Do you know of any experiments, which compared "flat" reflections to "ugly" reflections in a "musical" context? No noise or clicks please.
I
Yes, the Bech papers I have quoted used speech as a stimulus (and also noise). His first published paper used a simple model of speaker directivity that simply changed level with radiated angle, i.e. all reflections would have the same spectrum. The second time around he used the true measured response of the speaker at the radiation angles that corresponded to the reflections. For high elevation angles where the speaker was largly sucked out at mid frequencies he needed to raise its level to detect coloration.
In that case poor frequency response in that direction reduced degradation by the room.
David S.
I see a lot of people saying that the spectral response of a reflection "must" be flat, as if somehow nature demands it, but as Dave S has said, where is the proof that this is necessarily the case ?
Something that nobody seems to be taking into consideration is that many natural sound sources that we are trying to reproduce are not omni-directional, so why should our speakers be ?
Is a singer a completely omni-directional sound source with the same spectral balance around a full 360 degrees ? Of course not. A guitar strapped to the front of a person ? No. How about a Saxaphone ? No. I could come up with a dozen other examples.
The reality is that many of the common musical instruments (including vocalists) have a natural directivity characteristic that is omni-directional at low frequencies, (if they produce any low frequencies) but becomes progressively more directional at higher frequencies - much like speakers do, albeit not at quite the same rate.
Some instruments are more complex than just a single sound source and a directivity profile of course, for example a grand piano radiates from much more than just the opening, but the fact is even then the high frequencies are focused more in the direction where they reflect off the 45 degree open lid towards the audience.
If you record a vocalist from the front, then play that recording through a speaker that has a "traditional" directivity pattern (at the very least, baffle step transition from bass to midrange, along with a further narrowing in the treble) will that sound closer to an actual singer standing at that location in the room facing towards the listeners, or will an omni-directional speaker sound closer ?
My bet is that the directional speaker will sound closer to reality, because the omni-directional speaker is sending much greater high frequency content back towards the room corners behind it than a real singer standing there would.
(Simple test - turn the real singer to face into the corner and away from the listener - it will sound different, and so will turning the conventional directional speaker around, but by definition the omni won't change)
The reason flat power response sounds wrong is because most real sound sources don't have anything near a flat power response, yet we generally record the "primary" axis of an instrument. (For example we record a singer from the front, not from behind)
Take a recording made on the primary axis of an instrument (that has most of its high frequency content) and play it back through a true omnidirectional speaker and the power response at high frequencies will be much greater than the original sound source would have been if positioned in the same room.
The power response of a directional speaker wont be the same as the original source (since different source instruments will have different directivity characteristics) but it will be a closer approximation.
Quite apart from the direct to reflected ratio and comb filtering arguments, I think a true omni-directional speaker with flat power response (which seems to be the ultimate goal of a few people) is fundamentally flawed and misguided because it doesn't reproduce the fall in power response at high frequencies that most natural instruments and sound sources have, instead aiming for an unnatural flat power response.
It may sound pleasing on certain kinds of music in certain room configurations, if you have a certain taste in music reproduction, but I would argue that this is a case of it being Euphonic rather than accurate.
Something that nobody seems to be taking into consideration is that many natural sound sources that we are trying to reproduce are not omni-directional, so why should our speakers be ?
Is a singer a completely omni-directional sound source with the same spectral balance around a full 360 degrees ? Of course not. A guitar strapped to the front of a person ? No. How about a Saxaphone ? No. I could come up with a dozen other examples.
The reality is that many of the common musical instruments (including vocalists) have a natural directivity characteristic that is omni-directional at low frequencies, (if they produce any low frequencies) but becomes progressively more directional at higher frequencies - much like speakers do, albeit not at quite the same rate.
Some instruments are more complex than just a single sound source and a directivity profile of course, for example a grand piano radiates from much more than just the opening, but the fact is even then the high frequencies are focused more in the direction where they reflect off the 45 degree open lid towards the audience.
If you record a vocalist from the front, then play that recording through a speaker that has a "traditional" directivity pattern (at the very least, baffle step transition from bass to midrange, along with a further narrowing in the treble) will that sound closer to an actual singer standing at that location in the room facing towards the listeners, or will an omni-directional speaker sound closer ?
My bet is that the directional speaker will sound closer to reality, because the omni-directional speaker is sending much greater high frequency content back towards the room corners behind it than a real singer standing there would.
