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

[Thomas_A]

I have not followed the thread in detail due to vacation and bad internet connection, but now with better connection in London. I fully agree with a goal having an almost perfectly linear response on axis and a smooth off-axis dispersion. However, if the speaker is designed to be put in a small room, large room, near wall, toe in angle, listening angle, etc will show very different frequency response if you measure in a standard anechoic chamber. If you make double-blind listening of speakers that are intended for different rooms and placement, and then use the same room and placement for the test, the judgments will be different but there will be wrong conclusions of their quality based on the intended use in a room.

 

[Blumlein 88]

I have not followed the thread in detail due to vacation and bad internet connection, but now with better connection in London. I fully agree with a goal having an almost perfectly linear response on axis and a smooth off-axis dispersion. However, if the speaker is designed to be put in a small room, large room, near wall, toe in angle, listening angle, etc will show very different frequency response if you measure in a standard anechoic chamber. If you make double-blind listening of speakers that are intended for different rooms and placement, and then use the same room and placement for the test, the judgments will be different but there will be wrong conclusions of their quality based on the intended use in a room.

There is a little about position in the Toole book. When 4 speakers were tested in two different room positions the level of preference did change. The relative ranking of the 4 was the same (well two which were just nearly identical in position one were just nearly identical in position two with a slight preference for one vs the other in changed positions). There isn't much information about this however. I wouldn't doubt with extreme differences in size and shape the position might alter which is most preferred though apparently it would only change it slightly. The room position could change how good you think a speaker is by a significant amount. So good positioning matters. It appears it matters about equally as much for any given speaker. One could probably come up with some outliers where it might through things off.

 

[Thomas_A]

There is a little about position in the Toole book. When 4 speakers were tested in two different room positions the level of preference did change. The relative ranking of the 4 was the same (well two which were just nearly identical in position one were just nearly identical in position two with a slight preference for one vs the other in changed positions). There isn't much information about this however. I wouldn't doubt with extreme differences in size and shape the position might alter which is most preferred though apparently it would only change it slightly. The room position could change how good you think a speaker is by a significant amount. So good positioning matters. It appears it matters about equally as much for any given speaker. One could probably come up with some outliers where it might through things off.

Agreed that room size is a factor. If you have a small room with a desire to have the speaker near wall you may need a speaker with sloping response in the bass (measured) that use the wall gain support, and a wider dispersion to fill in more reflexes from the early reflections in the room. If you own a cinema you might much larger speakers that have a more narrow dispersion but need less compensation in the bass region due to the altered Shroeder frequency.

 

[Krunok]

Speakers are funamentally differnet from dacs and amps. A speaker would have need to have same radiation pattern to sound the same, and when you turn up the volume, capacity matters.

The experiment you refer to actually is a good example for how different those speakers sound - even when the frequency reponse and level is somewhat matched.

I think you missed this part: "As manufacturers will continue to build better and better speakers their spinorama charts would start to look alike", and spinorama charts represent what you called "radiation patterns", don't they?

Regarding turning up the volume I believe I mentioned "they will all be able to deliver mighty bass at decent SPLs."

The point of mentioning experiment with JBL and Kef LS50+subs was to remind you that, although those 2 speakers have very different directivity patterns, they still sound much more similar then one would expect.

Anyway, the point of my post was this: as more and more loudspeaker manufacturers start gettings things right meeting the modern loudspeaker SQ standards (like the ones spinorama has set) we can expect them to start sound alike as the differences between them would be getting harder and harder to hear.

 

[Krunok]

I am likely alone on this forum in viewing the above scenario as more dystopian and utopian ;-)

I certainly don't mean that such research, as HK et al are doing, shouldn't be done or that it shouldn't be used to advance speaker design. Quite the opposite! I want as much good science thrown at speaker design as possible, so there are ever more powerful tools and theories for the loudspeaker designers to advance the field.

So I want speakers that converge, as you describe above, to exist as a choice for myself, and anyone else.

