What role should listening play in speaker design

[Vuki]

There are many opinions on designing loudspeakers here and just a few expiriences.
All those advocating design without listening are in opinion camp, and all those with expirince are in measurement+listening camp.

 

[RexrothPigeon]

All those advocating design without listening are in opinion camp

I don't think anybody's "advocating design without listening" here. Can you find a post advocating this? This is the approach I find most reasonable:

What role should listening play in speaker design

Nobody said it is 100% random. You can't be twisting the argument so maliciously by mistake, or simply because you fell for Matt's debating tactics. In fact it is 100% non-random, in that one's cognitive biases will always work in the same directions at any one point in time (they will often...

www.audiosciencereview.com

@amirm wrote:

No one is saying don't listen. There are plenty of reasons to listen to a speaker. As I mentioned, even Harman uses blind listening tests with all the knowledge they have of the science.

What we are saying is that you better not use ad-hoc, sighted listening tests to override what you see in measurements. If measurements show flaws, use proper engineering to fix them.

All this aside, as someone who has compared measurements to sound of speakers across hundreds of them, I would say the reverse to you. If you are still an old school designer, optimizing the sound of a speaker with your ears per above, you will face a headwind when it comes to selling to us bunch. You can listen all you want. But don't screw up the measurements.

 

[Vuki]

I don't think anybody's "advocating design without listening" here. Can you find a post advocating this? This is the approach I find most reasonable:

What role should listening play in speaker design

Nobody said it is 100% random. You can't be twisting the argument so maliciously by mistake, or simply because you fell for Matt's debating tactics. In fact it is 100% non-random, in that one's cognitive biases will always work in the same directions at any one point in time (they will often...

www.audiosciencereview.com

@amirm wrote:

Most (maybe all) speaker designers use listening tests to evaluate their choices/compromises/solutions. They are not relying just on measurements and they almost never use blind listening test panel.

 

[Sokel]

There's also a great misunderstanding about measurements and higher price speakers.

While it was maybe true some years earlier that higher price/higher SPL speakers dictated a no-measurement area, all it took is Magico to publish some pics with a Klippel and their big tower.

That's an instant shift to the minds of blind believers, and even new-found faith zealots.
Mainstream comes always from the top, that's how it works.

 

[RexrothPigeon]

Most (maybe all) speaker designers use listening tests to evaluate their choices/compromises/solutions. They are not relying just on measurements and they almost never use blind listening test panel.

Yes, and if they allow their sighted listening "tests" to override the measurements, it is problematic because (A) their sighted listening is subject to the same biases as all sighted listening and (B) they are almost all past the age at which hearing (and also hearing judgment, as JJ and Dr. Toole have shown) is in decline.

 

[Sigberg Audio]

Somewhat interesting that many here seem to trust their own ears and preferences when it comes to dialing in the sound, but don't trust a professional loudspeaker designer to do the same.

Also it seems that some finds dialing in the sound to be a personal preference thing (that they don't want the loudspeaker designer to meddle with), and thus dismiss the idea that most of us have relatively similar preferences.

There seem to be some paradoxes hidden here, no?

 

[Audionaut]

Theoretically, what would happen if a speaker was designed without being listened to at all?

If you use high-quality, suitable drivers, you’ll end up with a very good speaker. There are even simulation programs available now that work very well, provided the TS parameters and other measurements for each individual driver are correct. This means you can use the drivers to see on your computer how a system will perform in a specific enclosure with a specific crossover setting, and the results will be 90% accurate.

 

[Audionaut]

What would happen if someone lined themselves up perfectly with a dart board, leveled there arm to the perfect angle, and threw a dart, without looking at all?

That analogy doesn't really hold up anymore. It's more like, instead of a person throwing the dart, you have a fairly sophisticated robot that you have to program with the correct data.
The dart's trajectory, shape, and weight; air temperature; air movement; the distance to and height of the dartboard; and so on.
If all the data is correct, it will work.

 

[Audionaut]

My best friend happens to love to listen to Chopin and has a syste tailored to that. Sounds great with Chopin, sounds awful for other stuff IMO

I can't even begin to imagine something like that, because it doesn't match my experience at all.
In my experience, a very good speaker sounds great with any kind of well-recorded music—and conversely, a poor-quality speaker completely fails with any kind of music.

