In addition the theory of temporal hearing rather than place theory seems to be more applicable at low frequencies according to the more modern research.
Brian Moore has a good book "An Introduction to the Psychology of Hearing" searching the title should lead to a version anyone can read.
It is obvious that you are not interested but much of your reasoning seems faulty to me and my own experience suggests that you are undervaluing the gains that you could have. There is some low hanging fruit that can be exploited without going to the lengths wesayso and I have.
In Stereo reproduction, with the cross talk between left and right channel, I'd agree. But not as a basic point of view that humans cannot hear phase information above 400 Hz.
Take this video from David Griesinger:
Here, he explains in detail how he researched the influence of phase on the proximity effect and on human speech. He's quite sure about our ability to use phase for localisation over a much wider frequency range than you just mentioned. I've tried a lot of things that Mr. Griesinger advocates, I haven't found a better source of information than his many talks and papers on this and other subject about human perception. I'll upload the transcript of this video so you can read it. It's an automated write-up, so it won't be coherent in itself (lol). But with the accompanying video, I think the point is well made. If not, just look at the many papers/presentations he wrote on this subject.
So few people to listen to him during presentation...
Anyway, i did not take Fluid statement as we are not able to hear phase info above 400hz: that's why i posted about localization: for the stereo illusion to take place we rely on phase info up to 1,5khz. Above that mainly intensity/level comes into play.
How i interpret Fluid's comment is that above 400hz other issues start to have significant effects on this if not outdoor or into an anechoic room( mainly room related issues in my point of view).
Edit: i'm sure most of the people into this conversation knows this but for someone reading this thread without previous inquiring into the subject it can seems like nitpicking as it's not. I hope it could help some to keep track into the subject.
Anyway, i did not take Fluid statement as we are not able to hear phase info above 400hz: that's why i posted about localization: for the stereo illusion to take place we rely on phase info up to 1,5khz. Above that mainly intensity/level comes into play.
How i interpret Fluid's comment is that above 400hz other issues start to have significant effects on this if not outdoor or into an anechoic room( mainly room related issues in my point of view).
Edit: i'm sure most of the people into this conversation knows this but for someone reading this thread without previous inquiring into the subject it can seems like nitpicking as it's not. I hope it could help some to keep track into the subject.
Last edited:
I listen when he speaks, so far he's been one of the biggest sources of useful information on human perception for me. What he says in this video goes against some common beliefs. But he's not talking about Stereo reproduction, he talks about the real deal. At 1.5 KHz we're already into the effects of cross-talk in a common Stereo triangle.
But our ability to hear phase stretches further than that (from the above transcript):
David Griesinger said:
Just for the record
.
I'm listening/reading to him too. It's just i'm amazed people into the industry seems to be too much into trends ( i've been into AES's presentation were crowd was huge about things i thought afterwards was futile...as being the flavor of the day).
Not enough time and life got in the way to further analyse his availlable writtings on his site... in 3 years i just scratched the surface of it.
Sure we have still not crack opened how our brain works despite the 'mains' principle are now mostly known... and we are still at the beginning of discovery about it imho.
Anyway, you already told me about this in private conversation we had about some function availlable into DRC-FIR: i was doubtful at the time ( it all seemed a bit hazey to me) , i'm not anymore.
The group of you which investigated all this and share/shared your results in the wild... you are just amazing people to me. ❤
(Apologize being sentimental. Had to face another issue this past week end which definetly killed my business: another robbery and destruction of my production tool. Time to move to something else... happily you keep my head above water thinking about what really matter to me! And you help keeping faith in human kind).
Not enough time and life got in the way to further analyse his availlable writtings on his site... in 3 years i just scratched the surface of it.
Sure we have still not crack opened how our brain works despite the 'mains' principle are now mostly known... and we are still at the beginning of discovery about it imho.
Anyway, you already told me about this in private conversation we had about some function availlable into DRC-FIR: i was doubtful at the time ( it all seemed a bit hazey to me) , i'm not anymore.
The group of you which investigated all this and share/shared your results in the wild... you are just amazing people to me. ❤
(Apologize being sentimental. Had to face another issue this past week end which definetly killed my business: another robbery and destruction of my production tool. Time to move to something else... happily you keep my head above water thinking about what really matter to me! And you help keeping faith in human kind).
Last edited:
@fluid, Basically I don't disagree with anything you are saying. Agreed, it's subjective. What you prefer may not be what I prefer.
@wesqayso, A lot here, but first thanks for the UT link. Just what I needed. First let me address the "one spot" issue. That really isn't a problem at higher frequency and for the direct sound IF the speaker has good directive. The problem comes in more in the modal region where moving left or right a couple of feet can change the response significantly.
