Recommended reading:
"Loudspeaker Measurements and Their Relationship to Listener Preferences: Part 1"
J. Audio Eng. Soc., Vol. 34, No. 4, 1986 April
"Loudspeaker Measurements and Their Relationship to Listener Preferences: Part 2"
J. Audio Eng. Soc., Vol. 34, No. 5, 1986 May
Here is a PDF made available (both articles):
https://pearl-hifi.com/06_Lit_Archi...blications/LS_Measurements_Listener_Prefs.pdf
"Loudspeaker Measurements and Their Relationship to Listener Preferences: Part 1"
J. Audio Eng. Soc., Vol. 34, No. 4, 1986 April
"Loudspeaker Measurements and Their Relationship to Listener Preferences: Part 2"
J. Audio Eng. Soc., Vol. 34, No. 5, 1986 May
Here is a PDF made available (both articles):
https://pearl-hifi.com/06_Lit_Archi...blications/LS_Measurements_Listener_Prefs.pdf
A more representative comparison might be to cover a window of typical listening positions plus, possibly, an average although not just an average. This will likely show some of the reasons a random arrangement of a small number of sources is not commonly used. A small number of sources in a specific array can work reasonably. A large number of random sources can also work reasonably well although at the price of the quality of the direct sound. Carlsson, for example, moved away from the Sonab arrangement towards one with a better defined direct sound in order to improve the imaging and perhaps tonal balance (lack reliable informaiton to say). In the examples I have heard this came at the price of the enhanced spaciousness which the Sonab speakers tended to provide.
"Necessity of good off-axis behavior is well-established science fact."
That one is true:
And if you go to the sources, you will find which aspects of "good off-axis behaviour"
(i.e. which concrete correlates of that behaviour) are really found as being decisive for
what is called "listener preference".
If so, that Toole would state FR "similarity between angles", or even the shape of any
single "under angle" FR being decisive, i am glad to learn from you by (re-)
interpreting or just citing the known and established sources.
Last edited:
Important excerpt:
"Listeners, it seems, like the sound of loudspeakers with a fiat, smooth wide band on-axis amplitude response that is maintained at substantial angles off axis."
Your three spaced wideband drivers can not maintain that. Period.
Recommended Reading:
"Method for predicting loudspeaker preference" US8311232B2
https://patents.google.com/patent/US8311232B2/en
Especially Table 13 (Page 16) is interesting IMO:
Model variables are listed with their estimated contribution to
listener preference.
"Method for predicting loudspeaker preference" US8311232B2
https://patents.google.com/patent/US8311232B2/en
Especially Table 13 (Page 16) is interesting IMO:
Model variables are listed with their estimated contribution to
listener preference.
Sure, but obviously that statement is not reflected in the decisive model variables
found:
The decisive model variable regarding "off axis bevaviour" is (sloppy speaking)
"boiling down" to smoothness of sound power over frequency (or smoothness
of DI over frequency, if you wish.)
A speaker achieving this in the way described in your citation may be "liked"
by the listeners, sure.
Edit: When accessed in the highly treated (by diffusers) listening lab at Harman e.g.
(Logic comes in here)
That statement from Toole above does - by no means - mean this being the only
way or even a necessary prerequisite to achieve the preference criteria derived from
the measured data and the listener's rankings.
Similarity (by itself) of FRs under (larger) angles is not found as a decisive criterion.
And let me kindly add:
If it would be, Toole very likely would have found it ...
Last edited:
What has been found and published in the Patent mentioned is this:
With the model variables "according to the meaning":
"similarity under angles" ?
It is just not there. There is no such variable found or confirmed as being decisive or even
contributing to listener preference. It may be just one way to achieve preferred values in
"SM_SP".
But the preference criteria are just that: Criteria for listener preference. Those variables
found are independent from the technology used: The listeners did not even access
"a loudspeaker" but more so the soundfield it created at their listening position in that
particular listening lab.
Right?
With the model variables "according to the meaning":
- average deviation on axis
- low frequency extension
- low frequency quality
- smoothness sound power
- smoothness on axis
"similarity under angles" ?
It is just not there. There is no such variable found or confirmed as being decisive or even
contributing to listener preference. It may be just one way to achieve preferred values in
"SM_SP".
