• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Why don't all speaker manufacturers design for flat on-axis and smooth off-axis?

svart-hvitt

Major Contributor
Joined
Aug 31, 2017
Messages
2,375
Likes
1,149
....mmm...you keep making this point. No-one is idolising, Toole et al are just being used as the go to reference because it is very accessible and there is very little , if any, evidence to contradict the research that has been performed. If you think the research is wrong say so and point out why with your supporting evidence. Otherwise you have no point. We are not all blindly following the individuals. The science speaks for itself.
I am not saying the Harman/Samsung research is wrong. But I wanted to flag my impression that many followers of that research camp have become museum watchers, more interested in preserving a status quo on audio research than asking new (stupid?) questions. I am not alone in thinking that progress may still be done in speaker research. Take a look at this excerpt from a 2017 JAES article:

«Loudspeaker specifications have traditionally described the physical properties and characteristics of loudspeakers: frequency response, dimensions and volume of the cabinet, diameter of drivers, impedance, total harmonic distortion, sensitivity, etc. Few of these directly describe the sound reproduction and none directly describe perception of the reproduction, i.e., takes into account that the human auditory system is highly non-linear in terms of spectral-, temporal-, and sound level processing (see, e.g., [3]). This disconnect between specifications and perception have made it challenging for acousticians and engineers (and consumers) to predict how a loudspeaker will sound on the basis of these specifications».
Source: http://www.aes.org/e-lib/browse.cfm?elib=18729

In other words, despite «Toole et al.» it is «...challenging for acousticians and engineers (and consumers) to predict how a loudspeaker will sound on the basis of these specifications».

Do you understand my point? The research that has become gospel among many at ASR still makes it difficult to predict how a speaker will sound. The Revel Salón vs M2 anecdote that suprised so many may be one indication that we still have difficulty predicting speaker sound?
 

Xulonn

Addicted to Fun and Learning
Forum Donor
Joined
Jun 27, 2018
Messages
688
Likes
1,797
Location
Boquete, Chiriqui, Panama
Do you understand my point? The research that has become gospel among many at ASR still makes it difficult to predict how a speaker will sound.
Huh? How does the acceptance of current SOTA robust, repeatable findings by psycho-acoustics researchers regarding listener preferences make research in sonic signature prediction difficult? It seems to me to be a different issue.

Also, it seems that you are confusing "gospel" with "current consensus." I don't think any of the many informed and audio-literate people here think that there is no need for further audio/loudspeaker research. That being said, I see the current consensus on loudspeaker/room acoustics/listener preferences as robust - and it has proven to be repeatable. I would guess that continuing new research looking at "sonic prediction" will add to the body of knowledge rather than invalidate previous research.

Also, I cannot envision an engineering/design > testing > marketing > sales > consumer environment where science and engineering can predict how all speakers will react to a practically infinite number of variations in people's listening environments. However, having standards and uniform testing standards and methods like the Canadian speaker manufacturers have had for decades can lead to a fair amount of uniformity in designing loudspeakers that sound good in multiple types of listening environments.

As a hobbyist fine woodworker, I lust for a pair of the highly rated (designed by Harman using Toole's findings?) Revel 206's and their absolutely beautiful gentle curves and walnut finish, but alas, even used, they cost more than I can afford. I may be able to afford a pair of less expensive , but supposedly not quite as "accurate or neutral" loudspeakers, such as the Goldenear Triton 7's But can I live with the "black sock on a fence post" design.

In addition, designing exotic loudspeakers (like the extreme deviation from a "box" in the photo below), and not focusing heavily on "accurate and neutral" sound reproduction will always exist. They may ignore room-compatible "predictive design," but if some people "prefer" them, that is just fine by me.

Alien Loudspeakers.jpg
 
Last edited:

svart-hvitt

Major Contributor
Joined
Aug 31, 2017
Messages
2,375
Likes
1,149
Huh? How does the acceptance of current SOTA robust, repeatable findings by psycho-acoustics researchers regarding listener preferences make research in sonic signature prediction difficult? It seems to me to be a different issue.

