Measurements & the stereo illusion

[Phelonious Ponk]

I'm at work and haven't listned to the demo. There is no system here worth listening to it on. But honestly, John, I suspect there are things I don't hear that others do. First of all, and you'll like this one:

You know that whole thing about perception and the human ability to accept and maybe even psycho-alter playback so that what we hear sounds more like what we hear in our natural environment? I think I may be pretty good at that. And I'll call it a blessing.

The second one you won't like as much:

I believe that many audiophiles are gearheads, tweakers, integrators, synergizers....this is much of their joy in the hobby and, as far as I can tell, a whole lot of them are not happy just being happy. I know them well; I used to be one of them. They've got to be tweaking integrating, synergizing, etc. It's the nature of the beast. As a result, they hear all kinds of things that nobody else hears. Not me, not their wives, not their friends, almost no one in blind listening tests... My decades in this game, from newsletters and Saturday mornings at the local shop to internet audiophile discussion boards, have left me believing that this class of audiophiles I once belonged to - and at this point it's the majority - and their suppliers honestly believe they hear all kinds of problems that don't exist and spend thousands of hours and dollars trying to solve them.

I think digital audio is the ultimate example. By design, it swept away hundreds of audio reproduction challenges. Audiophiles have, in turn, found new, esoteric, often immeasurable, and even when measurable often statistically inaudible, problems to fill the gap. Because they need the problems.

I don't expect you to join me in that belief, John, but there it is. And because that's what I have observed, my criteria for believing in one of tese problems is, as I said in another thread recently, quite simple: 1) I can hear it, or 2) Those who claim they can have verifiable evidence of its existence.

You guys drop the ball on #2 all the time. Once in awhile? Something we haven't figured out how to measure yet...maybe. All the time? Don't tell me it's my hearing or my system unless you can show me some game. That was a lot of words. I said it all more clearly and concisely in a post a few years ago on another board that went like this:


I SEE JITTER, PEOPLE...

I suspect this will get a reaction. I won't respond, because if I do your thread will be completely de-railed. MHO. YMMV. Enjoy it if you can.

Tim

 

[JackD201]

Tim,

Digital's problems exist and the people trying to fix them is the engineering community not the audiophiles. If they are doing it only for the audiophiles, that is another story altogether.

 

[jkeny]

I'm at work and haven't listened to the demo. There is no system here worth listening to it on.

I would be interested in your results when you get a chance to listen to the ICTD timing demo on a decent system

But honestly, John, I suspect there are things I don't hear that others do. First of all, and you'll like this one:

You know that whole thing about perception and the human ability to accept and maybe even psycho-alter playback so that what we hear sounds more like what we hear in our natural environment? I think I may be pretty good at that. And I'll call it a blessing.

The second one you won't like as much:

I believe that many audiophiles are gearheads, tweakers, integrators, synergizers....this is much of their joy in the hobby and, as far as I can tell, a whole lot of them are not happy just being happy. I know them well; I used to be one of them. They've got to be tweaking integrating, synergizing, etc. It's the nature of the beast. As a result, they hear all kinds of things that nobody else hears. Not me, not their wives, not their friends, almost no one in blind listening tests... My decades in this game, from newsletters and Saturday mornings at the local shop to internet audiophile discussion boards, have left me believing that this class of audiophiles I once belonged to - and at this point it's the majority - and their suppliers honestly believe they hear all kinds of problems that don't exist and spend thousands of hours and dollars trying to solve them.
........
I think digital audio is the ultimate example. By design, it swept away hundreds of audio reproduction challenges.

No mention of new problems that it might have created or do you think it is perfect?

Audiophiles have, in turn, found new, esoteric, often immeasurable, and even when measurable often statistically inaudible, problems to fill the gap. Because they need the problems.

You continually deny the fact that others hear things that you do not.

