Stereophile Review of AR3a...

[Zilch]

It's academic, an artificial listening environment specific to a particular genre of music no longer mainstream.

[Most concert halls have 8' ceilings.... ]

http://www.polkaudio.com/downloads/whitepa..._WhitePaper.pdf

[Carlspeak]

It's academic, an artificial listening environment specific to a particular genre of music no longer mainstream.

[Most concert halls have 8' ceilings.... ]

http://www.polkaudio.com/downloads/whitepa..._WhitePaper.pdf

Another cryptic post from the Zilchester. Could you pls. elaborate on your first sentence?

Also, Polk's figure 1 in his paper has be a bit skeptical. I see no reason there couldn't be IACT with the live performance as well as with the stereo reproduction with a SINGLE performer as shown. With multiple performers, the whole thing becomes more muddied - in theory at least.

[speaker dave]

That science remains in a primitive state. Unless and until it advances, the art of engineering better sound recording/reproducing systems cannot advance either. That is one of its bottlenecks. To demonstrate how primitive it really is, here's a paper writte by Leo Beranek in 2008;

http://www.leoberanek.com/pages/concerthalls2008.pdf

In it he explains how the coefficient of absorption in Sabines equation has to exceed 1 to reconcile it with Eyring's equation even though that makes no sense because 1 represents the coefficient of absorption for an open window which produces no reflections. He also defines the "Listener Envelopment Factor" LEV which takes into consideration reflections off of the back wall and ceiling as sigificant contributors. Suddenly after all these years, the ceiling reflections are important.

So we are told once again that the engineers and scientists are idiots and know nothing.

It has long been known that Sabine's equations were approximate and best used for rooms of low absorption, such as the concert halls that they were first developed for. Beranek just shows that with the right absorption coefficient you can get exactly the same answers between the two equations. An alpha higher than 1 isn't of great concern to anyone other than SM. It is well known that in some situations material absorptions will measure higher than 1. The recomendation is that materials be measured in situations similar to how they are used. Concert hall seats need to be measured in large blocks, for example.

The improvement in Lev calculations that Beranek quotes still relies heavily on Lateral Fractions, a measure of the relative strength of energy from the sides. The importance of this goes back to studies by Baron in the early 80's. It was important for finally defining why fan shaped halls were frequently less well liked than earlier shoebox shaped halls.

I don't know about fluid dynamics, but there are plenty of ray tracing programs that are both sophisticated and known to give good correlation with architectural measurements. I used CATT Acoustics and found it impressive. It is known to give better results than the simple equations for complex rooms. For example most concert halls need to be considered as two spaces coupled together (the orchestra shell and the hall proper). Ray tracing programs automatically deal with that.

Can someone here (SM perhaps) scale down Baranek's theory to a typical home listening environment and interpret what the results mean?

Beranek isn't talking about home listening rooms. Most of his papers and books have to do with defining the heirarchy of important parameters for concert halls. That is, what should we measure and what is a good numerical result? What is the right reverberation time and should it vary with frequency? What is the right proportion of early sound versus late sound? The same science of acoustics pertains to the home listening room but the goals are very different.

The best paper on that topic, to my mind, is by Sean Olive as quoted in Floyd Toole's book. He took the standard loudspeaker measurements and found the correlation of factors between what you could measure and the numerical ranking of a group of speakers. In the end he achieved great correlation with high weighting of the factors of on-axis response flatness, on-axis response smoothness, and bass extension. Power response smoothness was a minor factor and power response flatness was not desired.

You can always proclaim that no progress has been made, or you could read the paper and see the progress.

David

[Zilch]

It was Allison & Berkovitz's (and AR's, more generally) ill-founded reliance upon the applicability of Beranek's studies of concert halls to small rooms that led them astray in their approach to vintage designs....

[genek]

It was Allison & Berkovitz's (and AR's, more generally) ill-founded reliance upon the applicability of Beranek's studies of concert halls to small rooms that led them astray in their approach to vintage designs....

