The perfect speaker is room dependent - wide vs. narrow directivity and more

[Thomas_A]

How far away do you listen to? the closer the listening distance the more the elevation changes response at listening spot, angles to ear and toward first specular reflections change more. With 3m listening distance elevating speaker 10cm makes only ~2 degree difference and this shouldn't matter much unless very beaming speaker. If listening distance is only 1 meter then this would be roughly 6 degree change. Perhaps ear is very sensitive to issues or features of your speaker response, in room? Have you listened this in single speaker mono? Perhaps the stereo image suffers more than what simple change in angle and thinking change in response suggests. Interesting. I'm not sure how valuable it is to evaluate stuff with mono speaker if "qualities" don't translate for stereo.

Thomas_A, thanks, will read!
edit. nice, too late night to suck the info into imagination but shows how awesome hearing system is and how stereo systems really can't be optimized on all aspect. Perhaps I should stay in mono

I think the elevation is a nice illusion for film. I use only 4.2.2 system. For a center speaker I would probably go for one below and one above the TV. I don’t like when the dialogue comes below the persons. The rainbow effect from stereo places the dialogue perfectly on the screen.

 

[tmuikku]

Thanks thewas for the calculator .
In this example where the tweeter is 90 cm above the floor ( such as in my case with 8340 on the 63 cm high stand ) , one can see that the reflected sound is about 2.03 ms delayed.
If I compare my 8340 on my 50 cm high stand, the sound is becoming somewhat less clear, even If I angle the speaker so the direct sound reaching my ears are the same .



View attachment 221065

I suspect the change you are hearing is either HRTF stuff or change in vertical first reflections, most probably some of both and perhaps something else I cannot imagine currently If there is a woofer below tweeter, its path length difference is slightly shorter through floor than the tweeters and longer through the ceiling, It has different "reflection signature", its physical location is different. Perhaps the crossover just doesn't lend itself to elevated height so its the particular speaker you had just didn't work elevated for some reason? If it was point source I suspect sound would not change mucho, only about the HRTF. Have you tried the speaker on its side, sideways to make floor and ceiling reflections very similar? Anyway, thinking out loud.

 

[Lbstyling]

Everything is best done in moderation. Speakers in a school hall/gym/whatever you call those, really really bad.

On the other extreme, headphones. Even with crossfeed and convolution they still don't sound like speakers.

I thought that too, until I got my current phone.

Couldn't work out why my headphones sounded so crazy good with it.

Turns out it has Dolby Atmos stereo built in, and its set as on by default.

Sounds unbelievably good imo.

 

[Emlin]

First thing I did was turn Atmos off. It sounded crazy artificial. Maybe it would be good with Atmos source material (I haven't got any), otherwise I hate it.

 

[thewas]

For me there is something in the sound that tells me that reflections below 2 ms delayed sound is bad for sound quality, at least for mid and high frequencies.

In my experience those empirically acquired limits are not hard and also depend on many other factors like directivity, room reflectivity and listening distance, for example I often have more precise imaging on my desktop setups where the first reflections are much earlier than on my main setup, the only thing that can be safely said is that, keeping all other parameters same, the later the early reflections, the better.

 

[ppataki]

Hello,

I want to discuss here the differences of wide and narrow directivity of very good speakers in context with room acoustics and more.

One perfect speaker for all rooms. I don't think so.
On conclusion of the research at of @Floyd Toole and others are that a good speaker sounds good in any room. In general I will agree with this statement, but if you have a closer look at only very good speakers and different rooms and positions of the speakers and the listening position you will find that the perfect speaker is room dependent.

Listening distance
The listening distance is one of the most important parameter here. If you have a look at stereo you will find that you will need a amount of later reflections to get a feeling of "been surrounded by the sound". Omnidirectional speaker achieve this very well since there are more reflections of the room. But you will have a disadvantage with these speakers, the details of the sound will be masked by the reflections which makes the sound "soft", "washy" or "inexact". On the opposite side is a headphone. You have no additional reflections and the sound is "clear" or "exact" but you will have a poor or no feeling of being surrounded by the sound (binaural recording are different but I will not go into detail here). So there is a sweet spot of the ratio of direct sound to later reflected sound. It can be seen in the EBU recommendations on page 6 where the reverberation time at the listening position is recommended. I think this recommendations are valid.