 

[goat76]

He's at least 70 years old, so no, he doesn't have a good sense of how anything sounds

With all the experience Andrew Jones has with listening and pinpointing potential problems in a loudspeaker, I’m actually pretty sure you would walk away highly impressed by his ability to do that, if you had the chance to spend a few days with him.

Yes, his hearing is most likely not as good as it once was, but to say he “doesn't have a good sense of how anything sounds” is pretty disrespectful to all the knowledge and experience he has.
As his work is highly dependent on his listening ability, I’m sure he regularly checks his hearing so that he knows his shortcomings.

 

[Sigberg Audio]

We have people with pitch perfect voices, and we have people who have ears and a brain that can hear a song once, and then play that same song perfectly on the piano. It's not a stretch to think that some people can have a relatively good ear with regards to hearing if something sounds off, or if something sounds right when reproducing music through a loudspeaker.

Again not suggesting we shouldn't use measurements, but that a person who designs speakers for a living actually can trust his ears to quite some extent is not very far fetched. And yes we are aware of bias effects, but it does not come into play in the same way when you are tuning a speaker for development as when you are selecting which speaker to purchase in a store. Let's say all the components are already chosen and locked in, there's little conscious or subconscious reason to prefer one tuning over another.

 

[DanielT]

Yes, the problem is that ruler flat on-axis doesn't tell us everything we need to know.

An even flatter and more highly rated speaker (albeit at a higher price) is the Kii Three. While an impressive speaker in many ways, I must say (with great respect to the designer) it is not the best speaker I have ever heard. Can the reason for this be found in the measurements? Perhaps.

This is hypothetical. You are in the process of creating a new speaker model. You have a number of ready-made prototypes. You let a number of (potential customers) listen to those prototypes/versions and they think X sounds best but you Y. Do you go ahead and launch X or Y on the market? That is, do you give the market what it wants or do you give the market what you want the market to have (your vision)?

 

[Sigberg Audio]

This is hypothetical. You are in the process of creating a new speaker model. You have a number of ready-made prototypes. You let a number of (potential customers) listen to those prototypes/versions and they think X sounds best but you Y. Do you go ahead and launch X or Y on the market? That is, do you give the market what it wants or do you give the market what you want the market to have (your vision)?

If it is my own trusted group of listeners, I would potentially adjust towards X, if it was a random of group of .. untrained listeners if you will, I would likely go with Y. Not because I don't inherently trust the general population, but they come with their own bias, and with their own learned preference from their own systems.

I have never been in this exact situation, but I have been in a situation where people I know and trust have detected issues I did not. As I am sure you know, you get used to the sound of a new system relatively fast. So this is one of the reasons I have others come in to listen, to ensure I'm not gradually going off target without noticing. I also reset quite often by changing rooms and/or going back to listening to existing products. I've even purchased highly regarded competitor products to see how they both sound and measure in the same environment I'm testing in.

 

[Sigberg Audio]

The inherent bias at hifi shows has been interesting to witness as well, here in Norway the law of Jante is very much prevalent, which means anything new and successful is viewed with scepticism, especially if it's from Norway. So conservative audiophiles will by default not like a Norwegian active speaker company.

So during early days I got feedback at shows about the sound I knew not to be true, but rather the full effect of negative bias, which over the years have changed to grudging acceptance and finally praise.

 

[TimVG]

Quoting @kimmosto from a different thread some years ago

Some individual design with familiar drivers and concept does not necessarily need listening. I've done it few times also with commercial products, but then someone else has accepted or selected the best model and drivers by listening. XO tuning could be needed if data was not accurate enough, but not necessarily otherwise. Importance of listening tests increase exponentially with new - especially experimental concepts and drivers designer has not used before. Of course designer needs quite much (earlier) listening experience and knowledge in order to make proper conclusions, set targets and compromises with combination of measurement data and acoustical design.

This is in line with my original comment that it completely depends on the designer if listening is necessary. If you have the necessary experience with the system you are designing, if you have all the data (accurate), if you have a clear idea of what your targets are, and the whole thing is not doomed from the start by having to make large compromises in one way or another.. then where exactly would listening come in?

 

[RickS]

With all the experience Andrew Jones has with listening and pinpointing potential problems in a loudspeaker, I’m actually pretty sure you would walk away highly impressed by his ability to do that, if you had the chance to spend a few days with him.

