Measurements & the stereo illusion

[Phelonious Ponk]

I'm expressing a belief, an opinion. And that opinion does not include a number of the things John has attributed to it. I have no opponent here, micro, and John has no opponent who is actually in this conversation. But thanks for your contribution.

Tim

 

[j_j]

Amir, I gave an example, that I doubt you disagree with, where DBTs were later overturned by long-term, in-the-field usage. Simple as that!

Um, you do understand that DBT's for long-term medical side effects are not the same thing as subjective tests, yes?

 

[j_j]

You cannot just take random, "industry standard" measurements without a hypothesis or model of why you are doing these particular measurements - it's laughable!

I can certainly take a variety of "industry standard" measurements and apply them directly to the great deal we know about both monaural and binaural hearing, and come to supported conclusions.

Why is that laughable?

Hearing is not a mystery, it is reasonably well understood, and the sensitivities of the hearing apparatus at the periphery are also established to a fair-thee-well. That means that we know quite well what information actually gets to the brain, and what isn't even detected.

And even "mostly useless' measurements like SNR can, if large enough, provide assurance.

If I am characterizing something I will certainly be doing more than impulse response and SNR.

I see no support for your arguments about interchannel timing jitter, among other things, they don't account for the rather intense spectral distortion such jitter would cause. I will say that there are things, like scrape flutter in tape decks, that work that way due to physics, but if that's your point why don't you just say you PREFER stuff that goes through a tape deck? That's your choice, and nobody gets to argue what YOU prefer for your own listening.

You are, I trust, aware that there are things called "euphonic distortions". Now, the euphony is at the listener, so there is not agreement on what is euphonic (at least to a great part) across the whole spectrum of listeners.

 

[Robh3606]

Hello jkenny

It may be that having a very accurate reproduction of the stereo signal is NOT as good as maybe having a signal which has something added to iti.e interaural fluctuations may be beneficial to the illusion, who knows?

I look at this from the opposite side of the fence. I my mind at the reproduction side as an end user accuracy is exactly what you want and need. I don't want my equipment adding anything to the original mix.

Have you ever listened to Amused to Death by Roger Waters. I use it as a set-up disk when I position speakers. It is encoded with QSound.

http://qsound.com/spotlight/users/recording-artists.htm

There are some rather interesting effects on the CD. You end up with sound in every corner of the room even behind your head in some cases all encoded using this software. When I do not have the speakers set-up optimally not all the effects work. When I get them set-up right I get repeatable placement of the effects from each pair of speakers.

Once I get them set-up I also find that this gives me the best sound-stage from the speakers as well. So I know this works, well for me at least.

You might want to check out the software as to how it works and why it is able to do what it does. As I see it we have a very good handle on how to manipulate the music on the creation side to give us the desired sound-stage effects the artists want.

The one that sounds most realistic! We all have a reference point in audio - it's called the real world, it's not a preference! The illusion that most closely matches our in-built reference is the one that is "right". Isn't this the objective of stereo reproduction?

Well isn't setting things up to be the most realistic to your "reference" almost the definition of using personal preference to set things up. It's a personal qualitative assessment.

It's not like you take out a sound-stage meter and say OK that ones a 22 and this ones a 34 so all is well against some standard.

I also think that you should be looking at speakers not electronics. Thats where you are going to get the most bang for your buck as far as set-up and results. There are several things you can look for in speakers examples being smooth F/R smooth off axis response, cabinets with minimal diffraction, mirror imaged drivers and the most important of all pair matching. Could be that the better systems are simply more consistent with less variation from speaker to speaker which is why they tend to have more clearly defined depth and sound-stage.

As a hobbyist I have been building speakers for a while and the quickest was to improve imaging is pair matching drivers especially through the midrange into the upper octaves where the spatial clues lie. I always pair match my drivers. You would be surprised how much variation you can see depending upon the drivers.

Rob

 

[j_j]

As a hobbyist I have been building speakers for a while and the quickest was to improve imaging is pair matching drivers ...

Rob

THAT^^^^^ and in spades.

 

[Groucho]

As a hobbyist I have been building speakers for a while and the quickest was to improve imaging is pair matching drivers especially through the midrange into the upper octaves where the spatial clues lie. I always pair match my drivers. You would be surprised how much variation you can see depending upon the drivers.

How do you pair match the drivers? In terms of electrical or acoustic measurements?

 

[jkeny]

For those iterested in delving into this further have a look at this presentation http://www.davidgriesinger.com/Acou...mbre, Source Separation_talk_web_sound_3.pptx

He maintains that "engagement, near/far, pitch perception, timbre perception, direction detection & stream formation, all derive from the same property of sound - phase coherence of harmonics in the vocal formant range 630Hz to 4KHz

Interesting!!

