Dave and Dave
Griesinger also points out that the phase differences of the harmonics are important. That's why I am not so certain that phase is completely irrelevant, although, the data does indicate that it is not in the top three factors. Flipping the phase of a harmonic is a big change. Most phase errors would not be that extreme - or shouldn't be!
Around here I am always talking about small rooms. (Others may be different. I only deal with small rooms, larger rooms are an entirely different animal.)
Thanks for the correction of the Toole work. I was trying to remember the exact results, but seem to have over-stated the results. Still the implication remains the same. In regards to your work, it is quite interesting, but still it is your singular impression. You should do a larger blind study to verify what you are stating as "might not be". Based on the results that have been substantiated, I'll stick with my approach.
How one would raise the off axis response to compensate for an on-axis hole is hard to imagine since for single source systems the polar response is not a variable that one has much control over. It is determined by the sources acoustics. So if you are lucky enough to have an off-axis peak when there is an on-axis hole then "good for you", but one could never design-in such a thing.
Coincidentally, when my waveguides have an on axis hole because of edge diffraction they do tend to have more off-axis gain. Just lucky I guess
Griesinger also points out that the phase differences of the harmonics are important. That's why I am not so certain that phase is completely irrelevant, although, the data does indicate that it is not in the top three factors. Flipping the phase of a harmonic is a big change. Most phase errors would not be that extreme - or shouldn't be!
Around here I am always talking about small rooms. (Others may be different. I only deal with small rooms, larger rooms are an entirely different animal.)
Thanks for the correction of the Toole work. I was trying to remember the exact results, but seem to have over-stated the results. Still the implication remains the same. In regards to your work, it is quite interesting, but still it is your singular impression. You should do a larger blind study to verify what you are stating as "might not be". Based on the results that have been substantiated, I'll stick with my approach.
How one would raise the off axis response to compensate for an on-axis hole is hard to imagine since for single source systems the polar response is not a variable that one has much control over. It is determined by the sources acoustics. So if you are lucky enough to have an off-axis peak when there is an on-axis hole then "good for you", but one could never design-in such a thing.
Coincidentally, when my waveguides have an on axis hole because of edge diffraction they do tend to have more off-axis gain. Just lucky I guess
Again - not following exactly. Do you mean the increased distortion in our hearing mechanism or in the playback system? Then what if the playback system is very linear even to high SPL? The effect would go away?
That these kinds of "distortions" become more audible at higher SPLs is extremely interesting and important because in our test the playback system was always linear (or very nearly so) and yet there was an amplitude effect. The reality of that leads to a lot of questions about what we claim are high-SPL related "distortions" in loudspeakers.
While your observations are interesting, Dave, I'm having a tough time understanding this conclusion. Think of a speaker with a rising on-axis response in the HF and narrow dispersion, such that the power response is falling. Wouldn't that sound bright? It might not be so bad, but when put against a speaker with a flat on-axis response and a narrowing but wider dispersion that results in a similar power response, I would think that it would be identified as bright.
......Coincidentally, when my waveguides have an on axis hole because of edge diffraction they do tend to have more off-axis gain. Just lucky I guess
I have recently experienced the same thing. An on axis 1-2 kHz dip of 2-5 dB dissapeared (i.e. flattened out) when checked 30 deg. off axis
When dealing with round waveguides, an axial hole is inevitable. In my case the response actually rises off-axis, which makes perfect sense if you think about what the edge diffraction is doing. It is out of phase from the direct signal right on axis, but as the phase shifts off-axis the diffraction adds to the response. The lowest point is right on axis.
Diffraction is another one that has been similarly shown to be more audible at higher levels.
Even with group delay (just below the threshhold of audibility at the crossover), by reducing diffractions and maintaining a smooth DI I get the same quiet listening smoothness and detail character right up to the Xlim of the DE250 (I don't know what that is but it's over 110dB).
The same speaker would fall apart at various levels before then, despite amplifiers, orders of rolloff, and crossover integration.
I never thought about the distortion in the hearing mechanism, I was referring to that in the playback, but I think you're right, hearing mechanism distortion could also factor in. Its a very compelling theory they propose: The signal envelope is a function of the amplitude and phases of each frequency component in the signal. By varying the phase of particular components, the envelope amplitude may decrease to the point where other previously masked components now become audible. Its important to mention, they aren't talking about the electrical envelope, they mean the inner ear envelope.
This makes complete sense to me. Back in University a team I was on created a mainframe based basilar membrane FEM model which could model the membranes motion and envelope (with arbitrary stimulae and models several seconds of membrane motion) including hair cell stiffening etc etc. That was stone ages ago so I have no doubt this could (or has been) improved to the state that this theory can be modeled
I spend a lot of time listening to music in my work vehicle, mostly classical from the radio. It is not flat, it's not accurate a mixture of direct and reflected sound and lots of distortion and noise. Not the ideal listening environment, does this keep me from enjoying the music? No it does not, thanks to my brains ability to weed through crap I can get to the hart and soul of music and revel in its beauty!