Here I am asking about measurements that it seems haven't been done/aren't done on reproduction systems & my reasons for why I think these measurements might offer a possible measurement that could map to why there are audible differences between systems & what your posts so far have done is:
- deny any differences of any consequence because you have not heard them yourself
- deny that there are differences because no measurements have been yet shown

I hope you see the circular logic contained in your posts? It really leaves you no room for any progress, whatsoever.

Just one suggestion - This thread might progress faster & further without your posts - we know your views by now.

 

[Phelonious Ponk]

Tim,

Digital's problems exist and the people trying to fix them is the engineering community not the audiophiles. If they are doing it only for the audiophiles, that is another story altogether.

I'll respond to correct myself, Jack. You are right. Digital is not without real problems and the real engineering community endeavors to address them. IMO it's problems are far fewer and far less severe than many audiophiles, and theirmanufacturing community imagine. Have a nice thread, John.

Tim

 

[JackD201]

I'll agree that they are far less severe now than even 5 years ago. Digital is getting REALLY good and even dyed in the wool analog friends are acknowledging that. We have to give credit to the folks who accomplished that as well as the whiners who would not accept the status quo back then when voting with their wallets.

 

[jkeny]

I'll respond to correct myself, Jack. You are right. Digital is not without real problems and the real engineering community endeavors to address them. IMO it's problems are far fewer and far less severe than many audiophiles, and theirmanufacturing community imagine. Have a nice thread, John.

Tim

Tim,
I don't mean to exclude you from the thread but maybe your posts could be more productive. I thought with your first post here that you were actually going to discuss the topic but instead I heard your belief system being re-hashed. No need to post your beliefs again & again in every thread - we know what they are by now. But I'm very interested in any datapoints or technical info that you might have to offer

Yes, digital has progressed significantly particularly in the last 5 years or so, there's no denying that. As the level of reproduction improves maybe some phenomena emerge:
- hitherto masked issues begin to become audible
- the level of improvement available is the last 10%
- our expectations may be heightened as we approach nearer to the goal

Let's get back to the thread topic:
Is there any value in the tests & concept I presented or is it all just blah, blah?

 

[Groucho]

If we're talking about timing discrepancies affecting imaging and soundstage, isn't there a huge elephant in the room, that is the speakers? The inter-channel difference may, or may not, be all that great, but the absolute timing discrepancies (group delay etc.) that they introduce are huge. But DSP-oriented people simply measure these discrepancies, push a few buttons and they are gone!

 

[jkeny]

If we're talking about timing discrepancies affecting imaging and soundstage, isn't there a huge elephant in the room, that is the speakers? The inter-channel difference may, or may not, be all that great, but the absolute timing discrepancies (group delay etc.) that they introduce are huge. But DSP-oriented people simply measure these discrepancies, push a few buttons and they are gone!

Usually single driver speakers image far better than multi driver versions. If multi-driver speakers have been designed with time-aligned drivers & crossovers then I would imagine this is minimised.

Besides speaker timing differences will be fixed & not varying. I am talking about the possibility of FLUCTUATIONS in timing occurring in the reproduction system in a random way. But the fluctuation is the important aspect of it because I posit that we are much more sensitive to changing timing than to fixed timing differences.

I'm really working backwards from an observation that is generally reported - better sounding systems show improvements in a more solid, 3D sound stage. What could give rise to this observation? Looking towards psychoacoustics leads us to ICLD & ICTD (& IC) as the factors that are responsible for sound stage. So I ask the question - have these measurements been done on ordinary reproduction systems (not codecs) & with what results?

 

[Groucho]

Usually single driver speakers image far better than multi driver versions. If multi-driver speakers have been designed with time-aligned drivers & crossovers then I would imagine this is minimised.

Besides speaker timing differences will be fixed & not varying. I am talking about the possibility of FLUCTUATIONS in timing occurring in the reproduction system in a random way. But the fluctuation is the important aspect of it because I posit that we are much more sensitive to changing timing than to fixed timing differences.