...and resulted in the design of speakers whose owners would prefer to expend many dollars and hours of effort to maintain rather than replace with modern speakers designed to newer and "better" scientific research. And even with all the new and "better" research on concert hall acoustics, no designer or architect seems to be able to build a new hall that can equal or surpass the sound of the best old halls basedt on "ill-founded" science without going back and implementing a six-figure redo after the shortcomings of their new efforts become obvious.

That irritates you, doesn't it? Otherwise, why spend so much of your time coming here and telling everybody how wrong their preferences are?

[Zilch]

That irritates you, doesn't it?

Not in the slightest; I have no issues with anyone's preferences....

[kkantor]

Not in the slightest; I have no issues with anyone's preferences....

Well, then, what justifies the selection of alternative measurement methodologies?? You seem to feel that certain measurement protocols are inherently superior and "more accurate" than others. It's circular reasoning to assert that speaker accuracy can be quantified objectively according to a specific measurement protocol, then go on to justify the choice of that particular measurement protocol because it is subjectively preferrable.

-k

[kkantor]

It was Allison & Berkovitz's (and AR's, more generally) ill-founded reliance upon the applicability of Beranek's studies of concert halls to small rooms that led them astray in their approach to vintage designs....

More data, less wank. On what basis do you assert:

1- That A&B had an "ill-founded reliance upon" Beranek's studies of concert halls?

2- That A&B's work was "astray"? Astray from what path? By what criteria?

Also, do you have a suggestion for what work they >should< have consulted?

- k

[soundminded]

So we are told once again that the engineers and scientists are idiots and know nothing.

It has long been known that Sabine's equations were approximate and best used for rooms of low absorption, such as the concert halls that they were first developed for. Beranek just shows that with the right absorption coefficient you can get exactly the same answers between the two equations. An alpha higher than 1 isn't of great concern to anyone other than SM. It is well known that in some situations material absorptions will measure higher than 1. The recomendation is that materials be measured in situations similar to how they are used. Concert hall seats need to be measured in large blocks, for example.

The improvement in Lev calculations that Beranek quotes still relies heavily on Lateral Fractions, a measure of the relative strength of energy from the sides. The importance of this goes back to studies by Baron in the early 80's. It was important for finally defining why fan shaped halls were frequently less well liked than earlier shoebox shaped halls.

I don't know about fluid dynamics, but there are plenty of ray tracing programs that are both sophisticated and known to give good correlation with architectural measurements. I used CATT Acoustics and found it impressive. It is known to give better results than the simple equations for complex rooms. For example most concert halls need to be considered as two spaces coupled together (the orchestra shell and the hall proper). Ray tracing programs automatically deal with that.

Beranek isn't talking about home listening rooms. Most of his papers and books have to do with defining the heirarchy of important parameters for concert halls. That is, what should we measure and what is a good numerical result? What is the right reverberation time and should it vary with frequency? What is the right proportion of early sound versus late sound? The same science of acoustics pertains to the home listening room but the goals are very different.

The best paper on that topic, to my mind, is by Sean Olive as quoted in Floyd Toole's book. He took the standard loudspeaker measurements and found the correlation of factors between what you could measure and the numerical ranking of a group of speakers. In the end he achieved great correlation with high weighting of the factors of on-axis response flatness, on-axis response smoothness, and bass extension. Power response smoothness was a minor factor and power response flatness was not desired.

You can always proclaim that no progress has been made, or you could read the paper and see the progress.

David

"So we are told once again that the engineers and scientists are idiots and know nothing."

In this field I think that would be overly generous. When you spend many tens and even hundreds of millions of dollars of other people's money and get failure after failure it begs the question who is the more ignorant and incompetent, the acoustic scientists who design the disasters or the people who entrust the money to them. Philharmonic hall at Lincoln Center has been such a disaster these last nearly 50 years and has had so much patron money poured into trying to fix it that it was renamed "philanthropic hall" by those in the business. Having failed there, Beranek went on to produce the disaster of Louse Davies Hall in San Francisco which was billed as "the perfect instrument for hearing symphony orchestras." As Genek has told us, it had to be reworked because the acoustics were so poor.

