Objectivists, Harman Testing, Reviewers, and Reality

[andromedaaudio]

Amirm do you have a picture of the M 2 LS you are talking so much about , i googled(revel M 2 ) but couldnt find any .
One point though you have revel and ML(part of the Harman kardon company iirc ) in your business portfolio, objectively seen. did you put them up against the more well known competition and what price points are we talking here

 

[amirm]

Amirm do you have a picture of the M 2 LS you are talking so much about , i googled(revel M 2 ) but couldnt find any .

It is not a revel product. It is part of JBL line but the design was a co-development with the Revel group. Here is the link: http://www.jblpro.com/www/products/recording-broadcast/m2#.VZAfiPlVhBc

One point though you have revel and ML(part of the Harman kardon company iirc ) in your business portfolio, objectively seen. did you put them up against the more well known competition and what price points are we talking here

Sure, you should assume bias in my comments about Harman products as I mentioned But no, I have not had a chance to compare them to non-Harman products. I can only compare it to the rest of the products in Harman line in side-by-side testing.

That said, based on design methodology, I am confident of how performant it will be against other products.

 

[andromedaaudio]

Okay sure JBL, compressiondriver /horn highs and the mid woofer for the mids and bass, is this the M 2 in room response ?
What is the X over point and what is the resolution (smoothfactor ), i am sorry if you mentioned it somewhere i didnt read the whole thread
2,7 kHz ??

I explained already Mark that because you have not studied how these measurements are made, you mistakenly assume what you need to buy the Audio Precision to perform them. We are talking about frequency response measurements. Here is one I performed which cost $60 for a mic, and a computer and free software I already have (there is an 80 Hz crossover here):

This one measurement tells a world about what is going on here. It not only characterized the loudspeaker but it characterized it in my room. It shows my room to have wide swings in low frequencies and since those are room dependent, any subjective observation I make about bass notes will not translate to anyone else. See how I just connected subjective experience with objective science? We also see that tiny dip due to directivity error around 2.5 Khz. Another loudspeaker which had a much bigger problem there, would show even a bigger trough. And with it, as listening tests predict, will cause lower listener preference. Again, all of this cost $60. This I believe is less than the sales tax you pay for an audio cable you buy.

To show pseudo-anechoic measurements, you take the loudspeaker outside, elevate it from the ground and make measurements there. Again, all with that one $60 measurement mic. No audio precision was necessary or even useful as the software that runs in there is pathetic compared to computer programs. Heck there are even iPad apps that measure these things now so if you are not computer savvy, you can still do this work.

As to me demanding what you do, yes, I get to do that when you put your reviewer hat on. The moment you do that, you are in the industry and just like a manufacturer, we can scream bloody murder if they do something we don't approve. You are no longer just a forum member. What's more you routinely speak on behalf of all reviewers, defending current practices. So that entitles us to even more formal requests that you step up your game, and find a way to make your reviews more provable and believable.

Of course, like many businesses, you are welcome to ignore our requests, demands, or whatever you call it. So in no way are you forced to do anything. At the same time though, you can't ask us to tone down what we like, when you are doing the exact opposite.

 

[amirm]

Okay sure , JBL compressiondriver /horn highs and the woofer for the mids and bass is this the M 2 in room response ?

No, that is my (modest) Revel loudspeaker with no correction (but with crossover). I don't have the in-room measurement of the M2. I will see if I can get it from my designer and post.

 

[Phelonious Ponk]

I was going to point out the issue with quoting FT on dipoles, which were surround speakers, not panels, but Amir got to it already.

I firmly believe in all the research work Harman does, and value the results they publish, but in some cases, the methodology results in improper outcomes. The case in point is the MartinLogan they tested. All the other speakers are point-source box monopoles of one flavor or another, and I see the speaker shuffler is a great way to compare amongst speakers of a similar topology and dispersion characteristic. An ESL is a dipole line source, a very, very different beast in terms of how it interacts with a room than a point-source box monopole. Testing it with no regards for how a significant portion of its output is managed (the dipole rear-wave) is akin to grabbing one of the Infinity’s and laying it on its side and testing that, with predictably bad results.