Thing about frequency dependent windows. So it's a technique that uses a trade off between how long the window needs to be get a reasonably accurate of the amplitude at a given frequency vs how much reflected sound is included. OK. So what it boils down to is frequency dependent widows software is a parametric frequency response analyzer. It gives you a bunch of knobs to twist and it's up to you to decide what's right or wrong. You end up with a response curve that is used for the bases of room correction but it only accounts for "part" of the room. I guess I'm just to old school for this. Here are several measurements I took this morning with different windows. If I assume that the widow is related to a frequency of 10 cycles, then 1 ms = 10k hz, 2ms = 5k, 4ms = 2.5k, 16 ms = 625, and 500 ms = 20 hz. Compare the response for different windows. For example compare the amplitude and phase at 2.5k for the 4, 16 and 500 ms results. Note that down to the limit the phase for all cases matches the 500 ms phase pretty closely. Data is smoothed 1/6 octave.
I guess this wiki from Audiovero helps: https://www.audiovero.de/acourate-w...ionen:td-functions:frequency_dependent_window
Watching the video that Mitch posted a page ago: https://www.diyaudio.com/community/...earization-by-fir.393435/page-32#post-7313962 might have solved that part too.
I'll include it:
To be able to determine what to correct and especially when to correct it, one would need to examine the effect of the room on the speaker. Not only at the listening spot, but also at various points of interest around it, as to be sure that the correction won't be a "one spot" solution. That sure is not as easy as just making a few tweaks with PEQ. The above mentioned frequency dependent window is a tool one can use to look at the "early" wavefront. At higher frequencies, you'll be able to go in, see the early output before the reflections have happened (much like a properly gated window) and if needed, base your EQ on that.
At lower frequencies, the room effects will be part of that early window, no matter what one does. But programs like DRC-FIR (a free tool), Audiolense and Acourate (from Audiovero) let you vary the frequency dependent window used for magnitude correction EQ and even use a different sized window for phase EQ(*). Next to this handy feature, they usually have a zillion other options that determines how the program needs to behave. In DRC-FIR (the free program) you can set just about any (of those zillion) variable(s) to your liking, but that also means a lot of legwork from it's user to determine what does what.
The Acourate and Audiolense products follow a more pré-determined path and are more automated, based on the choices of their respective makers. There are small differences between these programs, as far as the inner workings go, but that does not detract from their purpose.
Personally, I've studied my speaker in my room and what I could do with a tool like the above mentioned solutions. Mitch has been a big inspiration in my early days of figuring this stuff out and we've compared notes on several occasions. Same goes for fluid and I, and many many more users of these tools.
These tools come in handy but require work to get a meaningful result. It would take me a long time to go over it and explain all they do, but I won't in this thread. I do have a thread which is almost 400 pages long that has a lot of this stuff covered, open for anyone to read and you'd be surprised how many have already done that.
There's a thread on the full range side of this forum that has helped a number of people get started with DRC-FIR. But it's still just a tool and any tool is as good or as effective as the user that determines what to do with it. It does tend to take time though, as we tried to explain earlier.
(*) you can even limit the phase EQ to a certain bandpass, for instance to only use it on that low end.
@wesqayso, A lot here, but first thanks for the UT link. Just what I needed. First let me address the "one spot" issue. That really isn't a problem at higher frequency and for the direct sound IF the speaker has good directive. The problem comes in more in the modal region where moving left or right a couple of feet can change the response significantly.
Thing about frequency dependent windows. So it's a technique that uses a trade off between how long the window needs to be get a reasonably accurate of the amplitude at a given frequency vs how much reflected sound is included. OK. So what it boils down to is frequency dependent widows software is a parametric frequency response analyzer. It gives you a bunch of knobs to twist and it's up to you to decide what's right or wrong. You end up with a response curve that is used for the bases of room correction but it only accounts for "part" of the room. I guess I'm just to old school for this. Here are several measurements I took this morning with different windows. If I assume that the widow is related to a frequency of 10 cycles, then 1 ms = 10k hz, 2ms = 5k, 4ms = 2.5k, 16 ms = 625, and 500 ms = 20 hz. Compare the response for different windows. For example compare the amplitude and phase at 2.5k for the 4, 16 and 500 ms results. Note that down to the limit the phase for all cases matches the 500 ms phase pretty closely. Data is smoothed 1/6 octave.