Also, it seems that you are confusing "gospel" with "current consensus." I don't think any of the many informed and audio-literate people here think that there is no need for further audio/loudspeaker research. That being said, I see the current consensus on loudspeaker/room acoustics/listener preferences as robust - and it has proven to be repeatable. I would guess that continuing new research looking at "sonic prediction" will add to the body of knowledge rather than invalidate previous research.

Also, I cannot envision an engineering/design > testing > marketing > sales > consumer environment where science and engineering can predict how all speakers will react to a practically infinite number of variations in people's listening environments. However, having standards and uniform testing standards and methods like the Canadian speaker manufacturers have had for decades can lead to a fair amount of uniformity in designing loudspeakers that sound good in multiple types of listening environments.

As a hobbyist fine woodworker, I lust for a pair of the highly rated (designed by Harman using Toole's findings?) Revel 206's and their absolutely beautiful gentle curves and walnut finish, but alas, even used, they cost more than I can afford. I may be able to afford a pair of less expensive , but supposedly not quite as "accurate or neutral" loudspeakers, such as the Goldenear Triton 7's But can I live with the "black sock on a fence post" design.

In additin, designing exotic loudspeakers (like the extreme deviation from a "box" in the photo below), and not focusing heavily on "accurate and neutral" sound reproduction will always exist. They may ignore room-compatible "predictive design," but if some people "prefer" them, that is just fine by me.

View attachment 29705
When I pointed out the deficiencies of «current consensus», I referred to a JAES article by authors affiliated with SenseLab, University of Aalborg and Bang & Olufsen.

Do you think those authors raise the wrong questions?
 

Xulonn

Addicted to Fun and Learning
Forum Donor
Joined
Jun 27, 2018
Messages
688
Likes
1,797
Location
Boquete, Chiriqui, Panama
When I pointed out the deficiencies of «current consensus», I referred to a JAES article by authors affiliated with SenseLab, University of Aalborg and Bang & Olufsen.

Do you think those authors raise the wrong questions?
Yet another non -sequitur.

That was not my point - I acknowledged that there is a "current consensus" on the understanding of listener preferences in rigorous, scientific blind testing, and then I said that further research is warranted.

The research that has become gospel among many at ASR still makes it difficult to predict how a speaker will sound.
Your snide remark about ASR regulars and "gospel" is a bit weird - and that was what prompted me to respond.

Perhaps there is a language barrier issue, but I still d not see how discussions at ASR have a significant negative effect on the field audio/acoustic research and the direction that it will take.

Then again, you may be trying to say that people do not actually prefer in blind testing what they say prefer in blind testing, which is even more strange.
 

svart-hvitt

Major Contributor
Joined
Aug 31, 2017
Messages
2,375
Likes
1,149
Yet another non -sequitur.

That was not my point - I acknowledged that there is a "current consensus" on the understanding of listener preferences in rigorous, scientific blind testing, and then I said that further research is warranted.



Your snide remark about ASR regulars and "gospel" is a bit weird - and that was what prompted me to respond.

Perhaps there is a language barrier issue, but I still d not see how discussions at ASR have a significant negative effect on the field audio/acoustic research and the direction that it will take.

Then again, you may be trying to say that people do not actually prefer in blind testing what they say prefer in blind testing, which is even more strange.
I am not the first on ASR to use the word «gospel». In fact, I picked up the term after reading another user’s comment. What I had in mind is the observation that «a new scientific truth does not triumph by convincing its opponents and making them see the light, but rather because its opponents eventually die, and a new generation grows up that is familiar with it».

Language barrier or not, maybe more of a cultural bias where respect for authority is a factor?

On blind tests: See @MattHooper ’s earlier remark on the sweeter coca cola. A blind test has lots of potential biases too that may not be obvious at first sight. This is not to be interpeted as if I am against blind tests; on the contrary. I am just pointing out the danger of blind spots in empirical studies.
 

