What role should listening play in speaker design

[amirm]

Note from staff: This thread has been split from another thread and best effort was made to preserve the flow.

When we are talking whether a speaker has flat on axis response and well behaved off-axis, Andrew Jones (like other designers) leaves room for his own judgement over using measurements. He also throws dart at the accuracy of such measurements. Here is one of countless interviews with the same theme:

Are speakers really better when professional designers tweak them by listening?

 

[TimVG]

I don't know what that means. Extensive measurements? We are talking whether a speaker has flat on axis response and well behaved off-axis. In every interview, he leaves room for his own judgement over getting such measurements. He also throws dart at the accuracy of such measurements. Here is one of countless interviews with the same theme:

I'm unsure of what AJ is getting at in this interview. Is he arguing the concept of the preference of certain frequency response curve(s) targets, or is he arguing the capability of making accurate measurements? Both?

I believe it's safe to assume since we've seen cross measurements of NFS systems, manufacturers, hobbyists.. That there is enough correlation between all of them if the respective methods are used correctly. Sure, there can be some deviation, but nothing that will make or break the DUT.

I think the harsh (for some) truth is that is perfectly possible to design and build a very good loudspeaker without having listened to as much as 1 second of music. I'm not arguing listening and doing some tone shaping is not desirable, but technically it's not an absolute requirements if you're aware of your design target priorities.

Much, if not all (comsol), of the acoustic (and more) properties can be predicted beforehand. Even for advanced hobbyists there is Akabak/ABEC to predict much of the directivity behaviour before building anything. Then it is a matter of selecting the correct drive units that operate well within their intended passband, etc etc..

Or perhaps I'm simplifying things.

 

[Sokel]

I'm unsure of what AJ is getting at in this interview. Is he arguing the concept of the preference of certain frequency response curve(s) targets, or is he arguing the capability of making accurate measurements? Both?

I believe it's safe to assume since we've seen cross measurements of NFS systems, manufacturers, hobbyists.. That there is enough correlation between all of them if the respective methods are used correctly. Sure, there can be some deviation, but nothing that will make or break the DUT.

I think the harsh (for some) truth is that is perfectly possible to design and build a very good loudspeaker without having listened to as much as 1 second of music. I'm not arguing listening and doing some tone shaping is not desirable, but technically it's not an absolute requirements if you're aware of your design target priorities.

Much, if not all (comsol), of the acoustic (and more) properties can be predicted beforehand. Even for advanced hobbyists there is Akabak/ABEC to predict much of the directivity behaviour before building anything. Then it is a matter of selecting the correct drive units that operate well within their intended passband, etc etc..

Or perhaps I'm simplifying things.

It would be naive to think that Jones is not using measurements.
I would short of swallow it if he was designing speakers alone, but he's not.

He's designing from drivers up, some are famous about measuring really good, be it were he was working for TAD or a decade for Kef, etc.

You can't design a driver from ground up by intuition alone.
What he probably communicates at his interviews is what most people want to hear, and most people are short of allergic to heavy technical insights. He's probably going with the flow and his target demographic.

 

[TimVG]

It would be naive to think that Jones is not using measurements.
I would short of swallow it if he was designing speakers alone, but he's not.

He's designing from drivers up, some are famous about measuring really good, be it were he was working for TAD or a decade for Kef, etc.

You can't design a driver from ground up by intuition alone.
What he probably communicates at his interviews is what most people want to hear, and most people are short of allergic to heavy technical insights. He's probably going with the flow and his target demographic.

I don't think anyone is arguing the points you raise?

But from this interview it seems like he is arguing the validity of 1) the measurements themselves (as in they are never accurate enough), and 2) the correlation of the prior to listener preference.

 

[Mart68]

What he probably communicates at his interviews is what most people want to hear, and most people are short of allergic to heavy technical insights. He's probably going with the flow and his target demographic.

Exactly that. People don't like the idea that the equipment was designed using machines, they like to think the designer of it has got beyond that to some hidden truth. The 'Cult of The Designer' has been built up carefully by the industry over the years. Got to keep that narrative going.

 

[Sokel]

I don't think anyone is arguing the points you raise?

But from this interview it seems like he is arguing the validity of 1) the measurements themselves (as in they are never accurate enough), and 2) the correlation of the prior to listener preference.

It is odd, yes.
On the one hand says that measurements don't paint the whole picture and on the other does this:



...which pretty much correlates with the general preference and does well overall.

From my point of view, does pretty well at hiding the engineer (as he should) and presenting the marketing side, it's rare that one can do both.
We, that we like a deeper insight may not like it but that's the way the market goes, specially when prices are not on the budget side.

One can sell a story way easier than a research.

 

[TimVG]

It is odd, yes.
On the one hand says that measurements don't paint the whole picture and on the other does this:

View attachment 520179

...which pretty much correlates with the general preference and does well overall.

From my point of view, does pretty well at hiding the engineer (as he should) and presenting the marketing side, it's rare that one can do both.
We, that we like a deeper insight may not like it but that's the way the market goes, specially when prices are not on the budget side.

One can sell a story way easier than a research.