(Simple test - turn the real singer to face into the corner and away from the listener - it will sound different, and so will turning the conventional directional speaker around, but by definition the omni won't change)
The reason flat power response sounds wrong is because most real sound sources don't have anything near a flat power response, yet we generally record the "primary" axis of an instrument. (For example we record a singer from the front, not from behind)
Take a recording made on the primary axis of an instrument (that has most of its high frequency content) and play it back through a true omnidirectional speaker and the power response at high frequencies will be much greater than the original sound source would have been if positioned in the same room.
The power response of a directional speaker wont be the same as the original source (since different source instruments will have different directivity characteristics) but it will be a closer approximation.
Quite apart from the direct to reflected ratio and comb filtering arguments, I think a true omni-directional speaker with flat power response (which seems to be the ultimate goal of a few people) is fundamentally flawed and misguided because it doesn't reproduce the fall in power response at high frequencies that most natural instruments and sound sources have, instead aiming for an unnatural flat power response.
It may sound pleasing on certain kinds of music in certain room configurations, if you have a certain taste in music reproduction, but I would argue that this is a case of it being Euphonic rather than accurate.
Last edited:
"Take a recording made on the primary axis of an instrument (that has most of its high frequency content) and play
it back through a true omnidirectional speaker and the power response at high frequencies will be much greater than
the original sound source would have been if positioned in the same room.
The power response of a directional speaker wont be the same as the original source (since different source
instruments will have different directivity characteristics) but it will be a closer approximation."
Hello Simon,
let me give an example:
Richtwirkung von Musikinstrumenten - Akustik - Schall/Lärm - baunetzwissen.de
Clicking on that picture enlarges diagram.
Holding a violin:
How to Play A String Notes on the Violin Video ? 5min.com
Where does the prominent lobe around 1Khz, which is present in many good instruments point to ?
Where do the other lobes point to, which are important to the instruments timbre ?
In most cases of listening to an orchestra it will not be pointing directly at a listener in the audience.
It is the task of the concert hall architects and acousticians, to provide fusion of the instrument's and singer's sounds,
which differ in directivity patterns, and those usually are complex and very frequency dependent.
Is a loudspeaker wrong in not reproducing a solo violin in front of the home listener in a way to imitate
the dispersion pattern of the instrument ?
This class of instruments e.g. will exhibit a lobed but in gross omni like pattern e.g. :
Cymbal - Wikipedia, the free encyclopedia
I can follow your lines of thought nevertheless and i do not state it being completely wrong.
But if we want a loudspeaker to imitate directivity patterns of instruments and singers, i am
convinced we are in a wrong context, even if claiming to approximate "average directivity patterns"
of natural sound sources.
In doing so, the loudspeaker is always wrong, because every pattern will fail at certain classes of
instruments or in certain recording venue's situations.
We have to accept the loss of information about the soundfield in stereophonic recording.
Trying to re-establish something like spatial structure of the soundfield as a certain
loudspeaker designer thinks "it would be best" always is detrimental.
Loudspeaker directivity has to be discussed in the domain of interfacing the listening room solely,
not in a "recreating the original soundfield" domain IMO.
Sound sources having frequency dependent directivity patterns tell us something about their
distance even outdoors. A male voice e.g. will sound different at say 20 feet than standing
close. If the voice is known, we might even be able to estimate the distance without comparing
the situations. If such effects are exploited - e.g. in a radio play (even artificially) - that
distinct "voice at 20 feet distance" is what is intended and "encoded" on the recording.
The loudspeaker itself does best, if not telling us where it is located as a sound source itself.
Belonging to that task is the ability to hide its "real" distance. The loudspeakers have to project
phantom sources, which may appear at certain distances due to the cues present on the recording
(like said spectrum change with distance or IDT cues at the recording venue).
If seen very dogmatic - which i do not either - only a CD Loudspeaker is able do hide its own
distance to some extent ...
it back through a true omnidirectional speaker and the power response at high frequencies will be much greater than
the original sound source would have been if positioned in the same room.
The power response of a directional speaker wont be the same as the original source (since different source
instruments will have different directivity characteristics) but it will be a closer approximation."
Hello Simon,
let me give an example:
Richtwirkung von Musikinstrumenten - Akustik - Schall/Lärm - baunetzwissen.de
Clicking on that picture enlarges diagram.
Holding a violin:
How to Play A String Notes on the Violin Video ? 5min.com
Where does the prominent lobe around 1Khz, which is present in many good instruments point to ?
Where do the other lobes point to, which are important to the instruments timbre ?
In most cases of listening to an orchestra it will not be pointing directly at a listener in the audience.
It is the task of the concert hall architects and acousticians, to provide fusion of the instrument's and singer's sounds,
which differ in directivity patterns, and those usually are complex and very frequency dependent.
Is a loudspeaker wrong in not reproducing a solo violin in front of the home listener in a way to imitate
the dispersion pattern of the instrument ?