After 23 years with ESL’s, I know full well that to get the most out of them, one needs to very, very carefully manage the rear wave. And I arrived at the knowledge through many, many measurements (and yes, lots of listening to correlate), some of which led me to change my views on reflection, diffusion and absorption topics. I won’t bore everyone with the details, but a dipole line source can’t just be plopped into the same spot as a monopole box and be expected to deliver its best.

A fairer fight might have been to put up a monopole line-source like a Scaena, which might have just blown away all the others with the benefits of its line source topology.

So even if one is a firm believer in the value of research, measurement and such, one needs to also evaluate the methodology and the details. Therefore, I reject the results of the ML in that test as completely invalid, however, I find the rest of the test extremely informative and useful.

I don't have any problem believing that the Harman test are not ideal conditions for panels or other dipoles and therefore, imperfect. I'd love to see Linkwitz' speakers tested and compare those results. I don't think that's the problem with the ML's, though. Measured on all axis in and anechoic chamber the didn't do well, and they don't do well in listening either. The shorter MLs change sound dramatically if you just stand up. The taller ones change very audibly if you walk around off axis. Even when they are well-positioned in a good room.

Tim

 

[JonFo]

...
One point though you have revel and ML(part of the Harman kardon company iirc ) ...

Just a note that MartinLogan is not at all related to the Harman group, it is owned by Shoreview, which also owns Paradigm.

 

[JonFo]

... The shorter MLs change sound dramatically if you just stand up. The taller ones change very audibly if you walk around off axis. Even when they are well-positioned in a good room.

Tim

Yep, once outside the line array 'beam' most panel speakers sound breaks down. Matter of fact the usual untreated room deployment of most ESLs or panel speakers (Magies) leads to the infamous head-in-a-vize meme we hear so much about. But a correctly treated room (i.e. lot's of absorption) yields a very different listening experience, one with a wider soundstage and less one-spot, of course, this really only works for those willing to devote the effort to getting it right.
As big a fan of ML's as I am, I never recommend them unless the prospect is willing to invest the effort in room treatments. But if you are willing to commit, man the rewards are quite impressive.

 

[andromedaaudio]

No i didnt mean martin logan, i meant mark levinson which is a brand amir also carries.
Martin logan electrostats might have some anomalies in their response and its ability to blend in with a woofer, but their speed might give it the edge for some people , a trade off so to say , i havent heard their latest offerings by the way, the CLS for example

Just a note that MartinLogan is not at all related to the Harman group, it is owned by Shoreview, which also owns Paradigm.

 

[jkeny]

I've dipped in & out of this thread & haven't read all the posts so forgive me if I am asking something that has already been answered but are the Harmon measurements just one set of measurable aspects of a speakers performance? Are there other possible aspects of performance not represented in these measurements? I'm not talking about room interaction & stereo operation as these have been addressed (to some extent, anyway) but are there other performance aspects not represented in the graphs?

I also wanted to go back to something I said & Amir replied to (which relates to the above question) - I said that auditory perception is not concerned with accuracy but rather about getting a close enough approximation to reality in realtime - so we take shortcuts/assumptions/guesses to get there. How close that approximation has to be depends on what task is being addressed. For instance we hear a very loud bang & instantly react in survival mode. 99% of the time we are wrong in our evaluation of the threat but our evolution has taught us that false positives are more beneficial for survival than false negatives. In a more relaxed type of listening we are using more auditory cues to create & evaluate the auditory scene in front of us. Note, we are not using all the soundwaves that impinge on our ears & evaluating this overwhelming data stream but are using specific aspects that we have learned give us enough information in the shortest time for our task-specific needs. For instance we evaluate the listening environment quite quickly by it's reverberation/ambience etc. & seem to be able to use this evaluation to "hear through" most environments & probably adjust as we continue in our listening.