That is also what I believe. I've zero doubt Andrew Jones has forgotten more about speaker design than I've learned up until now.

 

[Arindal]

He also throws dart at the accuracy of such measurements. Here is one of countless interviews with the same theme

Thanks for sharing. We agree on this point, and I don´t quite understand his arguments against Klippel and truly anechoic measurements either.

We are talking whether a speaker has flat on axis response and well behaved off-axis. In every interview, he leaves room for his own judgement over getting such measurements.

That was my conclusion as well, having listened to just a small fraction of this particular interview (when target curves and perfect measurements are discussed).

He is basically saying, that any specific target curve can never be met with zero tolerance. Particularly not if you take both on- and off-axis behavior into account, so a loudspeaker designer is inevitably forced to make compromises here and there, and he personally seems to prefer the particular compromise, giving the maximum of subjective rating in listening tests with a large variety of recordings.

I would subscribe to the latter point of view, but surely not everyone does.

On the one hand says that measurements don't paint the whole picture and on the other does this:

...

...which pretty much correlates with the general preference and does well overall.

Excellent example. I guess we all agree on the fact that such loudspeaker behavior is correlating with the ideal of flat anechoic response, if viewed over broad frequency bands, and it surely looks like it was designed involving and following measurements. The exception is the behavior around 3.9K which I would assume is typical for a large dome tweeter in a conventional coaxial arrangement with some amount of narrow-banded cancellation dip. Cannot speak for AJ, but this looks exactly like what he is describing as a compromise which there are reasons for not to fix.

I think the harsh (for some) truth is that is perfectly possible to design and build a very good loudspeaker without having listened to as much as 1 second of music. I'm not arguing listening and doing some tone shaping is not desirable,

Wouldn´t such loudspeaker, ´designed solely by machines to perfection´, having never seen a listening room or developer´s ear, be inevitably a huge market success, as it could sound perfect and be cheaper at the same time, compared to all other products which have undergone a complicated process of iteration?

I wonder who is producing such speaker and where can I listen to it. Maybe you have an idea? Maybe there is a loudspeaker designer out there who confesses of having never listened to own speakers prior to launching the product.

If I understood AJ correctly, he does not use listening tests after the main technical development process achieving sufficiently close to perfect measurements, to do ´tone shaping´. But rather to identify flaws and imperfections which are result of the aforementioned compromises, but lead to audible flaws when listening to a number of music recordings. To me, this sounds reasonable, and has nothing to do with applying any sort of ´tone shaping´, or personal preference curve.

 

[TimVG]

I wonder who is producing such speaker and where can I listen to it. Maybe you have an idea? Maybe there is a loudspeaker designer out there who confesses of having never listened to own speakers prior to launching the product.

I'll let you know. Here's a 5" +1" sealed monitor to be used with a subwoofer as a living room TV setup. It has a hybrid crossover where the passive components handle the crossover, and DSP is used to shape the signal (like the JBL 7-series). Measurements were made, quick active crossover tested to verify the measurements under a couple of vertical angles. Now I'm assembling the crossover and when finished will make another couple of measurements to verify the results. I don't think much, if any, tweaking will happen - although it would be easy to do so.

 

[Marin]

It would be naive to think that Jones is not using measurements.
I would short of swallow it if he was designing speakers alone, but he's not.

He's designing from drivers up, some are famous about measuring really good, be it were he was working for TAD or a decade for Kef, etc.

You can't design a driver from ground up by intuition alone.
What he probably communicates at his interviews is what most people want to hear, and most people are short of allergic to heavy technical insights. He's probably going with the flow and his target demographic.

Yes I think you're right.
Andrew Jones knows very good how speaker market works based on how long he has been in business.
But first of all I think he means that he wouldn't kill the speaker from a sensitivity standpoint with 30+ part's of crossovers just so he,we can "pixel pip" at flat line..

 

[Marin]

He is basically saying, that any specific target curve can never be met with zero tolerance. Particularly not if you take both on- and off-axis behavior into account, so a loudspeaker designer is inevitably forced to make compromises here and there, and he personally seems to prefer the particular compromise, giving the maximum of subjective rating in listening tests with a large variety of recordings.

You hit the nail on the head here.
But above all I think that he don't approve that every curve should be dissected,or pixel piping.....every speaker designer nightmare, especially coaxial designs

 

[Marin]

It is odd, yes.
On the one hand says that measurements don't paint the whole picture and on the other does this:

View attachment 520179

...which pretty much correlates with the general preference and does well overall.

From my point of view, does pretty well at hiding the engineer (as he should) and presenting the marketing side, it's rare that one can do both.
We, that we like a deeper insight may not like it but that's the way the market goes, specially when prices are not on the budget side.

One can sell a story way easier than a research.

I would go so far and say that this model is especially for reviewer who will make a lot of pretty pictures of frequency response and directivity plot,considering how much this unit is stuffed with sheep wool,or something like that.

 

[Arindal]

Now I'm assembling the crossover and when finished will make another couple of measurements to verify the results. I don't think much, if any, tweaking will happen - although it would be easy to do so.