This class of instruments e.g. will exhibit a lobed but in gross omni like pattern e.g. :
Cymbal - Wikipedia, the free encyclopedia
I can follow your lines of thought nevertheless and i do not state it being completely wrong.
But if we want a loudspeaker to imitate directivity patterns of instruments and singers, i am
convinced we are in a wrong context, even if claiming to approximate "average directivity patterns"
of natural sound sources.
In doing so, the loudspeaker is always wrong, because every pattern will fail at certain classes of
instruments or in certain recording venue's situations.
We have to accept the loss of information about the soundfield in stereophonic recording.
Trying to re-establish something like spatial structure of the soundfield as a certain
loudspeaker designer thinks "it would be best" always is detrimental.
Loudspeaker directivity has to be discussed in the domain of interfacing the listening room solely,
not in a "recreating the original soundfield" domain IMO.
Sound sources having frequency dependent directivity patterns tell us something about their
distance even outdoors. A male voice e.g. will sound different at say 20 feet than standing
close. If the voice is known, we might even be able to estimate the distance without comparing
the situations. If such effects are exploited - e.g. in a radio play (even artificially) - that
distinct "voice at 20 feet distance" is what is intended and "encoded" on the recording.
The loudspeaker itself does best, if not telling us where it is located as a sound source itself.
Belonging to that task is the ability to hide its "real" distance. The loudspeakers have to project
phantom sources, which may appear at certain distances due to the cues present on the recording
(like said spectrum change with distance or IDT cues at the recording venue).
If seen very dogmatic - which i do not either - only a CD Loudspeaker is able do hide its own
distance to some extent ...
Last edited:
When I'm listening to a constant directivity horn, all I can hear is tweeter playing at the exact location of the speaker. It will certainly not hide it's own location.
On the other hand constant directivity omni like speaker I cannot localise in room, unless you are abnormally close to it.
Constant directivity has nothing to do with hiding the speaker, there are other more dominant factors taking over.
- Elias
CD enables a loudspeaker to hide its distance at least in free space.
This surely contributes to hiding the loudspeaker distance even
in appropriate listening environment.
Making the speakers "disappear" virtually and let only phantom sources
be there when listening to music surely has a lot more requirements.
Being omni e.g. is none of those, neither is "indirect radiaton" nor
excessive group delay.
This surely contributes to hiding the loudspeaker distance even
in appropriate listening environment.
Making the speakers "disappear" virtually and let only phantom sources
be there when listening to music surely has a lot more requirements.
Being omni e.g. is none of those, neither is "indirect radiaton" nor
excessive group delay.
My experience as well for the first two statements.
The conclusion however is wrong. Directivty, be it "constant" or not, does influence localizing the loudspeaker, but mostly due to the means to achieve that directivity (but also the directivity itself).
And in keeping with 2nd statement - yes, listener distance from source also "plays it's part".
Last edited:
What's the radiation pattern for the "CD"?
IMO (and experience), "CD" or not, the more narrow the directivity, the easier it is to localized the source (regarding both depth and lateral position).
What happens in free space if you measure the on axis frequency
response of a common (non CD) 2 way loudspeaker at different distances ?
What happens if measuring off axis responses compared to on axis response ?
Directivity has no meaning in freespace ?
______________
Maybe i am lacking knowledge of (some) basic concepts, one never knows.
On the other hand i love my prejudices, some of them i love even more than others.
Prejudices give hold and security in a world of rapidly changing views and opinions.
Last edited:
No, it is well founded, at least compared to very unflat reflections. Read Toole.
Read Toole. "Any reflections of those sounds (that) might not spectrally resemble the direct sound (will) degrade perceived sound quality".
You can't undo solid research just by saying it never happened, in order to win an argument on a discussion board.
Toole's bottom line, on the subject of this thread, is "It is difficult to ignore the small benefits (without apparent disbenefits) of using normal forward-firing loudspeakers with wide dispersion, good off-axis behaviour, and allowing the relevant areas of side wall to reflect."
Elias wrote;
“When I'm listening to a constant directivity horn, all I can hear is tweeter playing at the exact location of the speaker. It will certainly not hide it's own location.
On the other hand constant directivity omni like speaker I cannot localise in room, unless you are abnormally close to it.”
A few thoughts and remember I am only telling it like I see it from my point of view and observation.;
If you were to take the omni directional speakers outdoors and played one, you would find you are very easily able to hear what direction the sound is coming from.
Then, if you set up in stereo at the same angle and distances your used to, you will find the stereo image is vastly stronger than indoors.
It will however end at or can only fill in between the two speakers.