"It has long been known that Sabine's equations were approximate and best used for rooms of low absorption, such as the concert halls that they were first developed for. Beranek just shows that with the right absorption coefficient you can get exactly the same answers between the two equations. An alpha higher than 1 isn't of great concern to anyone other than SM."

Proof that the illogic of fudging the properties of a material that on paper absorbs more sound than an open window in order for the mathmatics to "work" doesn't provoke a major rethink of what is wrong with the equation because those who are so called experts haven't thought it through critically in the first place. That it doesn't bother them only proves that their thinking aparatus is numb. When a proposition is illogical and it doesn't bother people who are experts especially when their work product produces so many failures, what else could you call them? How can you take them seriously. Saying that the science is still primitive is as generous to them as you can get.

[kkantor]

The best paper on that topic, to my mind, is by Sean Olive as quoted in Floyd Toole's book. He took the standard loudspeaker measurements and found the correlation of factors between what you could measure and the numerical ranking of a group of speakers. In the end he achieved great correlation with high weighting of the factors of on-axis response flatness, on-axis response smoothness, and bass extension. Power response smoothness was a minor factor and power response flatness was not desired.

I think Toole and Olive's work is excellent. Clear protocols to establish preference ranking and discrimination thresholds were a landmark step forward in the field. But, their extrapolation into accuracy criteria is, IMO, a conceptual error. It is important not confuse the ability to group speakers into preference bins with either the inability to distinguish between the sound of the speakers within a bin, or with the ability to state what would constitute a design improvement beyond the top preference bin. While I am not completely familiar with all of Olive's recent work, I don't believe he's ever asserted either of these positions, himself.

-k

[genek]

Well, then, what justifies the selection of alternative measurement methodologies?? You seem to feel that certain measurement protocols are inherently superior and "more accurate" than others. It's circular reasoning to assert that speaker accuracy can be quantified objectively according to a specific measurement protocol, then go on to justify the choice of that particular measurement protocol because it is subjectively preferrable.

Consumer preference as determined by how speakers score in listening panels is a perfectly valid business reason for manufacturers to make what they make, but telling consumers that what they prefer is more "accurate" or has some other scientifically superior foundation is probably better marketing.

[Zilch]

They went in with a hypothesis presuming a dominant, diffuse, steady-state reverberant field beyond a relatively short critical distance in typical listening rooms, consistent with and according to the Beranek model of large listening spaces, upon which the fundamental design tenants relied.

It wasn't home.

We forget that this was uncharted territory at the time; they were only then figuring it out. There is a very real sense in which they didn't know what they were doing, which renders pioneering work definitive in and of itself....

[kkantor]

They went in with a hypothesis presuming a dominant, diffuse, steady-state reverberant field beyond a relatively short critical distance in typical listening rooms, consistent with and according to the Beranek model of large listening spaces, upon which the fundamental design tenants relied.

It wasn't home.

We forget that this was uncharted territory at the time; they were only then figuring it out. There is a very real sense in which they didn't know what they were doing, which renders pioneering work definitive in and of itself....

C'mon. I don't mind agreeing with you from time to time, but asserting Allison and Berkovitz "didn't know what they were doing" is plain silly. That's like saying Beranek or Olson, "didn't know what they were doing" because of Thiele and Small. Ford didn't know what he was doing because of the Prius. Newton didn't know what he was doing because of Einstein who didn't know what he was doing because of Bohr, etc. There were plenty of people at the time who didn't know what they were doing, but A&B have proven not be among them. Anyway, what makes you think we "know what we are doing" now? Have speakers gotten so much better sounding?

In terms of their specific reliance on Beranek's view of reverberation, I guess I need to re-read A&B. I don't get it as a foundational issue. Further, I believe that all successful loudspeaker designs are essentially and necessarily based on what the listener actually hears. If you dispute this, I again ask: what measurements can you justify, not as providing equal preference scores, but as objective standards of accuracy and signal integrity? I would ask this of A&B, et al, too, BTW.

-k

[kkantor]

Consumer preference as determined by how speakers score in listening panels is a perfectly valid business reason for manufacturers to make what they make, but telling consumers that what they prefer is more "accurate" or has some other scientifically superior foundation is probably better marketing.

