Our only chance of perceiving the directivity is in how it interacts with the room. For example, if a speaker is highly directional then it will downplay the reverberent field of a room. Even though natural sounds, such as conversation, stir up a certain amount of room reverberation, the speaker will stir up less.
Headphones are "direct injection" of the recorded sound. There are no local acoustics to stimulate and the directivity of headphone elements are irrelevant. For much of the frequency range the headphone/ear combination is operating in a pressure mode, below the first cavity resonances.
On the other hand, the stereo effect is expanded because crosstalk is nil. This gives a natural stereo panorama that stretches from far left to far right, but unfortunately through the middle of our head rather than out front. This takes most of the sounds of the recording and solidly places them somewhere, so we become much less aware of the headphones as a sound source.
Note that speakers can approach the same imaging. Nearfield listening is the first step, it increasese the direct field and reduces the reverberent field. Sitting between the speakers and aiming them at us at close distance, in my experience, is nearly identical to headphone listening. This also reduces crosstalk and pulls any blended image away from the speakers, but it can run into the same problems as headphones: sound inside the head. I am told that listening to speakers in an anechoic chamber also causes sound in the head.
David S.
In my audio room there's a very high direct to reverberant ratio. Apart from the floor reflection at around -16 dB, there are no reflections higher than about -22 dB. Reverberation time nevertheless is about 0.45 seconds. In this setup I've never experienced in-head localization. However, with speakers outside in the free field (literally!) and listening pretty close to the speakers I have experienced said effect.
Same thing happens when looking at a jpeg format image or mpeg movie : around the edges the sharpness is algorythmically defined .
( depends also on resolution of the visualisation media and from the compression ratio also-crop-ession)
I am also on a sound recording forum and the sad thing is that an increasing number of engineers has taken to panning only hard-left - Center - hard-right.
Almost impossible for me to tell the difference between these and mp3.
Maybe it comes from the experience, that amplitude panning anywhere else does not work very well ?
- Elias
With the high percentage of listening via headphones/earbuds, this seems like a stupid thing to do. Hard left and hard right always sounds unatural/uncomfortable via headphones.
David S.
Hi Keyser, that's interesting because I own and heard many narrow-directivity loudspeaker (econowave, unity,yorkville ..) and they definitely do not dissapear (maybe we need to define 'dissapear'?). They are very accurate and 'sharp' yes but seems to not 'blend' well with the environment.
Dipoles and omni, on the other hand simply seem to stand there doing nothing. In fact most impressive is my little pluto clone (omni), it's like two poles sticking out but not radiate sound!!?
Also.. a few reds help
It does sound unnatural to me but it is covered up a bit by mixing the drums down to stereo and panning those full wide (one stereo channel hard left, the other hard right).
We've practically lost the art of mic placement with the younger generation and good panning seems hell bent on following suit.
I've seen sketches done by Paul McCartney for some Wings songs in which he clearly showed where and how far back he wanted the individual instruments positioned in the final mix. We are losing that part of the art of arrangement and that is a large part of the appeal of good stereo reproduction to me.
We've practically lost the art of mic placement with the younger generation and good panning seems hell bent on following suit.
I've seen sketches done by Paul McCartney for some Wings songs in which he clearly showed where and how far back he wanted the individual instruments positioned in the final mix. We are losing that part of the art of arrangement and that is a large part of the appeal of good stereo reproduction to me.
I think the only bullet proof way to make the disappearing act is to break the convention of speakers being the dominant radiation sources at the listening position. The room is there to help ! Currently I'm using succesfully the room side walls as a projection screens of the stereophonic sound image. It works very well, and the speaker (projector) is not localisable for a stereo signal.
- Elias
- Elias
That is exactly also my experience with any source with high directivity at high freqs. They do not disappear.
Nowadays I've abandoned speakers with high directivity at high freqs as a dead end which cannot be improved any further in this regard.
- Elias
Personally I always found the exact opposite.
The speakers I have with hf dispersion greater than 90deg tend to always be there while the ones 90deg or narrower disappear quite easily.
I also find that a wide dispersion does nothing for my ability to localize different instruments precisely. It all seems to merge into ill-defined 'mush'.
Superficially they appear to 'fill the room' better but it was always at the expense of definition.