Check this sound clip from slide 3 where there are 4 different groups of phase coherence - the sound power & spectrum of each group is the same
http://www.davidgriesinger.com/Acoustics_Today/ten371_130_88r_ten.mp3

 

[microstrip]

I have said it before, but think is is useful to repeat it now - measurements are only relevant if they establish a correlation with sound properties. Given good instruments and imagination any one can get different figures for some entity between two units. But however finding something that shows a systematic correlation with properties of the "stereo illusion" is a different matter.

IMHO, unless the stereo illusion is first debated and we settle on how it is created the debate is very difficult. We have two components - the easily predictable, due to to the amplitudes and phases of the two channels, and the part that relies more on the listener perception and his interpretation of small cues. The second part can only be debated in terms of statistical analysis, as two people can easily have different interpretations of the some cues and weight them differently.

 

[jkeny]

I have said it before, but think is is useful to repeat it now - measurements are only relevant if they establish a correlation with sound properties. Given good instruments and imagination any one can get different figures for some entity between two units. But however finding something that shows a systematic correlation with properties of the "stereo illusion" is a different matter.

Yes, have you tried that sound clip? What is the difference you hear between the first group of sound Vs the last group?

IMHO, unless the stereo illusion is first debated and we settle on how it is created the debate is very difficult. We have two components - the easily predictable, due to to the amplitudes and phases of the two channels, and the part that relies more on the listener perception and his interpretation of small cues. The second part can only be debated in terms of statistical analysis, as two people can easily have different interpretations of the some cues and weight them differently.

His slide 56 sums it up nicely headed "Stereo localisation is an illusion based on fuzzy data" & the point "generated by our brains desire for certainty & our willingness to guess" & "the brain must guess the location based on an average of conflicting cues"

 

[j_j]

How do you pair match the drivers? In terms of electrical or acoustic measurements?

Both, if possible. Electrical affects crossover more than one would want, sometimes.

 

[j_j]

His slide 56 sums it up nicely headed "Stereo localisation is an illusion based on fuzzy data" & the point "generated by our brains desire for certainty & our willingness to guess" & "the brain must guess the location based on an average of conflicting cues"

On the other hand, for a left/right soundstage, it is possible to put in enough cues that the auditory system can integrate them into a sense of direction. This process is not as mysterious as some people seem to want to make it.

ITD and ILD create great deal of the sense of direction, because that's what happens in a natural setting, which is where we learn. In general, we expect natural experiences (if you want to hear signals that just don't sound right, add some contradictory cues to something, and yeah, it sounds wrong even if you can't say why), and the two speakers in a stereo setup provide a substantial amount of that particular information, especially when the reduction in loudness (usually called precedence effect, but there are two different effects, one peripheral and one CNS-based involved) 1 millisecond after sound onset in a given critical band is considered. That mechanism alone suffices to provide some very useful cues to the auditory system.

 

[Groucho]

Both, if possible. Electrical affects crossover more than one would want, sometimes.

In terms of logistics how do you go about this? Do you have to buy ten of each driver or speaker, test them exhaustively and send the rejects back?

 

[jkeny]

On the other hand, for a left/right soundstage, it is possible to put in enough cues that the auditory system can integrate them into a sense of direction. This process is not as mysterious as some people seem to want to make it.

If you are referring to me, forgive me if my stumblings through this area are seeming to make it mysterious - it's partly because some of this is new information to me. But from my limited reading it seems to me that it is by no means fully agreed as to the exact operation of localisation & there are some areas still in question. Maybe you can outline what are the outstanding issues & what are the generally agreed models of operation?

ITD and ILD create great deal of the sense of direction, because that's what happens in a natural setting, which is where we learn. In general, we expect natural experiences (if you want to hear signals that just don't sound right, add some contradictory cues to something, and yeah, it sounds wrong even if you can't say why), and the two speakers in a stereo setup provide a substantial amount of that particular information, especially when the reduction in loudness (usually called precedence effect, but there are two different effects, one peripheral and one CNS-based involved) 1 millisecond after sound onset in a given critical band is considered. That mechanism alone suffices to provide some very useful cues to the auditory system.

Agreed that we all have our real-world listening experiences as the reference point that we use to evaluate audio reproduction. In stereo reproduction we seem to have a facsimile which is good enough to appeal to our sense of hearing but still doesn't give us an accurate fascimile of what we hear in nature. So we are working with a system that gives us some, but not all of what we hear in nature. My initial post was a question about an area where I believed there are audible differences (you may disagree) & I asked some questions about the possible reasons for these audible differences.