Larry
Larry
Earl, in my experience, its usually the opposite situation to deal with. An off axis sound power peak is mitigated (in small room) by a small on axis hole. For example, an off axis peak in sound power (eg tweeter omni near xover) very often needs a very slight (dB or two) of reduction in on axis in the same frequency range.
I found it also depends upon crossover summation. For example, an xover that sums flat on axis but with 90 deg phase difference between high and low pass (on axis) often needs a bit of toning down around xover because the power thrown into the room at xover may be too high due to the tweeter being very omni at the same frequency range. Alternatively, an in phase xover (eg LR4) with a deep reverse notch has maximum summation on axis, less sound power thrown into the room off axis at xover and helps compensate for the tweeters extra sound power around these frequencies. When designers go for "maximum reverse notch" in phase xovers I think this is what they're attaining without even knowing it.
I've never really found a need to add an on axis peak to fill an off axis hole (eg woofer beaming). To me that creates an audible colouration.
Admittedly, this isn't blind and is a small sample group (ahem) but its been consistent from design to design for me. I try and attain maximum audible neutrality (of course, very hard to discern without a reference).
Last edited:
The complete disaster is a simple 3 way 1st order passive XO. The something reasonable is a 3 way Neville-Thiele 2nd order digital XO. Both using XO points of 500 Hz and 5 kHz.
When generating the digital XO, one can enter in delays to time align the acoustic center of each driver. The procedure for finding the acoustic center of each driver is in the article I linked to in an earlier post.
The delays entered in above are part of the procedure to figure out driver acoustic center offsets. Suffice to say, with these delay values the driver misalignment effect is clearly audible – just as audible as this: https://www.youtube.com/watch?v=UQOkSF8auFc
What’s the resolution of this method? The resolution is 1 sample. 1/44100 = 0.0226 milliseconds or 0.68 centimeters. The technique can measure a minimum of 0.68cm driver misalignment. The higher the sampling frequency, the higher the resolution.
Once time aligned, and verified using the technique from the article, I can enter in different delay values, regen the filters without changing any other parameters, measure the step response to verify that it is not time aligned by increments of 0.68 cm.
With the digital audio music player I use (JRiver), I can A/B the filters while listening to music in real time. I can hear an audible difference within a few samples driver offset. From a subjective point of view the difference to my ears affect both imaging depth and tone quality.
Double blind ABX testing is really hard to do properly. I have a fair bit of experience setting one up here: http://www.computeraudiophile.com/content/520-fun-digital-audio-%96-bit-perfect-audibility-testing/
I am currently figuring out way to set this test up so folks can ABX (binaural) recordings of the speakers with different driver time misalignments to test for audibility, similar to the bit-perfect audibility test above.
Mitch, can you design a conventional crossover, i.e., with phase warp, and then compare the same crossover with phase adjusted so that all frequencies arrive at the same time? I did this test and I could not tell the difference between flat phase and normal phase (normal meaning what you get with crossovers). In JRiver, you can program the crossover and then use convolution to flatten the phase. Then ask someone else to switch the convolution in and out.
This certainly pushes my credibility button. To accept this I would have to see blind confirmation with multiple listeners. It just does not fit with any psychoacoustics that I know of. You are claiming a detection threshold of .05 ms. - that's a very small change. I think that a lot of people will have trouble with that as well as I.
Wow Larry, that's great. Then why bother with anything we are talking about around here if none of it will make any difference to you? I know lots of people who could care less about "sound quality" - my son for example, but he doesn't hang around an audiophile forum. He doesn't care.
That was my starting assumption as well. Most of the tests I did either shelved or dipped out sections of the direct response. Then the later arriving (reflected) response could easily be used to fill back in the steady state response. It goes the other way too: you can have a mild peak or rise in one element and correct it with less energy with the other.
One thing to think of that wasn't immediately obvious to me. If one element has a bump, it can be no greater than 3dB to be correctable. We are not talking about serial connected filters but parallel paths. Assuming equal levels of direct and reflected sound, a 3 dB peak in one needs an infinite hole in the other to just compensate. A 4 or 5 dB peak can not be compensated.
Back to rising or falling responses, we know that flat axial response and falling power response can sound good. Such speakers did best in the early Toole study. This does not preclude a gently rising response from being counterbalanced by a more strongly falling power response or a falling axial response being balanced by a flatter off axis power. Of course, when you are depending on off axis power to flatten your perceived response balance you need to get the room acoustics just right, but it can be done.
David
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.