Cosmik

Major Contributor
Joined
Apr 24, 2016
Messages
3,075
Likes
1,944
Location
UK
@amirm , you keep saying I am irrational.
There's an irony of course. Rationalism is a very specific philosophy:
Rationalism is the philosophy that knowledge comes from logic and a certain kind of intuition—when we immediately know something to be true without deduction, such as “I am conscious.” Rationalists hold that the best way to arrive at certain knowledge is using the mind’s rational abilities. The opposite of rationalism is empiricism, or the view that knowledge comes from observing the outside world.
Check out the bolded bit. I'm always trying to argue things based on common sense or deduction or intuition, but everyone else needs data, experiments, listening tests before they can consider an argument. Therefore, I think I am on pretty safe ground to claim most of you are irrational :) - or at least not rationalists.
 

Floyd Toole

Active Member
Audio Luminary
Technical Expert
Industry Insider
Joined
Mar 12, 2018
Messages
192
Likes
1,147
@amirm , please don’t reduce this discussion to frequencies below transition area only, where everyone is in agreement on benefits of room EQ.

In his «enticing marketing story» paper, @Floyd Toole wrote:

«For decades it has been widely accepted that a steady-state amplitude response measured with an omnidirectional microphone at the listening location in a room is an important indicator of how an audio system will sound. Such measurements have come to be known as generic “room curves,” or more specific “house curves.” That belief has a long history in professional audio, and now it has penetrated consumer audio with stand-alone products and receivers in- corporating automated measurement and equalization capa- bilities. The implication is that by making in-situ measurements and manipulating the input signal so that the room curve matches a predetermined target shape, imperfections in (unspecified) loudspeakers and (unspecified) rooms are measured and repaired. It is an enticing marketing story».
Source: https://www.audiosciencereview.com/...-theory-without-measurement.7127/#post-162524

I read this as a critique of software packages like Audiolense and Acourate, not an endorsement of said and related solutions. Maybe @UliBru can correct me if he read Toole as an endorsement?

Olive et al. published a paper on room correction a decade ago:

http://www.aes.org/e-lib/browse.cfm?elib=15154

That paper opens up the door for potential benefits of room correction. As far as I understand, the paper does not write off the potential benefits of room EQ as a low frequency issue only. So there is a potential inconsistency here between @Floyd Toole and Olive et al. because Toole writes off target curves while Olive et al. found benefits when applying target curves.

Olive et al. wanted at that time (2009) to investigate further:

«The authors are currently investigating these research questions and hope to present the fruits of this labor in upcoming papers».

AFAIK, these results were never published. The Olive et al. paper from 2009 is AFAIK the only AES paper on room correction software.

@Floyd Toole uses strong words in his «enticing marketing story» paper. But I cannot see that his strong opinion is supported by evidence. Where is the evidence for writing off room EQ above transition area? I must admit I am a but surprised that a science journal published such strong statements without solid evidence. Without evidence, Toole’s paper is more of an opinion piece, not science.

So I stand by my observation that Olive et al. found possible evidence for beneficial room EQ above transition area, while Toole ridicules such corrections. For some reason Olive et al. didn’t publish their later research on room EQ, as promised in their 2009 article.
It is one thing to consider broadband, tone-control adjustments of in-room response, and quite another to allow narrow-band measurements and equalization of a mixture of minimum-phase and non-minimum-phase fluctuations. The latter is widely permitted, if not encouraged, in many current algorithms. Narrow-band measurements might reveal resonances - by far the dominant problem in loudspeakers - but if you look at Section 4.6.3 in my book, you will see that this is likely only with low-Q - i.e. broadband - fluctuations. Also to be seen are fluctuations caused by non-minimum-phase acoustical interference, not correctable by equalization. It would have been nice to continue that research but other objectives intervened, and I retired. Still, it is blatantly evident that resonances are easily found in comprehensive anechoic data, but it is also blatantly evident that such data are very hard to find. Once the 3D loudspeaker sound has been radiated into a normally reflective listening space detailed analysis of loudspeaker performance is greatly degraded. The gratifying consolation is that listeners for 40 years have routinely recognized the most neutral loudspeakers in double-blind tests conducted in normal rooms, and these loudspeakers are also recognizable in comprehensive anechoic data - no in-room measurements or EQ are required except at low frequencies where small room resonances and adjacent boundary issues demand attention.

