Some subjective listening impressions-
Again, this is with NO REAL PROCESSING!!! again, other than PEQ there is no other digital experimentation and NO REAL PROCESSING/FIR!!!
Last night, just for laughs, I high passed the arrays with some basic in-line filters (capacitors) I had lying around.
To my ears, It definitely cleaned up the arrays at high volume playback.
Ill see if I can measure/confirm some of this in the distortion values of REW.
These filters are rated as 150Hz high pass into 8oHms, 1st order.
I realize I can do this with the miniDSP-HD,
but since I'm using it as a DAC...going from the DSP to my pre-amp,
which outputs to to L+R mains and L+R sub, I did not want to high pass digitally, depriving the subs of any native bass frequencies.
Another reason I didn't high pass with the DSP is I had hoped the small sealed enclosures would have restricted excursion, but I don't think that's happening much.
After all this, and my complaints about a lack of upper bass/mid bass response (again, other than PEQ there is no other digital experimentation and NO REAL PROCESSING/FIR!!!), one could come to the conclusion that I may have been better off building somewhat larger, and therefore ported enclosures (arrays).
That may have been a better starting point. They certainly would have dug deeper, and perhaps I'd be having less trouble blending them with my subs...
This is what led to my prior "musing" of building another set of small bass arrays to go next to these.....a kludge arrangement to having not built ported enclosures to begin with...
Again, this is with NO REAL PROCESSING!!! again, other than PEQ there is no other digital experimentation and NO REAL PROCESSING/FIR!!!
Last night, just for laughs, I high passed the arrays with some basic in-line filters (capacitors) I had lying around.
To my ears, It definitely cleaned up the arrays at high volume playback.
Ill see if I can measure/confirm some of this in the distortion values of REW.
These filters are rated as 150Hz high pass into 8oHms, 1st order.
I realize I can do this with the miniDSP-HD,
but since I'm using it as a DAC...going from the DSP to my pre-amp,
which outputs to to L+R mains and L+R sub, I did not want to high pass digitally, depriving the subs of any native bass frequencies.
Another reason I didn't high pass with the DSP is I had hoped the small sealed enclosures would have restricted excursion, but I don't think that's happening much.
After all this, and my complaints about a lack of upper bass/mid bass response (again, other than PEQ there is no other digital experimentation and NO REAL PROCESSING/FIR!!!), one could come to the conclusion that I may have been better off building somewhat larger, and therefore ported enclosures (arrays).
That may have been a better starting point. They certainly would have dug deeper, and perhaps I'd be having less trouble blending them with my subs...
This is what led to my prior "musing" of building another set of small bass arrays to go next to these.....a kludge arrangement to having not built ported enclosures to begin with...
You're welcome 🙂
I have checked their website and I see no mention of a Frequency dependent window option. I can understand that it works nicely with your QSC setup, but unless it actually does have a frequency dependent window option then it is not the same. If it does have the option and you can show me I'd like to see.
This matters much more for indoor measurement and tuning, the difference is not night and day but is an important part of the puzzle.
Maybe, the knowledge of how to use it properly is vital but there are differences between programs that do matter when trying to do certain things.
Very true and if the program is not automated in that step, getting that right is paramount.
Your experience is always valuable and you two may well be on the same page with your likes and dislikes, I just felt the need to address some points because they are important and I could see them getting lost.
I also try to limit my posting these days to where I think I have something useful to offer.
Good stuff fluid!
Yeah, their website doesn't make the frequency dependent correction very apparent.
It's in the Auto Magnitude and Auto Phase components.
The FirDesigner and FirDesignerM versions allow an unlimited number of bands in those components, each of which can have their own degree of smoothing applied.
By looking at the FIR mag and impulse response that will be generated, along with panels that show their windowing error, and wavelet, it's pretty easy to see when the Auto corrections need to be split into bands.
Auto also can take Coherence into account, backing off of corrections with low measurement Coherence (from Smaart, Systune etc)
Other things that make it my go-to program; are it's ability to work fully manual ala rePhase style or mix manual and Auto, the ability to define target curves via unlimited bands, and the ability to add manual Voicing corrections into the file in use, if measurements show such would help.