Short listening distance wide directivity
If you follow this path you will need a speaker with wide directivity if you listen at short listening distances. Linkwitz invented a speaker which creates a good feeling of being surrounded by the sound despite the short listening distance. which is very rare to get. This is done by having a omnidirectional speaker and put them in the middle of the room.

Long listening distance narrow directivity
If you have a longer listening distance you will need a speaker with a narrow directivity to get the recommended reverberation time at the listening position with the speaker. And the speaker have do be beam a lot to achieve the goal. E.g. at 2.5m a typical 2 way speaker with 17cm bass beams a lot to wide.

There are plenty other parameter which are important
How constant is the directivity is over frequency, speaker with narrow beam and extra backfiring speaker or dipol like characteristics, first side wall reflection absorb or not, side wall reflections at 60° ideal?, unsymmetric placement is bad but what differences are still okay...

How to get perfect sound?
And the question which speaker room combination is perfect and how should the ultimate goal look like? A "standard" speaker with narrow directivity in a studio like dry room? All horns oben back in a big room with diffusing walls? Line arrays (horizontal wide and vertical narrow diretivity)?

Best
Thomas

That EBU recommendation document is great!!
Do you know if there is one for living rooms too?
(I mean this one is rather for studios and professional environment)

 

[Kvalsvoll]

That EBU recommendation document is great!!
Do you know if there is one for living rooms too?
(I mean this one is rather for studios and professional environment)

Excellent document, but take note of the date: "May 1998". While the hifi world may seem to focus more on power-cords, some of us actually have done significant progress though the decades since then. We have a much better understanding of how room acoustics work, we have different ways to measure and describe performance.

And there is of course no reason for living rooms in ordinary homes to have different criteria compared to a studio, when both are used for the same purpose - reproduction of sound. The difference is that while the studio can be built to match a spec, the living room can not, due to practical and cost issues. But that does not change what the goal should be.

 

[ppataki]

Excellent document, but take note of the date: "May 1998". While the hifi world may seem to focus more on power-cords, some of us actually have done significant progress though the decades since then. We have a much better understanding of how room acoustics work, we have different ways to measure and describe performance.

And there is of course no reason for living rooms in ordinary homes to have different criteria compared to a studio, when both are used for the same purpose - reproduction of sound. The difference is that while the studio can be built to match a spec, the living room can not, due to practical and cost issues. But that does not change what the goal should be.

Do you know if there is an updated version available?

 

[Kvalsvoll]

Do you know if there is an updated version available?

Maybe someone else knows. I suspect they see this as more of a solved problem, and thus no need for much further research or updates.

It is a recommendation, not a strict specification for how to build a room. It is still up to the designer of a room to choose practical realization, and choose acoustic properties of the finalized room, and still be within recommended limits.

Just looking into how measurements and analysis of measurements have evolved, shows a huge difference. And this analysis can now be done much easier with a fraction of the cost, compared to 1998.

 

[thewas]

Mind you that same EBU recommendation (also some similar ITU one) specifies also some fixed operational room response tolerances which Toole correctly criticises, above transition frequency you want flat direct sound, not a predefined room response.

 

[Duke]

... above transition frequency you want flat direct sound, not a predefined room response.

I realize that this is by far the majority opinion on the subject, and is supported by a great deal of data and analysis. Only a fool would propose a different paradigm.

Two decades ago I left a growing and thriving industry to become a high-end home audio dealer and manufacturer, so clearly I have long since met the criteria for being "a fool". With that out of the way, here is the alternative paradigm I propose:

Given that the in-room response dominates our perception of the system's tonal balance, let's start by specifying a desired in-room response. And if we make the assumption that spectral disparity between the direct and reflected sound is undesirable, then the direct sound should track the desired in-room response. In other words, I'm proposing that the target response for the direct sound be an approximate match with the targeted in-room response... easier said than done.

 

[thewas]

I realize that this is by far the majority opinion on the subject, and is supported by a great deal of data and analysis. Only a fool would propose a different paradigm.

Two decades ago I left a growing and thriving industry to become a high-end home audio dealer and manufacturer, so clearly I already meet the definition of "a fool". With that out of the way, here is the alternative paradigm I propose:

Given that the in-room response dominates our perception of the system's tonal balance, let's start by specifying a desired in-room response. And if we make the assumption that spectral disparity between the direct and reflected sound is undesirable, then the direct sound should track the desired in-room response. In other words, I'm proposing that the target response for the direct sound be an approximate match with the targeted in-room response... easier said than done.