So my point to Amir was that the illusion of stereo reproduction is best achieved by trying to emulate these specific auditory cues that we subconsciously know & recognize exist in the real event being portrayed. This is why Toole says we have an exceptional ability to "measure" accuracy - he's wrong, IMO (but I'm not sure this is exactly what he said), we have an exceptional sensitivity to things that are not giving us the correct auditory cues i.e don't fit our expectation of the real world. The problem is we are not consciously aware of exactly what is amiss & so can easily grow to accept these mistakes as long as they are not too great - we are buying into this illusion, anyway, so a small mistake is easy to assimilate. However, I believe that there is also a ranking of auditory cues in our auditory perception & knowing the cues & their ranking will go a long way towards producing a believable illusion. We aren't there yet!

Now, one way of trying to achieve this is to try to make the reproduction system as accurate as possible but this is an impossibility as we are immediately hampered by the recording technology & the use of 2 just channels. The strive towards this accuracy is misdirected, in my opinion & has mislead us to the battle of measurements Vs listening. The underlying incorrect assumptions being that we know enough about auditory perception & that all we need to now focus on is being as accurate as possible in our reproduction - forgetting that this accuracy is impossible for 2 channel audio & forgetting that it's an illusion that is being created, not a facsimile of the actual event.

Although measurements are interesting as they give us glimpses into the performance accuracy of our reproduction equipment. If they are shown to correlate with audible preference, should be looked on as a datapoint for how important that aspect of the sound field is to our auditory perception.

So, for instance, what the Harmon measurements say to me is that in the real world we analyse the direct sound to categorise what auditory auditory objects are in the soundfield & also analyse the characteristics of reflected sound to further reinforce the perception of these objects. If this reflected sound has certain anomolies in it, like unexpected frequency differences, we are less "convinced" about our analysis of the direct sound - it is less of a realistic illusion. But, I suspect that this is only one aspect that goes into the creation of a realistic illusion & I don't know it's importance ranking. I believe that there may well be other more important aspects - hence my above question.

I've a lot more to say but this is too long anyway & may be considered somewhat off-topic!

 

[esldude]

I've dipped in & out of this thread & haven't read all the posts so forgive me if I am asking something that has already been answered but are the Harmon measurements just one set of measurable aspects of a speakers performance? Are there other possible aspects of performance not represented in these measurements? I'm not talking about room interaction & stereo operation as these have been addressed (to some extent, anyway) but are there other performance aspects not represented in the graphs?

I also wanted to go back to something I said & Amir replied to (which relates to the above question) - I said that auditory perception is not concerned with accuracy but rather about getting a close enough approximation to reality in realtime - so we take shortcuts/assumptions/guesses to get there. How close that approximation has to be depends on what task is being addressed. For instance we hear a very loud bang & instantly react in survival mode. 99% of the time we are wrong in our evaluation of the threat but our evolution has taught us that false positives are more beneficial for survival than false negatives. In a more relaxed type of listening we are using more auditory cues to create & evaluate the auditory scene in front of us. Note, we are not using all the soundwaves that impinge on our ears & evaluating this overwhelming data stream but are using specific aspects that we have learned give us enough information in the shortest time for our task-specific needs. For instance we evaluate the listening environment quite quickly by it's reverberation/ambience etc. & seem to be able to use this evaluation to "hear through" most environments & probably adjust as we continue in our listening.

So my point to Amir was that the illusion of stereo reproduction is best achieved by trying to emulate these specific auditory cues that we subconsciously know & recognize exist in the real event being portrayed. This is why Toole says we have an exceptional ability to "measure" accuracy - he's wrong, IMO (but I'm not sure this is exactly what he said), we have an exceptional sensitivity to things that are not giving us the correct auditory cues i.e don't fit our expectation of the real world. The problem is we are not consciously aware of exactly what is amiss & so can easily grow to accept these mistakes as long as they are not too great - we are buying into this illusion, anyway, so a small mistake is easy to assimilate. However, I believe that there is also a ranking of auditory cues in our auditory perception & knowing the cues & their ranking will go a long way towards producing a believable illusion. We aren't there yet!

