What is the ideal directivity pattern for stereo speakers?

[DBMandrake]

Originally Posted by speaker dave

My 0.4 alpha number is probably a little misleading in that it is a number I've found works for a well treated commercial theater of any size. It is probably significantly higher than the typical living room, although a purpose built listening room or home theater could easily hit it.

My living room is 96m cubed and has a mean alpha of .17 with the drapes open (0.69 secs. with Sabine equation) and alpha of .21 with drapes shut (.57 secs RT). An alpha of 0.4 would give .3 secs RT. These calculated RTs are with a spreadsheet model. Using CATT acoustics, a ray tracing program gave .5 secs drapes open and .42 sec drapes shut. Measurements are closer to the CATT acoustics numbers.

Nice room

Interesting, that's definitely a bit more live than what I have at the moment, but then your room is quite a bit bigger and more sparsely occupied.

Even with some significant damping material I am probably only neutralizing the hardwood floors and thin oriental rugs.

I've attached some pictures plus an image of the computer model, for anybody interested.

I can see what you mean in an earlier post where you talked about the left speaker being close to the wall and suspecting balance issues due to the proximity. That is indeed a very asymmetric layout you have there, (no doubt you don't have much choice about it, especially with a piano to contend with!) and you will definitely be getting left-right balance issues from that left speaker.

The rug will help a lot above about 1-2Khz, but there is going to be a lot of side-wall bounce in the low midrange that will do weird things to the stereo balance, if my own experiences are anything to go by.

I find that if both speakers are at least 1.5 metres from the side wall it doesn't matter if they're unequal distances, but once they're down to 1 metre or less unless both speakers are identical distances from their respective side-walls (within a few centimetres) the image is never perfectly balanced, and even then there is always tonal imbalance in the low midrange. (Just an equal tonal imbalance in left and right)

Not sure that I have any suggestions though, given your layout...

[Elias]

Originally Posted by DBMandrake

I agree - this is a major shortcoming of stereo as it is implemented in practice, (lack of standardisation) rather than stereo as a technology per se.

I disagree, as the lack of standardisation is not the biggest problem, but the true major shortcoming is that the two sound sources at +/- 30 degrees (stereo triangle) do not satisfy human psychoacoustics very well.

Originally Posted by DBMandrake

our brains have a remarkable ability to perceive something as convincing and real even if not all of the cues are present. As long as a large percentage of cues are present and there are no contradictory cues, the brain fills in the blanks.

Yes there should be no contradicting cues ! In case of stereo triangle, the fundamental contradiction comes from the two sound sources at +/- 30 degrees when there should be only one at the center (phantom panned at the center). There is no way to overcome this basic flaw of stereo, other than those famous unorthodox underground methods which break the classic principal rule of stereo triangle, and advance from this rebelness.


- Elias

[Elias]

Originally Posted by DBMandrake

Many of us experience perfectly good phantom images

I highly doubt they are perfect !

[Elias]

Originally Posted by erjee

The problem is that we have 14 subjects on 10 different speakers.

The number is higher than you would normally find in any scientific peer revieved paper !

Originally Posted by erjee

So for both cases the observations per speaker are consistent. If we then group by speaker instead of observer, N=10, 70% of the speakers is able put the high freq phantom image at the corerct location.

Not quite, since I tried with 6 configurations which all failed. So we have 15 setups, 14 subjects. So less than 50 % of the setups were capable of high freq phantom imaging. Tossing the coin

- Elias

[boris81]

Originally Posted by DBMandrake

I find that if both speakers are at least 1.5 metres from the side wall it doesn't matter if they're unequal distances, but once they're down to 1 metre or less unless both speakers are identical distances from their respective side-walls (within a few centimetres) the image is never perfectly balanced, and even then there is always tonal imbalance in the low midrange. (Just an equal tonal imbalance in left and right)

I too have discovered that acoustic symmetry is imperative for good image.