But the preference criteria are just that: Criteria for listener preference. Those variables
found are independent from the technology used: The listeners did not even access
"a loudspeaker" but more so the soundfield it created at their listening position in that
particular listening lab.
Right?
Last edited:
Me thinks,
you are neglecting that "one speaker" (being it the left or the right one) is not the only
loudspeaker radiating sound in a usual stereo setup.
In case of a central phantom source (or mono signal) both speakers radiate the same sound,
thus interfering with each other (constructively or destructively) at different points in space
(listening positions).
Even when neglecting the listening room for a while:
That two speaker arrangement cannot maintain any "similarity of frequency responses under
angles" that might have been (elaborately) constructed into the single speakers standing left
and right.
Well if that "similarity under angles" would be necessary or enjoyful, i would recommend
listening mono using just one loudspeaker.
For Toole's assessments regarding listener preference, there are in fact only single loudspeakers
used with programme in mono. This is because listeners have found to be more differentiated
and reliable in their rankings, when listening this way.
But still - even under these conditions - there is no contribution of "similarity under (larger)
angles" showing up in the results.
What we have (regarding the off axis contribution to listener preference) is "SM_SP" (as
smoothness of the sound power curve over frequency).
And i am sorry: That is what science says up to now (since nearly 40 years now, when
we also read carefully Toole's papers from 1986 i mentioned above).
If anyone sees something in the writing (e.g. from Toole), which is still unpublished(?) in
the results or underlying data, i am fine with that.
But as long as this seems not the case, i will stick with the results published and not with
any interpretations of those results from "outside the field".
Last edited:
It's not there explicitly. It's possible they did not consider it at all, and their analysis does not provide any information re. "similarity under angles" (weird terminology if you aks me). It's also possible they considered it to be covered by the power response (which I think is what they call "smoothness sound power" -- weird terminology!).
Anyways, it just dawned on me* your approach seems to use multiple wide-band drivers at horizonal and vertical offsets on the same baffle, all connected in parallel with no individual xovers/filters. That does sound like a recipe for disaster to me.
*I can be a bit slow sometimes...
@mbrennwa (you wrote)
"It's not there explicitly. It's possible they did not consider it at all, and their analysis does not provide any information re. "
Toole is born 1938 and has spent most of his professional life investigating listener
preference for loudspeakers and their correlates in measurable parameters.
Science is (truly) about being "explicit", especially when publishing results:
You are attributing Toole to "conceal" or "forget" or "overlook" (?) publishing explicitly
a key point in the results of his research over several decades?
Something in his results is assumed to be "implicitly there", which e.g. some members
of this forum have found and understood while he has overlooked it for decades?
Maybe because "so obvious", that he did neither need nor mind to publish that (as a)
key point?
I tend to believe that you are kidding.
"It's not there explicitly. It's possible they did not consider it at all, and their analysis does not provide any information re. "
Toole is born 1938 and has spent most of his professional life investigating listener
preference for loudspeakers and their correlates in measurable parameters.
Science is (truly) about being "explicit", especially when publishing results:
You are attributing Toole to "conceal" or "forget" or "overlook" (?) publishing explicitly
a key point in the results of his research over several decades?
Something in his results is assumed to be "implicitly there", which e.g. some members
of this forum have found and understood while he has overlooked it for decades?
Maybe because "so obvious", that he did neither need nor mind to publish that (as a)
key point?
I tend to believe that you are kidding.
Last edited:
@mbrennwa (you wrote)
"Anyways, it just dawned on me* your approach seems to use multiple wide-band drivers at horizonal and vertical offsets on the same baffle, all connected in parallel with no individual xovers/filters. That does sound like a recipe for disaster to me."
Fine, then we may discuss, where e.g. in the both "on axis" curves "desaster" is showing up when
visiting this post again:
The "desaster in pink" is the kind of desaster one is used to, being it loudspeakers from industry
or from within the DIY community.
The "desaster in green" is at least "suspicious", because there must be something wrong when
making use of "decorrelation" (under certain angles and certain wavelengths).
There must be something wrong even if the on axis FR is obviously very good and virtually free
of diffaction artefacts, which most common loudspeakers simply fail to achieve.
