It’s All a Preference

Well said. Agree.

You are starting point is good, but I don't agree where you end . I have sat through blind listening test with a group of people, evaluating speakers. The ones that measure properly did finish on top. The ones that did not, lost out. Now when I say measure, I mean the right measurements. Not just a random one. This is the type of measurement for speakers that correlates very highly with listening results:



From Dr. Toole's Room Acoustics book:

Dr. Toole said:

No matter how meticulously the playback equipment has been chosen and set up, and no matter how much money has been lavished on exotic acoustical treatments, what we hear in our homes and cars is, in spatial terms, a matter of chance.

How can we measure something that subjectively we react to as art? Measurements are supposed to be precise, reproducible, and meaningful. Perceptions are inherently subjective, evanescent, subject to various nonauditory influences within and surrounding the human organism. However, perceiving flaws in sound reproducing systems appears to be an activity that we are able substantially to separate from our critique of the art itself. We can detect flaws in the reproduction of music of which we have no prior knowledge and in which we fi nd no pleasure.

....

As consumers of these programs, we cannot know what was intended for the sound of any of these programs. We were not there when they were created. We may have been at performances by similar, or even the same, musicians, but they were likely to have been in different venues and possibly amplified. None of us ever placed our ears where the microphones were located to capture the sounds, nor would we want to; we were almost certainly at a distance, in an audience. A simple reproduction of the microphone signals cannot duplicate the experience.

Descriptors like pleasantness and preference must therefore be considered as ranking in importance with accuracy and fidelity. This may seem like a dangerous path to take, risking the corruption of all that is revered in the purity of an original live performance. Fortunately, it turns out that when given the opportunity to judge without bias, human listeners are excellent detectors of artifacts and distortions; they are remarkably trustworthy guardians of what is good. Having only a vague concept of what might be correct, listeners recognize what is wrong. An absence of problems becomes a measure of excellence. By the end of this book, we will see that technical excellence turns out to be a high correlate of both perceived accuracy and emotional gratification, and most of us can recognize it when we hear it.

Click to expand...

To that I reply: No Audiophile I know will declare the Bose System (whatever their thing with the little speakers) is great sounding ... So there is something that guides those choices.

mep said:

We like to talk about neutrality vice preferences like there is really a black and white choice to be made. Although we can say with a high degree of certainty that there are measurable differences between electronics and speakers of any type with some measuring *better* in ways that have been deemed to be better, in reality, none of them are neutral. And I mean neutral in the fact that no electronics sound exactly like live music and the circuit has zero colorations. We just aren’t there yet people.

You like SS? That’s a preference. You like tubes? That’s a preference. You prefer single-ended amps over push-pull amps? That’s a preference. You think electrostats sound better than box speakers? That’s a preference. You like digital and not analog? That’s a preference. I could go on and on, but I think you get the point. All of our systems are built upon our preferences.

So, given that no electronics or speakers are perfect, the sound and the gear we all buy are based on our preferences whether we care to admit it or not. While absolute fidelity may be our goal, we have no absolute fidelity at this point in time. I for one think there is much more work to be done in all links of the recording and playback chain before we can declare victory and say we have arrived and crossed the threshold of absolute fidelity.

Click to expand...

There is no absolute fidelity, that's true enough. But we have relative fidelity. We have the input from the recording and the output from the reproduction system. The closer they are to identical, the less the reproduction system adds to or subtracts from the recording, the higher its fidelity to the source material, the more neutral the system. At least that's what it was about before high fidelity became high end.

You have your Krells and your PLs. One sounds darker, the other sounds brighter. Neither is absolutely neutral, but one is probably closer than the other (it would be a remarkable coincidence if they were equally but differently wrong). You may prefer the one that is further from neutral. You may not like the sound of most of your recordings and prefer a less accurate reproduction of them, but that changes nothing but your choice. It doesn't change the fact that there is an ideal neutrality - perfect fidelity to the recording - and there is a relative neutrality that is right here, right now, being pursued by good audio engineers and vigorously denied by hundreds of audiophiles.

Tim

Phelonious Ponk said:

There is no absolute fidelity, that's true enough. But we have relative fidelity. We have the input from the recording and the output from the reproduction system. The closer they are to identical, the less the reproduction system adds to or subtracts from the recording, the higher its fidelity to the source material, the more neutral the system. At least that's what it was about before high fidelity became high end.

You have your Krells and your PLs. One sounds darker, the other sounds brighter. Neither is absolutely neutral, but one is probably closer than the other (it would be a remarkable coincidence if they were equally but differently wrong). You may prefer the one that is further from neutral. You may not like the sound of most of your recordings and prefer a less accurate reproduction of them, but that changes nothing but your choice. It doesn't change the fact that there is an ideal neutrality - perfect fidelity to the recording - and there is a relative neutrality that is right here, right now, being pursued by good audio engineers and vigorously denied by hundreds of audiophiles.