BUT...I ALSO happen to very much enjoy the variety that is out there in speaker design and sound. I've heard the HK speakers. But I also like different sounding speakers quite a bit and would be quite sad to see those choices removed by speakers being commodified to a single sound/design. It would be like everyone being issued the exact same car. There goes the joy of car enthusiasts who actually enjoy different types of cars for different reasons. I like variety of choice.

Now, of course the total commodification of speakers this way is highly unlikely. But philosophically speaking, I've seen objectivists who seem to have that goal as a good thing "I want a speaker that is neutral and tests with the best possible scores in blind tests...then I'm done and don't have to think about speakers anymore." That would be great for those people with that type of mindset. But....some of us aren't constituted that way, and we have scratches that need to be itched. Again, I've heard the Revel speakers, but I sure would be bummed if I could never listen to another pair of Quads again because that's a damned compelling listening experience too.

It all depends upon what you expect from your speakers. If you expect them to faithfully reproduce the sound then you can expect them to sound more and more neutral as is the case with modern amplifiers. If you want your speakers to add some specific signature to the sound they reproduce, well, that is similar to what tube amps are doing.

 

[Kvalsvoll]

I think you missed this part: "As manufacturers will continue to build better and better speakers their spinorama charts would start to look alike", and spinorama charts represent what you called "radiation patterns", don't they?

Regarding turning up the volume I believe I mentioned "they will all be able to deliver mighty bass at decent SPLs."

The point of mentioning experiment with JBL and Kef LS50+subs was to remind you that, although those 2 speakers have very different directivity patterns, they still sound much more similar then one would expect.

Anyway, the point of my post was this: as more and more loudspeaker manufacturers start gettings things right meeting the modern loudspeaker SQ standards (like the ones spinorama has set) we can expect them to start sound alike as the differences between them would be getting harder and harder to hear.

Similar speakers will start to sound very much alike, if they are developed using the same set of requirements. So, small bookshelves with dome + 6" woofer will all sound alike, while speakers with different size, driver arrangement, radiation pattern, all will sound different. There are very real physical limitations that prevent such a speaker to be able to sound like my F2, which has a very different radiation pattern.

And some manufacturers will continue to have their idea of how a perfect speaker must be, so even if this data set was sufficient to describe everything, and they have the technical ability to make a speaker that satisfies this requirement, we will see different sounding speakers.

The spinorama charts that shows DI and early reflections and listening window and sound power are not sufficient to fully describe the radiation pattern. My understanding is that this presentation form was developed to be able to display the information that was considered to be important, in a reasonably simple way, to be able to visualize the sound quality. The complete measurement set shows everything, provided the angle grid has sufficient resolution, but it is difficult to visualize and compare.

My understanding is that they did not provide a definite specification for how DI and ER should look like, so there is room for choice - both shape and angle coverage (narrow vs. wide). It does not specify frequency response of radiation, only that it should be smooth.

Speakers with different radiation sound different, even when frequency response amplitude measures equal. But they also measure differently - phase is different, decay profile is different, IR is different, GD is different. Only in a completely anechoic environment will they be equal (strictly, this is not true, but for fixed distance it is valid.), and even a very good room is far from anechoic.

The experiment mentioned was a good example because the author (@mitchco ?) did a good job of describing the set-up and experiment, then provided sound samples that actually show very audible differences, so you can hear it yourself. Of course it is not the same as being present in the room, but I think we understand there are limitations with such recordings, and they are also as I remember it discussed in the article.

Recently we have seen the release of some smaller speakers with much more controlled radiation at lower frequencies. Judging by how they are received, radiation matters.

Controlling radiation in the lower midrange gives improvements that can also be seen on measurements - frequency response is more predictable and more smooth, and does not vary so much between different rooms or placement. This is audible.

Radiation pattern affects tonality of transient sounds, so that it is not possible to equalize different speakers to similar tonal balance, because if they sound similar on continuous signals, they will sound different on transient signals.

A small speaker will never be able to deliver full-frequency range, full-scale sound. This is a given due to physical constraints, it is simply not possible to create low frequency sound at sufficient spl from very small physical size, it is also very difficult to achive the desired radiation control, even at higher frequencies.