The speakers I have with hf dispersion greater than 90deg tend to always be there while the ones 90deg or narrower disappear quite easily.
I also find that a wide dispersion does nothing for my ability to localize different instruments precisely. It all seems to merge into ill-defined 'mush'.
Superficially they appear to 'fill the room' better but it was always at the expense of definition.
But why is the speaker disappearing a requirement? If we bring a vocalist into your room, would they disappear or would you know exactly where the sound came from? I violin? A piano?
Isn't it the job of the recording to remove the sound source from hard left or hard right (and therefore at the speaker)?
David S.
Is this a hard reflection or diffuse reflection? If it is a hard reflection (and the preceeding direct response of the speaker is well suppressed) then you have just replace a sharp localization of the speaker source with a sharp localization of its reflected image. If it is a diffuse reflection you have increased the ASW (acoustic source width) of your new reflected source.
Lots of people like a Bose 901 effect (no disrespect intended) but just as many like sharper imaging.
David S.
Those "bad" habits can result in a funny effect in my dipole system. Usually the stereo stage starts along the front wall (which is 1 m behind the speakers) - extending beyond it. If the stereo mix is hard-panned to extreme left and right for the most part, the center phantom image stays at the front wall, but everything else is lined up along a half circle reaching to the left and right speaker position. I have recordings, where an additional effect (a cymbal for instance) has been added afterwards into a "good stereo" recording. If this effect is hard-panned to one channel exclusively, it will stick out as a strange artefact 1 m in front of the stereo stage.
Rudolf
Elias wrote;
Originally Posted by speaker dave
When the recording is dead and the speaker is directional, then the speaker becomes very obvious as a source.
But from the other threads we've learnt this should lead to pin point imaging. No ? When the reflective disturbances have been cleared out the phantom image is said to be rock solid. Are you telling now that you too have problems in perceiving phantom images but are perceiving the speakers ?
If by disappear you mean to produce a mono phantom without the right and left sources being noticeable as “sources” in addition, then we are talking about the same thing.
If you are refereeing to the posts I made and the assertions, then you have misunderstood what I was explaining.
If you stood out doors with your eyes closed and a person was talking to you from say 10 feet away, you can easily identify the direction because you have learned what the phase shift and timing means caused by the head / angle relative to the source. You can hear how high the source is in the air because your ears alter the frequency response you hear according to angle and you have learned all that transmogrification is how you judge height, not an acoustic flaw etc. You can tell how far away they are because the hf rolls off and the spectrum changes with distance from a talker, not as hf attenuation.
On the other hand, a recorded voice captures the spectrum and properties of the distance involved so when a loudspeaker reproduces that, the normal clues are replaced by the recorded ones.
To the degree a loudspeaker can present a sound field uniform enough so that what reaches the right and left ear is identical, then, your brain more easily accepts the aural distance captured in the recording. On the other hand, if the speakers radiate enough information for your brain to estimate how far away the source is (eyes closed) then, that source identity will compete with the image your trying to produce.
Fwiw, if you were outside, the directivity wouldn’t matter at all in fact the desirable condition is most easily accomplished with a small full range speaker on a large baffle, that will radiate like a single simple point over much of it’s band. They may be limited at both ends and have limited level and have other issues but the imaging can be stellar and if you roll the highest area (where it is in breakup) they also exhibit the “sound is coming from somewhere behind the driver” effect, often the signature of a simple radiation field.
Horns on the other hand are the hardest to get this effect with unless it covers a very wide band (the reason many hifi folks run horns over a much wider bandwidth than in commercial use). Being larger, they are automatically farther apart than direct radiators when used in a multi-way system. Also, one has to examine what horn shape radiates like a simple point source over a wide band.
When I started on this with the Unity and the later Synergy horns, it only seemed weird to me that this property existed and was so audible and yet nothing in any of the TEF measurements showed anything other than gradually all the sources combined into one time and space.
The microphone is always in one place though, it does not detect any difference between that space and another space a head width away.
It is in the difference between what one speaker projects to both ears, that allows one to localize the source.
While the direction is easy to hear, it is often the clues generated AT the loudspeaker, that also allow you to easily hear how far away it is OR hear them as a right and left source when playing a true mono signal.
