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

[kaka89]

I suggest before making such statements you familiarize yourself with the way Harman spinorama charts are generated.
Here is a short summary, more you can find here and here.

Great sharing.
However I don't understand why the author prefers greater early reflection. This will lead to greater room sound from the side walls, and it shouldn't be good in a living room.

 

[Krunok]

Great sharing.
However I don't understand why the author prefers greater early reflection. This will lead to greater room sound from the side walls, and it shouldn't be good in a living room.

View attachment 29629

Well, I'm not an expert on this but as I understand it early reflections are important part of how we percieve music. I believe they form most of what we call "soundstage" and for that reason I believe it is important their FR is flat. The situation is probably different with late reflections.

I'm sure @Floyd Toole can provide much better answer..

 

[JanRSmit]

You just wanted to write ipsilateral in a post!

Well, I'm not an expert on this but as I understand it early reflections are important part of how we percieve music. I believe they form most of what we call "soundstage" and for that reason I believe it is important their FR is flat. The situation is probably different with late reflections.

I'm sure @Floyd Toole can provide much better answer..

Reflections later than about 6msec are not added to the direct sound by our hearing system. This is one reason to have the speaker front at least a meter off the wall bevind the speaker.

 

[napilopez]

I think this what you stated is what is measured for the total sound power.

I don't think so... The sound power description comes a bit later. Before the above angles, the book says:

"The early reflections curve is an estimate of all single-bounce first reflections in a typical listening room. Measurements were made of early reflection “rays” in 15 domestic listening rooms. Figure 5.7 shows an example of the horizontal reflections in one of the rooms.

From these data, a formula was developed for combining selected data from the 70 measurements in order to develop an estimate of the first reflections arriving at the listening location in an “average” room (Devantier, 2002). It is the average of the following:

• Floor bounce: average of 20 °, 30 °, 40 ° down
• Ceiling bounce: average of 40 °, 50 °, 60 ° up
• Side wall bounces: average of ± 40 °, ± 50 °, ± 60 °, ± 70 °, ± 80 ° horizontal
• Front wall bounce: average of 0 °, ± 10 °, ± 20 °, ± 30 ° horizontal
• Rear wall bounces: average of 180 °, ± 90 ° horizontal

"

There's also this image, which includes the above angles minus the "rear wall bounces portion:

With the following description:

Figure 5.8: A visual portrayal of the angular ranges considered in calculating the listening window and early reflections in the front hemisphere of a forward-firing loudspeaker. The early-reflections calculation includes the horizontal portion of the listening window.

 

[napilopez]

Great sharing.
However I don't understand why the author prefers greater early reflection. This will lead to greater room sound from the side walls, and it shouldn't be good in a living room.

View attachment 29629

Toole's book suggests that for at-home listening, people tend to prefer some sidewall reflections. It contributes to the soundstage, and our brains are good at separating the reflected room from the sound of the speakers above the transition frequency. For mixing, a dead-ish room is sometimes preferred, but even for that, not always. Ceiling reflections appear to be more problematic.

Well, while we're at it, everyone should go by Toole's book, because I feel like I'm just regurgitating everything already in it. =]

 

[MattHooper]

Discussions about MLs are a long lasting topic on various forums. Just take a look at this one. You will also find some familiar faces there..

Thanks for the link. I'll have to go through that one. Looks like some spicy reading.

 

[MattHooper]

That guy stumbled across a super deal on some Quad ESL63's. He called me up and told me what he had, and said "you need to hear these". I went to his house, and heard them and have never heard a single thing that impressed me like those did. I became good friends with him. And immediately decided I would have some ESL 63's. I did some two months later. Kept them for over a decade. I've got a couple Harman designs now and they are impressive in the sound quality for money quotient. And for just plain sounding good quotient. But they aren't the kind of thing to cause a reaction like those Quad ESL63s.

That last line is an interesting point that resonates with me.

I'd heard high end speakers here and there (and grew up listening to my Dad's KEF 105.2 speakers powered by Carver amps).