It is also evident, to those who have read the papers and book, that proper interpretation of room curves is possible only if one has comprehensive anechoic data on the loudspeakers - permitting the recognition of potential frequency-dependent directivity problems, which cannot be corrected by equalization. See Figure 5.4, Section 5.6 and others throughout the book. With such data, room curves and room EQ can be useful, but the marketing of these algorithms places no limitations on their application. Mystery loudspeakers in mystery rooms are rendered "perfect" at the press of an icon - that is a good marketing story.

If a room curve were a definitive descriptor of loudspeaker timbre there should be a single target. But, there are many. I have read the instruction manuals for some of the popular algorithms and they basically recommend starting with a "default" target, and some offer alternatives. If the customer is not happy with the result, it is suggested that they open a friendly user interface permitting changes to the target curve. How is one to know what changes are needed unless one listens to music? So, the process from that point onwards is a subjectively guided tone-control exercise. Once set, the tone control adjustments are fixed, not available to tweak as necessary to cope with program variations, "circle of confusion" problems. In fact, the "circle" is now part of the room EQ - it is absolutely not a "calibration" at all. The so-called "Harman curve" is as described in Figure 12.4: the most likely steady-state room curve that one may measure from subjectively-highly-rated loudspeakers in double-blind tests - that's all. It is not the target curve for equalizing flawed loudspeakers. That cannot work reliably.

Incidentally, I see that you now refer to the research as Harman/Samsung. Actually, the Harman research group reports to Harman International, as do I, not Samsung. Samsung has its own research group, largely expatriates of Harman, and they direct themselves. It is friendly competition.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
23,615
Likes
41,309
Location
Seattle Area
Language barrier or not, maybe more of a cultural bias where respect for authority is a factor?
For sure. I detect that tone in your posts all the time where "local" work is prefered to what is developed elsewhere. If we were discussing Genelec research, I am sure you would be cheering us on.

I am fine with this motivation if it leads to something good. But it doesn't. It is argument after argument, protest after protest.

Be a friend of knowledge and information, not where it comes from.
 

svart-hvitt

Major Contributor
Joined
Aug 31, 2017
Messages
2,375
Likes
1,149
For sure. I detect that tone in your posts all the time where "local" work is prefered to what is developed elsewhere. If we were discussing Genelec research, I am sure you would be cheering us on.

I am fine with this motivation if it leads to something good. But it doesn't. It is argument after argument, protest after protest.

Be a friend of knowledge and information, not where it comes from.
I think you read me somewhat wrong. If it were a «local» work vs US work thing, how could I come to grip with this statement by Ilpo Martikainen, Genelec founder:

«We are most interested in how the speaker performs, i.e. sounds. This includes lots of listening tests as well. However, there is lot of research evidence of what makes the speaker sound good and how this correlates with measured performance, i.e. specs. I warmly recommend Floyd Toole’s book “Sound Reproduction: The Acoustics and Psychoacoustics of Loudspeakers and Rooms”, it is an excellent representation of this subject».
Source: https://www.community.genelec.com/forum/-/message_boards/message/914379#/

What has been the point of my last few posts is to ask if there’s more to speakers design than «frequency response, frequency response and frequency response».

And this is in the spirit of the thread isn’t it, which basically asks the question why not every speaker is designed in the same way. Is the heterogenity of speaker design just marketing or is it based on true factors of sound reproduction quality too?

In a previous thread I asked the question if the focus on frequency response has gone too far in some cases, where small speakers have been marketed as «small big speakers». If there is too much focus on a good thing - frequency response - could it be that we forget about things like headroom and distortion? Could it be that clever marketing guys misuse Toole and others to market speakers that score high on the most obvious speaker qualities, while being silent about other qualities? Shouldn’t an informed buyer of gear be taught that frequency response will only bring you thus far, but not all the way to the «top»?
 

