AVS Forum banner
Status
Not open for further replies.
1 - 19 of 19 Posts

·
Registered
Joined
·
1,611 Posts
Discussion Starter · #1 ·
Can you measure...

the sound stage presented to you?


Everyone knows what happens when put that nice RPTV in between your left and right speakers. Also, what about when you spead your speakers further apart, move them further out from the front wall into the room, toe them in/out? It will all affect the presented sound and image, but can that image be measured?
 

·
Registered
Joined
·
707 Posts
Nope. Not really.


In fact, there is not even any solid concensus on what causes "good imaging" and "bad imaging". It involves phase response, dispersion, power response into the room, diffraction effects, frequency response, and reflected sound, but there are as many "theories" (actually, wild-ass guesses) as there are speaker designers. There isn't even any agreement about which side(s) of the speaker the sound should come out of, which should tell you how little is really understood.
 

·
Registered
Joined
·
1,028 Posts
Imaging is a product of your brain processing sound waves. The sound can be measured easily enought, but so far there is no software or instrument which can include all of the time, phase, amplitude, and other intangibles and measure the soundstage that you can easily hear.


In fact, research in robotic technology is focusing on localizing direct sounds and researchers are having quite a hard time with that. A far less daunting challenge than measuring a perceived soundfield like your stereo gives.
 

·
Registered
Joined
·
251 Posts
In order to really analyze this, you would need a way to solve three dimensional partial differential equations with the appropriate boundary conditions for your room. This is impossible. You could simulate it with a computer and a few thousand sensors, but it would be a lot of work. I would imagine people have taken some rudimentary stabs at this problem, but I have not searched the literature for it.


If you would like to, I suggest you search the archives of the Journals of the Audio Engineering Society, and the Acoustical Society of America. If you live near a major university with an Engineering or Physics department, they will probably have these in their library.


Tim


ps >> by the way, if you find anything I will be very interested
 

·
Registered
Joined
·
1,611 Posts
Discussion Starter · #6 ·
To follow up:

...and I'm not trying to start anything here...

If none of this can be measured, then why do people have to see the "measurements" of a cable to believe that it makes a difference. I'm not talking about any psuedo effect here or how one would "perceive" there to be a difference. Just plain and simple. All I ever see when a thread is posted about cables is "show me - measure it". I'm not saying I believe one way or the other, it just seems to be the mostly used arguement against a "better" cable. Unless of course, cables don't make a difference in sound stage, depth, and imaging.
 

·
Registered
Joined
·
5,897 Posts
Quote:
Originally posted by video321
To follow up:

...and I'm not trying to start anything here...

If none of this can be measured, then why do people have to see the "measurements" of a cable to believe that it makes a difference. I'm not talking about any psuedo effect here or how one would "perceive" there to be a difference. Just plain and simple. All I ever see when a thread is posted about cables is "show me - measure it". I'm not saying I believe one way or the other, it just seems to be the mostly used arguement against a "better" cable. Unless of course, cables don't make a difference in sound stage, depth, and imaging.
You made a pretty wild leap there. The guys above are not implying that there are no measurable differences between different sounding speaker/room setups, but simply that we don't currently understand how to translate all the stuff we can measure (i.e., phase, group delay, diffusion, absorption, reflection, diffraction, loudspeaker dispersion, frequency response, etc., etc., etc.) into a consistant theory of how to arrive at the "best" soundstage.


If two speaker setups sound different, believe me, appropriate measuring equipment will detect differences at the listening position. What those measurements mean may be up for discussion, but the fact that we can measure them is not.


When talking about a cable or something similar, it doesn't matter if we understand how the different things we can measure will affect the final sonic presentation, it only matters that when we can hear a difference, we can measure a difference. In this case, the instruments are more sensitive than our ears.
 

·
Registered
Joined
·
5,897 Posts
BTW, I would think finite element approximation of the closed form partial differential equations would be the most productive route to take. This method has been very successful in a variety of problems, from structural response and heat transfer to electrical networks and fluid behavior.


I believe D.E. mentioned once that he used some FE based software for acoustical modeling. I'm not sure what software, or what its capabilities and limtations are, but I suspect that with increased computing power we will start to see useful simulation and prediction capabilities emerge for fairly complex acoustical environments.
 