Judging from the directivity - and I do not mean to predict particular sound quality or character solely based on measurements - I see most of very common issues one could imagine in a compact DIY speaker, which will make compromises and voicing by ear mandatory. Following AJ´s philosophy, this is actually the moment when the machines and interns are done, and the loudspeaker developer´s work begins.

 

[TimVG]

Judging from the directivity - and I do not mean to predict particular sound quality or character solely based on measurements - I see most of very common issues one could imagine in a compact DIY speaker, which will make compromises and voicing by ear mandatory. Following AJ´s philosophy, this is actually the moment when the machines and interns are done, and the loudspeaker developer´s work begins.

It depends on how you look at it. You can take different approaches: wider for lower baffle step, better CTC optimised for far field listening, etc. However this one will be rather inconspicuous, good in nearfield with a decent amount of vertical leeway, just a little soundpower dip due to vertical reflection average and short CTC. But yes, it is a standard thirteen in a dozen bookshelf (with some care taken to ensure good directivity within its limits). But that's the nice thing about DIY - you can build whatever you like.

 

[Marin]

I'll let you know. Here's a 5" +1" sealed monitor to be used with a subwoofer as a living room TV setup. It has a hybrid crossover where the passive components handle the crossover, and DSP is used to shape the signal (like the JBL 7-series). Measurements were made, quick active crossover tested to verify the measurements under a couple of vertical angles. Now I'm assembling the crossover and when finished will make another couple of measurements to verify the results. I don't think much, if any, tweaking will happen - although it would be easy to do so.

View attachment 520201

View attachment 520202

Why are you killing the speaker sensitivity?
There in the upper bass?

 

[TimVG]

Why are you killing the speaker sensitivity?
There in the upper bass?

Better for far field application taking in consideration overall directivity of this speaker. Look at the shadow slope in SP curve.

If I were to use it nearfield I would bring up that area maybe 1 dB

Full disclosure this is a repurposed cabinet from a different project so choices have to be made.

 

[PristineSound]

I don't know what that means. Extensive measurements? We are talking whether a speaker has flat on axis response and well behaved off-axis. In every interview, he leaves room for his own judgement over getting such measurements. He also throws dart at the accuracy of such measurements. Here is one of countless interviews with the same theme:

It almost sounds like Andrew Jones is asking is the "truth" really the truth with measurements.

There is a dabble into the circle of confusion, saying he doesn't really know what the source is, etc.

Overall, I am not sure what he is really trying to say. . .and frankly I am not sure if I would count on a pair of speakers tuned by solely the hearing of an elderly gentleman. I just hope he measures and have other younger audience to help him do the listening tests.

 

[a4eaudio]

It almost sounds like Andrew Jones is asking is the "truth" really the truth with measurements.

Is he "asking" that? I think he trying to inform the audience that the "measurements" are NOT the truth. An anechoic chamber is not the truth. A Klippel NFS is not the truth. Plotting a bunch of colorful lines that are not the truth, and adding circles and arrows to the lines, isn't all that informative. An more informative/honest representation would be to plot just error bands of the CTA-2034 curves. Kind of like polling that A is approved by 46% with +/- 3% error. The 46% is pretty meaningless, but it is meaningful that A is approved by 43 to 49% with a 99% probability. But just like a reader of a poll wants to see that 46% point estimate, the reader of a speaker review wants to see that single solid line that they can wrap their head around and pretend their is some certainty about it.

 

[PristineSound]

Is he "asking" that? I think he trying to inform the audience that the "measurements" are NOT the truth. An anechoic chamber is not the truth. A Klippel NFS is not the truth. Plotting a bunch of colorful lines that are not the truth, and adding circles and arrows to the lines, isn't all that informative. An more informative/honest representation would be to plot just error bands of the CTA-2034 curves. Kind of like polling that A is approved by 46% with +/- 3% error. The 46% is pretty meaningless, but it is meaningful that A is approved by 43 to 49% with a 99% probability. But just like a reader of a poll wants to see that 46% point estimate, the reader of a speaker review wants to see that single solid line that they can wrap their head around and pretend their is some certainty about it.

I think the word "truth" needs to be defined here.

An industrial precision measurement tool is more accurate than human ears can ever be, period, full stop. Also, @amirm can you confirm is the NFS is not accurate down to 20hz? I thought you may have mentioned this somewhere, I may be wrong, anyway, one of Andrew's big argument is that bass is hard to measure accurately.

If Andrew is talking about preference, that is a different story. But there is enough studies that shows most people prefer a linear frequency respond along with a constant directivity. Sure there are some who may prefer something different, but you design for the majority, the minority can EQ to what they like.

 

[amirm]

Not sure if I understand that correctly, but I rather read in his statements that his point is that while measurements can be accurate (or "true" to a microphone), that does not necessarily translate to "truth" in the sense of sound quality, because the latter is too complex and involving too many compromises compared to a single microphone at a single point.

So what is the answer? That we believe his ears and preferences???

Fact is that we need a standard. We are lucky that the preferred on-axis response is flat. There is zero excuse in producing a speaker with serious deviations from that. Certainly not when it is one designer's opinion.