Floyd Toole

Active Member
Audio Luminary
Technical Expert
Industry Insider
Joined
Mar 12, 2018
Messages
192
Likes
1,147
I think you read me somewhat wrong. If it were a «local» work vs US work thing, how could I come to grip with this statement by Ilpo Martikainen, Genelec founder:

«We are most interested in how the speaker performs, i.e. sounds. This includes lots of listening tests as well. However, there is lot of research evidence of what makes the speaker sound good and how this correlates with measured performance, i.e. specs. I warmly recommend Floyd Toole’s book “Sound Reproduction: The Acoustics and Psychoacoustics of Loudspeakers and Rooms”, it is an excellent representation of this subject».
Source: https://www.community.genelec.com/forum/-/message_boards/message/914379#/

What has been the point of my last few posts is to ask if there’s more to speakers design than «frequency response, frequency response and frequency response».

And this is in the spirit of the thread isn’t it, which basically asks the question why not every speaker is designed in the same way. Is the heterogenity of speaker design just marketing or is it based on true factors of sound reproduction quality too?

In a previous thread I asked the question if the focus on frequency response has gone too far in some cases, where small speakers have been marketed as «small big speakers». If there is too much focus on a good thing - frequency response - could it be that we forget about things like headroom and distortion? Could it be that clever marketing guys misuse Toole and others to market speakers that score high on the most obvious speaker qualities, while being silent about other qualities? Shouldn’t an informed buyer of gear be taught that frequency response will only bring you thus far, but not all the way to the «top»?
I have said, and I repeat: Frequency response is the most important parameter in any audio component. If it is wrong, nothing else matters. The meaning translates perfectly into simple input-to-output electronics - all (normally) minimum-phase, so the time domain is also intact. Obviously this is a good start. With loudspeakers we encounter the complications of a three-dimensional sound field; both radiated by the loudspeaker and in a much, much more complicated form, auditioned by listeners in rooms. So, "Toole" gets misused frequently when people think simplistically that a loudspeaker is describable with a single "frequency response" curve. It isn't, and the ways in which it isn't constitute a large part of my papers and book.

But, as you, and countless others in different forums, have commented: But, what about . . .? Non-linear distortion is a common one - sorry but we have only a few laboratory measurements of complex metrics that show signs of agreeing with what we hear. Conventional HD and IMD metrics are not reliable - only zero is confidence inspiring. Then come power handling/sensitivity/power output - i.e. how loud will it play before complaining or dying? At present these are not totally reliable specs, many are from the marketing departments, not the engineers. "Headroom" is a meaningless factor - if a device can play loud enough without audible degradation how much "headroom" does one need - and never use? The psyche is consoled no doubt, but that has nothing to do with science.

Bandwidth is a big factor. Low bass extension clearly correlates with elevated sound quality ratings. In loudspeakers there is a strong cost and size correlation with this factor, so one can have a timbrally very neutral small blue-tooth loudspeaker for listening at low levels and close distances that absolutely fails at the higher sound outputs required for distance listening. Is this a failed design? Of course not, it is just a well engineered, purpose built product, the sound quality merits of which are well described by frequency response - not a single curve, though :) Does it play loud enough for any person or situation - try it and see, is the only reliable test at the moment.
 

napilopez

Senior Member
Joined
Oct 17, 2018
Messages
321
Likes
750
Great to see @Floyd Toole offer a response.

Speaking purely anecdotally, I think most of us have gone through a situation where the automated Room EQ in, say, an A/V receiver seems to just make speakers sound worse, even if the resulting graph looked "prettier." I was puzzled by this until I read Toole's book.

I think that speaks to the idea that two ears and a brain can identify the sound of a loudspeaker 'through' the sound of a room, at least above the transition frequency.

One fun experiment would be to turn Room EQ on its head and see how it affects the (quasi) anechoic measurements of a speaker. It's been established that EQ could be useful if applied on anechoic measurements to address minimum-phase issues. So perhaps a good Room EQ system would be able to separate the sound of a speaker from the sound of the room and attempt to act on the speaker.

It would be interesting to see if some of the more renown Room EQ systems like Dirac/Trinnov/Lyngdorf actually made meaningful improvements to a speaker's anechoic performance. This Dirac paper, at least, acknowledges that some reflections (sidewall) can be good and others bad, suggesting its software takes some care on what it corrects for. Though it mainly seems to be about minimizing the impact of front and rear reflections.
 