It's also helpful to have measurement averaging built into the program.
Lately, I've been playing with VituixCAD's FIR building capacity, which i believe from the look of the impulse responses generated, is probably using pure impulse inversion.
I say that because the impulses look incredibly messy compared to the same corrections made with FirD. But they dang sure work, at least electrically.
Both programs produce super flat mag and phase...well VCAD is perfect, whereas FirD is near perfect.
I think this impulse difference highlights the need for a program to apply judicial correction, and completely agree for the need to have frequency dependent correction.
I'm about to start some listening tests of the VCAD files...will see what a truly messy impulse making flat mag and phase sounds like lol.
Hey, here's a related aside. And it gets to why i'm a fan of steep xovers.
When i make corrections to a driver, it's usually to get mag and phase flat thru xover down to -40dB.
Generally, on even a half-decently behaved driver, only one Auto band with one level of smoothing is needed across the driver's response range....when xover is steep (96dB oct, lin phase of course).
When i try, say 24dB oct xovers, getting response corrected down to -40dB almost always requires making extra Auto bands for the slower rolloff upper and lower response tails, to be able to soften the correction in the tails to avoid filter error. Hope that made sense.
Last edited:
It doesn't have to be the limited movement that made things better. It could very well be the balance was better. I'd say use what you've got for a while and learn from that. Then you may get new ideas to improve.
Get busy measuring, that will tell you what you've got so far...
I'm not convinced you'd need ported enclosures, bigger enclosures? Maybe that would have helped, bringing down the resonance frequency. Speaking for myself, there was a good reason to build free standing arrays. As that allowed me to play with placement and toe in etc. to optimize the results with regards to the room. I've never regretted that part.
I am trying to understand this, don't shoot me but unless there is another function this is not frequency dependent windowing but frequency dependent smoothing. I can see how that might be similar in effect but the result might be quite different, I have tried to do something similar manually but I couldn't improve on my baseline processing so I abandoned it as too much effort. This though seems quite easy.
What DRC does is set windows based on sample lengths (width) at different frequencies which can be changed through the configuration file. This is an example
Code:
L - Band: 0, 20.0 Hz, width: 9500, FIR, convolution...
L - Band: 1, 25.2 Hz, width: 6804, FIR, convolution...
L - Band: 2, 31.7 Hz, width: 5013, FIR, convolution...
L - Band: 3, 40.0 Hz, width: 3764, FIR, convolution...
L - Band: 4, 50.4 Hz, width: 2865, FIR, convolution...
L - Band: 5, 63.5 Hz, width: 2203, FIR, convolution...
L - Band: 6, 80.0 Hz, width: 1706, FIR, convolution...
L - Band: 7, 100.8 Hz, width: 1329, FIR, convolution...
L - Band: 8, 127.0 Hz, width: 1040, FIR, convolution...
L - Band: 9, 160.1 Hz, width: 816, FIR, convolution...
L - Band: 10, 201.8 Hz, width: 642, FIR, convolution...
L - Band: 11, 254.5 Hz, width: 506, FIR, convolution...
L - Band: 12, 320.5 Hz, width: 400, FIR, convolution...
L - Band: 13, 403.4 Hz, width: 317, FIR, convolution...
L - Band: 14, 508.8 Hz, width: 251, FIR, convolution...
L - Band: 15, 641.8 Hz, width: 199, FIR, convolution...
L - Band: 16, 809.5 Hz, width: 158, FIR, convolution...
L - Band: 17, 1017.9 Hz, width: 126, FIR, convolution...
L - Band: 18, 1288.2 Hz, width: 100, FIR, convolution...
L - Band: 19, 1620.0 Hz, width: 80, FIR, convolution...
L - Band: 20, 2041.6 Hz, width: 64, FIR, convolution...
L - Band: 21, 2590.0 Hz, width: 51, FIR, convolution...
L - Band: 22, 3265.6 Hz, width: 41, FIR, convolution...
L - Band: 23, 4127.9 Hz, width: 33, FIR, convolution...
L - Band: 24, 5148.3 Hz, width: 27, FIR, convolution...
