I have done graphs showing the settling time changes vs load variation, one is in my gallery.
It is also possible to show time shift vs cable impedance, I just didn't post that.
A 10 uSec delay by itself is pretty much nothing to humans if it is just one range of frequencies being affected buried in a mono program.
It is interchannel delay where we sensitive at the 2 to 5 uSec level. That is why I said use one amp channel, verify speakers produce a perfect central image, then split one zip and recheck the central image. Only a small band of frequencies we can discern at that time level, and it will happen only if the sensitive range has a low impedance.
If you detect nothing, you are done.
Jn
It seems unfair to say that something exists, and then tell someone to check it themselves because they might not perceive it. Are you aware of ANYONE who is able to perceive such a change at that level?
Also - 5 us is a distance traveled by sound of 1.7 mm. I definitely don't think I can tell any changes in the stereo image when I very, very slightly move my head.
Also - 5 us is a distance traveled by sound of 1.7 mm. I definitely don't think I can tell any changes in the stereo image when I very, very slightly move my head.
Last edited by a moderator:
Human sensitivity to ITD at these levels has been documented by neuro researchers for the last 50 years give or take. So it is real.
My test is not a "centering test" per se that requires locking the head in a "vice" so to speak.
It is a relative positioning test. It is designed to spot an ITD timeshift of some of the content. It relies on the listener determining if some of the content has shifted relative to the content that did not shift, either because electrically nothing happened, or humans are not sensitive to it.
I am unaware of any test measurement that is capable of discerning this using speakers, we have not progressed equipment-wise to the level of localization capability we humans have.
Think of this as a two source test with the sources behind an opaque screen. one is in the middle of the screen, one can move. You are being required to discern if the moving source is to the left, the right, or centered with the stationary source. You are not leaving the room only to return and determine the absolute position of the moving source. You are trying to discern relative position of two simultaneous sounds.
If the test shows a null, you are done. I suspect it costs roughly 5 dollars worth of wire to end the concern any individual would have on their system being sensitive to cables. Either yes or no.
Imagine if that were a standard test applied to all those wonderful cables out there, only to find that most cables regardless of cost made no difference? And that any differences were simply an unexpected consequence of system interaction with L, R, and C. ..no fancy materials, no fancy copper, no fancy geometries...just L, R, C.. what an elegant world it would be.
In my work, the most important thing is metrics..the measurement of what you are looking at. No proper measurement, you cannot improve. Using my home mill, my lathe, if I cannot measure my results to 25 microns accuracies, I cannot make something to that accuracy.. At work, motion control position accuracies of 1 micron are impossible if you don't have the encoder resolution.
jn
My test is not a "centering test" per se that requires locking the head in a "vice" so to speak.
It is a relative positioning test. It is designed to spot an ITD timeshift of some of the content. It relies on the listener determining if some of the content has shifted relative to the content that did not shift, either because electrically nothing happened, or humans are not sensitive to it.
I am unaware of any test measurement that is capable of discerning this using speakers, we have not progressed equipment-wise to the level of localization capability we humans have.
Think of this as a two source test with the sources behind an opaque screen. one is in the middle of the screen, one can move. You are being required to discern if the moving source is to the left, the right, or centered with the stationary source. You are not leaving the room only to return and determine the absolute position of the moving source. You are trying to discern relative position of two simultaneous sounds.
If the test shows a null, you are done. I suspect it costs roughly 5 dollars worth of wire to end the concern any individual would have on their system being sensitive to cables. Either yes or no.
Imagine if that were a standard test applied to all those wonderful cables out there, only to find that most cables regardless of cost made no difference? And that any differences were simply an unexpected consequence of system interaction with L, R, and C. ..no fancy materials, no fancy copper, no fancy geometries...just L, R, C.. what an elegant world it would be.
In my work, the most important thing is metrics..the measurement of what you are looking at. No proper measurement, you cannot improve. Using my home mill, my lathe, if I cannot measure my results to 25 microns accuracies, I cannot make something to that accuracy.. At work, motion control position accuracies of 1 micron are impossible if you don't have the encoder resolution.
jn
Last edited:
Couldn't tell ya now. I did all the analysis back in the 2000 to 2004 timeframe. Several computers and hard drives ago. However, I believe David Gresinger had that information on his website. for some reason, I always remember 11-11-11, I think it is the date of his paper, nov 11 2011.
A simple geometric analysis can be used to determine it. IOW, how far off axis a point would be 12 feet away such that there is a 5 uSec path length difference ear to ear.
I think I did post all the graphs on audio asylum (prop heads IIRC) back then, or maybe audioholics. The relatively blank "stares" received informed me I was in the wrong place.