·
Registered
Joined
·
1,028 Posts
video321-


I think you are exactly right. Since there is no instrument which can measure all of the subtle variables in a sound field and apply the proper weighting to all of those variables and mimic the processing performed by the brain to create a virtual soundstage; I believe to conclude that the differences in cables (which we can already measure) are "inaudible" is very premature.
 

·
Registered
Joined
·
1,652 Posts
Yes, you can measure soundstaging and imaging. Your measurement tool will be humans, and because they are humans, you will need to calibrate and control certain variables which are obviously different than if you are measuring with a scope or voltmeter. The only requirement for a measurement is that it be repeatable, and there's no rule that says you have to use some electronic gizmo to perform the measurement.


Perceptual coders are tested this way (with humans), and one of the reliably measured parameters has to do with the perceived size of the soundstage. NRC subjective speaker tests and its derivatives also have quantitative measurements for soundstaging and imaging.


--Andre
 

·
Registered
Joined
·
10,200 Posts
The leap from soundstage to cables is absolutely ludicrous. Cables are electrical components. Their purpose is to deliver not sound signals, but electrical signals to the only piece of equipment in a stereo system that is actually intended to generate sound---the speakers. If two cables successfully deliver measurably identical electrical signals (both voltage and current) to the same set of speakers, then you can be darn sure that the sound the speakers produce will be identical in both cases.


Speakers are probably the only piece of equipment that must unavoidably be evaluated in the a subjective fashion. Sure you can do anechoic frequency response measuerements, on-axis, off-axis, etc., and make significant design decisions based on those measurements. But in the end you've gotta place a pair of ears in front of them, because it's true we don't understand fully how we hear.


That's just not so with cables; they can be completely evaluated in an electrical context. Even if we use our ears to detect problems with improperly manufactured or chosen cables, we can always trace the causes back to measurable electrical effects. Still, I think people are fooled into believing otherwise because they're the next link in the chain. Guilt by proximity perhaps.
 

·
Registered
Joined
·
1,028 Posts
I agree that if you deliver the EXACT same electrical signal to a speaker there will be no change in sound between two items, whether they be components or cables, but...


Different cables do NOT deliver identical electrical signals to speakers. Easily measurable differences in impedence, capacitance, and inductance will change the complex signal. In addition there are other effects, very difficult or impossible to meaure, like phase distortion, IM distortion, and skin effect which also change the signal.


But again- I agree that differences in speakers are at least one order of magnitude greater than differences in cables.
 

·
Registered
Joined
·
10,200 Posts
Quote:
Easily measurable differences in impedence, capacitance, and inductance will change the complex signal.
I've never disagreed with this. But note that impedance includes capacitance and inductance (and resistance), so you've really only mentioned one thing here.
Quote:
In addition there are other effects, very difficult or impossible to meaure, like phase distortion, IM distortion, and skin effect which also change the signal.
Adding "skin effect" to this list demonstrates a certain confusion about the science involved. Phase distortion and IM distortion are effects with direct bearing on audible results (which, by the way, can be measured). Skin effect is much farther removed than that. From an electrical or audio engineering perspective, skin effect matters in that it affects certain electrical properties a speaker cable---for example, it alters its impedance, which in turn can cause effects like magnitude and phase distortion (which are also easily measurable).


Put another way, you don't measure skin effect per se, you measure the effects it has on the electrical properties of the cable and deduce its presence from the results.


So skin effect is a good two levels "deeper" than the other effects you've mentioned. The levels look something like this:

--- composition: metal, stranding, crystal structure, etc.

--- low-level: skin effect, electric field structure, etc.

--- bulk electrical effects: impedance, nonlinearity, noise sensitivity

--- audible consequences: frequency response (magnitude/phase distortion), harmonic/IM distortion, noise

None of these lists are exhaustive, of course.


I think this "level confusion" as I've termed it above is pretty common in the cable industry. Although from a marketing perspective it's probably quite useful because it allows them to create a longer list of potential problems that cables might have. But in the end, it's completely sufficient to stay on just one of the last two levels---and in each case, that level is quite measurable. Any marketing wizard who spends too much time focusing on those first two levels honestly should be suspect.
 