L - Band: 25, 6485.1 Hz, width: 22, FIR, convolution...
L - Band: 26, 8186.3 Hz, width: 18, FIR, convolution...
L - Band: 27, 11099.8 Hz, width: 14, FIR, convolution...
L - Band: 28, 13502.4 Hz, width: 12, FIR, convolution...
L - Band: 29, 17234.6 Hz, width: 10, FIR, convolution...
F - Band: 30, 20000.0 Hz, width: 9, FIR, completed.Attachments
I didn't want to bring the above up, but that's basically what has been essential in packages like Audiolense, Acourate and DRC-FIR.
You get to determine the window length, shape and size and can alter it to act different at different frequencies. The most time I've spend is to learn what that can do for perception. I've exaggerated the results to hear what it does, zooming in to the (for me) base properties I needed. A lot of work? Yes it is.
But is it really a punishment to listen to a lot of music? No way! It was a cool way to learn the game for me. And I'm not nearly done learning.
Of the above mentioned packages DRC-FIR is the one where you do everything almost manually, giving it a lot of control but its a lot of work too. However what I had to offer was a pré-baked solution, so to speak. All it takes is time... the software is free...
You get to determine the window length, shape and size and can alter it to act different at different frequencies. The most time I've spend is to learn what that can do for perception. I've exaggerated the results to hear what it does, zooming in to the (for me) base properties I needed. A lot of work? Yes it is.
But is it really a punishment to listen to a lot of music? No way! It was a cool way to learn the game for me. And I'm not nearly done learning.
Of the above mentioned packages DRC-FIR is the one where you do everything almost manually, giving it a lot of control but its a lot of work too. However what I had to offer was a pré-baked solution, so to speak. All it takes is time... the software is free...
Last edited:
Hi fluid,
Yes, my take is they are different implementations to achieve the same goal.....goal being to use an appropriate level of frequency resolution across the spectrum.
Like shown below in the DRC comparison chart to conventional smoothing.
Looks to me that the DRC method achieves frequency dependent resolution via frequency dependent windowing, or rather maybe better termed frequency dependent FIR file size.
(Quick question: does DRC allow or use different widowing ala rectangular, Hamming, cosine etc on each individual band, or does it use a universal window setting across all bands?)
Whereas like you say, FirD method lets you divide the bandwidth into however many sub bands you want, and apply different levels of smoothing to each sub band. And yes, it is quite easy.
The real-time visual tools that FIrD has to determine when to make sub bands, and what smoothing to apply are very helpful.
Sometimes, I've even found it helps to skip any correction at all in small narrow bands. Measurements back that up because they inevitably lack Coherence there, with or without correction.
Which method ultimately works better? Who knows? Would be interesting to find out.
For multiways, DRC's complexity just doesn't seem to be worth it for me🙂
If i were only working on a line array, i might give it a try, especially given that I'd need a FIR file to span the entire freq range.
Multiways of course allow different FIR file sizes for each section, to help jumpstart the correct resolution process if needed..
I've been toying with idea of a way to let us all compare our FIR programs.
Just spitballing an idea here, but what if we all took the same full range measurement from a TC9, agreed on a common high pass filter, and then each built a correction FIR.
I would love to see how the various FIR files impulse responses compare, particularly in terms of their cleanliness for starters.
Attachments
Hi wesayso, That's really cool, to learn a tool so deeply. And glad for you, the results you're getting 🙂
I guess one of the reasons i don't dig deeper into such, is in the end i feel like many of the things i try with processing, ultimately are just different forms of EQ. And are just another form of knob twisting to find what pleases me most.
So I look for the method that gives me the cleanest repeatable measurements, and then tend to leave it at that.. of course, provided it sounds right 😀
I suppose that is where I disagree on the goal, I think getting the response tuned based on what happens in time is the real trick. Smoothing inevitably comes from windowing but they are not interchangeable.
You can see similar things in REW by using it's FDW vs smoothing and setting the range to only see the effect on specific frequencies.
There is no setting to change it, the only option is sliding filtering or 1/n octave band filtering. I use sliding.