Edit: Well, you can use DSP to "correct" interference issues, if you did not understand Toole
properly and insist on making things even worse.
Ah, here we got it "we need similarity in FR (direct sound) under angles" (that is missing in the
proposal): But who says that? It has just been said, that "science says that".
Fortunately "science" - in the results - does not say anything like that:
Science says - up to now - that it is "SM_SP" that is relevant regarding the off axis response.
"SM_SP" on the other hand, ist also quite good with the concept proposed (and if done right).
"Anyways, it just dawned on me* your approach seems to use multiple wide-band drivers at horizonal and vertical offsets on the same baffle, all connected in parallel with no individual xovers/filters. That does sound like a recipe for disaster to me."
Fine, then we may discuss, where e.g. in the both "on axis" curves "desaster" is showing up when
visiting this post again:
Regarding "issues" with direct - on axis? - sound:
Who has designed and/or built conventional speaker configurations can tell easily "which curve is which" i guess ...
Who has designed and/or built conventional speaker configurations can tell easily "which curve is which" i guess ...
The "desaster in pink" is the kind of desaster one is used to, being it loudspeakers from industry
or from within the DIY community.
The "desaster in green" is at least "suspicious", because there must be something wrong when
making use of "decorrelation" (under certain angles and certain wavelengths).
There must be something wrong even if the on axis FR is obviously very good and virtually free
of diffaction artefacts, which most common loudspeakers simply fail to achieve.
Edit: Well, you can use DSP to "correct" interference issues, if you did not understand Toole
properly and insist on making things even worse.
Ah, here we got it "we need similarity in FR (direct sound) under angles" (that is missing in the
proposal): But who says that? It has just been said, that "science says that".
Fortunately "science" - in the results - does not say anything like that:
Science says - up to now - that it is "SM_SP" that is relevant regarding the off axis response.
"SM_SP" on the other hand, ist also quite good with the concept proposed (and if done right).
Last edited:
Yes, but if it was that easy, why are most loudspeakers built differently?
And:
We want to stick with exactly those errors and misbehaviour of loudspeakers
(in the room) that we are used to.
Remember: Coherence and strong Interference/Combing are sisters regarding
loudspeaker/room interaction.
In fact most loudspeakers do not even "get along" with their own baffles,
and that's where "room interaction" really starts, right?
But:
We do not want any other kind of misbehaviour, even if it would be more
forgiveable in realistic (stereo) listening scenarios/setups.
And:
We want to stick with exactly those errors and misbehaviour of loudspeakers
(in the room) that we are used to.
Remember: Coherence and strong Interference/Combing are sisters regarding
loudspeaker/room interaction.
In fact most loudspeakers do not even "get along" with their own baffles,
and that's where "room interaction" really starts, right?
But:
We do not want any other kind of misbehaviour, even if it would be more
forgiveable in realistic (stereo) listening scenarios/setups.
Last edited:
You are comparing apples and oranges. All loudspeakers which were subjects to Tool's analysis were multi-way loudspeakers (frequency bands divided between drivers with different size and construction), with no two or three drivers sharing the same and whole mid and high spectrum (your case). Your loudspeaker will exhibit comb-filtering off axis. No one of the loudspeakers from the Tool's analysis have comb-filtering issues.
Decisive criterion for pin-point location of the instruments and vocals in the stereo sound stage - do not use two or more speakers outputting the same signal across the whole mid and high frequency band. Even the infamous Tekton tweeter array has band-pass filter.
@Sonce (you wrote)
"No one of the loudspeakers from the Tool's analysis have comb-filtering issues."
You are sure regarding that?
No one (of the Testlings seen by Toole/Harman) had "multiple drivers in same frequency range"?
No one (of the Testlings seen by Toole/Harman) had baffle diffraction issues (interference) leading to "fluctuating" response (on axis, off axis)?
You seem to know even more than you reveal due to that point.
@Sonce (you wrote)
"Decisive criterion for pin-point location of the instruments and vocals in the stereo sound stage - do not use
two or more speakers outputting the same signal across the whole mid and high frequency band. "
But we were talking about "listener preference" (mainly according to Toole/Harman) up to now:
Who is "comparing apples and oranges" here?
Toole does (intentionally) not access "stereo imageing" when investigating listener preference.