·
Registered
Joined
·
251 Posts
"But again- I agree that differences in speakers are at least one order of magnitude greater than differences in cables."


More like 4 or 5.


"I've searched- found nothing."


That is surprising. I thought I had seen a few papers about this topic. Although we sometimes disagree, Bill, I want to assure you that I know that "soundstage" is a very real thing. Acoustic waves are very complicated, though. I am ashamed to say that acoustics was the only class I got a "B" on in undergrad.


"BTW, I would think finite element approximation of the closed form partial differential equations would be the most productive route to take. This method has been very successful in a variety of problems, from structural response and heat transfer to electrical networks and fluid behavior."


This is absolutely true. Electromagnetic fields are very similar to sound fields, and discrete computer approximations to the analytical solution of PDE is very, very common, and works very well. These simulations are used all of the time for modeling antennas, which are basically just loudspeakers for EM waves. Think about the problem Sprint has when they put up 300 cell phone towers in Los Angeles and they want to make sure they have good coverage. Finite element is not even the only approach, just the simplest and probably the most common. When I said solving these equations was impossible, I meant in the analytical form.


But, even though JBL and B&W probably do a lot of acoustic modeling, the days where there will be software available for home use are probably several years away. But, right now, TacT is doing room correction with one microphone. With a lot more processing power, you could set up a few dozen microphones in a three dimensional array, and do room correction on that. But, that would be pretty expensive.


And why go to all that trouble when you can already get enormous benefits to your system just by changing speaker cables?
 

·
Registered
Joined
·
10,200 Posts
Quote:
With a lot more processing power, you could set up a few dozen microphones in a three dimensional array, and do room correction on that. But, that would be pretty expensive.
You really don't need multiple microphones, since you could just take samples at each position sequentially. (The microphones are not left in place, but used only during a calibration process.) But it's not entirely clear how you should do the mathematics. If you try to correct at multiple locations simultaneously, the resulting system of equations is overdetermined. Therefore, you must decide: what is the best way to compromise? In general it seems you would want to create a nice, large "sweet spot." But what does that mean mathematically? Do you minimize the total frequency response error summed over all of the microphones? Do you minimize the peak frequency response deviation within a certain audible frequency rage? I dunno. I'm sure people have tried it, and I wonder what they have done and how well it worked.


What would make it interesting is if you could also place multiple full-range speakers on each channel---say, for example, 3 speakers per channel. With that approach you could perform perfect correction at exactly 3 locations, or you could create a much wider sweet spot. Of course, with 21 full-range speakers for a 7.1 system, the room would get crowded fast :) (Of course, you could place the multiple, independent drivers into the same cabinet.)


Another possibility that doesn't increase the number of speakers is to allow some "cross-feeding" of the signals across the multiple surround speakers. For example, the left-front signal could be fed to the left-rear or even the right-front speakers in order to help cancel room reflections. But I suspect that this would only be feasible if you have a good binaural model of the human head sitting there in the sweet spot.


Disclaimer: all this is speculation; I haven't done any of this room correction stuff before. I just like churning it around in my head, because I'm a geek :)
 

·
Registered
Joined
·
5,897 Posts
I would think multiple speakers per channel would suffer from comb filter effects. Room correction in general is a very interesting topic though.
 

·
Registered
Joined
·
10,200 Posts
Very good point, Bigus, you're right. Hmm, maybe the surround-speaker option is the only manageable alternative.
 

·
Registered
Joined
·
251 Posts
Michael,


Correct, the resulting set of equations would be overdetermined. But, once you acknowledge that the mathematics you are performing is only approximate, you would not want a square matrix anyway. The more equations the better; performing a least-squares approximate solution will minimize the mean-square error. Now, which error you choose to minimize is still up for debate. I would imagine you would try to perform the minimization on the deviation from an ideal impulse response.


"Disclaimer: all this is speculation; I haven't done any of this room correction stuff before. I just like churning it around in my head, because I'm a geek."


Ditto for me.


Tim
 
1 - 19 of 19 Posts
Status
Not open for further replies.
Top