Yes, his hearing is most likely not as good as it once was, but to say he “doesn't have a good sense of how anything sounds” is pretty disrespectful to all the knowledge and experience. As his work is highly dependent on his listening ability, I’m sure he regularly checks his hearing so that he knows his shortcomings.

Maybe, but rather we did not focus too much on one specific designer. My point is that we usually do not know and, since hearing loss often happens gradually, many do not know unless they are tested. Earlier, I struggled to find data from a scientific study, but finally found this*…

Along with why speaker measurements are essential, this is an excellent study showing how much bandwidth variability there is in human hearing and, along with other biases, hearing often has significant objective inconsistencies. While does not mean our hearing is “bad”, it does mean it is inherently untrustworthy for making conclusive claims. Toss in room influence and potential differences in source quality, subjective review results have little chance of repeatability. Without repeatability of results, conclusions based on them are untrustworthy. Whether Amir, AJ or someone much younger, individual subjective reviews are interesting conversation at best. That said, measurements need to be repeatable to be trustworthy as well!

*source is https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0243001

 

[Sigberg Audio]

Quoting @kimmosto from a different thread some years ago


This is in line with my original comment that it completely depends on the designer if listening is necessary. If you have the necessary experience with the system you are designing, if you have all the data (accurate), if you have a clear idea of what your targets are, and the whole thing is not doomed from the start by having to make large compromises in one way or another.. then where exactly would listening come in?

But for instance the point of "especially experimental concepts and drivers designer has not used before" - this is likely to be the case a lot of the time. So a situation where there are so many known parameters that you don't need to listen (according to Kimmosto) would be the exception rather than the rule.

 

[TimVG]

But for instance the point of "especially experimental concepts and drivers designer has not used before" - this is likely to be the case a lot of the time. So a situation where there are so many known parameters that you don't need to listen (according to Kimmosto) would be the exception rather than the rule.

I mean, it's not because I've not yet used a certain model of driver before that it would impossible for me to assume its behaviour based on measurements
I would interpret "unfamiliar driver" more as something with radically different behaviour in one way or another.

Anecdotal: when I was playing around with filters for the prototype in my avatar, I had 4 presets - and I could swear they all sounded a bit different, and one was definitely my favorite. Fast forward a couple of years, I lost my measurement data, no way to extract it from the miniDSP - I totally forgot which was which.. and all of a sudden they all sounded the same.

Listening is fun of course, but I've come to the conclusion it is not always necessary during design. Then again, when I do it, it's not my job that is at stake

 

[Mort]

Do you use ad-hoc, sighted listening tests to override what you see in measurements?

 

[Newman]

I looked it up a few days ago and the typical person in their 70's can hear up to 8-10 kHz. If AJ is missing the ONE octave with very little musical content 10-20kHz, this line of argument seems like scaremongering.

It's not that simple.

Floyd Toole: "My decision to stop participating in the double-blind listening tests was based on my feeling that it was becoming more difficult - perhaps aided by the fact that loudspeakers being evaluated were getting better and more similar sounding. But also, because the computer generated statistics on my subjective ratings indicated that I was becoming less reliable - the ratings showed increased variability. None of this was surprising, as we had learned many years earlier that this is the result of hearing loss, and my hearing was not getting any better. It is important to know that the listening tests I was participating in were multiple-loudspeaker (3 or 4 at a time) monophonic comparison tests, the purpose of which was to identify resonant colorations and other timbral defects. The almost universal result of such tests is a preference for timbrally neutral, i.e. flat frequency response (on-axis) loudspeakers. As discussed a few times in these forums, the recordings are simply test signals to stimulate and reveal problems. With increasing hearing loss it becomes more difficult to recognize these problems, even when this is the only task at hand.

"...Simple loss of high frequencies is an obvious problem in criticizing sound quality - if it cannot be heard it cannot be commented on. But there is much more. Lost broadband hearing loss, a common defect, means that subtleties at all frequencies are not heard. Then there are the changes in critical bandwidths that occur with age and hearing loss, meaning that how the music itself is perceived changes. Binaural hearing also degrades, meaning that how a listener combines direct and reflected sounds in the room - the soundstage and imaging - also change. Anyone - myself included now - who experiences greater difficulty conversing in restaurants or pubs has already become perceptually handicapped." - link

In summary, getting older as a speaker designer leads to many dimensions of lost acuity in aspects of loudspeaker performance.

cheers