On ANY speaker where you can identify a tweeter or midrange source, you are for sure radiating an interference pattern instead of a single source.
A source that radiates a smooth spherical segment, only has clues to it’s direction and not physical depth.
If you ever heard an ESL-63, you know that the image can sound like it is past / beyond the source location because it radiates few clues to it’s physical location.
Same for a tiny fostex on a big flat baffle, it also has a simple radiation pattern over a fairly wide band. If you can easily hear where the source is (distance from you) it can only be because your ears are not getting exactly the same information. This spatial difference can be from many things, diffraction being one of many ways .
We DO NOT hear like a microphone does, we hear from two ears and our brain discards and interprets the inputs into “an image”. How much our brain discards is hard to explain but an easy way to demonstrate it is to do what we were doing at work to evaluate our and our competitors speakers. I will put a link at the bottom that demonstrates what you hear is partly tied to what you see and know.
The DSL speaker generation loss test;
Take a good measurement mic and speaker and add a multi-track recorder. We put these on a tower away from reflections in the poormans anechoic space on a quiet day.
Play music through the speaker with the mic a meter or two away and record the mic signal and test music on separate tracks.
Play the recording of the speaker, through the speaker and then record it’s mic signal. We recorded the music track direct as well as a reality check.
MANY loudspeakers sounded pretty bad on the first pass, the reason is the mic only picked up the signal in one spot, not two so you now hear many things you couldn’t hear live because your hearing system ignored it.
By two generations many speakers sound bad, by three generations all but the best are unlistenable, by four generations even my best passive xover Synergy horns are sounding bad.
There are no measured results but the acoustic caricature of the speaker’s warts are unmistakable.
One can also do this in a room and then you auralize both the speakers warts and room problems as one would expect, a high degree of directivity produces less generation loss, all other things equal.
An important point though, if you have a speaker with directivity even if it’s only up high, one finds the center of the beam is normally the loudest and so there is a benefit from aiming that at the farthest listening position at the opposite side. This way the natural shape of the beam helps to make the SPL more constant as you sit from far right to far left and the sweet spot wider. In commercial sound, this is how a horn can make a near constant spl over a large distance, use the right lobe shape and air at the farthest seat.
Lastly, how we hear is not only not like a microphone but without out our awareness, our brain fills in a great deal and this includes what we already know and what we see. The bewildering results with “blind testing” come into sharper focus when you consider the Mcgurke effect in the link. The ONLY time you hear reality is when your eyes are closed.
From a great documentary about the senses, enjoy
Try The McGurk Effect! - Horizon: Is Seeing Believing? - BBC Two - YouTube
Best,
Tom Danley
Danley Sound Labs
Danley Sound Labs, Inc. | Facebook
“When I'm listening to a constant directivity horn, all I can hear is tweeter playing at the exact location of the speaker. It will certainly not hide it's own location.
On the other hand constant directivity omni like speaker I cannot localise in room, unless you are abnormally close to it.”
A few thoughts and remember I am only telling it like I see it from my point of view and observation.;
If you were to take the omni directional speakers outdoors and played one, you would find you are very easily able to hear what direction the sound is coming from.
Then, if you set up in stereo at the same angle and distances your used to, you will find the stereo image is vastly stronger than indoors.
It will however end at or can only fill in between the two speakers.
On ANY speaker where you can identify a tweeter or midrange source, you are for sure radiating an interference pattern instead of a single source.
A source that radiates a smooth spherical segment, only has clues to it’s direction and not physical depth.
If you ever heard an ESL-63, you know that the image can sound like it is past / beyond the source location because it radiates few clues to it’s physical location.
Same for a tiny fostex on a big flat baffle, it also has a simple radiation pattern over a fairly wide band. If you can easily hear where the source is (distance from you) it can only be because your ears are not getting exactly the same information. This spatial difference can be from many things, diffraction being one of many ways .
We DO NOT hear like a microphone does, we hear from two ears and our brain discards and interprets the inputs into “an image”. How much our brain discards is hard to explain but an easy way to demonstrate it is to do what we were doing at work to evaluate our and our competitors speakers. I will put a link at the bottom that demonstrates what you hear is partly tied to what you see and know.
The DSL speaker generation loss test;
Take a good measurement mic and speaker and add a multi-track recorder. We put these on a tower away from reflections in the poormans anechoic space on a quiet day.
Play music through the speaker with the mic a meter or two away and record the mic signal and test music on separate tracks.
Play the recording of the speaker, through the speaker and then record it’s mic signal. We recorded the music track direct as well as a reality check.