I can see why you like it, it matches in with SMAART, dual channel measuring and tweaking as you measure it, all your favourite things 😀
That is a good question and I have tried my best to find out. I started with help from wesayso and a template that was close to his. I then went all around the houses trying to find a "better" way to do it, surely I could find something else to improve on. I got quite creative and learned a lot about REW and DRC during the process, multipoint measuring, vector averaging and all that stuff.
What I have come back to is pretty much what I had at the start🙂 The difference is that I correct to flat with DRC so I have a level playing field based on time and then I use shelving filters to set the tonal balance by listening. It is within an nats whisker to measure the difference between a DRC target aiming for the same and the shelving but the target never sounds quite as good.
I can see how tis long term non repeatable fiddling doesn't appeal to you, it doesn't appeal to me either but I have not been able to do it better in any other way.
The complexity comes from learning the program. Once you know what it does and how it does it the complexity vanishes. In the ABEC help the program is described as simple and straightforward, and it really is once you understand how to drive it. When you first look at it though you think who the bleep thinks this is simple 🙂 Having said that DRC is designed to operate as a full bandwidth "room correction" (I really hate that phrase) tool. But it is flexible enough to do much more although I wouldn't recommend anyone use if for that purpose unless they already know the program well.
All of my comments here apply to line arrays in room because that is how I have used it, and this is a thread about line arrays in a room after all. When I build my next speaker maybe I will find a different path gives me the best results. You have the jump on me there for sure 😉
Interesting idea that would allow some comparison but ultimately I think it wouldn't achieve what you want. I don't have a preset for it and without listening to the speaker and tuning it in my room I couldn't guarantee if I would like what I came up with. I have tried to help people do this remotely with somewhat mixed results.
This is very true, but the results of using EQ based on different measurements, windows and all the rest does give different results which is why I am trying to point out the differences between how the programs function. Which you like most is your choice but if I bought a Granny Smith because you told me you like apples the best I might get a surprise if you made that statement based on eating a sweet red apple 🙂
I do hope some of this is of value to the OP...
I’m glad for all the discussion.
Most of it is Chinese to me, but I enjoy reading it and trying to learn more.
I did finally get up the panel absorbers. They are very easy to put up or take down. Now it’s time to get back to playing with REW in an attempt to improve the arrays some more…
Meanwhile…my son still enjoys it as is LOL
Most of it is Chinese to me, but I enjoy reading it and trying to learn more.
I did finally get up the panel absorbers. They are very easy to put up or take down. Now it’s time to get back to playing with REW in an attempt to improve the arrays some more…
Meanwhile…my son still enjoys it as is LOL
Attachments
Rooms looking great !
Nice work on the panels.
Our kids are often the beneficiaries of our audio obsession huh ?
My 4 sure have enjoyed it...
Your fine looking young man appears so too 🙂
Glad all the FIR talk hasn't been an issue...pls say when/if it is
Nice work on the panels.
Our kids are often the beneficiaries of our audio obsession huh ?
My 4 sure have enjoyed it...
Your fine looking young man appears so too 🙂
Glad all the FIR talk hasn't been an issue...pls say when/if it is
Thanks Mark!
Right after I mounted the panel absorbers, I realized I did not really heed Rons advice at trying to get away from Symmetry…
They are easy enough to take up and down, perhaps I’ll try to stagger (offset) them??
I am a bricklayer by trade, it’s in my DNA to keep everything plum and level, with uniformity and symmetry…
I’ll have to get my head out of the box LOLOLOLOL…
I think I will he’d Ron’s other advice and leave everything as is for a while……
and play with REW adjustments.
Right after I mounted the panel absorbers, I realized I did not really heed Rons advice at trying to get away from Symmetry…
They are easy enough to take up and down, perhaps I’ll try to stagger (offset) them??
I am a bricklayer by trade, it’s in my DNA to keep everything plum and level, with uniformity and symmetry…
I’ll have to get my head out of the box LOLOLOLOL…
I think I will he’d Ron’s other advice and leave everything as is for a while……
and play with REW adjustments.
Measure first! If you can get away with the symmetry without it creating much of a problem, you've actually came out ahead. Each room is different, lets see it first.