Now, one way of trying to achieve this is to try to make the reproduction system as accurate as possible but this is an impossibility as we are immediately hampered by the recording technology & the use of 2 just channels. The strive towards this accuracy is misdirected, in my opinion & has mislead us to the battle of measurements Vs listening. The underlying incorrect assumptions being that we know enough about auditory perception & that all we need to now focus on is being as accurate as possible in our reproduction - forgetting that this accuracy is impossible for 2 channel audio & forgetting that it's an illusion that is being created, not a facsimile of the actual event.

Although measurements are interesting as they give us glimpses into the performance accuracy of our reproduction equipment. If they are shown to correlate with audible preference, should be looked on as a datapoint for how important that aspect of the sound field is to our auditory perception.

So, for instance, what the Harmon measurements say to me is that in the real world we analyse the direct sound to categorise what auditory auditory objects are in the soundfield & also analyse the characteristics of reflected sound to further reinforce the perception of these objects. If this reflected sound has certain anomolies in it, like unexpected frequency differences, we are less "convinced" about our analysis of the direct sound - it is less of a realistic illusion. But, I suspect that this is only one aspect that goes into the creation of a realistic illusion & I don't know it's importance ranking. I believe that there may well be other more important aspects - hence my above question.

I've a lot more to say but this is too long anyway & may be considered somewhat off-topic!

One thing related to some of your comments is how we can adapt to inaccurate sound until we ignore it or even come to think it correct over time. So a question is if someone listening for years on inaccurate speakers they have grown used to would have the same preference as most other people? I believe many would answer no. Some research that might lead to one to think the answer instead is yes involved height and directional perception. Researchers tested people's perception of height in sound. Then inserted molds into the out ear to alter the shape which altered height perception. At first upon retesting height perception was erratic and inaccurate. After many hours of wearing the ear inserts the brain apparently learned how to process the new sound cues, and height perception improved. In more hours it eventually equalled the original perceptual accuracy. Things sounded 'normal' and perceptions were accurate despite the flawed input.

So when people buy and cater to speakers, how much of their performance is really fixed and how much is the owner's brain adapting?

Getting back to my original question about whether owners of inaccurate speakers will have the same preference in a Harman type comparison. Researchers removed the molds and retested. Unlike when molds were first inserted, everyone's perception was immeddiately correct and accurate. There was no period of hours learning it back the old way. That leads me to think even owners of inaccurate speakers would recognize and prefer results similar to everyone else. I know it isn't a direct fit, but it support the idea somewhat.

The test not performed is putting the molded inserts back into the ear a few days later if the testees would also immediately have accurate perception with that or whether another dozen hours would be needed to acclimate again. My guess, is the brain developed a new pattern along with the natural one, and would recognize the old inserts which would allow it to have accurate perception immediately without relearning. But it is only a guess.

 

[jkeny]

One thing related to some of your comments is how we can adapt to inaccurate sound until we ignore it or even come to think it correct over time. So a question is if someone listening for years on inaccurate speakers they have grown used to would have the same preference as most other people? I believe many would answer no. Some research that might lead to one to think the answer instead is yes involved height and directional perception. Researchers tested people's perception of height in sound. Then inserted molds into the out ear to alter the shape which altered height perception. At first upon retesting height perception was erratic and inaccurate. After many hours of wearing the ear inserts the brain apparently learned how to process the new sound cues, and height perception improved. In more hours it eventually equalled the original perceptual accuracy. Things sounded 'normal' and perceptions were accurate despite the flawed input.

So when people buy and cater to speakers, how much of their performance is really fixed and how much is the owner's brain adapting?