I am pretty sure the reasons are what I described, your brain and ears assemble one image from two inputs who’s behavior changes with angle allowing detection of height, behind, up, down detection, something not possible with only two omni reference points. It is a learned system we are completely unaware of.
If you try a couple of the recordings I linked earlier, try them with headphones and consider these are the two front channels of a 5 channel recording. These cover approximately the width of your vision, the other channels take over at the edges of the image here.
It is possible to re-produce these in a living room, at a realistic level (including the fireworks) and have it sound like you were outside in my backyard. The source identity is only one part of the problem, how to capture a real stereo image is no trivial task, one I have been dreaming about /working on for about 40 years. Give a couple of the recordings a shot, remember these are a work in progress and be careful with the fireworks recording. That “sounds” soft because the average level is low but has very large peaks.
A home hifi company has also recently discovered a source has a localizable identity related to the radiation shape. Look up the KEF blade and cut through the hype. While it might not be Constant Directivity or preserve input waveshape or go loud enough for commercial sound like say an SH-50, it does claim to radiate as a simple single point (and may) with benefit to imaging.
Best,
Tom Danley
Danley Sound Labs
.
Originally Posted by speaker dave
When the recording is dead and the speaker is directional, then the speaker becomes very obvious as a source.
But from the other threads we've learnt this should lead to pin point imaging. No ? When the reflective disturbances have been cleared out the phantom image is said to be rock solid. Are you telling now that you too have problems in perceiving phantom images but are perceiving the speakers ?
If by disappear you mean to produce a mono phantom without the right and left sources being noticeable as “sources” in addition, then we are talking about the same thing.
If you are refereeing to the posts I made and the assertions, then you have misunderstood what I was explaining.
If you stood out doors with your eyes closed and a person was talking to you from say 10 feet away, you can easily identify the direction because you have learned what the phase shift and timing means caused by the head / angle relative to the source. You can hear how high the source is in the air because your ears alter the frequency response you hear according to angle and you have learned all that transmogrification is how you judge height, not an acoustic flaw etc. You can tell how far away they are because the hf rolls off and the spectrum changes with distance from a talker, not as hf attenuation.
On the other hand, a recorded voice captures the spectrum and properties of the distance involved so when a loudspeaker reproduces that, the normal clues are replaced by the recorded ones.
To the degree a loudspeaker can present a sound field uniform enough so that what reaches the right and left ear is identical, then, your brain more easily accepts the aural distance captured in the recording. On the other hand, if the speakers radiate enough information for your brain to estimate how far away the source is (eyes closed) then, that source identity will compete with the image your trying to produce.
Fwiw, if you were outside, the directivity wouldn’t matter at all in fact the desirable condition is most easily accomplished with a small full range speaker on a large baffle, that will radiate like a single simple point over much of it’s band. They may be limited at both ends and have limited level and have other issues but the imaging can be stellar and if you roll the highest area (where it is in breakup) they also exhibit the “sound is coming from somewhere behind the driver” effect, often the signature of a simple radiation field.
Horns on the other hand are the hardest to get this effect with unless it covers a very wide band (the reason many hifi folks run horns over a much wider bandwidth than in commercial use). Being larger, they are automatically farther apart than direct radiators when used in a multi-way system. Also, one has to examine what horn shape radiates like a simple point source over a wide band.
When I started on this with the Unity and the later Synergy horns, it only seemed weird to me that this property existed and was so audible and yet nothing in any of the TEF measurements showed anything other than gradually all the sources combined into one time and space.
The microphone is always in one place though, it does not detect any difference between that space and another space a head width away.
It is in the difference between what one speaker projects to both ears, that allows one to localize the source.
While the direction is easy to hear, it is often the clues generated AT the loudspeaker, that also allow you to easily hear how far away it is OR hear them as a right and left source when playing a true mono signal.
I am pretty sure the reasons are what I described, your brain and ears assemble one image from two inputs who’s behavior changes with angle allowing detection of height, behind, up, down detection, something not possible with only two omni reference points. It is a learned system we are completely unaware of.
If you try a couple of the recordings I linked earlier, try them with headphones and consider these are the two front channels of a 5 channel recording. These cover approximately the width of your vision, the other channels take over at the edges of the image here.