I find the last sound clip I posted interesting because it demonstrates how depth of sound stage may be heard & explained - which was something I referred to in my description of a better reproductions systems giving us a more 3D sound stage. I find this happens with better sources if the speakers are capable of reproducing it (i use a Jordan JX92S, single drive speaker pair, btw)

 

[j_j]

If you are referring to me, forgive me if my stumblings through this area are seeming to make it mysterious - it's partly because some of this is new information to me. But from my limited reading it seems to me that it is by no means fully agreed as to the exact operation of localisation & there are some areas still in question. Maybe you can outline what are the outstanding issues & what are the generally agreed models of operation?

Well, Jens Blauert has written a very nice book on the subject. While he and I don't agree in every particular, it's a very good place to start, and I have been far too lazy with writing, since I retired and decided I was happy not being annoyed at the world.

I find the last sound clip I posted interesting because it demonstrates how depth of sound stage may be heard & explained - which was something I referred to in my description of a better reproductions systems giving us a more 3D sound stage. I find this happens with better sources if the speakers are capable of reproducing it (i use a Jordan JX92S, single drive speaker pair, btw)

There are 3 sets of cues for distance. The nearest (up to a couple of head diameters) is HRTF's (which vary with both distance and direction). The next is floor bounce vs. direct signal. The final one is direct/diffuse ratio.

Having speakers that match each other well is critical to not creating a "false" diffuseness, for instance, and also for keeping an imagine stationary in one place and not moving around by note, etc.

 

[j_j]

In terms of logistics how do you go about this? Do you have to buy ten of each driver or speaker, test them exhaustively and send the rejects back?

I can usually find pairs that match well with acoustic parameters, and then check the electronic ones. Usually, they will match pretty well. Occasionally one must tweak a crossover slightly.

Not the ideal solution, but as good as I can do with mailorder parts, and that's pretty good indeed.

 

[JackD201]

Many manufacturers use a golden unit method. They build a speaker and every other speaker is matched to that golden unit for QC. What parameters are tested exactly or the methodology, I'm not too sure of. I do know that rejection rates from suppliers is stipulated in OEM contracts. That alone helps keep tolerances from unit to unit close.

Stereo is dependent on symmetry. There's an understatement right there!

 

[jkeny]

Well, Jens Blauert has written a very nice book on the subject. While he and I don't agree in every particular, it's a very good place to start, and I have been far too lazy with writing, since I retired and decided I was happy not being annoyed at the world.

Yes but last revised & published in 1997? I would hope that some progress has been made in our understanding since then? Any good recent overview papers of the various models & what issues remain to be teased out.

There are 3 sets of cues for distance. The nearest (up to a couple of head diameters) is HRTF's (which vary with both distance and direction). The next is floor bounce vs. direct signal. The final one is direct/diffuse ratio.

Having speakers that match each other well is critical to not creating a "false" diffuseness, for instance, and also for keeping an imagine stationary in one place and not moving around by note, etc.

How does that concur with the sound clip example posted which demonstrates how phase coherence differences give depth of sound - HRTFs?

 

[j_j]

Yes but last revised & published in 1997? I would hope that some progress has been made in our understanding since then? Any good recent overview papers of the various models & what issues remain to be teased out.

I would strongly suggest you start with Blauert. While there has been some progress made, as it were, you are in need of the basics, and they aren't going to change unless the physics of air propagation also change, which is somewhat likely.

How does that concur with the sound clip example posted which demonstrates how phase coherence differences give depth of sound - HRTFs?

Not sure why you ask. Enough incoherence and everything will sound far away, but I thought that was clear.

 

[jkeny]

I would strongly suggest you start with Blauert. While there has been some progress made, as it were, you are in need of the basics, and they aren't going to change unless the physics of air propagation also change, which is somewhat likely.

As I'm sure you are aware, the progress I'm talking about is in the area of psychoacoustics!

 

[Phelonious Ponk]

As I'm sure you are aware, the progress I'm talking about is in the area of psychoacoustics, not the physics!

You can't separate the the psychoacoustics from the physics, the stereo illusion from the cues that are the raw material the brain works with, particularly if you design to manipulate that illusion, to stimulate perceptions. These cues are not only the foundation of the stereo illusion, they are the only things recordings, playback equipment and even room acoustics can affect; a designer cannot rewire the brain's reaction, he can only give it cues to work with. Don't understand the cues? You don't understand the psychoacoustics.

Tim