Well said.

- k

[Zilch]

You and I can dissect it over beers; Howard and I have slogged through it pretty thoroughly here in the kitchen already.

For starters, paraphrasing: "Ignore the nasty anechoic off-axis response curves, that all gets smoothed out by the dominant integrating reverberant field. It's the total output power that counts, not the direct field; it all good by the time it gets to the listener. Diffraction from the grille and edge molding and interference between the drivers may look bad, but it's of no consequence once you third-octave smooth the average of measurements taken at seven locations each in 14 separate typical listening rooms."

Then, the "Ut, oh" moment: "Hmmm, looks like the wide-dispersion constant-directivity we strove to achieve and measured in the reverberant chamber got progressively eaten up by room absorption at higher frequencies, leaving the direct field more dominant than we supposed, and the low end isn't matching what occurs in the concert hall."

And, implicitly: "Perhaps we should rethink some of the fundamental premises we're working with here...."

The rest is history -- they were learning, figuring it out, as were their competitors. It was an exciting time, and you were there.

In a very fundamental sense, it's apparent after all that, and the intervening years of "progress," you agree with SM that we haven't yet even defined the problem, let alone found the answers. We don't know what we're doing either and need a new horse, as this one has gone kinda lame....

[speaker dave]

"So we are told once again that the engineers and scientists are idiots and know nothing."

In this field I think that would be overly generous. When you spend many tens and even hundreds of millions of dollars of other people's money and get failure after failure it begs the question who is the more ignorant and incompetent, the acoustic scientists who design the disasters or the people who entrust the money to them. Philharmonic hall at Lincoln Center has been such a disaster these last nearly 50 years and has had so much patron money poured into trying to fix it that it was renamed "philanthropic hall" by those in the business. Having failed there, Beranek went on to produce the disaster of Louse Davies Hall in San Francisco which was billed as "the perfect instrument for hearing symphony orchestras." As Genek has told us, it had to be reworked because the acoustics were so poor.

The highly publicized failures in a few of this centuries concert halls is not surprising if you know the process. Huge tradeoffs are made between the ego of the architect, the financial needs of the patrons and the advice of the acousticians. This is not to say that mistakes are never made.

Proof that the illogic of fudging the properties of a material that on paper absorbs more sound than an open window in order for the mathmatics to "work" doesn't provoke a major rethink of what is wrong with the equation because those who are so called experts haven't thought it through critically in the first place. That it doesn't bother them only proves that their thinking aparatus is numb. When a proposition is illogical and it doesn't bother people who are experts especially when their work product produces so many failures, what else could you call them? How can you take them seriously. Saying that the science is still primitive is as generous to them as you can get.

A fairly stupid argument. What is the (universally used) Eyring equation if not "a major rethink of what is wrong with the (Sabine) equation"? Numerous papers were written on ever more accurate equations for more specific cases, including for opposing walls of disimilar characteristics. In the end numerical methods have proven both accurate and useful.

Ignoring progress made does not entitle you to claim it doesn't exist.

I'm willing to cut Sabine a little slack, knowing that he fostered a branch of science over a century ago with nothing but organ pipes and a stopwatch.

I await your textbook to shed much needed light on the subject.

[soundminded]

The highly publicized failures in a few of this centuries concert halls is not surprising if you know the process. Huge tradeoffs are made between the ego of the architect, the financial needs of the patrons and the advice of the acousticians. This is not to say that mistakes are never made.

A fairly stupid argument. What is the (universally used) Eyring equation if not "a major rethink of what is wrong with the (Sabine) equation"? Numerous papers were written on ever more accurate equations for more specific cases, including for opposing walls of disimilar characteristics. In the end numerical methods have proven both accurate and useful.

Ignoring progress made does not entitle you to claim it doesn't exist.

I'm willing to cut Sabine a little slack, knowing that he fostered a branch of science over a century ago with nothing but organ pipes and a stopwatch.

I await your textbook to shed much needed light on the subject.