MANY loudspeakers sounded pretty bad on the first pass, the reason is the mic only picked up the signal in one spot, not two so you now hear many things you couldn’t hear live because your hearing system ignored it.
By two generations many speakers sound bad, by three generations all but the best are unlistenable, by four generations even my best passive xover Synergy horns are sounding bad.
There are no measured results but the acoustic caricature of the speaker’s warts are unmistakable.
One can also do this in a room and then you auralize both the speakers warts and room problems as one would expect, a high degree of directivity produces less generation loss, all other things equal.
An important point though, if you have a speaker with directivity even if it’s only up high, one finds the center of the beam is normally the loudest and so there is a benefit from aiming that at the farthest listening position at the opposite side. This way the natural shape of the beam helps to make the SPL more constant as you sit from far right to far left and the sweet spot wider. In commercial sound, this is how a horn can make a near constant spl over a large distance, use the right lobe shape and air at the farthest seat.
Lastly, how we hear is not only not like a microphone but without out our awareness, our brain fills in a great deal and this includes what we already know and what we see. The bewildering results with “blind testing” come into sharper focus when you consider the Mcgurke effect in the link. The ONLY time you hear reality is when your eyes are closed.
From a great documentary about the senses, enjoy
Try The McGurk Effect! - Horizon: Is Seeing Believing? - BBC Two - YouTube
Best,
Tom Danley
Danley Sound Labs
Danley Sound Labs, Inc. | Facebook
As said, Toole has a limited view on the subject./QUOTE]
EXTREMELY limited.
FRUSTRATINGLY limited.
UNACCEPTABLY limited.
No my conclusion is right, which is specifically about constant directivity, not just any directivity. The point is, if it can be missed, constant directivity does not determine if a speaker can hide itself or not. Directivity of course is the key, as you share the same experience, but it's just about high or low directivity that matters, not that it needs to be constant.
- Elias
- Elias
My experience as well for the first two statements.
The conclusion however is wrong. Directivty, be it "constant" or not, does influence localizing the loudspeaker, but mostly due to the means to achieve that directivity (but also the directivity itself).
"Constant" Directivity is a subset of Directivity.
Your conclusion:
"Constant directivity has nothing to do with hiding the speaker, there are other more dominant factors taking over."
In that a "constant" form of directivity is form of directivity - your conclusion is wrong.
..but yeah, I get what you intended.
I've done this before (on a putting green well away from trees). Both with directional and "omni" loudspeakers.
In every case the more directional speaker was easier to localize.
Note though that this was at a distance more similar to home listening (i.e. close proximity of about 9-10 feet).
The differences were even more apparent in mono than in stereo.
This shouldn't come as any surprise though. We do use difference checking from one ear and the other and can detect minute changes in pressure (particularly at higher freq.s) at various angles.
As much as I admire the work of Toole there are some items he vacilates on. He gives the oppinion that power response is important but then admits that it is smoothness of power response rather than a particular shape, that matters. He makes a big deal about the Early Reflection average (an average of lateral and vertical response points that represent typical early reflection angles) but he doesn't show that a particular curve is required. With some sophisticated computer processing his team was able to take multiple measurements and predict an in room response but not to prove that it was an important curve. Again the off axis averages are about smoothness because lack of it is revealing of universal resonance. In the end on axis flatness and smoothness always factor highest in his listening tests.
Note that the early reflection average is just an average and wouldn't represent a real response at any point where the various delays of each reflection will determine how they really add at a point in space. Again, flat reflected energy won't guarantee a flat summed response.
The solid research I rely on would be the work of Kates and Salmi and others that have shown that perceived response is determined by direct sound and subsequent reflections that fall within the ears integrating window, that the integrating window is long at low frequencies and short at high frequencies. Based on that, the exact response of every reflection is immaterial if the integrated energy hits the right balance.
I believe that having flat reflection energy is one of those audiophile ideals. Must be a good thin! Can't hurt can it? I haven't seen any research to back that up and I don't think that Toole provides any.
David S.
FWIW, the speakers that I've heard disappear the most were CD and narrow~70 degrees above 1.5Khz and cross-fired in front of me. My speakers now are considerably more broad patterned, but they are still toed in fairly tight(not all the way to 45 degrees though). Disappear well and image well. No gaps up front.
In the end there are too many variables for absolutely optimal. Narrow will be the most adjustable so arguably the most optimal. It can go from spacious to dry in virtually and typical room. Stick'em on lazy susan bearings and adjust to taste.
Dan
In the end there are too many variables for absolutely optimal. Narrow will be the most adjustable so arguably the most optimal. It can go from spacious to dry in virtually and typical room. Stick'em on lazy susan bearings and adjust to taste.
Dan
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.