I mentioned a similar test I did indoors. A helper stood behind me with a pair of fiberglass sections. They were roughly 18" x 36" x 4" of standard insulation grade. As I listened to the pair of speakers in front of me he raised and lowered the fiberglass pieces. Lowered they were at ear level and blocked most of the side arriving sound but not the direct sound from the speakers. Raised they were pretty much out of the way.
I like this better than going outdoors because you can quickly compare the "with and without" sidewall reflections.
The difference I remember was that the sound shrank to an at-the-speakers-and-between field with the fiberglass blocking the side reflections. It wasn't bad, but it was different. From what I've heard about listening to stereo in an anechoic chamber, that is similar (probably more extreme, though.)
Note that the side wall reflections were never obvious before but the sound was noticably different when they were removed.
Anybody can try this test.
David S.
Why not? Usually for the solo vocalist the microphone is in the singers mouth. The only logical, scientific and objective conclusion is the image should be huge in size ! If the reproduction system should have any level of accuracy, the image of the solo vocalist's mouth should be at least the size of the front wall !
It's a phrase ! I don't actually think brains are moulded 
As far as I know. But I could be wrong Please do not think auditory system is perfect !
I still maintain that the ideal directivity for a rectangular room is what is created by a constant directivity cornerhorn. It sort of blurs the distinction between an omnidirectional speaker and one with constant 90° directivity, since the very nature of a trihedral corner forces 90° radiation anyway. The woofer is acoustically close to the apex of the corner, and the corner itself sets the pattern from the Schroeder frequency upwards. It is blended with a 90° midhorn, set at ear level, snuggled back into the corner deep enough the walls act as extensions at the low end of its range. The blending mitigates the floor bounce notch and the vertical room modes.
The whole system radiates a 90° pattern all the way through the audio band, free from early reflections from the nearest boundaries. Below the Schroeder frequency, multisubs can be employed, so with this arrangement, you have have uniform coverage throughout the room. Surrounds set alongside the listeners add spaciousness. To me, this is the best arrangement, one that seems like it covers everything without compromise. The only real problem, as I see it, is that not every room has the right corners to support this arrangement. They have to be at the right distance apart, and have to be unobstructed for a few feet from the apex. But if you have a room like that, it really seems ideal to me. I've always found it to be a magical setup.
The whole system radiates a 90° pattern all the way through the audio band, free from early reflections from the nearest boundaries. Below the Schroeder frequency, multisubs can be employed, so with this arrangement, you have have uniform coverage throughout the room. Surrounds set alongside the listeners add spaciousness. To me, this is the best arrangement, one that seems like it covers everything without compromise. The only real problem, as I see it, is that not every room has the right corners to support this arrangement. They have to be at the right distance apart, and have to be unobstructed for a few feet from the apex. But if you have a room like that, it really seems ideal to me. I've always found it to be a magical setup.
Markus wrote;
On Why we hear phantom images?
“Good question, what's the answer?”
The answer is we hear by localizing the source of sound. While we are not aware of it, our ears have a very difference response shape depending on the incoming angle and direction. Instead of hearing all these effects, what we have learned is what direction it is etc.
Imagine you’re on the top of tower above the ground and imagine a stationary real sound coming from directly in front of you, what is reaching each ear is essentially the same signal and your ear’s pinna response allows your brain to decide how high it is in position and completes the image.
You have no trouble at all hearing the direction of this sound.
If you were enclosed with walls of three sides, you would have a greater difficulty hearing the direction of the source because now you have information unrelated to the origin of the sound.
If you play this through headphones, the image floats in the center of your head because you are lacking the artifacts the ear and head shape cause with airborne sound coming from in front of you.
Play back a mono signal on a pair of speakers that have little character of their own and do this outdoors or where there is little reflected sound and the phantom image of it also appears directly in front of you. If you have not heard a phantom image so real that it sounds like there must be a center speaker, try to imagine that is possible with a coherent broadband source outdoors and indoors with coherent very directional speakers..
So far, none of this involves reflected sound at all and is only *harmed by them.
By harmed, I mean so far as the strength of the phantom image, like noise, your hearing system has to work overtime to hear, to decide what to lock on to.
An analogy. Consider you’re in a huge room with no absorption.
On a stool in front of you is a speaker playing a human voice saying random words. Up close, there is no problem hearing the words and as you back away, the SPL goes down at the inverse square rate but the words intelligible. AS you get farther way, the SPL falls off but at less than the inverse square and the words are getting harder to understand.
Very far away, your in the reverberant field and the SPL is constant and while it sounds like natural speech, you can’t make out any words.