I'm really working backwards from an observation that is generally reported - better sounding systems show improvements in a more solid, 3D sound stage. What could give rise to this observation? Looking towards psychoacoustics leads us to ICLD & ICTD (& IC) as the factors that are responsible for sound stage. So I ask the question - have these measurements been done on ordinary reproduction systems (not codecs) & with what results?

Well even if passive drivers are time aligned, there are still time delays which vary with frequency. So, if we have a sliding trombone going up in one speaker and another going down in the other, we've effectively got a varying inter-channel delay - which is what you're talking about. Maybe it doesn't destroy the illusion of a soundstage, but it certainly doesn't reproduce the soundstage. Only DSP-correction can eliminate this i.e. give you an impulse at the output in response to an impulse at the input.

 

[jkeny]

Well even if passive drivers are time aligned, there are still time delays which vary with frequency. So, if we have a sliding trombone going up in one speaker and another going down in the other, we've effectively got a varying inter-channel delay - which is what you're talking about. Maybe it doesn't destroy the illusion of a soundstage, but it certainly doesn't reproduce the soundstage. Only DSP-correction can eliminate this i.e. give you an impulse at the output in response to an impulse at the input.

Your example is not relevant to reproduction of the soundstage. The rising trombone in one channel will/should have a time delayed version of the same rising trombone in the other channel but time delayed & at a lower level - exactly what the microphone picked up (depending on how the mic placements were arranged). So this would give us the sound stage illusion & the placement of this trombone in the sound field. If there are no fluctuations in level or time differential between channels, then the sound stage will be stable. If there are fluctuations, the sound stage will not be as stable.

I'm talking about inter-channel time & level delays (ICTD & ICLD) of the SAME signal impulse which is what gives the illusion of sound stage.

 

[Groucho]

Your example is not relevant to reproduction of the soundstage. The rising trombone in one channel will/should have a time delayed version of the same rising trombone in the other channel but time delayed & at a lower level - exactly what the microphone picked up (depending on how the mic placements were arranged). So this would give us the sound stage illusion & the placement of this trombone in the sound field. If there are no fluctuations in level or time differential between channels, then the sound stage will be stable. If there are fluctuations, the sound stage will not be as stable.

I don't see what you're saying here. If, depending on the placement of the microphones, we have a strong piccolo in one channel, and a strong double bass in the other without much leakage to the other channel, then if the speakers introduce frequency-dependent delays (which were not picked up by the microphones because they weren't there), we will have a new, but stable, inter-channel delay when we reproduce the recording. However, if the piccolo slides down and the double bass slides up, the frequency-dependent delays (caused by the speakers and not in the original recording) will be varying in different, maybe opposite, directions, thus causing a varying inter-channel delay at the listener's position. So I'm suggesting that unless the 'group delay' is constant at all frequencies, real stereo music with different varying frequency components left and right, will induce varying inter-channel delays for each component, thus making the sound stage a bit 'spongey'. Only DSP-active can reasonably claim to solidify that sound stage, or at least aim to reproduce it as intended.

Edit: just noticed the bit about the same impulse. I guess I'm thinking about audio 'holography' in general. Will think about it...

 

[Groucho]

OK, so thinking about what you've said, any frequency-dependent delays introduced by the speakers would not significantly vary the relative timing of the impulses reaching each ear from a single sound source, so not affecting the left-right sound stage, but it would introduce a sponginess in the front-back sound stage in that different frequencies in the impulse would be delayed differently and therefore appear to be coming from different distances to the listener (which wouldn't have happened during the recording). From another viewpoint, you could just say that non-time corrected speakers would smear transients and detail. In your original post you also say:

Another factor often mentioned with better reproduction systems is that they expose more subtle detail in the sound - is this one reason why they produce a more realistic sound stage - it goes down to a lower signal level & so is more like what we hear in nature & the illusion is more real?