Getting back to my original question about whether owners of inaccurate speakers will have the same preference in a Harman type comparison. Researchers removed the molds and retested. Unlike when molds were first inserted, everyone's perception was immeddiately correct and accurate. There was no period of hours learning it back the old way. That leads me to think even owners of inaccurate speakers would recognize and prefer results similar to everyone else. I know it isn't a direct fit, but it support the idea somewhat.

The test not performed is putting the molded inserts back into the ear a few days later if the testees would also immediately have accurate perception with that or whether another dozen hours would be needed to acclimate again. My guess, is the brain developed a new pattern along with the natural one, and would recognize the old inserts which would allow it to have accurate perception immediately without relearning. But it is only a guess.

I know you copied this post into the thread I just started & I answered there but I think it's probably worth my repeating my answer here, just for completeness.

Yes, a good point. Often we don't realise something is wrong until it is changed or the fault removed. But again, I have to go back to the fact that it is an illusion we are buying into so we suspend a lot of what we expect from a real world audio event & settle for something less real. This is where the individual differences arise - just how much do we each settle for & in what way. There is no doubt, for instance, that loudspeakers are the biggest offenders when it come to distortion but yet we are able to hear through this & recognise much smaller distortion further upstream, quiet easily.

But one thing that strikes me - would we not have seen the effects of this pre-conditioning with "bad speakers" in the Harmon test - would they not have been more varied as a result? For instance, I believe students (I'm making assumptions about their listening transducers which is probably fairly correct considering the number of Beats headphones sold) showed the same preference as everybody else, no?

 

[esldude]

I know you copied this post into the thread I just started & I answered there but I think it's probably worth my repeating my answer here, just for completeness.

Yes, a good point. Often we don't realise something is wrong until it is changed or the fault removed. But again, I have to go back to the fact that it is an illusion we are buying into so we suspend a lot of what we expect from a real world audio event & settle for something less real. This is where the individual differences arise - just how much do we each settle for & in what way. There is no doubt, for instance, that loudspeakers are the biggest offenders when it come to distortion but yet we are able to hear through this & recognise much smaller distortion further upstream, quiet easily.

But one thing that strikes me - would we not have seen the effects of this pre-conditioning with "bad speakers" in the Harmon test - would they not have been more varied as a result? For instance, I believe students (I'm making assumptions about their listening transducers which is probably fairly correct considering the number of Beats headphones sold) showed the same preference as everybody else, no?

Yes, I think that is correct. That the results are pretty firm in nearly everyone having the same preferences so I think the 'filters' our brain construct for whatever speakers/headphones each person uses were quickly discarded when sources were hidden, and in an unfamiliar room.

 

[jkeny]

Yes, I think that is correct. That the results are pretty firm in nearly everyone having the same preferences so I think the 'filters' our brain construct for whatever speakers/headphones each person uses were quickly discarded when sources were hidden, and in an unfamiliar room.

Right, so we have all developed the same internal auditory map. Which is why I branched off & started another thread which is relevant to this whole area of "Objectivity, Harmon testing, Reviewers & Reality" & asked a question that puzzles me.

 

[DaveC]

On the K2 vs M2, they have different dispersion patterns so it's not going to be easy to get a totally fair comparison under most indoor conditions. I'd guess the M2 is around 120 degrees vs 90 degrees for the K2s. At 120 degrees you're going to get more issues with early reflections, but what I've found is sometimes people prefer these reflections even though they do significant damage to the detail and soundstage of the recording, people simply don't know any better. Given a typical semi-live room experienced listeners tend to recognize and appreciate the speaker with a tighter pattern more often, but so do others given some time for their ears to adjust and pointing out the differences so they can hear and appreciate them.