But hearing my friend's Quad 63s for the first time was a true shock. It really sounded like nothing I'd heard before - that sense of "no box" like a truly transparent window on to the recording. When I got my Quad 63s I had a great many friends, musicians, acquaintances hear them and they had a similar reaction: like nothing they'd heard before.

I don't find even the HK products produce that sensation, they sound like really good box speakers. But I'm always down for listening to a good panel speaker for what they seem to do in a unique manner. (Though, as I've mentioned, ultimately I prefer box/dynamic speakers).

 

[MRC01]

... That guy stumbled across a super deal on some Quad ESL63's. He called me up and told me what he had, and said "you need to hear these". I went to his house, and heard them and have never heard a single thing that impressed me like those did. I became good friends with him. And immediately decided I would have some ESL 63's. I did some two months later. Kept them for over a decade.
...

That's how I felt when I heard Mag 3.6/R for the first time. That was 20 years ago and I still listen to them every day.
More precisely, I listen to music through them, which is an important distinction as some audiophiles really do listen to the equipment not the music.

 

[MattHooper]

Toole's book suggests that for at-home listening, people tend to prefer some sidewall reflections. It contributes to the soundstage, and our brains are good at separating the reflected room from the sound of the speakers above the transition frequency. For mixing, a dead-ish room is sometimes preferred, but even for that, not always. Ceiling reflections appear to be more problematic.
]

I can modulate the reflectivity (in upper frequencies) of my side walls via thick curtains. I really like the option because sometimes I do prefer the airier, more brilliant and "wider soundstage" sound of exposing the walls more, though sometimes I prefer the more focused and somewhat more timbrally nuanced sound of cutting down those reflections.

 

[napilopez]

I can modulate the reflectivity (in upper frequencies) of my side walls via thick curtains. I really like the option because sometimes I do prefer the airier, more brilliant and "wider soundstage" sound of exposing the walls more, though sometimes I prefer the more focused and somewhat more timbrally nuanced sound of cutting down those reflections.

That's neat! No such option here, unfortunately.

My main takeaway was that so long as the sidewall reflections aren't dramatically changing the timbral qualities of the speakers - the way some absorption panels might - they seem to be fine or even preferable for most listeners. Which is good news for us folks who haven't invested in room treatment or aren't able to for aesthetic reasons =P

If I were to treat somewhere, it'd be above me, but my listening area is below an exposed wooden frame that seems to do a decent job diffusing sound on its own.

 

[MattHooper]

In listening tests, as RE Greene points out, even professionals get used to a particular sound. This is one element of the "circle of confusion". As he points out, "detail" or "transparency" can be seductive but wrong. It starts off sounding good, perhaps, but over time, the sound wears you down and you go out looking for your next speaker. I don't see why the Harman tests should be immune from this, and nor do they address all rooms, all speaker placements, all types of music, all volume settings, etc.

I have some sympathy with your line of questioning. (But then, I'm a bit of a philosophy-geek....through with an empirical bent....)

As you point out, HK is testing for preference. Clearly, human preference CAN be tested for (it's done all the time). But the applicability to that OUTSIDE the test scenario seems to get messy.

This is not exactly like testing for mere audibility, as one might do with, say, changes between audio codecs (or DACs, speaker cables etc). If you can't detect an audible difference under the best controlled conditions, there isn't much of a case to be made that you can detect it in anyway in uncontrolled conditions.

Whereas testing for preference isn't such a straight line. After all, what is the point in testing for preference in the first place? One obvious use case is helping design speakers that people will prefer over other speakers. And also the consumer can take the information to heart about "what type of speaker design people tend to prefer under blind test conditions" and use that in his/her own buying decisions.

But also, presumably, both the speaker designers, and consumers, are seeking information that will help guide them to a speaker purchase leading to long-term satisfaction.

Well...how tightly do the data from the blind testing conform to, or predict, actual consumer long-term satisfaction with a speaker?