Last edited:

MRC01

Addicted to Fun and Learning
Joined
Feb 5, 2019
Messages
748
Likes
648
Location
Pacific Northwest
I have said, and I repeat: Frequency response is the most important parameter in any audio component. If it is wrong, nothing else matters.
...
Non-linear distortion is a common one - sorry but we have only a few laboratory measurements of complex metrics that show signs of agreeing with what we hear. Conventional HD and IMD metrics are not reliable - only zero is confidence inspiring.
...
Then come power handling/sensitivity/power output
...
Bandwidth is a big factor. Low bass extension clearly correlates with elevated sound quality ratings. ...
Across these factors, even if everyone agrees FR is the most obvious and important, individual weighting and preferences must vary. For example, I notice distortion more than lack of the lowest bass octave (likely due to the kind of music I listen to). Someone else may find the opposite, or rank power output most important after FR. Or imaging, or some other factor.

Which leads to endless debate, even among people who agree that FR is the most important factor.
 

March Audio

Major Contributor
Manufacturer
Joined
Mar 1, 2016
Messages
4,604
Likes
5,059
Location
Albany Western Australia
Which would imply a lot of the room modes cannot be fixed either, since a great deal of them are not minimum phase behavior. A general statement that room modes can be corrected by room correction is simply wrong.
I repeat my my statement which you snipped.about generalising.
 

March Audio

Major Contributor
Manufacturer
Joined
Mar 1, 2016
Messages
4,604
Likes
5,059
Location
Albany Western Australia
I am not saying the Harman/Samsung research is wrong. But I wanted to flag my impression that many followers of that research camp have become museum watchers, more interested in preserving a status quo on audio research than asking new (stupid?) questions. I am not alone in thinking that progress may still be done in speaker research. Take a look at this excerpt from a 2017 JAES article:

«Loudspeaker specifications have traditionally described the physical properties and characteristics of loudspeakers: frequency response, dimensions and volume of the cabinet, diameter of drivers, impedance, total harmonic distortion, sensitivity, etc. Few of these directly describe the sound reproduction and none directly describe perception of the reproduction, i.e., takes into account that the human auditory system is highly non-linear in terms of spectral-, temporal-, and sound level processing (see, e.g., [3]). This disconnect between specifications and perception have made it challenging for acousticians and engineers (and consumers) to predict how a loudspeaker will sound on the basis of these specifications».
Source: http://www.aes.org/e-lib/browse.cfm?elib=18729

In other words, despite «Toole et al.» it is «...challenging for acousticians and engineers (and consumers) to predict how a loudspeaker will sound on the basis of these specifications».

Do you understand my point? The research that has become gospel among many at ASR still makes it difficult to predict how a speaker will sound. The Revel Salón vs M2 anecdote that suprised so many may be one indication that we still have difficulty predicting speaker sound?
Your impression is wrong and actually pretty patronising.

I don't know where you get this impression that people aren't still open to looking at new information on the subject. The problem is that there appears to be little new evidence to look at. You are conflating the two.

Show us the new research which contradicts or extends upon the Toole et al work.

Otherwise it's just a meaningless commentary "oohhh there might be something else / additional going on here". Well that is actually no surprise to any of us.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
23,615
Likes
41,309
Location
Seattle Area
What has been the point of my last few posts is to ask if there’s more to speakers design than «frequency response, frequency response and frequency response».
The research comprising listening tests basically says "NO." Here is Sean Olive's paper: A Multiple Regression Model For Predicting Loudspeaker Preference Using Objective Measurements: Part I-Listening Test Results

1563587639730.png

94% of the comments are about tonality of the speakers.

Moreso, if you read the paper, you will see that what listeners meant by "distortion" was often correctness of the frequency response, not distortion itself.

1563587781121.png

Sean calls this the "halo" effect. Good sounding speaker is associated with a speaker that has less distortion.

Now, blast a speaker at very high volumes and distortion obviously matters. It is just that we have much bigger fish to fry at nominal listening levels.
 