So far as hearing the information, there is no point where a decrease in the direct to reflected sound ratio improves anything related to hearing information.
Now, in these situations, one finds that the greater the speakers directivity, the larger the distance is where you can still make out the words, the greater the direct to reflected ratio is at any distance.
With the stereo image, you’re not using words, you using an acoustic image and this requires similar things if the goal is relaying information..
best,
Tom Danley
On Why we hear phantom images?
“Good question, what's the answer?”
The answer is we hear by localizing the source of sound. While we are not aware of it, our ears have a very difference response shape depending on the incoming angle and direction. Instead of hearing all these effects, what we have learned is what direction it is etc.
Imagine you’re on the top of tower above the ground and imagine a stationary real sound coming from directly in front of you, what is reaching each ear is essentially the same signal and your ear’s pinna response allows your brain to decide how high it is in position and completes the image.
You have no trouble at all hearing the direction of this sound.
If you were enclosed with walls of three sides, you would have a greater difficulty hearing the direction of the source because now you have information unrelated to the origin of the sound.
If you play this through headphones, the image floats in the center of your head because you are lacking the artifacts the ear and head shape cause with airborne sound coming from in front of you.
Play back a mono signal on a pair of speakers that have little character of their own and do this outdoors or where there is little reflected sound and the phantom image of it also appears directly in front of you. If you have not heard a phantom image so real that it sounds like there must be a center speaker, try to imagine that is possible with a coherent broadband source outdoors and indoors with coherent very directional speakers..
So far, none of this involves reflected sound at all and is only *harmed by them.
By harmed, I mean so far as the strength of the phantom image, like noise, your hearing system has to work overtime to hear, to decide what to lock on to.
An analogy. Consider you’re in a huge room with no absorption.
On a stool in front of you is a speaker playing a human voice saying random words. Up close, there is no problem hearing the words and as you back away, the SPL goes down at the inverse square rate but the words intelligible. AS you get farther way, the SPL falls off but at less than the inverse square and the words are getting harder to understand.
Very far away, your in the reverberant field and the SPL is constant and while it sounds like natural speech, you can’t make out any words.
So far as hearing the information, there is no point where a decrease in the direct to reflected sound ratio improves anything related to hearing information.
Now, in these situations, one finds that the greater the speakers directivity, the larger the distance is where you can still make out the words, the greater the direct to reflected ratio is at any distance.
With the stereo image, you’re not using words, you using an acoustic image and this requires similar things if the goal is relaying information..
best,
Tom Danley
speaker dave,
Would you please respond to post my post 849?
And maybe to Earl's post 851, where he talks about his discussing the issue I raised with Floyd Toole.
This may or may not be true, but has it been thoroughly tested? As Tom Danley says it, how can you say X doesn't matter if you've never had the chance to judge with or without X. A fair comparison would be something like a Revel Ultima Salon compared to for example Geddes Summa, both supported with multiple subs. Maybe the Summa should have some EQ to make the direct sound as smooth as the Salon's.
Of course this comparison is still a difficult one to draw hard conclusions from, as both speakers may work best under specific acoustic settings, but it's better than the no mans land we seem to be in now.
In his book, Toole hardly speaks of speakers with high directivity. On page 381 he mentions some high directivity speakers. The measurements can be found in figure 18.8. He says both sound very good, but he doesn't mention blind tests - something he does do when he speakers of for example the speaker of figure 18.6, a few pages earlier.
At page 398 he speaks of two speakers of which the measurements can be found in figure 18.17. One a conventional cone-dome speaker, the other a speaker with horns for mids and highs. He says in blind tests they performed statistically the same, with some people preferring the cone-dome and others the speaker with horns.
So let's assume they do indeed sound similarly good under the test conditions. What were those conditions? They probably weren't at sound levels you might occasionally listen to at home, because it might put the subjects' ears at risk. But I have an idea which speaker might win if listened to at levels above 90 dB (unweighted).
I admire your skepticism. It's good to have well-grounded opposing views.
With respect to the part of the quote I put in bold, you are probably referring to something like the speakers I've built. To be honest, I have no idea if extending CD behavior as low as my design does is useful, but to what frequency do you think it is?
Would you please respond to post my post 849?
And maybe to Earl's post 851, where he talks about his discussing the issue I raised with Floyd Toole.
This may or may not be true, but has it been thoroughly tested? As Tom Danley says it, how can you say X doesn't matter if you've never had the chance to judge with or without X. A fair comparison would be something like a Revel Ultima Salon compared to for example Geddes Summa, both supported with multiple subs. Maybe the Summa should have some EQ to make the direct sound as smooth as the Salon's.