From what I've read elsewhere Toole and Olive have found statistically significant correlations for wide high frequency dispersion to listener preference, which explains the M2's waveguide/horn, but this is one area where I can appreciate and understand the results but I disagree and personally, I greatly prefer a speaker with a more controlled dispersion pattern. With a wide dispersion speaker I find high levels of first reflection absorption to be critical, and a more damp room is required just in general. This is an issue with folks who have systems in the living rooms as they are less likely to want to put up lots of room treatment, so despite the preference tests I think a controlled dispersion loudspeaker to be more likely to perform well in the average home installation, therefore the owners will be more likely to stay happy.

Also, if the M2 has a single compression driver playing from 900-20kHz and uses DSP to get a flat response I think that will have some issues associated with it. Not that I think DSP can't be transparent, just that there would need to be a lot of correction if the driver they use is similar to every other one I've seen data on. i prefer the K2 going with the beryllium diaphragm tweeter crossed in at 15kHz, this is a much better option than trying to get a larger CD to play flat to 20 kHz.. imo of course.

 

[esldude]

On the K2 vs M2, they have different dispersion patterns so it's not going to be easy to get a totally fair comparison under most indoor conditions. I'd guess the M2 is around 120 degrees vs 90 degrees for the K2s. At 120 degrees you're going to get more issues with early reflections, but what I've found is sometimes people prefer these reflections even though they do significant damage to the detail and soundstage of the recording, people simply don't know any better. Given a typical semi-live room experienced listeners tend to recognize and appreciate the speaker with a tighter pattern more often, but so do others given some time for their ears to adjust and pointing out the differences so they can hear and appreciate them.

From what I've read elsewhere Toole and Olive have found statistically significant correlations for wide high frequency dispersion to listener preference, which explains the M2's waveguide/horn, but this is one area where I can appreciate and understand the results but I disagree and personally, I greatly prefer a speaker with a more controlled dispersion pattern. With a wide dispersion speaker I find high levels of first reflection absorption to be critical, and a more damp room is required just in general. This is an issue with folks who have systems in the living rooms as they are less likely to want to put up lots of room treatment, so despite the preference tests I think a controlled dispersion loudspeaker to be more likely to perform well in the average home installation, therefore the owners will be more likely to stay happy.

Also, if the M2 has a single compression driver playing from 900-20kHz and uses DSP to get a flat response I think that will have some issues associated with it. Not that I think DSP can't be transparent, just that there would need to be a lot of correction if the driver they use is similar to every other one I've seen data on. i prefer the K2 going with the beryllium diaphragm tweeter crossed in at 15kHz, this is a much better option than trying to get a larger CD to play flat to 20 kHz.. imo of course.

I seem to recall Sean Olive saying they tested this and it was one of their surprises. Removing too much first reflection was generally perceived negatively. That was the results of some of their testing. That was part of what lead them to learn your off axis response needs to be free of ripples and track the forward response. Removing first reflections was a negative and having off axis response anomalies with some first reflection stuck out like a sore thumb.

 

[DaveC]

I seem to recall Sean Olive saying they tested this and it was one of their surprises. Removing too much first reflection was generally perceived negatively. That was the results of some of their testing. That was part of what lead them to learn your off axis response needs to be free of ripples and track the forward response. Removing first reflections was a negative and having off axis response anomalies with some first reflection stuck out like a sore thumb.

Yeah... it makes me want to consider having two different horns, one for the folks who like their 1st reflections and one for those who would prefer to hear more direct sound. The issue is most people are not familiar with the massive improvements getting rid of first reflections makes and are familiar with hearing a conventional cone and dome speaker in a live room. It takes a while for the brain to adjust to the sound of a directive speaker or even a conventional speaker with a lot of 1st reflection absorption, which has a similar sonic effect. But, the issue is that's what it takes to realize the full potential of stereo in terms of soundstage and resolution, which go hand-in-hand at that point.