As far as I'm aware, at this point, it doesn't. It only predicts what people will prefer on the specific blind-test conditions. The first thing we'd want to worry about is the "Pepsi-Challenge" effect, where as everyone knows (at least in lore) Coke re-designed their product to be sweeter like Pepsi, because Pepsi kept winning blind challenges. New Coke being sweeter then fared well in blind tests against Pepsi. But New Coke failed in the marketplace, for various reasons (it's not a straight-line effect story), one of which is plausibly that many who had always preferred coke to pepsi drank quite a bit of coke, and they found the "less obviously sweet" original formula faired better when drinking such quantities. The test tended to favor a flavor profile that stood out when taken in small amounts like that blind test. (This resonates with me, as I can enjoy a little bit of a very sweet drink, but it wears on me fast if I have much of it, hence I actually choose to drink mostly less sweet-tasting drinks).

So, it seems at least possible there is a Pepsi-challenge effect accounting for some of why people prefer a certain speaker design in blind test conditions. (I've read a lot of what Floyd Toole has written, watched his talks, but if this was addressed in studies I missed it).

But even presuming a Pepsi-challenge effect is not the problematic variable, there is still the question of how the blind-test data transports to the "real world" of consumer speaker purchases. Does it predict with much accuracy which speakers consumers, audiophiles especially, will find to produce long term satisfaction? As far as I can tell: No.

Audiophiles who are very picky about seeking the sound that pleases them, seem to have found long term satisfaction from a wide range of speaker designs. Some are avid ribbon-speaker devotees, others electrostatic "you'll-take-my-Quads-from-my-cold-dead-fingers", others horns, other narrow dispersion, others wide dispersion...it's just all over the map.

Yes...the preference for various speakers can be due to any number of variables and influences. But...that's the problem of moving from blind-testing in the lab to predicting for a consumer "you'll find long term satisfaction with X design instead of Y design."

 

[amirm]

But hearing my friend's Quad 63s for the first time was a true shock. It really sounded like nothing I'd heard before - that sense of "no box" like a truly transparent window on to the recording. When I got my Quad 63s I had a great many friends, musicians, acquaintances hear them and they had a similar reaction: like nothing they'd heard before.

That is the general impression of them. Alas, that impression doesn't go away regardless of what you play. It homogenizes the music that way. Some music should not sound that way. It shouldn't be open, wide, tall, etc. It should be focused.

If all you listen to is some kind of orchestral music that matches that signature, it is fine. Otherwise, one you are sensitized to that overlay, it becomes a negative, not positive that it was the first time you heard them.

 

[MRC01]

IME, dipole panels can sound big and open, or tight and focused, depending on how they're set up in the room. When set up right, they can sound big and open, or tight and focused, depending on the recording without any changes to room setup.

 

[MattHooper]

That is the general impression of them. Alas, that impression doesn't go away regardless of what you play. It homogenizes the music that way. Some music should not sound that way. It shouldn't be open, wide, tall, etc. It should be focused.

If all you listen to is some kind of orchestral music that matches that signature, it is fine. Otherwise, one you are sensitized to that overlay, it becomes a negative, not positive that it was the first time you heard them.

I agree...but I'm not sure in the same respect.

If you mean to say "Quads homogonize the differences between recordings" I'd strenuously disagree. One of the revelations when listening to the Quads for me, and I know many others, was how distinct and particular it made each recording. My Quads were chamelions in that way. It was box speakers that seemed to homogonize recordings in comparison. And I found that any type of music played through them was well served and balanced (within the confines of a sense of hearing what was in the mix, instrumental size/balance etc).

But if you mean there is some other general homogonizing effect, then I'd agree. As I've said, I abandoned electrostatics because I found that, while they told me intellectually a great deal about the nature of each recording, they did not move air in the palpable way I desired. In THAT sense, all recordings lacked in that aspect.

But then...I have never found a single hi fi system, no matter the speaker used, that did not ultimately homogonize sound.
Which is why I'm not ideologically disposed to denigrate those seeking the "sound" or "flavor" that suits their desires best.