I am not sure I understood correctly what you write, but in my (of course limited) experience you want ideally an on-axis flat sound loudspeaker with such directivity that automatically gives you your preferred listening position response just through its directivity, room reverberation and listening distance.

 

[Duke]

I am not sure I understood correctly what you write, but in my (of course limited) experience you want ideally an on-axis flat sound loudspeaker with such directivity that automatically gives you your preferred listening position response just through its directivity, room reverberation and listening distance.

Yes, what you are describing is what I understand to be the paradigm of the majority, as informed by state-of-the-art research.

What I propose is a bit different: That the direct sound (which is usually but not necessarily the on-axis response) have approximately the same frequency response as the target in-room response. So the direct sound and the in-room response curves would both have approximately the same shape, whether that shape be flat or gently-downward-sloping or something else.

 

[MerlinGS]

Given that the in-room response dominates our perception of the system's tonal balance, let's start by specifying a desired in-room response. And if we make the assumption that spectral disparity between the direct and reflected sound is undesirable, then the direct sound should track the desired in-room response. In other words, I'm proposing that the target response for the direct sound be an approximate match with the targeted in-room response... easier said than done.

I'm not sure I am understanding you correctly. Are you suggesting speakers be designed with the direct sound approaching "the desired in-room response"; i.e. its anechoic measurement should not approach flat as much as possible but instead slope like "the desired in-room response"? If that is your suggestion, then I'm not sure that is a useful approach. For now, let's ignore the research that suggests speakers should be designed with a flat and linear response. The problem is most speakers will not be listened to in rooms approaching an anechoic environment; i.e. rooms have furniture, rugs, curtains, and not to mention varying dimensions and shapes. If you have a speaker with a flat frequency response and excellent directivity, you then can apply the eq to meet the room response you desire. However, if you have a speaker with a sloping frequency response (let's say a Harman curve for the sake of argument) when measured in an anechoic room, it will then have a very poor sound in most rooms since the absorption characteristics of most rooms will affect the sound of the speaker further increasing its downward slope. And unlike the flat frequency speaker, the anomalies you now have may be more difficult to eq (since a lot of boost in the frequency response will be required).

 

[Duke]

I'm not sure I am understanding you correctly. Are you suggesting speakers be designed with the direct sound approaching "the desired in-room response"; i.e. its anechoic measurement should not approach flat as much as possible but instead slope like "the desired in-room response"?

Yes, that's exactly what I'm suggesting! And I realize it's highly counter-intuitive.

The problem is most speakers will not be listened to in rooms approaching an anechoic environment;

Imo "rooms approaching an anechoic environment" are NOT desirable for home audio.

If you have a speaker with a flat frequency response and excellent directivity, you then can apply the eq to meet the room response you desire.

And once you have EQ'd the room response to your target, you have SIMULTANEOUSLY EQ'd the direct sound such that it is NO LONGER flat. If you had to boost the top end to achieve your desired in-room curve (which is likely because normally-furnished rooms are more absorptive at short wavelengths than at long ones), NOW your direct sound is TOO BRIGHT. So you have to find a compromise. My proposal results in LESS spectral discrepancy between the direct and reflected sound to begin with, and therefore less discrepancy even AFTER you have EQ'd the in-room sound to meet the response you desire.

However, if you have a speaker with a sloping frequency response (let's say a Harman curve for the sake of argument) when measured in an anechoic room, it will then have a very poor sound in most rooms since the absorption characteristics of most rooms will affect the sound of the speaker further increasing its downward slope

The room will have the same effect on the reflections regardless of their initial spectral content.

In order to accomplish what I'm proposing, the speaker would need to have LESS high frequency energy than normal in the direct sound, but NOT less high frequency energy than normal in the reflections (and probably MORE than normal, if that's what it takes to arrive at our preferred in-room curve).

An argument could be made for a system which allows EQing the direct and reflected sound spectra independently, insofar as is feasible.

And unlike the flat frequency speaker, the anomalies you now have may be more difficult to eq (since a lot of boost in the frequency response will be required).

I'm not proposing LESS high frequency energy than normal in BOTH the direct and reflected sound. I'm proposing the SAME amount of high frequency energy in both, and that the desired in-room curve be the target for BOTH. Implicit in my proposal is a radiation pattern which can accomplish this... something easier said than done.

