
10,270
This is the most important takeaway IMO. It never really occurred to me before. But duh, there it is smacking me in the head. You just have to stop and think about it critically for a second. When you use any IR that's not reflection free, you are running your amp sim into a cab sim with reflections from some other room and via your playback system you're adding reflections from the room you're in.
The point is finer than that. The problem with a 200ms IR isn't that it includes reflections, it's that it ONLY includes early reflections. Early reflections outside of the context of full ambience sounds weird (for the technical reasons provided by Jay).

Even Jay says he applies reverb or other ambience effect to his reflection free IR when recording - our ears are very used to hearing a recording of an instrument in a room played back in a different room.

I totally get that reflection free IRs are the way to go if you're trying to get "amp in the room" sound/feel from your (properly placed) FRFR system.

I can also see using a reflection free IR in combination with an long IR of a room being a good way to model a room-mic'ed cab.

What I still can't wrap my head around is how they might be used if someone was trying to model a close-mic'ed cab. You'd have to add a filter to account for the mic, including proximity effects, at the least.
 
Last edited:

Jay Mitchell

Member
Messages
5,648
The point is finer than that. The problem with a 200ms IR isn't that it includes reflections, it's that it ONLY includes very early reflections.
No. "very early reflections" are those that occur within 30ms of direct or less. Reflections at and beyond ca. 50ms are easily perceived as "echoes."

Even Jay says he applies reverb or other ambience effect to his reflection free IR when recording -
And you just stated (correctly, for once) above that a 200ms IR can't capture ambience. You're arguing at cross-purposes with yourself here.

our ears are very used to hearing a recording of an instrument in a room played back in a different room.
But - this is what you stubbornly refuse to get - our ears are not very used to hearing a recording of an instrument in a room played back in an incomplete portion of a different room.
 

Jay Mitchell

Member
Messages
5,648
What I am trying to figure out is this: is there a better way to model a cab mic'ed within an inch of the grill with microphone X than by properly capturing a 20ms IR with microphone X placed within inches of the grill.
You should ask someone who pursues such efforts, then. You are polluting this thread with OT and nonproductive debris. Please stop.
 

Chocol8

Member
Messages
314
What I still can't wrap my head around is how they might be used if someone was trying to model a close-mic'ed cab. You'd have to add a filter to account for the mic, including proximity effects, at the least.
That’s a point that I think is still an unsolved issue. If we can simulate the room reflections in post processing, the main problem is how do you get a reflection free close mic’ed IR? (Assuming you want a close mic’ed sound)

I think you are combining reflection free IR with far field ground plane reflection free IR where the later is merely a subset of the former.

The problem as I currently see it, is in capturing a reflection free close mic’ed IR and I doubt that is insurmountable. If digitally correcting for the reflections is too difficult, you need to capture the IR in free space or a truly or nearly so non-reflective room. Now the latest revelation that all we need is 20 ms and not 200 means you only need (I think!!!) to get the mic and cab 3.5 meters from a reflective surface rather than 35 meters. That means a room at least 7 meters wide and long with at least 7 meter ceiling height. That is a much more manageable size as many gyms, warehouses etc. would suffice. I don’t really know how to go about capturing the IR in 20 ms yet so my understanding could be completely off here.

The other approach you mentioned of taking a far field IR and processing it into a close mic’ed IR seems more difficult if not completely impractical.
 

Jay Mitchell

Member
Messages
5,648
Here's some more "theory:"





This data is from a stock Helix cab, and it's actually pretty interesting. It appears that the first ~3ms are from a raw IR and that what comes after has been lowpass filtered (reduced detail) and possibly decimated. This would be the "ultrares" concept applied to a short IR. From what I understand, the stock cabs in the Helix use less cpu power than do imported IRs, which the appearance of this IR tends to support.
 
Last edited:

burningyen

Member
Messages
14,623
This data is from a stock Helix cab, and it's actually pretty interesting. It appears that the first ~3ms are from a raw IR and that what comes after has been lowpass filtered (reduced detail) and possibly decimated. This would be the "ultrares" concept applied to a short IR. From what I understand, the stock cabs in the Helix use less cpu power than do imported IRs, which the appearance of this IR tends to support.
That is surprising, because the Helix's stock cabs don't sound lowpass filtered to my ears.
 

Fractal Audio

Member
Messages
1,115
Lots of good info here.

One minor thing I might add is that in the vast majority of cases I've found that IRs are very nearly minimum-phase.

Here is a plot of a the raw vs min-phase IR for a randomly selected IR:


As you can see they are nearly identical. The min-phase IR has been delayed by 23 samples so that it aligns with the raw IR.

MPT'ing an IR imparts almost no perceptible change to the tone and makes mixing IRs vastly easier.
 

Mark Al

Member
Messages
838
Here's some more "theory:"





This data is from a stock Helix cab, and it's actually pretty interesting. It appears that the first ~3ms are from a raw IR and that what comes after has been lowpass filtered (reduced detail) and possibly decimated. This would be the "ultrares" concept applied to a short IR. From what I understand, the stock cabs in the Helix use less cpu power than do imported IRs, which the appearance of this IR tends to support.
The phrase "reduced detail" v.s. "ultraRes" seems contradicts each other.... :) I am not familiar with the "ultraRes" concept, could you elaborate it a bit, or link something I can read-up? This is not the same "ultraRes" from Fractal right? @cliffc8488 :)

Hypothetically, with the reduced detail approach, what's the pro and cons of doing this? Is it to achieve certain/better sound/tones or is it potentially purely for pragmatic purposes, like saving disk spaces?

Lastly, from the graphs, would you be able to observe reflection information from the Helix stock cab?

Thanks!
 

Jay Mitchell

Member
Messages
5,648
The phrase "reduced detail" v.s. "ultraRes" seems contradicts each other....
It makes more sense than it sounds like at first.

Hypothetically, with the reduced detail approach, what's the pro and cons of doing this?
Pro - for any given IR length, it saves cpu horsepower.
Con - when used as L6 is apparently using it here, it can run the risk of obscuring some details of the sonic signature of a cab.

Lastly, from the graphs, would you be able to observe reflection information from the Helix stock cab?
Not definitively. The reduced rate of decay from ca. 3ms onward - and the peaks that follow - could be due to early reflections.
 




Trending Topics

Top