Of course this comparison is still a difficult one to draw hard conclusions from, as both speakers may work best under specific acoustic settings, but it's better than the no mans land we seem to be in now.
In his book, Toole hardly speaks of speakers with high directivity. On page 381 he mentions some high directivity speakers. The measurements can be found in figure 18.8. He says both sound very good, but he doesn't mention blind tests - something he does do when he speakers of for example the speaker of figure 18.6, a few pages earlier.
At page 398 he speaks of two speakers of which the measurements can be found in figure 18.17. One a conventional cone-dome speaker, the other a speaker with horns for mids and highs. He says in blind tests they performed statistically the same, with some people preferring the cone-dome and others the speaker with horns.
So let's assume they do indeed sound similarly good under the test conditions. What were those conditions? They probably weren't at sound levels you might occasionally listen to at home, because it might put the subjects' ears at risk. But I have an idea which speaker might win if listened to at levels above 90 dB (unweighted).
I admire your skepticism. It's good to have well-grounded opposing views.
With respect to the part of the quote I put in bold, you are probably referring to something like the speakers I've built. To be honest, I have no idea if extending CD behavior as low as my design does is useful, but to what frequency do you think it is?
Of course it will sound artificial, because it is not natural in any way.
Even if you image yourself on a plain field, without any lateral reflections, you will never find a situation in this world that there would be exactly two identical sound sources making the exact same sound at the exact same moment (the stereo). It simply do not exist in this world.
BUT, when you add lateral reflections, the situation is not imaginary anymore but becomes a real life case, and thus more natural.
The psychoacoustics is the one has to bend under the marketing agenda !
- Elias
Sorry Tom but this doesn't answer the question. I'm well aware of this and the surrounding literature:
An externally hosted image should be here but it was not working when we last tested it.
Actually there is no explanation why we perceive a single phantom image instead of two separate events.
Because VER increase ASW, and ASW is the key to realistic image. Without large ASW the sound is coming from two singular near distance points the two tweeters, or in your case the throaths of the two horns or "waveguides", in the freq range dominant to spatial hearing. That would be the most unrealistic case.
As pointed out by David S. that leads to image of artificial type.
Since I did get that, I haven't had to suffer from artificial images any more but they turned into more realistic.
- Elias
markus wrote;
“Actually there is no explanation why we perceive a single phantom image instead of two separate events.”
Well perhaps you didn’t like my explanation, but that was one. It is why a rose smells like a rose, it is partly the arrangement of molecules reaching your olfactory organ AND it is what you already know about how roses smell. Your ears are the same way, hearing works based on the signals reaching our two ears AND what we have learned throughout our lives about how things sound coming from different directions. In the case of headphones most of those cues are not present and so the image has an in head quality as your ears don’t hear the direction from normal cues.
Best,
Tom Danley
Danley Sound Labs
“Actually there is no explanation why we perceive a single phantom image instead of two separate events.”
Well perhaps you didn’t like my explanation, but that was one. It is why a rose smells like a rose, it is partly the arrangement of molecules reaching your olfactory organ AND it is what you already know about how roses smell. Your ears are the same way, hearing works based on the signals reaching our two ears AND what we have learned throughout our lives about how things sound coming from different directions. In the case of headphones most of those cues are not present and so the image has an in head quality as your ears don’t hear the direction from normal cues.
Best,
Tom Danley
Danley Sound Labs
Quite right but that doesn't explain the "why". Here's one of the better attempts: http://www.hauptmikrofon.de/theile/ON_THE_LOCALISATION_english.pdf
Last edited:
Actually, I agree. I had read something and not thought it out all the way. Thanks for the correction.
But the inter-aural crosstalk needs to happen only once. If it happens during recording, then it must not happen again, with different time delays, during playback.
Yes! It should happen only once.
And actually in the beginning it did happen only once, as Blumlein the inventor of stereo, as far as I know, proposed a crossed figure of eight microphone to capture the soundfield in a singular point. No cross talk at the recording stage. But then other smart engineers come along and thought "Hey, what happends if we capture the soundfield at two points in space, you know, human's ears are also not one singular point", and here we go! If aliens would come to observe the earth, they would be surprised to see the variety of recording techniques invented by all the smart engineers. It has left us a complete mess, that is known as stereo. sigh.
- Elias
Molding defect ?
It's not only words from Elias. From an evolutionary point of view, with perspective of selection of the best adapted, perceiving something that doesn't exist (the phantom localisation) drives to a stalemate. We still have this imperfect feature because there is no evolutive pressure on this axis, i.e. mankind has never been confronted to dual sound sources till the apparition of stereo.