 

[jkeny]

OK, so thinking about what you've said, any frequency-dependent delays introduced by the speakers would not significantly vary the relative timing of the impulses reaching each ear from a single sound source, so not affecting the left-right sound stage, but it would introduce a sponginess in the front-back sound stage in that different frequencies in the impulse would be delayed differently and therefore appear to be coming from different distances to the listener (which wouldn't have happened during the recording). From another viewpoint, you could just say that non-time corrected speakers would smear transients and detail. In your original post you also say:

Let's just step back a bit in order to reduce the complexities & variables that have to be considered by eliminating consideration of the speakers & their imperfections. If we perceive a different soundstage by ONLY changing the source (keeping all other devices the same, including the speakers) then I assume that something has changed in the analogue waveform reaching the speakers. My simplistic view is that ICTD and/or ICLD has changed. The interchannel relationship of the signals have changed in either timing or level.

What I was asking about was a test that analysed/measured the stereo analogue output stream in regard to the stability of this interchannel timing & level (ICTD & ICLD).

Some might say that the Diffmaker test is doing exactly this but is it? Maybe those who know the workings of Diffmaker can answer this? I'm aware that for instance, it may not be necessary to examine both channels as I presume that examining one channel & it's comparison to the input signal should reveal some change in transfer function but what level does Diffmaker work to - 90dB nulls? What about timing differences?

Some questions that occur to me - what level of difference in timing audibly shifts the sound stage - the demo I linked to showed that 0.22mS is audible but is this the threshold of audibility? What if it is a fluctuating timing difference rather than a fixed timing difference - does this make the threshold of audibility lower i.e. do we perceive a more diffuse, less solid sound stage with a less than fully stable timing differential between the channels?

Is this jitter? I propose that a fluctuating timing difference may not show up as jitter in a FFT. One reason being that this fluctuation in ICTD would be as a result of changes in transfer function being brought about by the system handling input signals which were complex & non-repeating i.e music. So a single tone or even multitone signal may not exercise the system sufficiently to elicit variations in transfer function.

The same considerations apply to ICLD - what is the threshold of audibility?

 

[jkeny]

this aint really rocket science or new ideas. The timing and phase relationships that come off the source would be maintained all the way till the speakers spit. However, your two stereo channels cant even achieve a null of much better than 40 db, and if you throw the speakers in too, now the null is (i am guessing here) another 15 to 20 db worse..and that I think is conservative.

Sorry, I'm not with you here - can you explain what you mean?

The term group delay is applicable here also, ie does each channel affect the amplitude and phase equally in the real word....the obvious answer is no. Hopefully, while each channel does not hold gain and phase lock, they dont drift or wander about....but the speakers will do much more damage. One instruement that can be used specifically to check phase (timing) is a vector voltmeter. In radar and some other stuff, we used this meter often to verify back up channels against the primary channel.

Again I need some more explanation here to understand what you are saying.

The common wisdom is that phase, for most folks, is not terribly audible but I do grant you that it is (along with amplitude) vital to get the best reproduction from the source for stereo, or true binaural.

I guess it depends on what you mean by phase not being terribly audible - ICTD is used for localisation of sounds & without it we would be terrible hunters/survivors & would most likely be extinct as a race by now

 

[j_j]

Well first let's squash a myth that seems to have infinite legs. There is a claim out there, echoed by any number of people who should have known better, that the best resolution in ITD (that's Interaural Time Difference) and/or ICTD (Inter Channel Time difference) is one sample duration. That is preposterously wrong, even though audiophile writers published hither and yon, even in IEEE Spectrum, have claimed that to be a problem. If it was true, your cell phone, cable modem, HDTV, dialup modem, and lots of other stuff wouldn't stand a snowball's chance of working.