 

[NorthStar]

http://www.herronaudio.com/images/Measurements.pdf

Thank you Myles. That is a good read and nice format to let each industry person speak their point of view on the relevance of measurements. Out of all of those though, I only counted two people who commented on loudspeaker measurements. For now, let me discuss one of them whose work I know,Siegfried Linkwitz, from his AES paper, Room Reflections Misunderstood? It is a great easy to read paper that I hope membership seeks out and reads. Let's see what he says in your article first:

"Siegfried Linkwitz, VP, R&D,
Audio Artistry

I would have to say that measurement of
the speaker's frequency response is the
technical measurement that comes closest
to predicting the sound of a loudspeaker. If
you see a graph of a 20 to 20,000 kHz ±2 dB
on-axis, anechoic response you can be fairly
sure that the speaker has a smooth
sound characteristic.

However, this information
alone does not necessarily imply
that the sound is neutral. transparent and
dynamic, and that it would dis play those
characteristics in your room.
First of all, it takes more than the anechoic
on-axis response to describe t he
sound you hear. When you listen in a room,
your ears not only receive the on-axis sound
but also, with a slight delay, the sound that
has bounced around in your room and been
modified by it. Thus, the anechoic response
must also be measured at angles up to at
least ±60 degrees horizontally and ± 15
degrees vertically off-axis, in order to assess
the contribution of sound radiation in these
directions to the overall sound, when the
speaker is placed in a room.

Simply averaging the different anechoic responses over a
"listening window" or normalizing them to
the on-axis response is not as relevant a
description as is plotting the actual frequency
response curves at different angles on the
same graph.

The other half of the equation, which is
also related to a speaker's frequency
response. is its phase response.
[...]
However. I am not
convinced that the typical amounts of
phase shift due to crossovers that are
observed in these higher frequency regions
have audible effects. On the other hand, I
am convinced that phase shift is very
important at the low frequency end of the
spectrum."

You can read the rest of his write up in the article but it doesn't change the summary of it exposed above. That is, he is 100% in agreement with Dr. Toole's research that it is both on-axis and off-axis frequency response measurements that are great predictors of loudspeaker sound. As I mentioned, the research/Harman's measurements are for the entire hemisphere around the loudspeaker, and predictions are made by weighted combination of each class of reflections. Sound bouncing from behind the loudspeaker for example, has less impact on its sound than on-axis. Or first side reflection. I have also left the part about phase in there which above bass frequencies, it is not an audible concern (agrees with research from Vanderkooy).

For confirmation of how well he is aligned with Dr. Toole's research, let's look at what he says in the AES paper I referenced earlier:

"Furthermore,
it had been found that loudspeakers with wide
horizontal dispersion were preferred in listening tests
[4]. What is known scientifically about loudspeakers
and rooms for sound reproduction has been reviewed
by F. E. Toole [11]."

Notice how he completely differs to work of Dr. Toole with nary of an objection. The reference is: "[11] Floyd E. Toole, “Loudspeakers and Rooms for Sound Reproduction – A Scientific Review”, J. Audio Eng. Soc., vol. 54, pp. 451-476 (2006 June)"

Here is another point of agreement:

"For example, the
author and others had observed that wide horizontal
dispersion in the high frequency range of a
loudspeaker is subjectively desirable and increases
fidelity [4]."

And the reference: "[4] F. E. Toole, “Loudspeaker Measurements and Their Relationship to Listener Preferences”, J. Audio Eng. Soc., vol. 34, part 1, pp. 227-235 (1986 April); part 2, pp. 323-348 (1986 May)"

As I said, while this research may seem new to members here, it is accepted as foundation of acoustic and sound reproduction science by many researchers and designers. So it is no wonder that a reference that you pick Myles as a counterpoint to mine, happens to be completely supportive of the research I have been sharing. Yes, Linkwitz says he listens to loudspeakers too to find problems . Nothing different has been said. We still listen. But measurements of loudspeakers when made according to a system that takes into account how it acts in a room and our perception, is a strong predictor of our preferences.

I'd like to also read Part 2 of that link just above.

* This entire thread is a good read. ...Great discussion, exchanges, info, supports, ...all for the best...of audio reproduction, from the music we love listening to.