To make it clear, lets imagine the same defect for the view : sitting in front of your stereo speakers, you would see only one. Any good?
Or with the famous caveman : progressing carefully in a dark cave, weapon in hand...suddenly a big bear roars in front of him...our guy instinctively hits the beast in the belly. Bad luck, this was not one bear, but two, disposed on the summits of a triangle and roaring together. Our man is dead. Phantom image issue.
It's not only words from Elias. From an evolutionary point of view, with perspective of selection of the best adapted, perceiving something that doesn't exist (the phantom localisation) drives to a stalemate. We still have this imperfect feature because there is no evolutive pressure on this axis, i.e. mankind has never been confronted to dual sound sources till the apparition of stereo.
To make it clear, lets imagine the same defect for the view : sitting in front of your stereo speakers, you would see only one. Any good?
Or with the famous caveman : progressing carefully in a dark cave, weapon in hand...suddenly a big bear roars in front of him...our guy instinctively hits the beast in the belly. Bad luck, this was not one bear, but two, disposed on the summits of a triangle and roaring together. Our man is dead. Phantom image issue.
I responded to 851 in 858.
See my comments on blocking the side wall reflections with fiberglass. I actually feel this is better than comparing different speakers in that we are absolutely sure the direct response is the same for both cases and we can make an instant changeover.
It's a fair comment to say that Toole's tests don't publish extensive data on the variety of designs that the diy forum likes. I don't think that disqualifies his conclusions but I do understand the argument of those that feel that way.
An important point you'll find when you start measuring loudspeakers the only systems that measure well are fairly conventional in construction: 2 and 3 way cone and dome systems. You won't find a line array, an electrostatic or any other exotic technology system that measures as well as a premium conventional system, such as a B&W 801. This used to be true of horns as well but the CD designs of the 80s through now are considerably better and many horn based systems will be close to the best conventional technology systems. I base this on the my own personal measurements, the measurements of Toole, and the vast number of John Atkinson measurements in Stereophile.
High end speakers generally suck.
A rationalization perhaps, but if I were Toole, searching for correlation between measurements and good sound, I'd stop including exotic speakers in the test groups after it became obvious that most sounded as bad as they measured.
There are a few electrostatics in Toole's book other than the Quad ESL63. They have very poor frequency response both in terms of flatness and smoothness. I have measured the Beveridges and seen curves of a number of Martin Logans and Acoustats and Apogees. They all ranged from mediocre to bad. If high directivity were to sound good, these examples wouldn't show it. Poor frequency response would disqualify them.
Personally I think the ESL 63 measures pretty well. It is certainly the only electrostatic I've seen that had half decent response. Whenever it is placed into a test such as this it is couched as a speaker of very good performance but alternative technology or directivity (Salmi used it in a 3 way comparison of loudspeakers of different directivities. He thought they all sounded closely similar in an anechoic chamber but different in their interaction with a live room). Yes, it could have a little better bass extension and be slightly flatter in bass/mid/treble balance but there are many worse speakers out there.
Toole appropriately uses it as an example of a speaker that differs primarily in directivity. He also is careful to split the test into a individual perceptual factors. The Quad was marked down for its imaging in mono tests. When ranked in stereo tests it nearly caught up with the group. Exploring it farther he had the panel rank order the speakers by music type. For pop music the Quad was again ranked fairly low. The assumption is that pop studio recordings contained components of steered mono so the poor performance seen in mono tests came back (for jazz it was top place).
Now you can argue that a directional dipole and a directional forward firing speaker (say a cardioid bass cabinet and CD horn) may both have equal directivity yet give different listener preferences since their polar patterns put energy in different directions (rearward nulls vs. side nulls). I can't exclude that possibility, but if the Quad was less liked due to its high directivity I would think it unlikely that another similarly directional system was better even with a different kind of narrow directivity.
Another way to look at it is that Tooles' tests generally show that wide dispersion is preferred to medium dispersion. How likely is it that narrow dispersion is liked more. That is: medium not so good, wide better, but narrow, on the opposite side of medium, even better. It's possible but it doesn't seem very likely to me.
Once you get to the 200Hz and below region I think all that matters is the in-room frequency response. It possible for a dipole bass system (nulls to the side, or a cardioid bass system (null to the rear) can be put into a particular room and give a better performance than an omni bass system due to fortuitous interaction with room modes. I just don't know if you can generalize that a particular pattern is universally better.
My comment was more that a system with a CD horn crossed in at 1000 Hz isn't a fully constant directivity system. On the other hand, if power response isn't a reliable indicator of quality then no particular d.i. curve is a prerequisite of good sound.
David S.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.