Edit: He uses - as already mentioned above - single loudspeakers in mono for that.
Highly rated Loudpeakers in that field usually also excel in stereo reproduction.
Regarding "pin point imageing":
I assume you being aware of stereo - at least common intensity stereophony - simply not working
above (say) 3 Khz regarding mid phantom sources? There is "little to none" contribution of tweeters
in that field.
But ability to "pin point image" (stereo) is something often mentioned, when discussion on different
matters seems difficult (for some reason).
Recommended Reading (stereo, imageing):
"A New Approach to the Assessment of Stereophonic Sound System Performance"
Bennet, Barker , Edeko (1985)
Figure 5 (Page 318) is very interesting in this context.
Here i found a link recently:
http://decoy.iki.fi/dsound/ambisonic/motherlode/source/A new approach to the assessment of stereophonic_Bennett et al_1985.pdf
"No one of the loudspeakers from the Tool's analysis have comb-filtering issues."
You are sure regarding that?
No one (of the Testlings seen by Toole/Harman) had "multiple drivers in same frequency range"?
No one (of the Testlings seen by Toole/Harman) had baffle diffraction issues (interference) leading to "fluctuating" response (on axis, off axis)?
You seem to know even more than you reveal due to that point.
@Sonce (you wrote)
"Decisive criterion for pin-point location of the instruments and vocals in the stereo sound stage - do not use
two or more speakers outputting the same signal across the whole mid and high frequency band. "
But we were talking about "listener preference" (mainly according to Toole/Harman) up to now:
Who is "comparing apples and oranges" here?
Toole does (intentionally) not access "stereo imageing" when investigating listener preference.
Edit: He uses - as already mentioned above - single loudspeakers in mono for that.
Highly rated Loudpeakers in that field usually also excel in stereo reproduction.
Regarding "pin point imageing":
I assume you being aware of stereo - at least common intensity stereophony - simply not working
above (say) 3 Khz regarding mid phantom sources? There is "little to none" contribution of tweeters
in that field.
But ability to "pin point image" (stereo) is something often mentioned, when discussion on different
matters seems difficult (for some reason).
Recommended Reading (stereo, imageing):
"A New Approach to the Assessment of Stereophonic Sound System Performance"
Bennet, Barker , Edeko (1985)
Figure 5 (Page 318) is very interesting in this context.
Here i found a link recently:
http://decoy.iki.fi/dsound/ambisonic/motherlode/source/A new approach to the assessment of stereophonic_Bennett et al_1985.pdf
Last edited:
Yes, I am sure - no one had multiple drivers in the same frequency range, so they didn't have comb-filtering issues.
All of tested loudspeakers by Toole had baffle diffraction issue, more or less. Your loudspeaker have no diffraction on-axis, but have diffraction off-axis because your speaker cabinet has sharp edges.
"Comparing apples to oranges" is not strong enough to describe what you are intending to compare - you are comparing hazelnuts to watermelons!
Excerpt from the above paper: "Each loudspeaker cabinet housed a single type 8P unit produced by Goodmans Loudspeaker Ltd."
So not three fullranges in the cabinet, but only one - single! Also, paper is comparing phantom image of two stereo loudspeakers widely separated, not comparing single loudspeaker with three (or two) fullranges in the cabinet. Your one loudspeaker has issues while it is single, with two of your loudspeakers in stereo configuration - issues will be double!
@Sonce (you wrote)
"Your loudspeaker have no diffraction on-axis, but have diffraction off-axis because your speaker cabinet
has sharp edges."
You got it:
On axis baffle/edge diffraction artefacts are neutralized. The off axis edge diffraction component itself is
also more diffused (more sources, each source creates less sound pressure than a single one had to create,
diffraction patterns more distributed in space and time).
This loudspeaker can truly afford his own baffle edges (even when left unrounded) in contrary to most
common loudspeakers that cannot even afford their own baffle edges, because those act detrimental in
the direct sound already.
I am very happy, you got the "hang of the concept" (sometimes things take a while).
@Sonce (you wrote)
"you are comparing hazelnuts to watermelons!
Excerpt from the above paper: "Each loudspeaker cabinet housed a
single type 8P unit produced by Goodmans Loudspeaker Ltd."