 

[Blumlein 88]

That is the general impression of them. Alas, that impression doesn't go away regardless of what you play. It homogenizes the music that way. Some music should not sound that way. It shouldn't be open, wide, tall, etc. It should be focused.

If all you listen to is some kind of orchestral music that matches that signature, it is fine. Otherwise, one you are sensitized to that overlay, it becomes a negative, not positive that it was the first time you heard them.

Never became a negative to me in 12 years or so I owned them. I didn't find them to homogenize music either.

Among non-audiophiles speakers that play loud cleaner than they have usually heard and with real solid deep bass (not loose marshmellow wallowing bass) impresses them. And Quads impressed them. The Quads were cool looking and different, but it was the sound that drew non-audiophile's attention to them in the first place. Comments like, "it just sounds so real", or "it's like the music is here, you don't hear a stereo box". People would come in for other reasons and get transfixed sometimes when they weren't even there to listen to music.

One could attribute that to many things. I think it was lack of a box, and the quasi-point source nature of it.

 

[MRC01]

... One of the revelations when listening to the Quads for me, and I know many others, was how distinct and particular it made each recording. My Quads were chamelions in that way. It was box speakers that seemed to homogonize recordings in comparison. ...

My experience as well, with planar dipoles.

 

[MattHooper]

Never became a negative to me in 12 years or so I owned them. I didn't find them to homogenize music either.

Among non-audiophiles speakers that play loud cleaner than they have usually heard and with real solid deep bass (not loose marshmellow wallowing bass) impresses them. And Quads impressed them. The Quads were cool looking and different, but it was the sound that drew non-audiophile's attention to them in the first place. Comments like, "it just sounds so real", or "it's like the music is here, you don't hear a stereo box". People would come in for other reasons and get transfixed sometimes when they weren't even there to listen to music.

One could attribute that to many things. I think it was lack of a box, and the quasi-point source nature of it.

Yup.

Many of my guests had experience listening to bigger, much more powerful speakers with "slam" and all the things one would think the average person would be impressed by. And yet every single person expressed amazement by the Quads....so many "it sounds so real!" comments.
People knew sound could be impressive, but they often never considered music could sound "real," and the Quads did that for many listeners.

(And btw what sounds "real" I think is context-driven to a degree. The Quads remove the box sound and so lacking that they sounded "more real" to people. But if you turn your attention to other aspects of sound, you can decide the Quads sound "less real" than certain other speakers - e.g. if a sense of dynamic, palable presense is what you cue in on, some horn speakers may sound "more real" than the Quads).

 

[Krunok]

Reflections later than about 6msec are not added to the direct sound by our hearing system. This is one reason to have the speaker front at least a meter off the wall bevind the speaker.

Ok, so you move speakers away from the wall and reflections got "later". Let's suppose speakers were at 0.5m from the wall behind and now are 1m so the sound travels 1m more than before, which is 3ms. Does that make the sound better?

 

[svart-hvitt]

I have some sympathy with your line of questioning. (But then, I'm a bit of a philosophy-geek....through with an empirical bent....)

As you point out, HK is testing for preference. Clearly, human preference CAN be tested for (it's done all the time). But the applicability to that OUTSIDE the test scenario seems to get messy.

This is not exactly like testing for mere audibility, as one might do with, say, changes between audio codecs (or DACs, speaker cables etc). If you can't detect an audible difference under the best controlled conditions, there isn't much of a case to be made that you can detect it in anyway in uncontrolled conditions.

Whereas testing for preference isn't such a straight line. After all, what is the point in testing for preference in the first place? One obvious use case is helping design speakers that people will prefer over other speakers. And also the consumer can take the information to heart about "what type of speaker design people tend to prefer under blind test conditions" and use that in his/her own buying decisions.

But also, presumably, both the speaker designers, and consumers, are seeking information that will help guide them to a speaker purchase leading to long-term satisfaction.