It is possible to re-produce these in a living room, at a realistic level (including the fireworks) and have it sound like you were outside in my backyard. The source identity is only one part of the problem, how to capture a real stereo image is no trivial task, one I have been dreaming about /working on for about 40 years. Give a couple of the recordings a shot, remember these are a work in progress and be careful with the fireworks recording. That “sounds” soft because the average level is low but has very large peaks.
A home hifi company has also recently discovered a source has a localizable identity related to the radiation shape. Look up the KEF blade and cut through the hype. While it might not be Constant Directivity or preserve input waveshape or go loud enough for commercial sound like say an SH-50, it does claim to radiate as a simple single point (and may) with benefit to imaging.
Best,
Tom Danley
Danley Sound Labs
.
I think when human perception is included the stereo theory becomes fragile. For example, in the case of center panned stereo signal I can hear different freqs coming from different directions, namely low freqs come from the phantom center position and high freqs come from the two locations which are the tweeters. I may have mentioned this earlier
This is very bad when believable imaging is desired. It does not happen. The Rayleigh's doublex theory of spatial hearing do not explain this. And why it's a problem because stereo is based on the assumption that the doublex theory is the basis of phantom imaging ! It does not work !
It's a hard wall reflection. I have arranged some free space on the side walls. So ray tracing can be used to determine the reflection angles in most of the freq range. However, there is more than one reflection of course ! If there would be only one reflection, then yes the walls would act like virtual single source speakers. I think the time difference between the first and second order side reflections (floor -> side wall and ceiling -> side wall) are small enough to fit inside time window to determine ASW. And yes, ASW is increasing with this concept. Increased ASW only adds for realism in my opinion. I cannot believe in images that appear to come from a small hole in space. With this concept the images fill the whole front wall and partly side walls also. These seems to be height as well in some recordings, but it must be some sort of HRTF trick. It's much better than conventional stereo triangle where sound appears to come from a thin line drawn between the speakers (if they can get out from the speakers in the frst place).
- Elias
I think when human perception is included the stereo theory becomes fragile. For example, in the case of center panned stereo signal I can hear different freqs coming from different directions, namely low freqs come from the phantom center position and high freqs come from the two locations which are the tweeters. I may have mentioned this earlier
.... - Elias
I have some experiences (impressions) which might be similar to the above.
Back when I was fiddling with convetional multiway stereo speakers, in some bad situations of room acoustics or some embarracing stage of xover tuning, the images could be strangely rendered.
For example, the lead vocal in most pop music is at the center. Ideally it should be a well-integrated body, as if a real person is there.
Somehow, in those bad situations, I've heard vocal images which seemed facing sideway. He/she is not singing towards me, but facing elsewhere. This happens when the HF and the lower frequencies are not in good convergence, with slightly more HF on one side, which sheds more 'light' on that side, thus the effect of 'spitting' sounds towards that direction, instead of straight ahead.
This might be caused by imperfection balance between 2 channels, or asymmetry of room acoustics. Usually I managed to fix that problem by matching the speakers and room acoustics more carefully. Then it could return normal and 'precisely' converged, like adjusting a 3 lens projector. However, the 'feel' of precision collapses right away when I leave the sweet spot.
So I think that's the intrinsic flaw of conventional multiway stereo speakers, which are inevitably in non-perfect balance/symmetry in all acoustic aspects in most positions in room. And the interactions between the 2 sources and environment are not the same as the real things. So it can never be natural sounding.
BTW, I did it many times, listening with eyes closed to ordinary talking of families in room, musical instruments played indoor, live concerts... etc. I've never heard pin-point images. The real things (within reasonable distances) always sound bigger then they appear.
A friend of mine once had serious problems with panning/stereo imaging in the studio he worked in.
The imaging would not translate correctly to other systems. They tried everything and even contemplated changing the mixing console.
Eventually they did trace the problem to a rogue asymmetrical hard wall reflection (actually a plastic folder s.o. put in the wrong place). The monitors were Dynaudio Acoustics M2, it would have been much less of a problem if the monitors would have had narrower treble dispersion.
IME wide dispersion requires very careful room treatment and absolute symmetry of the room to image precisely.
The imaging would not translate correctly to other systems. They tried everything and even contemplated changing the mixing console.
Eventually they did trace the problem to a rogue asymmetrical hard wall reflection (actually a plastic folder s.o. put in the wrong place). The monitors were Dynaudio Acoustics M2, it would have been much less of a problem if the monitors would have had narrower treble dispersion.
IME wide dispersion requires very careful room treatment and absolute symmetry of the room to image precisely.
HI Tom - Happy Thanksgiving.
While I think that $30,000 for a pair of decent loudspeakers is consumer gouging, there is much in that design that is appealing. The low diffraction cabinet is something that I have been doing for a long time and I believe that it is critical to the "disappearing act". But so are nearby room reflections and diffractions, which are almost never dealt with.
I saw that they are using the top plate magnet trick on the woofers - a dual magnet with one placed on the top plate. This makes significant improvments in the field structure (wastes a lot of magnet energy doing so of course, so its impractical with Neo these days).
The balanced and connected woofers is a nice technique.
I would still prefer the much lower thermal modulation of a compression driver (read dynamics) to the midrage/tweeter combination that they are using.
Hi Earl
How have you been, hope things are well.
This is kind of a weird concept, if you never heard it, I can understand being skeptical of the idea that a speaker can disappear as a source or the origin of a sound and then produce the strong impression of a phantom .
Its funny, people are aware of diffraction and other transitions near the tweeter and some are aware of the baffle step effect but really, a horn is the ultimate, “fully conscious” way to deal with where the sound goes after you have produced the disturbance.
The shape of yours doing it in a least perturbing way . From that it is easy to imagine that if you could see pressure at an instant in time, that the bubble would be nice and smooth and facing one speaker squarely would present both ears with a very similar/identical information and so not convey the “here I am 8 feet away” localization, just direction..
Hey I meant to reply in that other thread about the noise limit. The fellow there was talking about doing response / room correction and basing it on an impulse measurement. The potential issue i was thinking of is visible if one averages a few measurements and you get a somewhat different lf response than one measurement. It is in part the noise issue that limits how far down the slope (relative to the flat area) you get valid data.
I watched a “pro” do measurements where the trustworthy response ended less than -20 relative to the flat (PSST fella, that subwoofer really doesn’t do that).
If one uses such a measurement to DSP “correct the room”, it does that (usually) with an impulse response…which sometimes has limited dynamic something or other. I mean if you’re going to convolve or use the magnitude and phase, I would think you would want at least 40dB or more of normal slope.
Of course, when you have directivity on your side the room reflection problems become much smaller, placement options a bit larger and you can hear more of the recordings image (if it has one).
Best,
Tom Danley
How have you been, hope things are well.
This is kind of a weird concept, if you never heard it, I can understand being skeptical of the idea that a speaker can disappear as a source or the origin of a sound and then produce the strong impression of a phantom .
Its funny, people are aware of diffraction and other transitions near the tweeter and some are aware of the baffle step effect but really, a horn is the ultimate, “fully conscious” way to deal with where the sound goes after you have produced the disturbance.
The shape of yours doing it in a least perturbing way . From that it is easy to imagine that if you could see pressure at an instant in time, that the bubble would be nice and smooth and facing one speaker squarely would present both ears with a very similar/identical information and so not convey the “here I am 8 feet away” localization, just direction..
Hey I meant to reply in that other thread about the noise limit. The fellow there was talking about doing response / room correction and basing it on an impulse measurement. The potential issue i was thinking of is visible if one averages a few measurements and you get a somewhat different lf response than one measurement. It is in part the noise issue that limits how far down the slope (relative to the flat area) you get valid data.
I watched a “pro” do measurements where the trustworthy response ended less than -20 relative to the flat (PSST fella, that subwoofer really doesn’t do that).
If one uses such a measurement to DSP “correct the room”, it does that (usually) with an impulse response…which sometimes has limited dynamic something or other. I mean if you’re going to convolve or use the magnitude and phase, I would think you would want at least 40dB or more of normal slope.
Of course, when you have directivity on your side the room reflection problems become much smaller, placement options a bit larger and you can hear more of the recordings image (if it has one).
Best,
Tom Danley
I am not skeptical at all. In fact my speakers do this better than any others that I have ever heard, so its certainly doable. I rank speakers on their ability to disappear. Its a critical aspect of the design.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.