DDF

Senior Member
Joined
Dec 31, 2018
Messages
366
Likes
402
What has been the point of my last few posts is to ask if there’s more to speakers design than «frequency response, frequency response and frequency response».
Perhaps ironic to your line of inquisition, it's actually "perceived frequency response" that is still the largest unknown in audio. This isn't a design problem, but instead one of human cognition. While some rudimentary studies are available, we can't predict with a high degree of accuracy how our perceived timbral balance will be affected by the number, delay, relative phase and direction of delayed auditory images (room reflections).

So we have very useful rules of thumb. For example, Harman (Dr Toole or Dr Olive) theorize that reflections similar in spectral make up to the direct sound will enhance auditory resolution and thresholds (giving a "second peak" at the signal, as they say).

But so many design choices simply boil down to "frequency response". Where to place drivers and ports? Well, what frequency regions do you want the boundary and driver-to-driver cancellations to occur, and from which directions (floor, ceiling, walls) & how do you want your on axis response to compare to the DI? These are just different recipes for changing the room contribution. Monopole, bipole, dipole, higher order (cardiod etc) or some mix? Same thing. How far to space drivers? Ditto.

But of course there's more to speaker design than this. Pushing a driver too hard reveals itself in distortion testing and in listening during the design phase. The objections from a tweeter crossed too low can be measured and heard, requiring crossover re-adjustment. Been there, done that. And as Dr Toole repeats, resonances are highly audible (though they show up as frequency response issues).

So to perceive that the common view is that only "frequency response" matters is a bit of a red herring. It's not true. What is true is that it doesn't matter enough: these questions have been only meagerly researched over the last 100 years.
 

svart-hvitt

Major Contributor
Joined
Aug 31, 2017
Messages
2,375
Likes
1,149
The research comprising listening tests basically says "NO." Here is Sean Olive's paper: A Multiple Regression Model For Predicting Loudspeaker Preference Using Objective Measurements: Part I-Listening Test Results

View attachment 29710
94% of the comments are about tonality of the speakers.

Moreso, if you read the paper, you will see that what listeners meant by "distortion" was often correctness of the frequency response, not distortion itself.

View attachment 29711
Sean calls this the "halo" effect. Good sounding speaker is associated with a speaker that has less distortion.

Now, blast a speaker at very high volumes and distortion obviously matters. It is just that we have much bigger fish to fry at nominal listening levels.
You illustrate my point nicely. Previously, I quoted a JAES paper from 2017:

«Loudspeaker specifications have traditionally described the physical properties and characteristics of loudspeakers: frequency response, dimensions and volume of the cabinet, diameter of drivers, impedance, total harmonic distortion, sensitivity, etc. Few of these directly describe the sound reproduction and none directly describe perception of the reproduction, i.e., takes into account that the human auditory system is highly non-linear in terms of spectral-, temporal-, and sound level processing (see, e.g., [3]). This disconnect between specifications and perception have made it challenging for acousticians and engineers (and consumers) to predict how a loudspeaker will sound on the basis of these specifications».
Source: http://www.aes.org/e-lib/browse.cfm?elib=18729

You show no interest in such research. In the process, you make the research of Harman Kardon gospel, i.e. an authority only fools would question.

Why is it that you find the quoted JAES article of no relevance?
 

svart-hvitt

Major Contributor
Joined
Aug 31, 2017
Messages
2,375
Likes
1,149
Your impression is wrong and actually pretty patronising.

I don't know where you get this impression that people aren't still open to looking at new information on the subject. The problem is that there appears to be little new evidence to look at. You are conflating the two.

Show us the new research which contradicts or extends upon the Toole et al work.