Tim

Click to expand...

How do you know what ideal neutrality sounds like Tim?

tomelex said:

We have come a long way with that word (preferences) haven't we! Thank you Ethan, even your most vocal detractors have come around to some things over time.

As far as more work in the playback chain, I think stereo, with image expansion, and other stuff, could sound more exciting, but it is playing around within the limitations, and yes, things like putting a board between your head and to the center point between the two speakers helps some songs sound "different"r, or sometimes firing the speakers at the rear walls or into the corners, all kinds of crazy stuff, has had some interesting "effects" on some songs, and maybe there is some boredom with stereo after all these years, and I always reming folks, listen in mono for a couple days, then flip that switch to stereo, and wow, what a weird sound at first, until you adapt to it again.

Tom

Click to expand...

There's a difference between preferences and fidelity. For one, fidelity comes first.

tomelex said:

Good point Amir,

Tom

Click to expand...

amirm said:

You are starting point is good, but I don't agree where you end . I have sat through blind listening test with a group of people, evaluating speakers. The ones that measure properly did finish on top. The ones that did not, lost out. Now when I say measure, I mean the right measurements. Not just a random one. This is the type of measurement for speakers that correlates very highly with listening results:

From Dr. Toole's Room Acoustics book:

Click to expand...

Should be pointed out that the study was carried out with four speakers available for comparison, where a natural hierarchy revealed itself as that with the least distortion.
Most people pick a preferred speaker without having the opportunity to do side-by-side comparisons. It's quite a job to quickly move another speaker into the correct position for such a comparison to be valid, times two for the pair, times four for the models. Sean Olive has developed a system for doing that at the press of a button in his test rooms, but I've only seen a serious attempt at doing the same with rotating platforms in one chain of stores (B&O).

We listen to our preferred distortion, and only become aware of qualitative nuances through direct and valid comparison.

Soundproof said:

Should be pointed out that the study was carried out with four speakers available for comparison, where a natural hierarchy revealed itself as that with the least distortion.

Click to expand...

They have done a lot more of these tests than the four. Indeed, Sean Olive wrote an AES paper on how to correlated the 70-point anechoic chamber measurements to preferences in blind tests: A Multiple Regression Model for Predicting Loudspeaker Preference Using Objective Measurements: Part II - Development of the Model: http://www.aes.org/e-lib/browse.cfm?elib=12847

"A new model is presented that accurately predicts listener preference ratings of loudspeakers based on anechoic
measurements. The model was tested using 70 different loudspeakers evaluated in 19 different listening tests. Its
performance was compared to 2 models based on in-room measurements with 1/3-octave and 1/20-octave
resolution, and 2 models based on sound power measurements, including the Consumers Union (CU) model, tested
in Part One. The correlations between predicted and measured preference ratings were: 1.0 (our model), 0.91 (in*
room, 1/20th-octave), 0.87 (sound power model), 0.75 (in-room, 1/3-octave), and ?0.22 (CU model). Models based
on sound power are less accurate because they ignore the qualities of the perceptually important direct and earlyreflected sounds.
The premise of the CU model is that the sound power response of the loudspeaker should be flat,
which we show is negatively correlated with preference rating. It is also based on 1/3-octave measurements that are
shown to produce less accurate predictions of sound quality. "

Most people pick a preferred speaker without having the opportunity to do side-by-side comparisons. It's quite a job to quickly move another speaker into the correct position for such a comparison to be valid, times two for the pair, times four for the models. Sean Olive has developed a system for doing that at the press of a button in his test rooms, but I've only seen a serious attempt at doing the same with rotating platforms in one chain of stores (B&O).

We listen to our preferred distortion, and only become aware of qualitative nuances through direct and valid comparison.

Click to expand...

The point of Harman research is that we can, with high confidence predict the in-room listener preference if we have the right measurement data. It this case, it calls for a speaker that has smooth off-axis response. Such angled radiations from the speaker hit the room surfaces and combine with the direct sound. If their quality is poor, then they modify the direct response of the speaker and result in lower listener preference.

KeithR said:

How do you know what ideal neutrality sounds like Tim?

Click to expand...

Subjectivists don't, Keith. If you believe in measurements you can define neutrality. If you don't, it will always be completely relative, including the language we use to describe it. Whose to say an old Altec lansing isn't more "natural" than a Revel Salon if there is no accepted metric to measure by? A nod is as good as a wink to a blind horse. I'm lucky. Up to a point (I don't like flat FR headphones), the stuff that measures well is usually what I prefer, so I get to be both an objectivist and a subjectivist. More tools in the box.