We know
. We do use speakers with good directivity. And where it falls short we use room treatment. These frequency dependent windows can help tackle the "problem" in the modal region. You'd be surprised how even the response can be over a wide area if enough care is used. Yes, it takes time. Yes it's a lot of manual labor, making more than one set of measurements in one spot etc. But it's worth the trouble. No holes in the bass output due to floor or other in-room troubles one can encounter, making use of multiple time aligned sources that make up the whole. Can be done as mono, or if you're lucky and have enough sources, even as stereo down to the low frequencies. But simple, it's not. I'd be lying if I didn't look at what Geddes did, which was inspiring but I use it just that little bit different, using those frequency dependant views on the output. Much shorter windows are used for phase manipulation and one needs to learn to read the room to find out what needs to be left alone (where another source can pick up from there). No forcing the curve in shape, as you simply can't EQ a null. But using multiple sources in the modal region allows you to blend them and use their strong points. Using shaping EQ as well as phase manipulation at these frequencies can do a lot if one is willing to do the leg work. Does it sound better? Heck yeah. I didn't get there in one day.Despite all the hard work, it has my preference in listening. I have a family, so a one spot wonder won't do me any favours. I use this for Stereo + Home Theatre and put care into good seats for all members of the household (meaning a couch of 3).
Last edited:
Hey guys, got no time for low hanging fruit......I'm all busy picking an orchard !!
Here's the new orchard I have to pick
Now that it's warm enough to start testing, i've added bass-reflex ports to syn11 and put it back together.
been waiting for this since November...STOKED i am.
(Won't post any more about it on this thread...will do so on the syn11 i started a while back.)
Although, as happy as I am to get going with this project again, it does come at a cost.
My 3 ch LCR rig has dropped down to 2 ch stereo, to provide the pair of 18"s needed for syn11.
The LCR rig has been the best indoor sound I've ever achieved, maybe ever heard.
But I'm more than willing to set the LCR aside to explore a path that I think might lead to even greater potential.
This is what's fun for me, and really doesn't allow my stopping to try to milk all the potential out of a particular design.
I mean heck, my syn/sub stacks are on dollys because I try so many placements and setups. When I read books that say distances to speakers in the stereo triangle, need to be within 1/4" of each other, for good DSP correctable measurements....I'm like no thank you...tis a complete non-starter for me.
Fluid, I hope you and wesayso can see continued DIY experimentation is what like to do, and recognize I think this process is what has been giving me the sound i enjoy. MyFi as you say, wesayso.
It's a sound very different from usual home stereo...i think it's more likely to be heard in large studios or live-sound venues, where sound quality and unconstrained dynamic linearity are paramount. Its crazy clean, crazy powerful. The LCR rig has a sweet zone about 15 feet wide.
Eventually, will I get where maximising in-room performance is my goal?...no doubt.
I already cited a couple of reasons..such as I'm out of ideas to pursue, or I can't get affordable quality plywood anymore.
But for now, it's time to put the mics on the mast and start taking some measurements !
Now that it's warm enough to start testing, i've added bass-reflex ports to syn11 and put it back together.
been waiting for this since November...STOKED i am.
(Won't post any more about it on this thread...will do so on the syn11 i started a while back.)
Although, as happy as I am to get going with this project again, it does come at a cost.
My 3 ch LCR rig has dropped down to 2 ch stereo, to provide the pair of 18"s needed for syn11.
The LCR rig has been the best indoor sound I've ever achieved, maybe ever heard.
But I'm more than willing to set the LCR aside to explore a path that I think might lead to even greater potential.
This is what's fun for me, and really doesn't allow my stopping to try to milk all the potential out of a particular design.
I mean heck, my syn/sub stacks are on dollys because I try so many placements and setups. When I read books that say distances to speakers in the stereo triangle, need to be within 1/4" of each other, for good DSP correctable measurements....I'm like no thank you...tis a complete non-starter for me.
Fluid, I hope you and wesayso can see continued DIY experimentation is what like to do, and recognize I think this process is what has been giving me the sound i enjoy. MyFi as you say, wesayso.
It's a sound very different from usual home stereo...i think it's more likely to be heard in large studios or live-sound venues, where sound quality and unconstrained dynamic linearity are paramount. Its crazy clean, crazy powerful. The LCR rig has a sweet zone about 15 feet wide.
Eventually, will I get where maximising in-room performance is my goal?...no doubt.
I already cited a couple of reasons..such as I'm out of ideas to pursue, or I can't get affordable quality plywood anymore.
But for now, it's time to put the mics on the mast and start taking some measurements !
Whose book has nonsense in it like that? This thread was based on a faulty premise just like that one. It makes no difference to me what anyone chooses to do with their own time and system but when bad information isn't challenged it begins to take root and starts to be accepted just because.
mitchba's book, "Accurate Sound Correction Using DSP", several times talks about the need to keep mic-to-stereo speaker distances within 1/4".
And to hold to an equilateral triangle within 1/4" tolerances, because every mixing/mastering control room he's encountered had been configured that way.
I fully realize the book is trying to lay out theoretical best practices with that, and do not mean or want to disparage his fine book in any way. (It's a great resource imo). And I don't begin to want to sink into a debate about this book...
I think the 1/4" config recommendation, and maintaining room symmetry to the same spec with regards to speakers and their boundaries, is ideal for acoustic reasons. And both are entirely beneficial, independent of whether or not to use DSP for room correction. (Just like I think all acoustic solutions to room correction should be attempted prior to DSP.)