Edit: @MerlinGS , the quotes in our signature lines are conceptually somewhat similar, in that they leave the door open to unconventional paradigms.

 

[Kvalsvoll]

Mind you that same EBU recommendation (also some similar ITU one) specifies also some fixed operational room response tolerances which Toole correctly criticises, above transition frequency you want flat direct sound, not a predefined room response.

Yes, they have some ideas on target response that sort of looks like something that originates from defects in typical speakers - beaming highs - rather than what the target really should be. If the speaker has sort of wide radiation across most of the frequency range, and then beams at high frequencies, the response will fall off in the top, if the room has enough reflected energy to affect the response in the range where radiation is wider.

When the speaker is right, the response will be close to flat, regardless whether the room is towards dead or very live. Same speaker, 4 very different rooms, one has no acoustic treatment (lift below 100hz is manually adjusted on the bass-system):

 

[thewas]

What I propose is a bit different: That the direct sound (which is usually but not necessarily the on-axis response) have approximately the same frequency response as the target in-room response. So the direct sound and the in-room response curves would both have approximately the same shape, whether that shape be flat or gently-downward-sloping or something else.

So something like a wide band constant directivity I guess? I have heard mainly wide radiating ones trying to achieve it (like omnis), but as you probably know they usually sound too diffuse in typical rooms and listening distances. Others try to keep a higher constant directivity down to the bass which many (including myself) like more but as you this is usually not easy or cheap to achieve.

When the speaker is right, the response will be close to flat, regardless whether the room is towards dead or very live. Same speaker, 4 very different rooms, one has no acoustic treatment (lift below 100hz is manually adjusted on the bass-system):

As above, could you please show the directivity of such a loudspeaker?

 

[Duke]

So something like a wide band constant directivity I guess? I have heard mainly wide radiating ones trying to achieve it (like omnis), but as you probably know they usually sound too diffuse in typical rooms and listening distances. Others try to keep a higher constant directivity down to the bass which many (including myself) like more but as you this is usually not easy or cheap to achieve.

There are several ways to minimize the spectral discrepancy between the direct and reflected sound.

Perhaps the most simple and elegant method would be to have "constant directivity" down to a usefully low frequency, along with a particular off-axis angle where the direct sound approximately matches the power response over most of the spectrum, and that angle would be the recommended listening axis.

As to how low the speaker should maintain a "constant directivity" characteristic, the impression I get from Griesinger and Geddes is that above 700 Hz is where constant directivity matters the most from a psychoacoustics standpoint. Imo constant directivity down to the Shroeder frequency (150-300 Hz ballpark) would be excellent, and even lower would be even better.

 

[Kvalsvoll]

As above, could you please show the directivity of such a loudspeaker?

No. This is not a place I can or want to use to show that.

But I don't need to. This post from @Bjorn shows much of what is important, with several examples and graphs:
https://www.audiosciencereview.com/...arrow-directivity-and-more.15171/post-1257068

For best performance, it is necessary to control directivity across the whole frequency range, including low bass. The most difficult part to get right is the important 100-500hz range. Low bass is easy, when you have a separate bass-system with units placed close to boundaries.

In a room with proper acoustic treatment, all speakers will be closer to flat. A speaker with the right pattern will be closer to flat in any room.

One issue I have observed with non-treated spaces is that they tend to have too much high frequency absorption, so that the spectral distribution of reflected energy has a loss at high frequencies. This can be improved to some extent by increasing the radiation at high frequencies, such as mounting a rear radiating tweeter. Treated rooms generally does not have this problem, because they are made to preserve reflected energy at high frequencies, but rooms size makes a difference here, especially if the back wall (behind the listener) is far away.

 

[gnarly]

Imo constant directivity down to the Shroeder frequency (150-300 Hz ballpark) would be excellent, and even lower would be even better.

I share that opinion, and have made rather large synergy horns trying to achieve that.
By Keele's classic pattern control formula, lowest horizontal I've achieved is about 230Hz (48" wide, 90 deg conical)

Gives a weight, an authority to the lower midrange, that is difficult to completely match with EQ on some of my smaller synergy builds.

For best performance, it is necessary to control directivity across the whole frequency range, including low bass. The most difficult part to get right is the important 100-500hz range. Low bass is easy, when you have a separate bass-system with units placed close to boundaries.

Amen concerning the difficulty of pattern control in the 150-500Hz range !

How do you say low bass is easy though?
Only way I know would be subs fully in the corners....?