Tim

amirm said:

They have done a lot more of these tests than the four. Indeed, Sean Olive wrote an AES paper on how to correlated the 70-point anechoic chamber measurements to preferences in blind tests: A Multiple Regression Model for Predicting Loudspeaker Preference Using Objective Measurements: Part II - Development of the Model: http://www.aes.org/e-lib/browse.cfm?elib=12847

"A new model is presented that accurately predicts listener preference ratings of loudspeakers based on anechoic
measurements. The model was tested using 70 different loudspeakers evaluated in 19 different listening tests. Its
performance was compared to 2 models based on in-room measurements with 1/3-octave and 1/20-octave
resolution, and 2 models based on sound power measurements, including the Consumers Union (CU) model, tested
in Part One. The correlations between predicted and measured preference ratings were: 1.0 (our model), 0.91 (in*
room, 1/20th-octave), 0.87 (sound power model), 0.75 (in-room, 1/3-octave), and ?0.22 (CU model). Models based
on sound power are less accurate because they ignore the qualities of the perceptually important direct and earlyreflected sounds.
The premise of the CU model is that the sound power response of the loudspeaker should be flat,
which we show is negatively correlated with preference rating. It is also based on 1/3-octave measurements that are
shown to produce less accurate predictions of sound quality. "


The point of Harman research is that we can, with high confidence predict the in-room listener preference if we have the right measurement data. It this case, it calls for a speaker that has smooth off-axis response. Such angled radiations from the speaker hit the room surfaces and combine with the direct sound. If their quality is poor, then they modify the direct response of the speaker and result in lower listener preference.

Click to expand...

Amir, the Harman testing procedure doesn't allow for valid evaluation of dipole speakers.

Not to be argumentative but their testing is meant to sell their speakers and prove the validity of their design. It doesn't mean that other approaches aren't valid.

And color me skeptical about that "acoustically" tranparent curtain.

FrantzM said:

To that I reply: No Audiophile I know will declare the Bose System (whatever their thing with the little speakers) is great sounding ... So there is something that guides those choices.

Click to expand...

Not exactly true

http://www.tonepublications.com/review/we-review-the-bose-901/

MylesBAstor said:

Amir, the Harman testing procedure doesn't allow for valid evaluation of dipole speakers.

Not to be argumentative but their testing is meant to sell their speakers and prove the validity of their design. It doesn't mean that other approaches aren't valid.

And color me skeptical about that "acoustically" tranparent curtain.

Click to expand...

The testing is not meant to "prove the validity of their design," it is meant to test listener response to a variety of designs, including their own, and inform future designs. And I'm sure Harmon measured the effect of the curtain, or used material they understood the acoustic properties of.

Tim

MylesBAstor said:

Amir, the Harman testing procedure doesn't allow for valid evaluation of dipole speakers.

Not to be argumentative but their testing is meant to sell their speakers and prove the validity of their design. It doesn't mean that other approaches aren't valid.

And color me skeptical about that "acoustically" tranparent curtain.

Click to expand...

98.85% of speakers are not dipoles ... As for thecurtain lack of acoustical transparency , that would affect all the speakers equally wouldn't it? Same material after all ...

FrantzM said:

98.85% of speakers are not dipoles ... As for thecurtain lack of acoustical transparency , that would affect all the speakers equally wouldn't it? Same material after all ...

Click to expand...

Possibly but it could also be masking subtler information and details too that some speakers may have and others don't eg reducing them all to the same level like communisim.

mep said:

Amir-While I understand the point you are trying to make, I don’t agree with it. Do you think that Dr. Toole assumes that all people would rate each piece of gear the same from best to worst and everybody would plunk their money down and buy the same gear that rated best?

Click to expand...

All? No. And there is nothing about what people "buy." People buy for many reasons beside audio quality. And that is part of the problem they have tried to solve. By putting the speakers behind curtains, they eliminate anything but the notion of the comparative sound quality as judged by listeners. I have sat through that test twice with a group of others. In every case, majority votes for the speakers that have the best measurements and the same as what you see in the chart (even though the speakers do change). Clearly there is a common thread there as far as what we prefer as good sound. Think about it. That is what you do when you go to buy a speaker. You listen and decide on fidelity. The only difference here is that they don't let you see the brand, price, and look of the product. And they show you alternatives in the exact same spot. These are things that you cannot do when you go t a stereo store.

[ I don’t think so. I think Dr. Toole is talking about people being able to hear gross aberrations vice hearing variations upon a theme of goodness which always boils down to preference. Dr.Toole even stated in the paragraph you quoted that preference must be considered as ranking in importance.