Well...how tightly do the data from the blind testing conform to, or predict, actual consumer long-term satisfaction with a speaker?

As far as I'm aware, at this point, it doesn't. It only predicts what people will prefer on the specific blind-test conditions. The first thing we'd want to worry about is the "Pepsi-Challenge" effect, where as everyone knows (at least in lore) Coke re-designed their product to be sweeter like Pepsi, because Pepsi kept winning blind challenges. New Coke being sweeter then fared well in blind tests against Pepsi. But New Coke failed in the marketplace, for various reasons (it's not a straight-line effect story), one of which is plausibly that many who had always preferred coke to pepsi drank quite a bit of coke, and they found the "less obviously sweet" original formula faired better when drinking such quantities. The test tended to favor a flavor profile that stood out when taken in small amounts like that blind test. (This resonates with me, as I can enjoy a little bit of a very sweet drink, but it wears on me fast if I have much of it, hence I actually choose to drink mostly less sweet-tasting drinks).

So, it seems at least possible there is a Pepsi-challenge effect accounting for some of why people prefer a certain speaker design in blind test conditions. (I've read a lot of what Floyd Toole has written, watched his talks, but if this was addressed in studies I missed it).

But even presuming a Pepsi-challenge effect is not the problematic variable, there is still the question of how the blind-test data transports to the "real world" of consumer speaker purchases. Does it predict with much accuracy which speakers consumers, audiophiles especially, will find to produce long term satisfaction? As far as I can tell: No.

Audiophiles who are very picky about seeking the sound that pleases them, seem to have found long term satisfaction from a wide range of speaker designs. Some are avid ribbon-speaker devotees, others electrostatic "you'll-take-my-Quads-from-my-cold-dead-fingers", others horns, other narrow dispersion, others wide dispersion...it's just all over the map.

Yes...the preference for various speakers can be due to any number of variables and influences. But...that's the problem of moving from blind-testing in the lab to predicting for a consumer "you'll find long term satisfaction with X design instead of Y design."

The point you make, that one risks testing for a specific test procedure rather than general speaker qualities, is as good as it’s an obvious one. Another point, made by @oivavoi recently, that findings from just one research group, would stand stronger if they were supported by other research groups, is also a good one.

In other fields of research, where much more resources are spent on research than in the audio field, these points would have been making the head lines more often. My point is, I find it interesting that quite a few ASR members do not share these views. On the contrary, it’s as if these views are ridiculed. To me, it’s an indication that audio science is a bit immature, lacking in curiosity, with a need to push certain research camps instead of taking a broader view.

 

[amirm]

RE Green makes a few suggestions starting at the part titled "How Things Can Go Wrong".

He is mistaken about the listening tests we are discussing. First, the order is truly random so which comparison you are making won't enter the equation across many trials. From Sean Olive's paper:

Second, the switching time between speakers is 3 to 4 seconds. When I took the test, that did a very effective job of making you forget the previous speakers and just judge the speaker on its own merit. I focused on the voice for example and tried to imagine how natural it sounded to me.

I attended two sessions at Harman. First one was with a bunch of high-end audio dealers and the second, organized for top acoustics experts in US. In both trials, almost everyone voted the same, picking the JBL speaker as the best sounding. So prior knowledge of sound made no difference.

FYI I read through Green's paper. It is full low resolution frequency response charts that we know are not good practice in predicting listener preference. There is also one reference to listening tests and that reference doesn't support much of what he says, but is indeed in support of Harman's research. That frequency response is by far the most important thing in listener preference for speakers. Here is the AES paper: http://www.aes.org/e-lib/online/browse.cfm?elib=11885

Listening Room Influence on Loudspeaker Sound Quality and Ways of Minimizing It


The test setup was such:

When the speaker had flat on-axis response, the difference between electrical path and the one going through the speaker and microphone was almost not there! Listeners had a hard time telling the two paths apart indicating that distortions, phase, etc. play no role in our detection of sound. To wit, these are the conclusions stated: