The eye pays little attention to the time domain, which makes it possible to use time-redundancy to lossy-compress video (portions of video frames with no motion in them are discarded with little to no apparent loss). We're not going to notice that the grain structure of the sky isn't changing; visually, it might even be an improvement.
For the ear, time-delayed reflections are translated into distance and space, a very different system than the eye uses to estimate distance. The ear hears into things, while the eye looks at surfaces.
For the ear, time-delayed reflections are translated into distance and space, a very different system than the eye uses to estimate distance. The ear hears into things, while the eye looks at surfaces.
This leads to a fairly simple conclusion:
You can not use one single measurement to determine the quality of a speaker.
I can think of about ten parameters that are important for loudspeakers. Part of the reason loudspeakers are difficult to design is the requirement to cover nearly three decades; a three-decade bandwidth is difficult enough for an RF antenna without bandswitching, much less a transducer that also has to have reasonably low distortion.
OK, now we're agreed on the need for a constellation of parameters, how should they be weighted? Now we're back in the subjectivist soup again, since different listeners may well have different perceptions, and almost certainly will have different musical tastes.
I think I can agree that an ideal monaural loudspeaker should be as close reciprocal of a perfect microphone as possible
but it is not enough - because what kind of a microphone and what kind of a speaker? omni? dipole? cardioid? ?
furthermore - in stereo/multichannel - what kind of microphone array and what kind of an equivalent loudspeaker array?
moreover - where is the listener in all of it? what exactly are we trying to get for him? are we trying to put the listener as if in the original soundfield in place of the microphone array as in case of dummy head recordings for headphone binaural? or what are we trying to do?
in case of conventional loudspeaker stereo triangle it seems that we simply do not know at all what we are doing - it is just using some old tricks that happen to work somehow creating some kind of picture - somewhat more or less realistic for some, not so for others
there is no science behind this technology - it is just hit-and-miss, a chimpanzee technology
Last edited:
"putting the science back into speakers" is a meaningless phrase if applied this broad. i am sure professionals like tom go at their business with quite a scientific approach, so in this case nothing needs to be put back. in other cases, it boils down to the individual taste. i figured out what i like in audio in a quite long process of build/discard and could not be be less interested if science told me what i like isnt the right thing. and companies that make money with consumer stuff, they would probably only bother if the science increased their money spend per money earned ratio. in the end, science is allready where it needs to be.
A thing that I find is that my reactions to music are almost entirely visceral and emotional, the intellect has very little to do with it, and that's why I like it.
I suspect it is because of this that a lot people think that there must be some sort of ghost in the recording and reproduction equipment, when as a scientist all that I can say is that we have never found one and that it is the performance and the composition etc.where the ghost lies, and not in the machines.
There is of course a huge industry that has sprung up by taking advantage of this belief and many pundits willing to pay lip service to it.
In fact it has grown so large that some people now contend that it must be true because so many people seem to believe it.
A great many people also believe that the earth is flat and six thousand years old, so does this mean that it is in the places where this is the consensus?
The commercial imperative is what drives the audio industry and when you have such a situation a science does dominate, and that is marketing psychology.
The guru of the market psychologists is B.F. Skinner the inventor of behaviorism.
He said that since we are all being conditioned all the time anyway it was incumbent upon the powers that be to see that we were conditioned in the right way, and foe him the right way is consumer capitalism, and the marketers concur. Marketers use this as their rational, and have spent vast amounts of money in research into the subject, and it does work, it is also true that most people think that it doesn't work on them.
The audio industry only wants this particular science in loudspeakers because they sell more of them that way, Bose for instance employs hoards of science Phd' s and there consumer outlets are a tour de force in marketing psychology, they use the physical sciences to design the cheapest ones that give the most profit margin, and then psychology to sell them, and they are only one of the more glaring examples.
I would suggest that this is a bit of science that needs to be taken out and not put in, but as we all know there is fat chance of that.
Anyway I now climb down from my soap box and await comments with baited breath.
rcw
I suspect it is because of this that a lot people think that there must be some sort of ghost in the recording and reproduction equipment, when as a scientist all that I can say is that we have never found one and that it is the performance and the composition etc.where the ghost lies, and not in the machines.
There is of course a huge industry that has sprung up by taking advantage of this belief and many pundits willing to pay lip service to it.
In fact it has grown so large that some people now contend that it must be true because so many people seem to believe it.
A great many people also believe that the earth is flat and six thousand years old, so does this mean that it is in the places where this is the consensus?
The commercial imperative is what drives the audio industry and when you have such a situation a science does dominate, and that is marketing psychology.
The guru of the market psychologists is B.F. Skinner the inventor of behaviorism.
He said that since we are all being conditioned all the time anyway it was incumbent upon the powers that be to see that we were conditioned in the right way, and foe him the right way is consumer capitalism, and the marketers concur. Marketers use this as their rational, and have spent vast amounts of money in research into the subject, and it does work, it is also true that most people think that it doesn't work on them.
The audio industry only wants this particular science in loudspeakers because they sell more of them that way, Bose for instance employs hoards of science Phd' s and there consumer outlets are a tour de force in marketing psychology, they use the physical sciences to design the cheapest ones that give the most profit margin, and then psychology to sell them, and they are only one of the more glaring examples.
I would suggest that this is a bit of science that needs to be taken out and not put in, but as we all know there is fat chance of that.
Anyway I now climb down from my soap box and await comments with baited breath.
rcw
rcw, I think your comments are a breath of fresh air. Feel free to exhale now. 
Science is just a tool. Like a paint brush. When someone can use a tool very well, we call them an artist. There's no dichotomy here, just different levels of skill. If someone claims to use "more science" in designing loudspeakers, it's either an admission of a lack of skill, or a marketing tool to snare the unwary.
Good speakers speak for themselves.
Do I need to quote the old ads from Edison where listeners claimed the Victrola sounded "just like the performance" (circa 1910)? Loudspeakers surpassed recording techniques back in the '60s. Now it's just a question of matching the recording to a speaker that makes it sound the best. The search for "one speaker for all" is silly, IMO, and the goal of "high fidelity" is long outdated.
Science is just a tool. Like a paint brush. When someone can use a tool very well, we call them an artist. There's no dichotomy here, just different levels of skill. If someone claims to use "more science" in designing loudspeakers, it's either an admission of a lack of skill, or a marketing tool to snare the unwary.
Good speakers speak for themselves.
Do I need to quote the old ads from Edison where listeners claimed the Victrola sounded "just like the performance" (circa 1910)? Loudspeakers surpassed recording techniques back in the '60s. Now it's just a question of matching the recording to a speaker that makes it sound the best. The search for "one speaker for all" is silly, IMO, and the goal of "high fidelity" is long outdated.
Last edited:
I take Toole to mean the outward appearance of the loudspeaker package. A little like the plate photographed in "Super Duper Great Cuisine" ... the plate full of wax art. This follows with your psychological/marketing theory pretty well.
Hi rcw, Lynn, all
Actually I think the issue can be clearer than it seems .
While microphones can often have a “pattern” measurement microphones do not, they are omni directional and capture / convert the pressure at that point in space to a voltage signal.
Also, while capturing a stereo image itself is very problematic, that is not the case with a mono system and here a single measurement mic can make a VERY realistic capture of an acoustic event and one has the advantage in making that kind of recording at home it that there is no processing. Best of all and unlike commercial recordings, you were there for the original event. Commercial stereo recordings are normally “created” at the mixdown stage and reflect the skill of the engineering producing a convincing / believable “stereo image”.
The counterpart of a measurement mics lack of directivity would be listening to a loudspeaker outdoors where there are no room reflections, a loudspeakers directivity is essentially irrelevant when there is no room, all you hear, the only part that reaches your ears is the direct radiation.
In that simplest of all configurations, a generation loss recording using a loudspeaker and measurement microphone clearly demonstrate / auralize a loudspeakers flaws and “what is not accurate” is what limits the number of generations one can go before it sounds bad.
While any electronic component can tolerate many generations, a loudspeaker is vastly more inaccurate and even very good ones will only pass 2-4 generations before being lame, some sound bad on the first pass..
Rather than Heisenberg or Einstein, I believe the late Dick Heyser had one of the best views of the situation. He believed that our frames of reference are something we apply to make it easier for us to understand, nature has no frames of reference. For him, sound was a 3 dimensional pressure gradient which propagated at the local speed of sound in the directions the gradient describes.
Hi Lynn, THD made perfect sense in radio where any signals outside the assigned frequency were “bad news”, THD does not reflect our ears sensitivity to the distortion as we hear it where auditory masking makes distortion close to the fundamental very hard to hear while high orders much more audible.
Add to that our ears frequency response which is not flat, the fact that music is not a single note but a spectrum and one sees why THD % is not very useful. For example mid band, 10% 2nd harmonic is inaudible even with a single pure tone and even less so with music which is mostly even harmonically related in it’s content...
So far as square waves, most speakers can produce a square wave if you search for the right frequency and right microphone location. That is not the same thing as a speaker reproducing a square wave unilaterally over a broad bandwidth, over a wide variety of mic locations which has been my focus (preserving time).
I have not found “all pass” filters to be useful, any “named” crossover slopes above first order produce an “all pass” phase response at summation where the lower frequency is delayed relative the higher portion and so what would be really useful is the inverse of an all pass filter which unfortunately can’t be made with passive parts.
This is a time dispersion problem, a single impulse input produces a longer lasting impulse at the microphone, make the upper and lower source in different locations then you also have a variable path length from each source making the sum position dependant as well..
What is required for a generation loss recording is a loudspeaker which does not disperse a single impulsive signal in time, that does not add or subtract from the amplitude of the input signal, that tracks the dynamic nature of the input signal and one which radiates a simple pattern that has a high degree of similarity in a variety of physical locations. To the degree it preserves the signal, it can produce a tolerable signal over more generations. Compared to any part of the electronic chain, loudspeakers are not faithful , are the weakest link by far.
The eye / ear / brain perception system can’t be ignored either, what we “hear” is strongly related to what we see and already know. Google the McGurke effect for nice examples and a demonstration of why we only hear reality with our eyes closed.
I think Dirk makes an important point about creating an illusion of a real event. There are a number of parts of that task such as how to capture a stereo image.
The point of reducing that to a simpler situation with a single measurement microphone, one that encompasses the speakers root ability to preserve / reproduce the input signal is what one has when you make the generation loss recordings I am talking about.
Once a loudspeaker can preserve / reproduce the signal to a high degree, it does sound more realistic. Once one has a loudspeaker that radiates a simple envelope, the clues that identify the speakers location in depth are also removed and that has a powerful positive effect on the stereo image. For example, a hifi speaker like a Quad esl-63 radiates a spherical patch and so has very little source location identity, the sound often sounds like it’s coming from behind the speaker, being able to “hear” where a mid and hf driver is, is the inverse of that ideal of not hearing the depth location source itself..
That ability to locate the source depth continues on when you have stereo instead of a mono phantom image, one has a mono phantom image plus a tight and left source. If the speaker radiates a simple balloon (not an interference pattern from multiple sources) the mono phantom image is much more real , there is no sense of a right and left source even though they are playing an identical signal. While a small full range driver has other flaws and limitations, it can produce that kind of radiation pattern over a fairly wide band.
Best,
Tom
Actually I think the issue can be clearer than it seems .
While microphones can often have a “pattern” measurement microphones do not, they are omni directional and capture / convert the pressure at that point in space to a voltage signal.
Also, while capturing a stereo image itself is very problematic, that is not the case with a mono system and here a single measurement mic can make a VERY realistic capture of an acoustic event and one has the advantage in making that kind of recording at home it that there is no processing. Best of all and unlike commercial recordings, you were there for the original event. Commercial stereo recordings are normally “created” at the mixdown stage and reflect the skill of the engineering producing a convincing / believable “stereo image”.
The counterpart of a measurement mics lack of directivity would be listening to a loudspeaker outdoors where there are no room reflections, a loudspeakers directivity is essentially irrelevant when there is no room, all you hear, the only part that reaches your ears is the direct radiation.
In that simplest of all configurations, a generation loss recording using a loudspeaker and measurement microphone clearly demonstrate / auralize a loudspeakers flaws and “what is not accurate” is what limits the number of generations one can go before it sounds bad.
While any electronic component can tolerate many generations, a loudspeaker is vastly more inaccurate and even very good ones will only pass 2-4 generations before being lame, some sound bad on the first pass..
Rather than Heisenberg or Einstein, I believe the late Dick Heyser had one of the best views of the situation. He believed that our frames of reference are something we apply to make it easier for us to understand, nature has no frames of reference. For him, sound was a 3 dimensional pressure gradient which propagated at the local speed of sound in the directions the gradient describes.
Hi Lynn, THD made perfect sense in radio where any signals outside the assigned frequency were “bad news”, THD does not reflect our ears sensitivity to the distortion as we hear it where auditory masking makes distortion close to the fundamental very hard to hear while high orders much more audible.
Add to that our ears frequency response which is not flat, the fact that music is not a single note but a spectrum and one sees why THD % is not very useful. For example mid band, 10% 2nd harmonic is inaudible even with a single pure tone and even less so with music which is mostly even harmonically related in it’s content...
So far as square waves, most speakers can produce a square wave if you search for the right frequency and right microphone location. That is not the same thing as a speaker reproducing a square wave unilaterally over a broad bandwidth, over a wide variety of mic locations which has been my focus (preserving time).
I have not found “all pass” filters to be useful, any “named” crossover slopes above first order produce an “all pass” phase response at summation where the lower frequency is delayed relative the higher portion and so what would be really useful is the inverse of an all pass filter which unfortunately can’t be made with passive parts.
This is a time dispersion problem, a single impulse input produces a longer lasting impulse at the microphone, make the upper and lower source in different locations then you also have a variable path length from each source making the sum position dependant as well..
What is required for a generation loss recording is a loudspeaker which does not disperse a single impulsive signal in time, that does not add or subtract from the amplitude of the input signal, that tracks the dynamic nature of the input signal and one which radiates a simple pattern that has a high degree of similarity in a variety of physical locations. To the degree it preserves the signal, it can produce a tolerable signal over more generations. Compared to any part of the electronic chain, loudspeakers are not faithful , are the weakest link by far.
The eye / ear / brain perception system can’t be ignored either, what we “hear” is strongly related to what we see and already know. Google the McGurke effect for nice examples and a demonstration of why we only hear reality with our eyes closed.
I think Dirk makes an important point about creating an illusion of a real event. There are a number of parts of that task such as how to capture a stereo image.
The point of reducing that to a simpler situation with a single measurement microphone, one that encompasses the speakers root ability to preserve / reproduce the input signal is what one has when you make the generation loss recordings I am talking about.
Once a loudspeaker can preserve / reproduce the signal to a high degree, it does sound more realistic. Once one has a loudspeaker that radiates a simple envelope, the clues that identify the speakers location in depth are also removed and that has a powerful positive effect on the stereo image. For example, a hifi speaker like a Quad esl-63 radiates a spherical patch and so has very little source location identity, the sound often sounds like it’s coming from behind the speaker, being able to “hear” where a mid and hf driver is, is the inverse of that ideal of not hearing the depth location source itself..
That ability to locate the source depth continues on when you have stereo instead of a mono phantom image, one has a mono phantom image plus a tight and left source. If the speaker radiates a simple balloon (not an interference pattern from multiple sources) the mono phantom image is much more real , there is no sense of a right and left source even though they are playing an identical signal. While a small full range driver has other flaws and limitations, it can produce that kind of radiation pattern over a fairly wide band.
Best,
Tom
RCW- I love the anarchic soapbox rant! psuedo science or high school physics has been perverted and dumbed down strategically to market to the more 'general' public. Maybe science is a poor term. I for one WOULD like to see better engineering in the DIYer scene. To almost echo Sretens comments, the membrane to surround to frame juncture could use better engineering in many cases. Cost saving again.
Tom Danley,
I appreciate your argument but find much to disagree on there. Using a measurement microphone to define and capture the true requirement is just not possible to any great extent. Yes we can move that microphone, or better yet place a loudspeaker outside, that is discounting the ground plane for now and put it on a turntable and do polar response to try and measure what is happening. But musical instruments and even the human voice are not what I would want to think of as point source producers. Perhaps at specific frequencies but not on a broad-band scale. So whether a direct radiator, electrostatic, waveguide or any other form we are not reproducing the original sound using the same type of radiation pattern. Perhaps a plasma driver comes closer, but even that does not have the same distributed radiation as say a drum or tuba and the sound coming from distributed surfaces on the original producer. So unless we can reproduce that radiation pattern we are only capable of simulating as close a analogous system as we can physically create. A pulsating sphere would we closer to reality but I haven't seen one of those yet. Baring the MBL attempt we haven't gotten there yet.
Steven
I appreciate your argument but find much to disagree on there. Using a measurement microphone to define and capture the true requirement is just not possible to any great extent. Yes we can move that microphone, or better yet place a loudspeaker outside, that is discounting the ground plane for now and put it on a turntable and do polar response to try and measure what is happening. But musical instruments and even the human voice are not what I would want to think of as point source producers. Perhaps at specific frequencies but not on a broad-band scale. So whether a direct radiator, electrostatic, waveguide or any other form we are not reproducing the original sound using the same type of radiation pattern. Perhaps a plasma driver comes closer, but even that does not have the same distributed radiation as say a drum or tuba and the sound coming from distributed surfaces on the original producer. So unless we can reproduce that radiation pattern we are only capable of simulating as close a analogous system as we can physically create. A pulsating sphere would we closer to reality but I haven't seen one of those yet. Baring the MBL attempt we haven't gotten there yet.
Steven
actually in a reverberant space (99% of existing listening rooms) pulsating sphere is not good at all
the Beveridge solution - a pulsating cylinder acoustically integrated into a side wall - is much better, here is (concisely) explained why: white_paper
a flooder that is an extremely short, upfiring and (sufficiently) coincident speaker (and which can be seen as an extension of the idea behind Snell "Type 1") in the side wall placement comes close enough to this ideal (uniform 180 degrees lateral dispersion and no audibly significant floor reflection)
Last edited:
it is indeed very important for spatial realism, this and low IACC
a flooder (see previous post for description) - even a single one in a mono configuration - has as little location identity as possible, it just literally disappears (for a listener's position in the reverberant field that is for a normal position in a normal listening room) and the Beveridge placement leads to low IACC
ps. actually a single stereo speaker (stereo bipole with direct sound blocked) achieves basically the same as in this case floor reflections localize the sidewalls as THE "sound sources" which is perfectly coherent with our perception of the room as a whole - "there is a wall - there should be a reflection"
Last edited:
On the other hand, how does a flooder relate to science ? As was revealed, it is a concept discovered by accident by a music lover. Doesn't sound very scientific
on the third hand
a scientific explanation can be given for it - as to why exactly it is better* and why conventional stereo is worse - the answer can be found in the books (Toole and the rest)same applies to SSS, doesn't it?
*ps. "better" for music reproduction at home
I sincerely believe that music reproduction at home and PA/sound reinforcement are really two separate worlds, environments, conditions, goals - everything is fundamentally different
Last edited:
Hi Kindhornman
They used to say if everyone agreed, then no one is thinking very hard.
In that vein, indulge my explanation for a bit and consider conducting the experiment I suggest.
If you set up a stereo outdoors in a configuration similar to what one has in the living room, one finds that in addition to anemic bass, the stereo imaging is normally much improved over the typical “in room” performance.
The room reflections DO NOT help preserve the stereo image, they harm it, they reduce the size of the near field where the direct sound dominates and are signals which compete with the recorded / direct energy.
Aside from an excuse to also have a bbq, this outdoor arrangement is something everyone interested in stereo should try and experience. I know everyone has a listening space indoors and that is not how we use the system, but it does demonstrate how room effects perturb / harm the stereo image, it is a point of reference as the only easy way to experience “your system” “without” the normal reflections present in a room.
Also, the way a musical instrument radiates, it’s radiation balloon is not what we are trying to reproduce, our ears only occupy two points in space and as a result we cannot hear or be effected by sound radiated off axis, to the sides or rear in that outdoor condition.
The lack of reflected or off axis energy reaching ones ears do not harm your ability to hear those instruments outdoors either and loudspeakers that have attempted to re-create an instruments radiation balloon have been failures, a violin or guitar body makes a great source but a lousy loudspeaker.
In fact, I believe what we want as a reproducer (and not producer) is a speaker which has the minimum of it’s own properties.
How do we hear?
Well, leaving out the physiology, we hear the left / right planes based on the amplitude AND time phase of the arrivals to each ear. Our only auditory connection to the outside world has come through our hearing / brain mechanism which assembles the single mental acoustic image out of those two inputs. So, we know that a sound coming from our right arrives first at the right ear and your head being an obstruction and the distance from right and left ears produces a time delay / phase shift AND low pass filtering (the head as an obstruction).
With two points of reference you cannot measure more than one plane, left right so how do we hear up / down? Your outer ears alter the frequency response that gets inside, in fact if you measure your ears Pina response, you find it’s a different looking curve at any angle the sound arrives.