"What is the (universally used) Eyring equation if not "a major rethink of what is wrong with the (Sabine) equation"? "

You make it too easy. If Eyring's equation was right and all of the rest of the theory was right, then blunders like Lousie Davies Hall wouldn't happen, that is unless the most respected acousticians like Beranek who was instrumental in the design was more interested in collecting his fee then he was in forever owning one more failure that he should have known in advance wouldn't work because the theory told him it wouldn't...if it was right. You can chalk up his successes like Tokyo more to chance than to theory. Davies Hall happened after Philharmonic hall. Was this the same mistake over again that he agreed to be a party to? Are his professional ethics that weak? I don't think so.

"Ignoring progress made does not entitle you to claim it doesn't exist."

Acknowledging that one model that doesn't work has been replaced with another that still doesn't work doesn't mean the problem has been solved by a long shot. There is still far further to go than has been achieved to date. They are still walking into the woods, not out of it.

I haven't noticed the world beating a path to your door to grab up your designs either.

[speaker dave]

For those that are interested in reading recent research on subjective evaluation, the following blog by Sean Olive is highly recommended.

http://seanolive.blogspot.com/

Scroll down and go through the past archives. His bit on what source material gave the best scores in subjective evaluations (pink noise, followed by dense pop/rock recordings) was interesting. He also gives a talk on room correction systems and how they were ranked. A loudspeaker plus a room correction system that is highly ranked in a comparison test is a good indicator of what a highly regarded speaker (sans correction) would need to be.

To Ken: is a high rank in preference testing an indicator of accuracy? I don't know if it is or not, but I suspect that both you and I would accept consumer preference in the marketplace over some hard to define notion of accuracy. (Not that I wouldn't strive for accuracy...)

And a last word on concert halls. Yesterday I went to my first concert in Toronto's new Koerner Hall at the Royal Conservatory of Music. It is considered both physically beautiful and excellent sounding. Clearly the result of dumb luck on the part of the architects and acousticians.

David

[speaker dave]

"I haven't noticed the world beating a path to your door to grab up your designs either. "

Maybe not, but I've been thinking about a speaker along the lines of a Bose 901 with a bunch of added tweeters. That will make them sit up and take notice.

David

[Zilch]

I haven't noticed the world beating a path to your door to grab up your designs either.

No, huh?

[You're not paying attention, obviously.... ]

[genek]

And of course, it required the listener to lock himself into a specific position, usually at the sweet spot

This would be the kiss of death for any speaker for me, regardless of its other qualities. Real musicians don't get up and move around the stage if I do, and I expect recorded ones to behave the same way.

[Zilch]

And they backed up that hypothesis by making scads of measurements that proved their point.

Their data DISPROVED the hypothesis, Howard.

Get past that; it has you locked into a '70s paradigm of how speakers work....

[speaker dave]

The problem with the whole direct field/reverberent field debate is that none of our observations truly prove our respective points. Most of the speakers we would point to have a fairly flat anechoic response and a gradually falling power response. The room adds increasing absorption (towards the high end) and so the end result without EQ is a room curve that falls off from LF to HF. Howard would say "gently falling in-room response. That's why it sounds good." Zilch and I would say "the underlying direct field is flat. That's why it sounds good." As long as the power response falls predictably, i.e. the directivity index rises gently, the direct field can be flat while the room curve follows popular notions of what in-room curves should be.

Direct field or room response, which is important?

The only way to get at the question is to try cases where the relationship between power response and direct field are very atypical. I've described several times the study by Lipshitz and Vanderkooy where they used a side firing dipole to independently manipulate the reverberant field (or the direct field). Their conclusion was that holes in the power response were benign. Flat power response was too bright, fixing the power response by messing up the direct response was bad, etc. etc.

I'm studying the history of the Cinema "X" curve. This is the traditional EQ for movie houses. Note that X stands for eXtended. The extended curve with more highs is rolled off a mere 10dB by 20kHz. People were happy with that until they realized smaller theaters sounded too dull when equalized to the curve, so somebody came up with a variable X curve that has various rolloffs based on room size. So if a rolled off high end is important, why can't we nail down how much to roll it off? How can that be an indeterminate yet important parameter? It seems a lot more plausible to me that, just like with our perception of source direction, the early or direct field is disproportionately important and the ear is able to largely ignore the late arriving sound.