Click to expand...

They use the word preference. That causes people to automatically decide that they are measuring taste. It is not so. If it were taste, then the results would not converge as highly as it does. Importantly, the measurements highly correlate with listening tests. What we think should be good, flat frequency response (combined direct and indirect) is what wins most of the time. We know we all hear frequency response variations. So the fact that this becomes a requirement for good sound, makes all the sense in the world.

When I sat through the test, and I listened, at first you are taken back that the task is hard in there is no true reference. You do have flavor A against flavor B. But the speaker that is designed by science, sounds more natural. It sounds more real. So you vote for that.

Speaker "B" is the B&W 702N. It has a dip in its overall sound in the 2K to 3K range. This is because there is too big of gap between its woofer size and tweeter. What makes it a marketing hit with that distinct look, does have a price. As the frequencies go up, the woofer gets too directional. Yet the frequency is too low for the tweeter to pick up. Eventually the tweeter does pick up the load and then all of a sudden the speaker becomes less directional so its response off axis goes up/gets better. But the damage is done in the most critical region of our hearing, the 2K to 3K range. So vocals, when compared to a speaker that doesn't have that dip, don't sound as good.

To me, when we have listening tests with science and measurements, then we have a very compelling case as our baseline. Put another way, you have to start believing something about the science. And that is a great starting position. There may be better religion, but it has the responsibility to prove that by changing the rules, it can still succeed. This is what I love about Dr. Toole's work. He creates and entire end-to-end story that hangs together. It may not be perfectly right. But it is a hell of a lot better position to start with than randomly assigning goodness to speakers.

MylesBAstor said:

Amir, the Harman testing procedure doesn't allow for valid evaluation of dipole speakers.

Click to expand...

You mean that it doesn't eliminate the back reflections? How many people do that? Is that written their marketing brochures that they only sound good that way? BTW, the graph is anechoic chamber measurements (unit "M"). There is no "back wall" there. The direct sound of the speaker is not flat. It is full of resonances. If I took a speaker with flat response, and made those variations to it, you would not prefer it better. When I listened to that speaker in the blind tests, it sounded all wrong. It had this flat, phasey response that simply was unnatural compared to cone speakers. I thought there was something broken in it. Then the curtains opened and I see it is the Martin Logan. My jaw dropped. I had this notion of these speakers having wonderful sound. A the time, we also carried a planar speaker. All of a sudden, I realized I hear the same sound out of our speakers and comparisons I had done. I grant you, if you took that speaker and treated the back/side walls heavily, and suppressed that poor side response, it would likely improve its sound. But people better be prepared to do that and know that at the end of the day, it is still not an accurate speaker.

Not to be argumentative but their testing is meant to sell their speakers and prove the validity of their design. It doesn't mean that other approaches aren't valid.

Click to expand...

Much of what you see here came from Dr. Toole and Olive's work while at the Canadian National Research Council. NRC was funded by the Canadian government to encourage investment in this area from the industry. It was there that they showed that we can correlate what is preferred by listeners as good sound with proper set of measurements. This work did lead Harman to hire some of the key people there like Dr. Toole and Olive. They replicated the research labs at Harman which then led to major change in their design philosophy for speakers. Unlike some other companies, they did not keep the information to themselves. They published their findings for all to learn and follow. I know few companies that invest hundreds of millions of dollars in audio and then let the world have the fruits of their labor. So while they are associated with Harman, their work here cannot be put down due to bias to sell speakers. Anyone can build speakers to match that kind of directivity and with it, have the blessing of their research with respect to it being good.

And color me skeptical about that "acoustically" tranparent curtain.

Click to expand...

Every experiment has some impact on what is being tested. A perfect test is not created. Surely the same applies in medicine. The curtain is pretty far away from the speakers so reflections back and forth to the speaker front is heavily minimized. And as noted, all speakers see the same thing. BTW, you can listen to the speakers afterward without the curtain. I have and it does not change the rankings.

At the end of the day, we either walk blind and go by our highly flawed and subjective evaluation of speakers in different rooms, different locations and different times, or we start with some sound foundation. That is what is being offered here. That speakers that lose in such tests happen to also have flawed measurements, should be an eye opener.

The temptation is strong to dismiss the tests. I used to be there. It took two sessions of sitting in those rooms and days of listening to the explanation and even much more researching the literature that has gotten me 90% of the way there.

You are right that their results, despite significance, are heavily dismissed by people as evidenced by the fact that Harman is not the #1 manufacturer of consumer speakers. People routinely and all the time buy flawed speakers by these standards. Few like me get to sit in that room and get all new insight as to what is good, and what is not -- both subjectively and objectively. That is a hugely missed opportunity for us as a community.