Simply put, the order of magnitude for resolving full-scale level time differences in a redbook CD is on the order of 1/(44100 * 2 * pi * 65536) give or take a factor of 2 or 5 depending on the filter tracking quality of the hardware. (assuming that things like proper reconstruction filters are actually in place, if not, i.e. if it's "filterless" meaning that there is no reconstruction filter either analog or digital then the interaural resolution is much, much worse, yes, really, because signal images (digital images, not stereo directionality) will worsen the time resolution as well as create ambiguity)

The threshold of hearing for ITD is reported to be 2, 5, or 10 microseconds, depending on who you believe, and what kind of utterly artificial, carefully designed stimuli one uses in order to make the test as easy and sensitive as possible. (so it's clear, that's a good thing to do) Clearly, the time resolution of a redbook CD is way, way, way under that.

Now, that 2, 5,or 10 microsecond result does say something about playback in the acoustic realm, because 1 foot of differential distance in speakers is about 1/1130 seconds, give or take, seconds, or approximately 1 millisecond per foot. http://www.sengpielaudio.com/calculator-speedsound.htm is very informative in that regard.

However, those results are also for very, very broadband signals that depend almost entirely on high-frequency signals that do not often exist in real material. Percussion is one exception, which is interesting for reasons far beyond the presence of high frequencies (the energy of some impulsive percussion is high enough that the fundamental nonlinearity of air results in a spectrum that changes with distance until the peak goes below 120dB SPL or so, which is a different but interesting problem). Now, http://sengpielaudio.com/calculator-air.htm also shows how much signal gets from the stage to you, and puts a very clear slant on what issues might relate to "frequency response" in a real setting. Try 30kHz and 72F, 40% RH for some informative suggestions on what you hear in a natural setting.

My point? Yes, time alignment matters. To give more detail, some informal tests using 48kHz samples showed that at 200Hz, 2 samples was about the just noticeable difference (JND) (using 1-critical-band wide Gaussian pulses for stimuli). At 500Hz, somewhat less. At 1kHz, around 6 samples is the JND. At 4kHz, you're back down to 1 sample, give or take, and using broadband signals with a strong onset, the result is well under one sample, as one can see from the reported results above. The 1kHz result is expected, because two mechanism in the ear are clashing horribly at 1kHz.

But how much? One must relate all the time delays to the time delay around the head, variously estimated at between .6 and .9 milliseconds (frequency, distance, different heads), in order to relate the delays to biological timing.

Well, that's a lot of basics. To the OP question in speakers...

What matters? Interchannel matching. This is almost always speakers, speakers, speakers. Crossovers, etc, do not often give matching as close as one might wish. Matching to well under 1dB is ideal, and building speakers with that kind of match can provide startlingly good imaging. Usually (but of course not always) doing frequency response (that would be first-arrival frequency response, not long-term-integrated room response, please, which is (*(* near meaningless, see Johnston and Smirnov, or Fejzo, Johnston, et al, both AES preprints for more info) matching will get the delay in the speaker down to a reasonably unimportant level. Then, of course, you want equal distances, and you want equal amounts of reflections (with "none" being the best) from each side, and you'll get good imaging.

Electronics should not be an issue. If interchannel delay is an issue, "time to launch" should be considered. No, just no. Should not be. Yes, I know some early CD players shared a dac and put a half-sample delay, which did move the image slightly sideways. Phasers armed and locked on, Captain!

Some folks claim amplitude-only panning is enough, based on some 1950's work that was ok as far as it went, but that did not consider biologically sensible time delays. That isn't true, but it's sufficient for a first approximation. You can do better with panning both time and amplitude, but there will then be huge arguments from the industry, because that's not mono-compatible, and yes, people still care. Don't ask me.

Envelopment, etc, is a completely different issue. The length of that article is more than I want to do after a good dinner of salmon and garden greens. Maybe some other day.

 

[j_j]

The common wisdom is that phase, for most folks, is not terribly audible but I do grant you that it is (along with amplitude) vital to get the best reproduction from the source for stereo, or true binaural.

Depends on what kind of phase. If you have 1 million degrees of phase shift at 20khz, and the phase shift from DC to 20kHz is exactly a straight line, that's a pure delay, and no, you won't hear that if it's applied to both channels equally.