For a speaker guy, inner ear measurements usually look like something is very wrong with whatever it is.
In reality, we know nothing else, we cannot hear any of those notches and changes in response shape because we have no other from a of reference. In fact all those measurement anomalies being the only thing we know, are exactly the same things that allow us to hear how high the sound source is or if it’s behind you.
Under some conditions, we can “hear” how far away a source of sound is too, this triangulation depends on differences in the signals reaching the right and left ears.
If one presents an identical signal to the right and left ears (lets say from straight ahead), the sound is obviously from straight ahead BUT you have no idea how far away it is.
When the speaker radiates no clues about its location in space, then how it was recorded governs how far away it sounds, if the highs are rolled off or some reverberation, then it sounds far away.
If it is recorded dry (nothing but a close mic) then it sounds close. When the loudspeakers radiates it’s own spatial clues, then that image competes with the recorded one.
It is those same spatial clues that make the right and left speaker audible as the source when a mono signal is fed to the system, without spatial clues, you only have a sound floating in front of you, the mono phantom.
So far as loudspeaker radiation, it is the small scale texture in the forward radiation pattern that conveys those clues, that provides the difference between right and left your ears need to triangulate..
A Speaker with a lot of source identity would tend to have a very rough texture if one viewed the front of the radiation balloon (or high resolution polar plot) while a “smooth” texture has none.
That is why some speakers like the ESL-63, the Manger and a small full range driver and very few other hifi speakers have the property of not having the location of the source depth be immediately obvious.
Except for all the other problems, a tiny single source or “other way” of radiating a portion of a sphere and a floor to ceiling homogeneous line source are the only forms of radiation which do not convey that complexity. In the perfect case, a perfect point source, perfect infinite plane wave and infinite length perfect line source are the only radiation forms which do not carry information about the source.
The complexity we don’t want for this comes from many things, edge diffraction and direct radiator sources which are too large to radiate simply and others. All of these things can radiate and produce the interference pattern that lets you hear it’s location in depth. In hifi, they often think of two drivers adding at crossover like two signals adding through resistors. While that idea is valid, it only happens when the two sources are less than about ¼ wavelength apart.
At a significantly greater distance, they radiate independently and produce a coarse interference pattern visible in the polar pattern as lobes and nulls. When you can hear there is a tweeter, a mid and low source in a 3 way speaker, you can hear that due to the independently radiating but interfering sources.
If you want to have a loudspeakers that produce a mono phantom image and not be obvious sources on the right and left, they cannot radiate very much source identity.
Oh yeah, the normal stuff too, they have to have a pretty flat or somewhat downward tipped frequency response, have low distortion, shouldn’t spread out the signal in time, ideally reproduce the input signal as a radiated pressure.
A demonstration of what I am saying is possible if you’re willing to experiment a little.
Obtain a pair of small full range drivers like some Fostex offer.
Mount them on a large flat baffle. Position them in normal stereo configuration but do it much closer than normal so that you are not close to side walls and making the SPL requirement easier. EQ these if you can but you don’t need it to hear this.
Play your favorite stereo recordings and enjoy.
These will radiate a simple hemispherical radiation pattern up to fairly high the combination of being small and large flat baffle prevents them from radiating an interference pattern to a few kHz or above.
Remember what I am talking about is how the speaker radiates and its effect on stereo here and not the limitations of the driver etc.
Best,
Tom
They used to say if everyone agreed, then no one is thinking very hard.
In that vein, indulge my explanation for a bit and consider conducting the experiment I suggest.
If you set up a stereo outdoors in a configuration similar to what one has in the living room, one finds that in addition to anemic bass, the stereo imaging is normally much improved over the typical “in room” performance.
The room reflections DO NOT help preserve the stereo image, they harm it, they reduce the size of the near field where the direct sound dominates and are signals which compete with the recorded / direct energy.
Aside from an excuse to also have a bbq, this outdoor arrangement is something everyone interested in stereo should try and experience. I know everyone has a listening space indoors and that is not how we use the system, but it does demonstrate how room effects perturb / harm the stereo image, it is a point of reference as the only easy way to experience “your system” “without” the normal reflections present in a room.
Also, the way a musical instrument radiates, it’s radiation balloon is not what we are trying to reproduce, our ears only occupy two points in space and as a result we cannot hear or be effected by sound radiated off axis, to the sides or rear in that outdoor condition.