[Elias]

Originally Posted by speaker dave

I'm looking at a Deutsche Welle document .

http://www9.dw-world.de/rtc/infotheq...perception.pdf

See figure 7 (Everybody read this paper because it is a good overview of what we are talking about!!). It shows a very linear dependence with 2ms of shift being offset with 16 dB of level.

Those numbers are quite different from e.g. here: See post # 470



- Elias

[Elias]

Originally Posted by DBMandrake

In fact some of the comments made by Markus regarding his perceptions with different amounts of toe in / out and ipsilateral reflection levels suggest that excessive early ipsilateral reflections may be the cause of imprecise phantom imaging at high frequencies, making the problem a speaker set-up issue, not any fundamental flaw in stereo, as you try to paint it.

and

Originally Posted by Rudolf

Isn't this an indicator, that systems with strong ipsilateral reflections will make tweeters even more audible as separate sources than the usual stereo triangle?

Rudolf

do not explain why stereolithic projection worked best for me even it produces most highest levels of ipsilateral early reflections of all configurations. The ipsilateral is not to blame. We are not in a war on ipsies !

Originally Posted by gedlee

Only the narrow directivity in a reverberant space has a decent compromise on both scenarios. It is excellent for the "They are here" and acceptable for the "You are there."

Narrow directivity at the treble range is certainly not acceptable since the sound will be coming from the tweeters ! Who will accept this ?

Originally Posted by Humdinger

I wouldn't be surprised to see that those who had a weak or tweeter based pinknoise had a crossover between 2kHZ and 6kHZ.

Are you ready for the surprise !?! There was no cross over at all ! (of course not a real sursprise since I explained it already carefully, it was FE126En)

Originally Posted by tnargs

Two media channels to satisfy two ears? I assume your car is an automatic because it has two pedals for two feet? I don't follow the logic of your claim in any way shape or form. The logic and rationale for the advantages of multichannel are extremely clear and widely agreed on by the best experts in the business.

It is true, you don't understand !


- Elias

[gedlee]

Originally Posted by DBMandrake

Still, the sweet spot widening effect seems to work with crossing in front of the listener instead of crossing behind, it's just now that I've done these calculations its left me unsure as to exactly what's behind it, and I'm now thinking that straight intensity/time trading can't be the sole explanation, even if it does help somewhat.

Perhaps other factors are also at work, ...

In short, I'm not sure exactly what's happening.

This is what I said several pages back. I did not expect the situation to meet the "ideal" but the fact is that it does work. The time-intensity tradeoff is certainly a big part of it, but it clearly cannot explain it all. Perhaps the data that Dave is using was for impulses or noise or some other signal that makes it extreme and that for music the ear becomes much more tollerant. Maybe it even learns to a certain degree.

I cheaked and Blauert's data is no where near as extreme as Dave was quoting (figure 3.5). His MAXIMUM level offset for a central image is about 9 dB, and that's for a 4 ms time difference. This is certainly different than what Dave was quoting.

[Radugazon]

Enlightening.

Dan

Superior Hearing...I wish I got some, but it was a joke. I'm not the only one in my house to hear schismatic HF.

Statistically, I see that many domestic installations have the speakers close to the boundaries, show big asymmetry, are stuffed by some furniture (ahh, a piano...), AND have some acoustic treatment, mostly absorbing for HF only.
Mine is just the opposite (rigorous symmetry__see boris, post 1179__, empty space, speakers only close to the floor, membrane like boundaries reflecting mainly HF, all this for the sake of side firing).

I think that here are the roots of the difference. But after all, it's not a problem, each of us is happy with the best adaptated configuration.

[dantheman]

Radugazon, I think your explanation is more probable than Elias's. Hopefully he does too and was just trolling. Not very good odds that everyone you brought in all had superior hearing. He's always argued that he's a 'superior hearing, high powered pinna' anomaly. Not many people wanting to build a speaker or playback system for the one in a million. Not good business.

Dan