I hear ya on the creation part, it's also a lot of fun! Gratifying work 🙂. But it doesn't have to be audio related for me,
I've worked on cars, motorcycles etc. Always something keeping me busy.
I hear ya on the creation part, it's also a lot of fun! Gratifying work 🙂. But it doesn't have to be audio related for me,
I've worked on cars, motorcycles etc. Always something keeping me busy.
Last edited:
Yep, sound like you, Ron, and I are birds of a feather, always working on something.
So you have the devil level plumb bob eyes too haha.
I was in elevator construction, and learned to spot out of plump first, then came level, square, etc. Both a blessing and a curse huh? 😉
But seriously, I really enjoy the construction trades..
Hey Ron, my 2nd greatest passion after audio is old two-stroke standup jetskis, hot-rodding and riding. (after getting into too much trouble with cars 😡). Even got a dyno at the height of modifying..
Here's to a fellow motorhead
Interesting....I've always thought getting the response tuned based on what happens in time, is simply a given standard task for any competent FIR generator.
If we are not screwing up, it seems like all we are really working with is mag and phase (time).
And that the goal for me anyway, is to know what mag and phase, can and cant be corrected, when and where in the frequency range.
I've come to the conclusion the only regions of speaker response that can be successfully corrected are the true minimum phase regions (which never include xover regions).
And the trick in fixing minimum phase response variation, once they are determined, is finding how much correction helps up to the point it hurts..
So inside those regions, it seems like it's a pretty straight forward back and forth, DFT to IFT, to correctly handle mag and phase (time), without alot of room for much methodological variation.
Beyond that, it appears any methodological differences are about psyche tunings or EQ preferences or such....but i just can't see where time is a factor..
To try to understand what you mean by the trick is in time...
can you give an example of some time issue/fix/optimization that could not be done in rePhase ? (if we took the manual time haha) 🙂
PS. I realize i'm always coming from the perspective of a speaker builder/manufacturer.
Yep, sound like you, Ron, and I are birds of a feather, always working on something.
So you have the devil level plumb bob eyes too haha.
I was in elevator construction, and learned to spot out of plump first, then came level, square, etc. Both a blessing and a curse huh? 😉
But seriously, I really enjoy the construction trades..
Hey Ron, my 2nd greatest passion after audio is old two-stroke standup jetskis, hot-rodding and riding. (after getting into too much trouble with cars 😡). Even got a dyno at the height of modifying..
Here's to a fellow motorhead
Yes! 😎
I built an elevator shaft last year (only three floors) and the elevator man (inspector) and I butted heads a bit.
Young fella came in with a laser measuring the variation of distance between the block walls.… he said there was no allowance for more than 3/32ths of an inch!!
The pair of guys that were actually building/installing the elevator were great though.!
We did use jack lines pulled from the opening in the roof, down to the bottom of the elevator pit for all four corners, (it was a retrofit in a vey old building) but it is still just concrete block work after all LOL.
I love hot Rods too!!!
About 12 years ago I built a Shelby Covra replica by Factory Five Racing. Lots of fun with a ridiculous power to weight ratio, but I did eventually tire of it and also very tiny for anyone over 5’10”.
Also used to ride/race motocross when I was younger and had the physical wares LOL.
Hi Mark,
I'll try and answer your question, based on what I've found (so far), as a user of both DRC, REW and a tool like RePhase...
For a start, the frequency dependent window function used in DRC-FIR is way different in operation than the one incorporated in REW...
Here's a comparison, done on an old measurement from ra-7's arrays early on (of which I kept a picture to show the windows):
The blue one (3rd from the top) is a 5 cycle frequency dependent window in REW
The top is a first option of a 5 cycle frequency dependent window from DRC-FIR.
Second from the top is another (default) option of a 5 cycle frequency dependent window in DRC-FIR.
The bottom graph is the second from the top DRC-FIR curve, after a 1/3 octave smoothing (done in REW).
In DRC, there is an opportunity to vary the number of cycles as frequency changes. Real handy with a source that has it's output arrive varied somewhat in time.