So i have no problem with 1/4" from a theoretical point of view....it's reality that gets in the way, imo.
And a very real concern that the rest of the advice/techniques are going to be similarly too specific, or too theoretical, to be practical or even necessary.
And when I study the steps and complexity of either Acourate or FIR-DRC, I get this sinking feeling...that they are so complicated, that they preclude fundamental understandings. And eventually turn in to trial and error, just like less complicated software.
Nothing I haven't said many times....just me..
Hey, completely shifting gears and thinking about FDW.....
Pls help me see how it differs from any other form of frequency selective measurement massaging, where the goal is to get best representative measurement for correction.
Seems to me, FDW is applied post-impulse capture, and therefore doesn't effect the actual time window used to capture the impulse.
I've read FDW is needed to ensure a long enough time window to allow both direct sound and tail reflections into low frequency measurements, while keeping relections out of high freq measurements.
Unless gating were applied to the original impulse capture, how can it not already have all the low freq info, both direct and reflected?
Seems like FDW is an elegant, continuous form of frequency dependent gating, where cycles are substituted for a series of frequency dependant gate timings.
Bottom line, doesn't it ending being another form of smoothing?
It's hard to see much pragmatic difference between FDW and fractional octave smoothing for example, particularly when fractional octave can be set multiband.
And to hold to an equilateral triangle within 1/4" tolerances, because every mixing/mastering control room he's encountered had been configured that way.
I fully realize the book is trying to lay out theoretical best practices with that, and do not mean or want to disparage his fine book in any way. (It's a great resource imo). And I don't begin to want to sink into a debate about this book...
I think the 1/4" config recommendation, and maintaining room symmetry to the same spec with regards to speakers and their boundaries, is ideal for acoustic reasons. And both are entirely beneficial, independent of whether or not to use DSP for room correction. (Just like I think all acoustic solutions to room correction should be attempted prior to DSP.)
So i have no problem with 1/4" from a theoretical point of view....it's reality that gets in the way, imo.
And a very real concern that the rest of the advice/techniques are going to be similarly too specific, or too theoretical, to be practical or even necessary.
And when I study the steps and complexity of either Acourate or FIR-DRC, I get this sinking feeling...that they are so complicated, that they preclude fundamental understandings. And eventually turn in to trial and error, just like less complicated software.
Nothing I haven't said many times....just me..
Hey, completely shifting gears and thinking about FDW.....
Pls help me see how it differs from any other form of frequency selective measurement massaging, where the goal is to get best representative measurement for correction.
Seems to me, FDW is applied post-impulse capture, and therefore doesn't effect the actual time window used to capture the impulse.
I've read FDW is needed to ensure a long enough time window to allow both direct sound and tail reflections into low frequency measurements, while keeping relections out of high freq measurements.
Unless gating were applied to the original impulse capture, how can it not already have all the low freq info, both direct and reflected?
Seems like FDW is an elegant, continuous form of frequency dependent gating, where cycles are substituted for a series of frequency dependant gate timings.
Bottom line, doesn't it ending being another form of smoothing?
It's hard to see much pragmatic difference between FDW and fractional octave smoothing for example, particularly when fractional octave can be set multiband.
So, what wrong with building a speaker with IIR only and when FR, distortion and directivity is OK, do a final FIR correction of phase. This was the premise. (As oppose to use FIR for "everything" i.e. XO, FR and phase...)
//
I became curious about these frequency dependent windows and was going to try something to look at the difference between using just a long window and frequency dependent windows. Then I decided it would take too long to do it manually and I really wasn't that interested. But Just a couple of weeks ago the developer sent me a new version of my design software, 11 updates since the last version I had. Turns out he implemented frequency dependent windowing. So I figured time to compare. It took measurements at about 1 M and about 3 M. One using a 400 msec window and a second using frequency dependent windows that were from 5 msec to 400 msec. I seems he allows 8 cycles for the frequency or interest. So the frequency dependent windowing starts at about 1600 Hz and below. At 400 msec the cut off is about 20 Hz, below that the 400 msec window is held constant. Below are the results. Top 1 M., bottom 3 M. Blue SPL and green phase = 400 msec window, Red SPL and blue phase = frequency dependent result. Other than some what I would call insignificant difference in the 800 to 1600 region of the 3 M result, which would pretty much disappear if stronger smoothing were used, there's not much going on here, IMO. Certainly not much in the modal region where frequency dependent windowing appears to do pretty much nothing.
Indeed when you apply a basic frequency dependent window with settings that don't introduce a real difference you don't see a difference. Holm impulse has a basic FDW routine, so does REW. They can introduce a difference much greater than what you show but nowhere near as well as being able to change the number of cycles with frequency as adding an exponent or other method does. When the only tool you have is a hammer everything looks like a nail.