 

[Krunok]

My understanding is that they did not provide a definite specification for how DI and ER should look like, so there is room for choice - both shape and angle coverage (narrow vs. wide). It does not specify frequency response of radiation, only that it should be smooth.

That is not my understanding. I believe clear guidelines have been established for he optimum spinorama charts. That would mean that, if all speakers would be designed having that same target in mind, small speakers would differ from large speakers not by DI and ER but only by LF extension, so when you add a subwoofers you would have a hard time to tell them apart.

Once industry accepts the same target for speakers all products will start to converge to that target and all speakers will start to sound more and more alike.

 

[MattHooper]

It all depends upon what you expect from your speakers. If you expect them to faithfully reproduce the sound then you can expect them to sound more and more neutral as is the case with modern amplifiers. If you want your speakers to add some specific signature to the sound they reproduce, well, that is similar to what tube amps are doing.

Well, I'm a rapscallion who uses tube amplifiers, so there you go

 

[Thomas_A]

That is not my understanding. I believe clear guidelines have been established for he optimum spinorama charts. That would mean that, if all speakers would be designed having that same target in mind, small speakers would differ from large speakers not by DI and ER but only by LF extension, so when you add a subwoofers you would have a hard time to tell them apart.

Once industry accepts the same target for speakers all products will start to converge to that target and all speakers will start to sound more and more alike.

So there is someone claiming that there is one and only one optimum dispersion pattern regardless of room size and distance to reflecting surfaces? If so I would like to see that research.

 

[Krunok]

So there is someone claiming that there is one and only one optimum dispersion pattern regardless of room size and distance to reflecting surfaces? If so I would like to see that research.

So there is someone claiming that optimum dispersion pattern depends on room size and distance to reflecting surfaces? If so I would like to see that research.

 

[Thomas_A]

So there is someone claiming that optimum dispersion pattern depends on room size and distance to reflecting surfaces? If so I would like to see that research.

Well of the research concluding these optima has been conducted in an optimum listening room, you will have problems to translate that to any room. There is research that defines the time window of lateral reflection to get optimum spatial cues, but if those criteria are not met, you cannot claim the same rules applies, unless the research has been done. I have not seen any though.

 

[Krunok]

Well of the research concluding these optima has been conducted in an optimum listening room, you will have problems to translate that to any room. There is research that defines the time window of lateral reflection to get optimum spatial cues, but if those criteria are not met, you cannot claim the same rules applies, unless the research has been done. I have not seen any though.

Your room is what it is, as well is mine. Speaker with optimum dispersion pattern may behave better in your room than in mine, but speaker witl less-than-optimum dispersion pattern will behave worse in both rooms than speaker with optimum dispersion pattern..

 

[Thomas_A]

Your room is what it is, as well is mine. Speaker with optimum dispersion pattern may behave better in your room than in mine, but speaker witl less-than-optimum dispersion pattern will behave worse in both rooms than speaker with optimum dispersion pattern..

So my question to that was whether the experiments in the different rooms have been performed?

 

[Krunok]

So my question to that was whether the experiments in the different rooms have been performed?

My guess is that manufacturers like Harman don't really perform measurements in different rooms. They aim to manufacture a speaker that would sound better in any room than speaker with worse measurements in anechoic environment.

 

[Kvalsvoll]

That is not my understanding. I believe clear guidelines have been established for he optimum spinorama charts. That would mean that, if all speakers would be designed having that same target in mind, small speakers would differ from large speakers not by DI and ER but only by LF extension, so when you add a subwoofers you would have a hard time to tell them apart.

Once industry accepts the same target for speakers all products will start to converge to that target and all speakers will start to sound more and more alike.

My understanding is that there are clear guidelines for how to measure and extract the charts, but no definition of how they should look.

And the charts from Harman speaker I have seen, are different, and they all have wider dispersion towards omni at low frequencies. Is that the best solution, or is it because that was possible to achieve? A dipole can be designed to have flat power, but that may not necessarily be the most desired response either.