Yet you'll have to ask yourself the question: when do you want that wave front to be right (or rather restored to something that should resemble right)? For me it is: when it first hits my ears. After that a lot will still happen. So I don't worry about room curves and tonal balance just yet.
As I've mentioned before, we need the room for an array like this, without the room it behaves like a finite array. So figuring out where in time we want the impulse correction is different from what it would be for correcting a single driver.
Which is why one can use a different number of cycles for different parts of the frequency's as present in the measured IR.
This can be done in an IIR fashion, at that point in time. Normal EQ can follow later, using longer windows to create room curves/ set tonality, but for that first correction DRC is targeting to restore the IR (at that point in time, set by the frequency dependent window) back to it's ideal shape, largely depending on all other variables available to you to steer that process. (such as dip limiting and peak limiting etc. and lots of little stuff)
Complicated? Ehh yeah, if you start fresh it is, it's a learning process.
If one would want phase correction, you can choose a separate (frequency dependent) window bound to it's own group of variables to steer that, totally separate from the magnitude correction. Handle with care 😉.
Then there is a separate process that can be turned on or off, where one can limit ringing etc. Which is partly dependent on other variables that are chosen in the earlier stages.
Besides all of the above, there still are a few other features I haven't mentioned.
A lot of control in a little package... I use it more as a speaker correction tool than it's name suggests: Digital Room Correction. What I've set up for arrays is way different from any of the standard templates. Lots of trial and error to find out what does what (and when) and how it changes perception. And I can tell you, it does change perception.
Try it with a pre-baked solution featuring a 4 or 5 cycle window correction and follow it up by an 8 cycle window and see (or rather hear) what changed.(*)
Your frequency curve won't look all that different (on first sight). What you'll hear will be. I've used the smallest size number of cycles I could get away with that still made me keep the most positive perceptional changes, thus it varies with frequency.
Even that part (the varying) is controlled by variables and different options with an almost unlimited variety of possibilities.
This above isn't exclusive to DRC-FIR. Acourate and Audiolense all have their own way to control the correction (when where and how etc.). In fact the author of Acourate, Dr. Uli Bruggeman has even been of influence in the development of DRC-FIR. That communication can partially be found on DRC-FIR's forums.
Basically, it is a lot of math bundled with a certain view on things, in a package to give control to it's end user. DRC-FIR being the least automated of the 3 I mentioned. Totally open to the end-user.
Use it with care, you might just get what you've asked for.
Which is the bigger question, isn't it? What do you want it to do? 🙂
That's it in a nutshell, so it should have become a little more obvious that it would be real hard to recreate or match this within REW. Should one want to. I have tried it and wasn't successful. The windowing alone means you're behind right from the start. But whatever I did never created something remotely comparable to what I got out of DRC, (judged as a listener).
That 'time window' was what made me interested in this specific package. Being somewhat able to make a distiction between direct and indirect sound within a room, for a speaker that needs that room to (complete it's) function.
For correcting a single driver one would have completely different demands.
Disclaimer;
(*) Mark, not targeted at you, I wouldn't dare ask you again to try it. 😀
More people (interested in this subject) might read this though. So it is a 'general' you.
I'll try and answer your question, based on what I've found (so far), as a user of both DRC, REW and a tool like RePhase...
For a start, the frequency dependent window function used in DRC-FIR is way different in operation than the one incorporated in REW...
Here's a comparison, done on an old measurement from ra-7's arrays early on (of which I kept a picture to show the windows):
The blue one (3rd from the top) is a 5 cycle frequency dependent window in REW
The top is a first option of a 5 cycle frequency dependent window from DRC-FIR.
Second from the top is another (default) option of a 5 cycle frequency dependent window in DRC-FIR.
The bottom graph is the second from the top DRC-FIR curve, after a 1/3 octave smoothing (done in REW).
In DRC, there is an opportunity to vary the number of cycles as frequency changes. Real handy with a source that has it's output arrive varied somewhat in time.
Yet you'll have to ask yourself the question: when do you want that wave front to be right (or rather restored to something that should resemble right)? For me it is: when it first hits my ears. After that a lot will still happen. So I don't worry about room curves and tonal balance just yet.
