Discussion in 'Digital & Modeling Gear' started by Jay Mitchell, May 17, 2019.
Not entirely IIRC.
So I guess those IR's in there were before you discovered how to make them fully reflection free?
Looking forward to trying your new IR's when I get them!
No. I've known how - and had the means - to collect reflection-free IRs since 1985.
Originally, ca. 2007, the length of IRs the Axe-Fx could accommodate was ~10ms (512 samples at 48kHz SR). Back then my R&D space - the one in which I acquired longer reflection-free IRs - was not always open and unobstructed. It was a sizeable project to clear that space and set up all my test gear. Because of the time it consumed, that work was reserved for actual speaker R&D (IOW, for revenue-generating endeavors). I acquired guitar cab IRs in a smaller space I used for production testing. I could capture reflection-free 10ms IRs in that space, so the 512-sample IRs I originally shared were indeed reflection free.
During that time, working in concert with AlbertA, I developed a workaraound that enabled us to double the IR length in the AFX to 20ms. In order to do so, I had to split every IR into two parts and implement a specific signal chain in the AFX that delayed the second half by exactly the right amount and added it to the first half. Shortly after that, Cliff revised the FW to accommodate 1024-point IRs.
I believe, but I cannot now be certain, that I windowed the second portions of the split IRs to remove reflections. I no longer have copies of those IR fragments, as they were rendered moot by the FW update. Furthermore, I don't recall whether I windowed the IRs I shared with Cliff for inclusion in the factory FW.
Since I relocated my company in 2013, I have a test space that always remains clear, and it is only a minor project to set up an R&D-level test session. I no longer have a need to acquire IRs in a smaller space. I now perform all my testing - even production QC testing - in this space.
@Jay Mitchell I've been thinking about the signal chain, from sine wave sweep to the recording of the IR.
Simplified chain: signal generator > amplifier > speaker > SOUND IN AIR > microphone > pre-amplifier > DAW... basically all of the transformations of the signal before deconvolution.
With regard to speaker IR's, when you use the term 'baked in' to describe room reflections within the first milliseconds of the capture affecting the subsequent frequency response of the IR, I understand that as an error in the response of the speaker under test.
But then I think about what I want, and/or don't want, to be 'baked in' with regard to the signal chain of the recording, it occurs to me that my choices are individual components: the speakers amplifier, the microphone, and the mic pre, as well as the knowledge that my signal generator and DAW are calibrated and doing what they are intended to do.
@AlbertA describes the care it takes to obtain the IR measurement of the SUT in #314.
IR's taken from the amplifiers are easy enough to do.
IR's taken from the transducers take a lot more care.
What I'm getting at is this: if the goal is to obtain a 'perfect reflection' of the speaker, that is but one component in the signal chain.
If great care is taken to accomplish that, it stands to reason you would want 'perfect reflections' of each component in the signal chain.
Which brings me back the the term 'baked in'; when I cook, I don't buy pre-mixed, I add to taste.
I don't like it when my paprika is already mixed in with the chili, or the sugar with the salt.
Assuming IR's replace each component in the signal chain when they are convolved, I want the individual IR ingredients in their "freshest" (best possible measurements) so I can do the cooking myself.
What are your thoughts on this?
Hasn't Jay already spoken to this in post #301?
And - as I have carefully pointed out - because you want the impulse response of the speaker, you must take steps to eliminate the effects of all the other components in the test chain. The mic, amp, and preamp are straightforward: quality examples of these components have near-ideal impulse responses and - this is important - impart no sonic signature of their own. Eliminating the acoustic signature of the test space is the challenging part. That's what this entire thread is about.
I'm wondering about what constitutes being 'baked in' for a combo amplifier IR.
A Deluxe Reverb amplifier was conceived as a combination of the amplifier and the speaker, they go together as a matter of convenience and portability.
Does that combination and resulting IR reflect the first two transformations in the chain; amplifier > speaker?
As opposed to a Marshall 4x12 cabinet where musicians use different amplifier heads to drive the same cabinet, depending on preference.
When you are obtaining an IR for a 4x12 cab, I'm assuming you would not use a tube guitar amplifier to do it, you'd use an amplifier that's designed for testing, the same way you would employ the best possible measurement microphone to obtain the response of the speaker in the room.
If I'm reading you right, if I make an IR of a Deluxe Reverb amplifier, I would first obtain the IR response of whatever speaker is in the cabinet, using a testing amplifier, and then I would convolve that with one of many IR's of fender amplifier tube circuits.
Capturing an IR of a tube amplifier is pointless cuz IRs can't reproduce non-linearities (i.e. compression, distortion, etc.).
To roughly replicate the non-linearities of an amp you would need multiple IRs of the amp captured at different input signal levels and an algorithm that interpolates them depending on the input signal (that's what Acustica Audio Nebula VST and, I think, Kemper do).
But I think it's better to leave those tasks to modelers imho.
You are making this far more complicated than it is.
And, when you're using an IR of the cab, it's pretty much always going to be in a modeler or software simulation. That's what will simulate the amp. The IR is there to simulate the cab.
You are not. Impulse responses are useful to capture linear properties of devices. Guitar amps are highly nonlinear. You can't use impulse responses to capture their properties.
This has gone 'way OT. Let's try to bring it back, please.
I see, that makes sense, thanks @dlc86.
I had an AxeFX III last summer and the amp modeling (...it's modeling, correct?) was what impressed me the most about it.
Ok, didn't mean to get off track, getting back to sound in the measurement space...
Can you even get into the far field in a chamber that small?
Based on the images of the chamber included with the IR's, 10 feet appears realistic.
One of the files has "4m" in the title. That could be possible with the mic and speaker near corners. In any case, the magnitude of the reflections is such that the effort to take that data in the chamber is wasted.
I was itching to try out Jay's IRs but only got back to my guitar and HX Stomp this afternoon so I may be a bit late to the party. My setup is simple: a modified late-70s Strat into the HX Stomp, and my patches usually consist of an amp, cab and a bit of reverb (I find the stock patches useless). I got the HX last month, I have used it exclusively over headphones thus far, and it is my first real modeller (apart from a truly awful PocketPod). I've tried the stock cabs and a few free IRs.
Now, to Jay's three samples:
A is an IR, similar to many other IRs. Sounds good as far as it goes, but can be tiring and weak, in particular with clean sounds. To be fair, these drawbacks only become apparent once compared to C.
B is weird, that's the only way I can describe it.
C stands out from all the stock cabs and any IRs I have downloaded through its clarity and presence. It makes the guitar sound more like a guitar than like something simulating a guitar. To test this IR, it put it into various patches I made (with the Stone Age, Deluxe, Twin, Litigator and Plexi models). It works best with the first four of these, but even with the Plexi it sounds good with a bit of high cut applied. I could probably live with just three IRs - this one, one built along similar lines for a Marshall 4x12 and one for a Bassman 4x10.
Caveat: I am not a tonehound or audio expert. All I am trying to do is get my crappy playing, warts and all, into headphones so that I can improve on it. IR C really gets me much closer to the feel of playing an amp while stuck under headphones. Thanks Jay!
One more thing re:the "anechoic" IRs. One of the photos in the zip file shows a computer screen with a display of the speaker's magnitude response. Had that display been of the log impulse squared or envelope time curve ("ETC") - IOW, exactly the same data but presented in a different format - the presence of problem reflections would have been intuitively evident to a competent practitioner, and measures could perhaps have been taken to address them. Worst-case, it would then become clear that taking further data without fixing the problem would be futile. My money says that, until I did so earlier today, nobody has pointed out the presence of obvious reflections in the "anechoic" IRs this person is sharing.