The lack of reflected or off axis energy reaching ones ears do not harm your ability to hear those instruments outdoors either and loudspeakers that have attempted to re-create an instruments radiation balloon have been failures, a violin or guitar body makes a great source but a lousy loudspeaker.
In fact, I believe what we want as a reproducer (and not producer) is a speaker which has the minimum of it’s own properties.
How do we hear?
Well, leaving out the physiology, we hear the left / right planes based on the amplitude AND time phase of the arrivals to each ear. Our only auditory connection to the outside world has come through our hearing / brain mechanism which assembles the single mental acoustic image out of those two inputs. So, we know that a sound coming from our right arrives first at the right ear and your head being an obstruction and the distance from right and left ears produces a time delay / phase shift AND low pass filtering (the head as an obstruction).
With two points of reference you cannot measure more than one plane, left right so how do we hear up / down? Your outer ears alter the frequency response that gets inside, in fact if you measure your ears Pina response, you find it’s a different looking curve at any angle the sound arrives.
For a speaker guy, inner ear measurements usually look like something is very wrong with whatever it is.
In reality, we know nothing else, we cannot hear any of those notches and changes in response shape because we have no other from a of reference. In fact all those measurement anomalies being the only thing we know, are exactly the same things that allow us to hear how high the sound source is or if it’s behind you.
Under some conditions, we can “hear” how far away a source of sound is too, this triangulation depends on differences in the signals reaching the right and left ears.
If one presents an identical signal to the right and left ears (lets say from straight ahead), the sound is obviously from straight ahead BUT you have no idea how far away it is.
When the speaker radiates no clues about its location in space, then how it was recorded governs how far away it sounds, if the highs are rolled off or some reverberation, then it sounds far away.
If it is recorded dry (nothing but a close mic) then it sounds close. When the loudspeakers radiates it’s own spatial clues, then that image competes with the recorded one.
It is those same spatial clues that make the right and left speaker audible as the source when a mono signal is fed to the system, without spatial clues, you only have a sound floating in front of you, the mono phantom.
So far as loudspeaker radiation, it is the small scale texture in the forward radiation pattern that conveys those clues, that provides the difference between right and left your ears need to triangulate..
A Speaker with a lot of source identity would tend to have a very rough texture if one viewed the front of the radiation balloon (or high resolution polar plot) while a “smooth” texture has none.
That is why some speakers like the ESL-63, the Manger and a small full range driver and very few other hifi speakers have the property of not having the location of the source depth be immediately obvious.
Except for all the other problems, a tiny single source or “other way” of radiating a portion of a sphere and a floor to ceiling homogeneous line source are the only forms of radiation which do not convey that complexity. In the perfect case, a perfect point source, perfect infinite plane wave and infinite length perfect line source are the only radiation forms which do not carry information about the source.
The complexity we don’t want for this comes from many things, edge diffraction and direct radiator sources which are too large to radiate simply and others. All of these things can radiate and produce the interference pattern that lets you hear it’s location in depth. In hifi, they often think of two drivers adding at crossover like two signals adding through resistors. While that idea is valid, it only happens when the two sources are less than about ¼ wavelength apart.
At a significantly greater distance, they radiate independently and produce a coarse interference pattern visible in the polar pattern as lobes and nulls. When you can hear there is a tweeter, a mid and low source in a 3 way speaker, you can hear that due to the independently radiating but interfering sources.
If you want to have a loudspeakers that produce a mono phantom image and not be obvious sources on the right and left, they cannot radiate very much source identity.
Oh yeah, the normal stuff too, they have to have a pretty flat or somewhat downward tipped frequency response, have low distortion, shouldn’t spread out the signal in time, ideally reproduce the input signal as a radiated pressure.
A demonstration of what I am saying is possible if you’re willing to experiment a little.
Obtain a pair of small full range drivers like some Fostex offer.
Mount them on a large flat baffle. Position them in normal stereo configuration but do it much closer than normal so that you are not close to side walls and making the SPL requirement easier. EQ these if you can but you don’t need it to hear this.
Play your favorite stereo recordings and enjoy.
These will radiate a simple hemispherical radiation pattern up to fairly high the combination of being small and large flat baffle prevents them from radiating an interference pattern to a few kHz or above.
Remember what I am talking about is how the speaker radiates and its effect on stereo here and not the limitations of the driver etc.
Best,
Tom
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.