As I've mentioned before, we need the room for an array like this, without the room it behaves like a finite array. So figuring out where in time we want the impulse correction is different from what it would be for correcting a single driver.
Which is why one can use a different number of cycles for different parts of the frequency's as present in the measured IR.
This can be done in an IIR fashion, at that point in time. Normal EQ can follow later, using longer windows to create room curves/ set tonality, but for that first correction DRC is targeting to restore the IR (at that point in time, set by the frequency dependent window) back to it's ideal shape, largely depending on all other variables available to you to steer that process. (such as dip limiting and peak limiting etc. and lots of little stuff)
Complicated? Ehh yeah, if you start fresh it is, it's a learning process.
If one would want phase correction, you can choose a separate (frequency dependent) window bound to it's own group of variables to steer that, totally separate from the magnitude correction. Handle with care 😉.
Then there is a separate process that can be turned on or off, where one can limit ringing etc. Which is partly dependent on other variables that are chosen in the earlier stages.
Besides all of the above, there still are a few other features I haven't mentioned.
A lot of control in a little package... I use it more as a speaker correction tool than it's name suggests: Digital Room Correction. What I've set up for arrays is way different from any of the standard templates. Lots of trial and error to find out what does what (and when) and how it changes perception. And I can tell you, it does change perception.
Try it with a pre-baked solution featuring a 4 or 5 cycle window correction and follow it up by an 8 cycle window and see (or rather hear) what changed.(*)
Your frequency curve won't look all that different (on first sight). What you'll hear will be. I've used the smallest size number of cycles I could get away with that still made me keep the most positive perceptional changes, thus it varies with frequency.
Even that part (the varying) is controlled by variables and different options with an almost unlimited variety of possibilities.
This above isn't exclusive to DRC-FIR. Acourate and Audiolense all have their own way to control the correction (when where and how etc.). In fact the author of Acourate, Dr. Uli Bruggeman has even been of influence in the development of DRC-FIR. That communication can partially be found on DRC-FIR's forums.
Basically, it is a lot of math bundled with a certain view on things, in a package to give control to it's end user. DRC-FIR being the least automated of the 3 I mentioned. Totally open to the end-user.
Use it with care, you might just get what you've asked for.
Which is the bigger question, isn't it? What do you want it to do? 🙂
That's it in a nutshell, so it should have become a little more obvious that it would be real hard to recreate or match this within REW. Should one want to. I have tried it and wasn't successful. The windowing alone means you're behind right from the start. But whatever I did never created something remotely comparable to what I got out of DRC, (judged as a listener).
That 'time window' was what made me interested in this specific package. Being somewhat able to make a distiction between direct and indirect sound within a room, for a speaker that needs that room to (complete it's) function.
For correcting a single driver one would have completely different demands.
Disclaimer;
(*) Mark, not targeted at you, I wouldn't dare ask you again to try it. 😀
More people (interested in this subject) might read this though. So it is a 'general' you.
Attachments
The context of what I am describing is measuring a line array in a room, any speaker in a room but primarily a line array because it cannot be successfully measured outside tuned and placed in a room and have it sound as good as it could.
You think I mean the time response of a speaker and while that is corrected that is not what I am describing either.
I am describing setting the envelope of the EQ corrections based on what the response is at different points in time based at different frequencies as measured.
At high frequencies the window is short (2.07ms) so it is only the speaker response being tuned, as you move down in frequency more of the room interaction is allowed as the window grows larger.
If you do this to one of your outdoor measurements it won't make much difference because there are effectively no reflections to filter out with the time window.
To see the difference between smoothing and windowing take any measurement you have made at the listening position of a speaker in a room, if you don't have one handy you need to make one to see the effect.
Duplicate it so you have two of the same, choose the frequency range of interest or leave it full band.
On one use different smoothing settings, on the other different window lengths or FDW in cycles.
So the real difference comes down to application, if you are forced (or choose) to measure in a room, variable windowing works differently to variable smoothing.
I don't like it anymore than you do that it ends up being only useful to that speaker in that room but the template works pretty well for anyone else with the same type of speaker if they can take a good measurement at the listening position.
Edit: wesayso beat me by two minutes 🙂