David

[kkantor]

Both Carver and Lexicon came up with a similar approach, although it was done electrically and not acoustically as Polk has done it, and while it often resulted in some remarkable results with at least some recordings, it also could sound a bit funny at times. And of course, it required the listener to lock himself into a specific position, usually at the sweet spot, although the Lexicon processor could move that sweet spot location to the left and right of the middle if the listener made the processor adjustments. If the listener moved out of the focus location dictated by the processing (this was possible if more than one person was listening to the program) the overall frequency response was compromised.

Live music certainly does have interaural crosstalk, but the idea with the Polk, Carver, and Lexicon approach was to deal with the fact that the whole stereo soundstage was being created by two, spaced-apart speakers. At a live performance all of the sound is coming from discrete points across the stage. There are no "phantom" images at a live performance, such as what you have at every location between the left and right speakers with stereo playback.

Howard Ferstler

Howard,

Someday, I hope we can talk about the real history of crosstalk cancellation. I think it is a very interesting story. One of my very first projects working as an intern for Berkovitz at AR was to analyze Damaske and Schroeder's huge, tapped-inductor-based crosstalk circuit, and develop a more modern signal processing equivalent. We wanted to add recursion, too. I developed my first approach around the hand-made Bucket Brigade tapped delay lines that AR had done with Reticon for the 16-ch reverberator. Joel Cohen at Sound Concepts got involved, and marketed the IR-2100 using a commercially available Reticon. Carver brought out Sonic Holography, which instead used opamp gyrators to model Damaske's inductor. Then Polk did it cleverly with their SDA, (which couldn't apply recursion, but had other advantages.) There were various lawsuits, which were, to my memory, all dismissed when it was found that JVC had beaten all of us to market with their, "BiPhonic" stuff. (Remember their earphone-microphones to make binaural recordings with your own head?) In Japan also, Roland was playing in the HRTF area with their "RSS" studio technology, and used some cross-talk cancellation in the algorithms. Let's not forget Zuccarelli's Holophonics or Hughs SRS or Circle Surround or Aura or the Convolvotron.

-k

[kkantor]

For those that are interested in reading recent research on subjective evaluation, the following blog by Sean Olive is highly recommended.

http://seanolive.blogspot.com/

Scroll down and go through the past archives. His bit on what source material gave the best scores in subjective evaluations (pink noise, followed by dense pop/rock recordings) was interesting. He also gives a talk on room correction systems and how they were ranked. A loudspeaker plus a room correction system that is highly ranked in a comparison test is a good indicator of what a highly regarded speaker (sans correction) would need to be.

To Ken: is a high rank in preference testing an indicator of accuracy? I don't know if it is or not, but I suspect that both you and I would accept consumer preference in the marketplace over some hard to define notion of accuracy. (Not that I wouldn't strive for accuracy...)

And a last word on concert halls. Yesterday I went to my first concert in Toronto's new Koerner Hall at the Royal Conservatory of Music. It is considered both physically beautiful and excellent sounding. Clearly the result of dumb luck on the part of the architects and acousticians.

David

On one hand, I agree that a high preference rank is a very important indicator of accuracy. On the other hand, it is already established that some number of alternative products with readily distinguishable sound can all fall within the most preferred bin, and so we haven't fully determined a scale of absolute accuracy, or clear design goals.

I just D/L'd Olive's paper on room correction. I'm not yet convinced it reveals that much in the way of loudspeaker design goals, at least in the areas we are talking about. OK, we know that peaks, dips and resonances are a bad thing, and that 1/3-octave, quasi-CB resolution is not enough. We know that increasing smoothness and bass extension will generally lead to a preferred speaker. Etc. I need to read and digest the paper carefully before I assert too much about it. I will say that I was surprized by small and similar selection of program material. With such a small selection of consistent material, any experiment can yield results that reflect an inadvertent selection bias arising from past experience with the products; or, similarly, results that reflect conditioned expectations on the part of listeners who have become familiar with the music as reproduced a certain way over many years.

-k