BTW, the work was done to determine the level of accuracy required to fix a sound source in 2D space relative to another source, both side to side as well as depth, and included IID as well. More for an overall system requirement rather than just cables. Note: it was interesting how the type of source made IID differences.. planar vs linear vs point source made significant IID changes.
PS..I will try to find the information next time I have access to the computers, I think the IT guy copied the old hard drives to the new ones over the years, and I do recall I had taken the excel graphs and converted to jpg's.
PPS. All the analysis considered just undithered ITD. Dithering makes humans even more sensitive and up to roughly 12Khz. How that would translate to a moving cone which is also outputting ITD sensitive information, I have no idea. It does rear the ugly head of level dependent ITD sensitivity though, another confounder.
jn
A simple geometric analysis can be used to determine it. IOW, how far off axis a point would be 12 feet away such that there is a 5 uSec path length difference ear to ear.
I think I did post all the graphs on audio asylum (prop heads IIRC) back then, or maybe audioholics. The relatively blank "stares" received informed me I was in the wrong place.
BTW, the work was done to determine the level of accuracy required to fix a sound source in 2D space relative to another source, both side to side as well as depth, and included IID as well. More for an overall system requirement rather than just cables. Note: it was interesting how the type of source made IID differences.. planar vs linear vs point source made significant IID changes.
PS..I will try to find the information next time I have access to the computers, I think the IT guy copied the old hard drives to the new ones over the years, and I do recall I had taken the excel graphs and converted to jpg's.
PPS. All the analysis considered just undithered ITD. Dithering makes humans even more sensitive and up to roughly 12Khz. How that would translate to a moving cone which is also outputting ITD sensitive information, I have no idea. It does rear the ugly head of level dependent ITD sensitivity though, another confounder.
jn
Last edited:
OPINION WARNING (CAUTION!)
JN,
Fully agree it is a system requirement. Various opportunities to diminish localization performance exist, perhaps starting as early in the reproduction chain as a dac clock.
I still don't get why everything around here has to be oversimplified or else a peer reviewed paper presented on it, simply because of some theory that gullible people could be prevented from spending money foolishly. That same person could be prevented from spending money wisely too, particularly if they are too scared to ever listen to anything that isn't optimized solely for AP figures of merit.
</OPINION WARNING>
JN,
Fully agree it is a system requirement. Various opportunities to diminish localization performance exist, perhaps starting as early in the reproduction chain as a dac clock.
I still don't get why everything around here has to be oversimplified or else a peer reviewed paper presented on it, simply because of some theory that gullible people could be prevented from spending money foolishly. That same person could be prevented from spending money wisely too, particularly if they are too scared to ever listen to anything that isn't optimized solely for AP figures of merit.
</OPINION WARNING>
That is actually one of the few audio things that has been scientifically proven (m5usevultiple tests + results independently confirmed in proper scientific experiements.
Humans can detect a 5µsec difference.
dave
A lot of this goes back to the 2007/8 papers by Milind N. Kunchur, Ph.D.
Homepage of Kunchur's Research Group
Information for prospective students
He wrote incorrectly about the poor time resolution of CD digital audio. This caused a lot of reaction in some forums and e-mail groups.
But it was felt that an under 10us shift might be possibly be noticeable, but any test might have uncontrolled variable problems.
* * * * * * * * * * * *
as to:
ITD - Interaural Time Differences
ILD - Interaural Level Differences
IID - Interaural Intensity Differences
I think that Floyd E.Toole, j.j. (James D. Johnston) and David Griesinger have all written about them.
Last edited:
Jan Nordmark, 1974. Data down to 1.2 uSec dithered, 2 to 5 uSec from 500 hz out to a few K. Dithered, the discernment goes out to 12 khz.
For reasons I forgot, I didn't seem to be thrilled with Kunchar. Not sure why.
ITD is used to find all sources, not just single sources. My test uses two sources with discernment between both as the metric.
jn
For reasons I forgot, I didn't seem to be thrilled with Kunchar. Not sure why.
ITD is used to find all sources, not just single sources. My test uses two sources with discernment between both as the metric.
jn
Last edited:
It would also depend on the amount of ITD. The test would be moving your head slightly and seeing if the image changed 
Here is an extreme example Stereo Perception, Sound Localization & Auditory Cues
Here is an extreme example Stereo Perception, Sound Localization & Auditory Cues
planet10, I would welcome a real answer to one of those lots of sound people can hear but cannot be measured as you claimed. It's been years and still no real answer from you other than dodging replies. I think we all would welcome your real straight answer.
That's because it's waiting for the cable burn-in to come (back) up first.
- Status
- Not open for further replies.