Otherwise it's just a meaningless commentary "oohhh there might be something else / additional going on here". Well that is actually no surprise to any of us.
One example:
«Loudspeaker specifications have traditionally described the physical properties and characteristics of loudspeakers: frequency response, dimensions and volume of the cabinet, diameter of drivers, impedance, total harmonic distortion, sensitivity, etc. Few of these directly describe the sound reproduction and none directly describe perception of the reproduction, i.e., takes into account that the human auditory system is highly non-linear in terms of spectral-, temporal-, and sound level processing (see, e.g., [3]). This disconnect between specifications and perception have made it challenging for acousticians and engineers (and consumers) to predict how a loudspeaker will sound on the basis of these specifications».
Source: http://www.aes.org/e-lib/browse.cfm?elib=18729
 

svart-hvitt

Major Contributor
Joined
Aug 31, 2017
Messages
2,375
Likes
1,149
I have said, and I repeat: Frequency response is the most important parameter in any audio component. If it is wrong, nothing else matters. The meaning translates perfectly into simple input-to-output electronics - all (normally) minimum-phase, so the time domain is also intact. Obviously this is a good start. With loudspeakers we encounter the complications of a three-dimensional sound field; both radiated by the loudspeaker and in a much, much more complicated form, auditioned by listeners in rooms. So, "Toole" gets misused frequently when people think simplistically that a loudspeaker is describable with a single "frequency response" curve. It isn't, and the ways in which it isn't constitute a large part of my papers and book.

But, as you, and countless others in different forums, have commented: But, what about . . .? Non-linear distortion is a common one - sorry but we have only a few laboratory measurements of complex metrics that show signs of agreeing with what we hear. Conventional HD and IMD metrics are not reliable - only zero is confidence inspiring. Then come power handling/sensitivity/power output - i.e. how loud will it play before complaining or dying? At present these are not totally reliable specs, many are from the marketing departments, not the engineers. "Headroom" is a meaningless factor - if a device can play loud enough without audible degradation how much "headroom" does one need - and never use? The psyche is consoled no doubt, but that has nothing to do with science.

Bandwidth is a big factor. Low bass extension clearly correlates with elevated sound quality ratings. In loudspeakers there is a strong cost and size correlation with this factor, so one can have a timbrally very neutral small blue-tooth loudspeaker for listening at low levels and close distances that absolutely fails at the higher sound outputs required for distance listening. Is this a failed design? Of course not, it is just a well engineered, purpose built product, the sound quality merits of which are well described by frequency response - not a single curve, though :) Does it play loud enough for any person or situation - try it and see, is the only reliable test at the moment.
Thanks for answering :)

Well, you give the obvious answer yourself:

FREQUENCY RESPONSE IS NOT THE SINGLE MOST IMPORTANT FACTOR.

Size is. Headroom is a derivative of size. And size and headroom are cost driving.

For what I know my MacBook Pro speaker is probably flat and smooth. But I would prefer a legacy and not so flat and smooth JBL horn speaker with a 15 inch woofer when listening to music. Do you see my point?

I am not making an absurd example here. We have in recent years experienced speaker producers claim ridiculous bass extension and talk of «small big speakers» that are flat and smooth but will fail miserably against some of the big legacy design speakers.

Again: This is not critique of your research per se. But your and others’ research has been misused (because marketing people and even some people at ASR say it’s only FR that matters) to sell speakers that fail for obvious reasons (they’re too small to deliver what they promised).

Further to illustrate my point: Take a look at Harman Kardon’s top speakers, the Salón and the M2. Both are BIG, and the Salón costs a lot. If FR were the only game in town, why spend so much on a Salón speaker if a Devialet Phantom has similar FR specifications?
 

March Audio

Major Contributor
Manufacturer
Joined
Mar 1, 2016
Messages
4,604
Likes
5,059
Location
Albany Western Australia
One example:
«Loudspeaker specifications have traditionally described the physical properties and characteristics of loudspeakers: frequency response, dimensions and volume of the cabinet, diameter of drivers, impedance, total harmonic distortion, sensitivity, etc. Few of these directly describe the sound reproduction and none directly describe perception of the reproduction, i.e., takes into account that the human auditory system is highly non-linear in terms of spectral-, temporal-, and sound level processing (see, e.g., [3]). This disconnect between specifications and perception have made it challenging for acousticians and engineers (and consumers) to predict how a loudspeaker will sound on the basis of these specifications».
Source: http://www.aes.org/e-lib/browse.cfm?elib=18729
Example of what? No-one is under the impression that we have all aspects of acoustic perception buttoned down, which is the thrust of your point.
 
Top Bottom