"
I did not mention the Bennet, Barker, Edeko Paper for any "comparison" regarding loudspeakers.
The paper is IMO more about "fundamental limits" of stereophony regarding higher frequencies
(say > 3Khz) and thus smaller wavelengths.
As can be seen, the contribution of a "usual" tweeter's frequency range to (exact?) imageing is
somewhat "overestimated" in the HiFi community anyhow, to say the least.
Best thing that a loudspeaker can do above the "some Khz" range is to behave "unobtrusive" with
regard to direct sound and also sound power, in order to not reveal the loudspeaker himself "as a
sound source".
Mitigating comb filter effects due to adjacent walls (by avoiding coherent wavefronts under angles
that would strongly interfere with mirror sources) also contribute to that goal.
Fortunately you know by now, how to make such a loudspeaker, that behaves "unobtrusive" -
especially in the "some" Khz range - and therefore does not disturb imageing by revealing ist own
presence in the room through loudspeaker specific acoustic artefacts, that other loudspeakers
exhibit to great extent.
I suggest you simply build a prototype yourself: The concept provided is "free".
"Your loudspeaker have no diffraction on-axis, but have diffraction off-axis because your speaker cabinet
has sharp edges."
You got it:
On axis baffle/edge diffraction artefacts are neutralized. The off axis edge diffraction component itself is
also more diffused (more sources, each source creates less sound pressure than a single one had to create,
diffraction patterns more distributed in space and time).
This loudspeaker can truly afford his own baffle edges (even when left unrounded) in contrary to most
common loudspeakers that cannot even afford their own baffle edges, because those act detrimental in
the direct sound already.
I am very happy, you got the "hang of the concept" (sometimes things take a while).
@Sonce (you wrote)
"you are comparing hazelnuts to watermelons!
Excerpt from the above paper: "Each loudspeaker cabinet housed a
single type 8P unit produced by Goodmans Loudspeaker Ltd."
"
I did not mention the Bennet, Barker, Edeko Paper for any "comparison" regarding loudspeakers.
The paper is IMO more about "fundamental limits" of stereophony regarding higher frequencies
(say > 3Khz) and thus smaller wavelengths.
As can be seen, the contribution of a "usual" tweeter's frequency range to (exact?) imageing is
somewhat "overestimated" in the HiFi community anyhow, to say the least.
Best thing that a loudspeaker can do above the "some Khz" range is to behave "unobtrusive" with
regard to direct sound and also sound power, in order to not reveal the loudspeaker himself "as a
sound source".
Mitigating comb filter effects due to adjacent walls (by avoiding coherent wavefronts under angles
that would strongly interfere with mirror sources) also contribute to that goal.
Fortunately you know by now, how to make such a loudspeaker, that behaves "unobtrusive" -
especially in the "some" Khz range - and therefore does not disturb imageing by revealing ist own
presence in the room through loudspeaker specific acoustic artefacts, that other loudspeakers
exhibit to great extent.
I suggest you simply build a prototype yourself: The concept provided is "free".
Last edited:
No, I will not build prototype of your loudspeaker, because I know what I would get. I have measured and tested several similar (vintage) loudspeakers with two, three and four widerange drivers covering the same mid-high frequency band. All of them sucks big time in the stereo image department, some strange fuzziness was apparent too.
As the old sayin - "There is no free lunch".
As the old sayin - "There is no free lunch".
Last edited:
No. Just two, three and four widerangers (plus woofer or two). Vintage studio monitors and others.
Had a thought about the whole stereo image above 3khz thing; doesn't the center image from a pair of sound sources, rely on the time that each source reaches the ear? Exact same signal, arriving at the exact same time, simulates a single center sound source? If that is true, then the shorter wave lengths of high frequency signals will have finer and finer head (and therefore ear) placement restrictions, yes? (as well as Head Related Transfer Function affects for crosstalk effect) How does SPL affect this? We are talking psychoacoustics here, so I'm not certain. I can only assume that at some frequency, this stops being able to be accomplished via directivity, no matter the means used to achieve that directivity? HRTF is mostly affected by amplitude, since timing from ear to ear is constant and we learn how that relates to locating sound for each of us individually. Or is the wavelength not related to timing?