It is possible to design radiation so that there is a very linear coverage inside a defined area, and a quite steep and abrupt reduction outside. If this is to be the requirement, small speakers will be impossible. If a more lean coverage with wider dispersion is the standard, it would exclude the option of designing for the more controlled pattern.

 

[Krunok]

My understanding is that there are clear guidelines for how to measure and extract the charts, but no definition of how they should look.

If that would be the case such charts would be pretty useless as you wouldn't be able to tell which speaker is better by looking at the charts.

I suggest you read this article.

 

[edechamps]

I believe clear guidelines have been established for he optimum spinorama charts. That would mean that, if all speakers would be designed having that same target in mind, small speakers would differ from large speakers not by DI and ER

While guidelines do exist for on-axis (should be flat, horizontal) and off-axis curves (should be smooth), I am not aware of clear guidelines for DI. I just found this in the discussion of Olive's model, section 3.2.3:

The ideal target slope for the on-axis and listening window curves (0 and –0.2) is identical for both test samples, which indicates that the on-axis curve should be flat, while the off-axis curves should tilt gently downwards. The degree of tilt varies among curves for Test One and the larger sample. Test One includes mostly 2-way designs whereas the larger sample includes several 3-way and 4-way designs that tend to have wider dispersion (hence smaller negative target slopes) at mid and high frequencies. This suggests that the ideal target slope may depend on the loudspeaker’s directivity.

In Olive's paper, general slope of ERDI (Early Reflections Directivity Index) and SPDI (Sound Power Directivity Index) was not particularly well correlated with preference (r around +/- 0.2).

One thing to note is that Olive's model attributes significant weights to Narrow Band Deviations (NBD) and Smoothness (SM) of the Predicted In-Room Response (PIR) - 20.5% and 17.5%, respectively. PIR is a curve that is derived from the on-axis and off-axis curves in ways that are described in another paper. Because PIR would be influenced by the loudspeaker's directivity, I think one might be able to conclude that directivity does have an impact on the final preference score as calculated by Olive's model, but given the relationship is quite indirect, it might be tough to determine what an ideal DI curve looks like.

 

[Kvalsvoll]

If that would be the case such charts would be pretty useless as you wouldn't be able to tell which speaker is better by looking at the charts.

I suggest you read this article.

There is a description for what to look for - generally, smooth is better. But there is no definition of the actual radiation pattern.

 

[Krunok]

While guidelines do exist for on-axis (should be flat, horizontal) and off-axis curves (should be smooth), I am not aware of clear guidelines for DI. I just found this in the discussion of Olive's model, section 3.2.3:



One thing to note is that Olive's model attributes significant weights to Narrow Band Deviations (NBD) and Smoothness (SM) of the Predicted In-Room Response (PIR) - 20.5% and 17.5%, respectively. PIR is a curve that is derived from the on-axis and off-axis curves in ways that are described in another paper. Because PIR would be influenced by the loudspeaker's directivity, I think one might be able to conclude that directivity does have an impact on the final preference score as calculated by Olive's model, but given the relationship is quite indirect, it might be tough to determine what an ideal DI curve looks like.

Yes, i read the same thing, but from that sentence ("while the off-axis curves should tilt gently downwards. ") I concluded that it is the more gentle, the better.

Maybe we can kindly ask Dr. @Floyd Toole for comment?

 

[edechamps]

Yes, i read the same thing, but from that sentence ("while the off-axis curves should tilt gently downwards. ") i concluded that it is the more gnetle, the better.

In Olive's paper, the absolute slope of the Early Reflections Directivity Index and the Sound Power Directivity Index are noted as ERDI_SL and SPDI_SL, respectively. Both anticorrelate with preference with r = ~-0.2. Which is consistent with the idea that the flatter the DI the better, but the small correlation probably doesn't mean that much. Also, this number is from data from the smaller test (Test One), not the large-scale study. One way to try to get to the bottom of the influence of ERDI/SPDI might be to do some analysis on the raw data.