If you have more than about 15 degrees or so of phase shift (that deviates from linear phase) in a critical band, AND you have critical material, you may indeed her a monophonic effect.

For ITD, interaural phase shift matters much below 500Hz, and not much above 2kHz. HOWEVER, interaural phase shift of the ENVELOPE of the signal is quite important over 2kHz.

So if you muck up the waveform at low frequencies, or the envelope at high frequencies, you might have a problem. Mucking it up in the 1kHz range isn't nearly as touchy.

 

[Phelonious Ponk]

Well first let's squash a myth that seems to have infinite legs. There is a claim out there, echoed by any number of people who should have known better, that the best resolution in ITD (that's Interaural Time Difference) and/or ICTD (Inter Channel Time difference) is one sample duration. That is preposterously wrong, even though audiophile writers published hither and yon, even in IEEE Spectrum, have claimed that to be a problem. If it was true, your cell phone, cable modem, HDTV, dialup modem, and lots of other stuff wouldn't stand a snowball's chance of working.

Simply put, the order of magnitude for resolving full-scale level time differences in a redbook CD is on the order of 1/(44100 * 2 * pi * 65536) give or take a factor of 2 or 5 depending on the filter tracking quality of the hardware. (assuming that things like proper reconstruction filters are actually in place, if not, i.e. if it's "filterless" meaning that there is no reconstruction filter either analog or digital then the interaural resolution is much, much worse, yes, really, because signal images (digital images, not stereo directionality) will worsen the time resolution as well as create ambiguity)

The threshold of hearing for ITD is reported to be 2, 5, or 10 microseconds, depending on who you believe, and what kind of utterly artificial, carefully designed stimuli one uses in order to make the test as easy and sensitive as possible. (so it's clear, that's a good thing to do) Clearly, the time resolution of a redbook CD is way, way, way under that.

Now, that 2, 5,or 10 microsecond result does say something about playback in the acoustic realm, because 1 foot of differential distance in speakers is about 1/1130 seconds, give or take, seconds, or approximately 1 millisecond per foot. http://www.sengpielaudio.com/calculator-speedsound.htm is very informative in that regard.

However, those results are also for very, very broadband signals that depend almost entirely on high-frequency signals that do not often exist in real material. Percussion is one exception, which is interesting for reasons far beyond the presence of high frequencies (the energy of some impulsive percussion is high enough that the fundamental nonlinearity of air results in a spectrum that changes with distance until the peak goes below 120dB SPL or so, which is a different but interesting problem). Now, http://sengpielaudio.com/calculator-air.htm also shows how much signal gets from the stage to you, and puts a very clear slant on what issues might relate to "frequency response" in a real setting. Try 30kHz and 72F, 40% RH for some informative suggestions on what you hear in a natural setting.

My point? Yes, time alignment matters. To give more detail, some informal tests using 48kHz samples showed that at 200Hz, 2 samples was about the just noticeable difference (JND) (using 1-critical-band wide Gaussian pulses for stimuli). At 500Hz, somewhat less. At 1kHz, around 6 samples is the JND. At 4kHz, you're back down to 1 sample, give or take, and using broadband signals with a strong onset, the result is well under one sample, as one can see from the reported results above. The 1kHz result is expected, because two mechanism in the ear are clashing horribly at 1kHz.

But how much? One must relate all the time delays to the time delay around the head, variously estimated at between .6 and .9 milliseconds (frequency, distance, different heads), in order to relate the delays to biological timing.

Well, that's a lot of basics. To the OP question in speakers...

What matters? Interchannel matching. This is almost always speakers, speakers, speakers. Crossovers, etc, do not often give matching as close as one might wish. Matching to well under 1dB is ideal, and building speakers with that kind of match can provide startlingly good imaging. Usually (but of course not always) doing frequency response (that would be first-arrival frequency response, not long-term-integrated room response, please, which is (*(* near meaningless, see Johnston and Smirnov, or Fejzo, Johnston, et al, both AES preprints for more info) matching will get the delay in the speaker down to a reasonably unimportant level. Then, of course, you want equal distances, and you want equal amounts of reflections (with "none" being the best) from each side, and you'll get good imaging.

Electronics should not be an issue. If interchannel delay is an issue, "time to launch" should be considered. No, just no. Should not be. Yes, I know some early CD players shared a dac and put a half-sample delay, which did move the image slightly sideways. Phasers armed and locked on, Captain!

Some folks claim amplitude-only panning is enough, based on some 1950's work that was ok as far as it went, but that did not consider biologically sensible time delays. That isn't true, but it's sufficient for a first approximation. You can do better with panning both time and amplitude, but there will then be huge arguments from the industry, because that's not mono-compatible, and yes, people still care. Don't ask me.

Envelopment, etc, is a completely different issue. The length of that article is more than I want to do after a good dinner of salmon and garden greens. Maybe some other day.

...

 

[jkeny]

Thanks for posting J_J
I stated this already & wonder what you make of it & if you have any comments?

Let's just step back a bit in order to reduce the complexities & variables that have to be considered by eliminating consideration of the speakers & their imperfections.

If we perceive a different soundstage by ONLY changing the source (keeping all other devices the same, including the speakers) then I assume that something has changed in the electrical signal of the analogue waveform being fed to the speakers.

My simplistic view is that ICTD and/or ICLD has changed. The interchannel relationship of the signals have changed in either timing or level.

 

[Groucho]

Let's just step back a bit in order to reduce the complexities & variables that have to be considered by eliminating consideration of the speakers & their imperfections. If we perceive a different soundstage by ONLY changing the source (keeping all other devices the same, including the speakers) then I assume that something has changed in the analogue waveform reaching the speakers. My simplistic view is that ICTD and/or ICLD has changed. The interchannel relationship of the signals have changed in either timing or level.

My take on it would be different: if I perceive a different soundstage by only changing the source, my simplistic view would be that the most likely explanation is that that I am reacting to my own biases. Soundstage in particular is one of those nebulous audiophile phenomena that are very 'malleable' I tentatively suggest. And if I have to get out of my chair and fiddle with cables etc. to disconnect one source and substitute another, I don't think I stand a chance of telling them apart. If I can switch between them instantly from the listening position and hear the piccolo and clarinet separated on one setting and merging together on the other, I would be more convinced. In which case I would be looking at making a stereo recording and comparing phase between the channels for various test tones. But I wouldn't commit to the effort without establishing that the effect was real, first. Is there any evidence to that effect?

(I expect someone will say that placebo is also "real", but I think they know what I mean, really ).

 

[Phelonious Ponk]

My take on it would be different: if I perceive a different soundstage by only changing the source, my simplistic view would be that the most likely explanation is that that I am reacting to my own biases. Soundstage in particular is one of those nebulous audiophile phenomena that are very 'malleable' I tentatively suggest. And if I have to get out of my chair and fiddle with cables etc. to disconnect one source and substitute another, I don't think I stand a chance of telling them apart. If I can switch between them instantly from the listening position and hear the piccolo and clarinet separated on one setting and merging together on the other, I would be more convinced. In which case I would be looking at making a stereo recording and comparing phase between the channels for various test tones. But I wouldn't commit to the effort without establishing that the effect was real, first. Is there any evidence to that effect?

(I expect someone will say that placebo is also "real", but I think they know what I mean, really ).

And you'd be golden if you could hear the difference. If you couldn't, it would be the switching system masking the difference between the two sources, or your bias in favor of not hearing any differences in anything, even blind, when you don't know what you're comparing. Evidently the bias against hearing differences is so strong that even in blind listening it trumps the bias in favor of hearing what you want to hear, when you can clearly see the choices.

Tim