This goes directly to what I said earlier: even when given the means to acquire good data, novices are prone to overlooking problems and taking bad data. I've encountered the syndrome many times over the past 30+ years. And, to be clear, when I use the word "novice" in this context, I am referring to anyone who lacks extensive experience in performing loudspeaker testing and measurement. People who fall into that category include studio recordists, electrical engineers, and DSP codewriters. Good judgement is the result of experience, and experience is the result of bad judgement.
This is a fascinating thread and I'm glad to see discussion around far-field IR's again (haven't really read anything about them for some time) as I've been interested in them for several years now but have had a distinct lack of ones to experiment with as they seem to be difficult and time consuming to acquire and therefore not available in general.
I'd love to audition the ones you've made available here Jay (PM sent) and have a few that Antcarrier made a couple of years ago and put up on the Fractal forum that I'm definitely going to revisit.
While clips don't tell the whole story, the ones in this thread have a very natural, 'real' cab tonality that I find very exciting and has re-ignited my curiosity.
Well crafted far-field IR's would greatly simplify my rig and I very much like the idea of the sound of only the cab and nothing else in my cab block. I'd like nothing more than to have a couple of 'real cab' IR's and get out of the near-field rabbit hole one can easily find themselves in. While I have efficient methods of auditioning the hundreds of near-field IR's I have it would be truly awesome to have IR's of cab/cabs as you'd hear them while playing. Indeed, I've been on that trail for several years now...
Eliminating the room/reflections, the mic, the variations of mic positioning, etc. from influencing the cab sound is a holy grail for me. While I have, and use, near-field IR's that I like there are simply too many variations and many 'artifacts' baked into them.
I also find it encouraging that users here are reporting that they don't have to go through EQ contortions with the far-field IR's that you generally have to do with near-fields, particularly in taming/shaping the low end.
Yep, both samples I posted are amp only (Helix BE-100) + delay, no Eq and no post prod.
Spoiler: HX Edit shot
I don't post a lot on TGP, but it's interesting that I found a post of mine from 2013 regarding far-field IR's (this topic brings me out of the woodwork):
Oct 1, 2013:
I'd certainly welcome a selection of properly acquired far field cabinet IR's as I'd love to have the true sound of a speaker cabinet as you would hear it from 'playing distance' with all of it's transducers interacting, etc. IOW, the true sound of a speaker cabinet.
I said it before and I'll say it again, I suspect there are a LOT of players who would also be all over 'pure', reflection free far-field IR's which are 'done right'...just the sound of the cab and nothing more. I agree, strongly, that this would be a game changer indeed...I'm surprised that, many years later, far-field IR offerings/tech are still very much 'in the background' and hasn't had wider experimentation.
I've just tried the 3 IRs Jay provided (Yeah, bit late to the game). I am using and Atomic Amplifire to load the IRs, although the amp simulating duties are done by an Origin Effects Revival Drive. Listening through some average monitors (looking forward to try them, mostly C, in rehearsal). For testing purposes I turned off the reverb on the unit.
B is just weird. It's like it's got a stuck phaser or flanger baked in! Don´t see how it could be used for anything (unless you're looking for that specific sound in a specific context).
A sounds pretty good, altough a bit bassy for me, but at the same time has a certain cocked-wah sound to it (albeit very mild). Kind of remains me of Brian May's sound, for some reason.
C is just great. It's the sound of rock and roll right there! Well balanced and natural sounding.
Not surprised with the quality of Jay's IRs, though, since I've been using his Marshall Cab IR since I found it.
Many interesting infos in those 19 pages, even if I'm not understanding most of it, it is quite technical aha
A question, having a "reflection free" IR shouldn't have any advantage if you are monitoring with headset/in ears right? The mic is capturing exactly what you would hear with your head at it's place and the in ear has no additional reflection
(Assuming it is a far field IR)
I may want to try those 3 IRs anyway, can they be used with an helix?