• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Perceptual Effects of Room Reflections

Duke

Active Member
Manufacturer
Forum Donor
Joined
Apr 22, 2016
Messages
246
Likes
392
Location
Princeton, Texas
Thanks for posting Figure 7.14. I looked at it, as you suggested... and maybe I'm looking through a skewed lense but this is one of the things I see:

While it is true that the ranking order did not change going from mono to stereo, I can't help but notice how much the Quad's score improved. IF the Quad (or some other dipole speaker) had scored up there with the others in mono, might it have passed up some of them (moved up in the rankings) in stereo? Obviously to the ears of some listeners, the stereo Quads did just that!

In other words, it looks to me like Figure 7.14 arguably raises the possibility that preference in mono does not NECESSARILY predict preference in stereo, at least if significantly different types of loudspeakers are involved.

And I realize that others may look at Figure 7.14 and come to the opposite conclusion. I did for years, until just now, because Tuga kept telling me to LOOK at it.
 
Last edited:

krabapple

Senior Member
Joined
Apr 15, 2016
Messages
355
Likes
294
In Toole's book there is much discussion of this test -- an examination of loudspeaker directivity and its resultant room reflections , not mono vs stereo per se -- including analysing the results further by different variables, and discussion of why the Quad performed the way it did.

From a detailed analysis of the data, Toole concludes that for *stereo* listening, 'the nature of the recordings themselves proved to be the overriding factor'. (Underlined by me)

This is why just posting a single figure and caption can be misleading. It's a pity AES articles aren't open source, and that Toole's book isn't free...but still.
 

Duke

Active Member
Manufacturer
Forum Donor
Joined
Apr 22, 2016
Messages
246
Likes
392
Location
Princeton, Texas
[David] Clark in 1983...

Response notches are almost inaudible if the notches are filled in by reflections within 10 ms.

Response notches are annoying if not filled in by reflections.
Enjoyed your article very much, Amir!

I wonder how applicable the above points are to the floor bounce notch?

Here's what occurs to me:

The floor and ceiling bounces are within 10 milliseconds of one another, each with its notch at a different frequency. Thus, perceptually, they might fill in one another's notches. And if so, this might be a partial explanation of why the floor bounce notch is perceptually relatively benign. Not that I dispute the standard explanation - that we are acclimated to it - but rather perhaps the whole story includes a perceptual partial filling-in of the floor-bounce notch by the ceiling bounce.
 

tuga

Major Contributor
Forum Donor
Joined
Feb 5, 2020
Messages
1,855
Likes
1,384
Location
Oxford, England
Thanks for posting Figure 7.14. I looked at it, as you suggested... and maybe I'm looking through a skewed lense but this is one of the things I see:

While it is true that the ranking order did not change going from mono to stereo, I can't help but notice how much the Quad's score improved. IF the Quad (or some other dipole speaker) had scored up there with the others in mono, might it have passed up some of them (moved up in the rankings) in stereo? Obviously to the ears of some listeners, the stereo Quads did just that!

In other words, it looks to me like Figure 7.14 arguably raises the possibility that preference in mono does not NECESSARILY predict preference in stereo, at least if significantly different types of loudspeakers are involved.

And I realize that others may look at Figure 7.14 and come to the opposite conclusion. I did for years, until just now, because Tuga kept telling me to LOOK at it.
Thank you for your comments.

I have not read a large significant chunk of the relevant literature and have limited technical knowledge, so I rely mostly on good old common sense and listening experience and try to apply simple logic, and try to learn as much as I can.

After analysing Fig. 7.14 I am convinced that, even though there are practical advantages to listening in mono (i.e. a stereo version of Harman speaker shuffler whould have been complicated to produce, as would one which positioned each single or pair of speakers optimally in the room) as well as benefits for critical assessment of particular aspects of performance, speakers should be evaluated for both preference and performance in stereo (and complemented with mono assessment sessions).

Further, I find that it would be more fruitfull if we used Harman's research as a starting point not as the definitive answer to all speaker-related questions. Because it isn't.
 

Duke

Active Member
Manufacturer
Forum Donor
Joined
Apr 22, 2016
Messages
246
Likes
392
Location
Princeton, Texas
After analysing Fig. 7.14 I am convinced that, even though there are practical advantages to listening in mono (i.e. a stereo version of Harman speaker shuffler whould have been complicated to produce, as would one which positioned each single or pair of speakers optimally in the room) as well as benefits for critical assessment of particular aspects of performance, speakers should be evaluated for both preference and performance in stereo (and complemented with mono assessment sessions).
Agreed. Listening in mono is much more practical and in most ways much more useful.

But one of the questions I have, as a speaker designer, is WHY did the Quads move up so much in perceived sound quality in stereo, even if their rank remained third out of the three? Was is because of improvements in spatial quality, and in turn was that because it was a dipole?

Spatial quality matters. How much does it matter?

In his book Toole refers to a study done by Wolfgang Klippel (yes, the same Klippel) in which he examined the relative contributions of spatial quality and sound quality to "naturalness" (how realistic the speakers sound), and "pleasantness" (how enjoyable they are).

He found that “naturalness” (realism and accuracy) was 30% related to sound quality (coloration, or the lack thereof); 20% related to tonal balance; and 50% related to the “feeling of space”.

“Pleasantness” (general satisfaction or preference) was 30% related to sound quality and 70% related to the “feeling of space”.

In other words, according to Klippel, the "feeling of space" was 50% of what made speakers sound realistic, and 70% of what made speakers enjoyable! Given that stereo plays a major role in "the feeling of space", imo evaluations which leave it out are incomplete. But they are NOT useless!! They are still extremely valuable, just not the complete picture. (I am not arguing that these percentages are absolutes; only that spatial quality matters enough that it should be fully included in any subjective evaluation which claims to be complete.)

Also, Geddes and I design speakers that are supposed to be aggressively toed-in, like at about 45 degrees. This results in the first significant sidewall reflection being the long bounce off the far side wall. That reflection arrives at the opposite ear, which decreases the interaural cross-correlation, which in turn enhances the perception of spaciousness. Any perceptual benefit that might arise won't show up unless the speakers are evaluated in stereo with the recommended setup geometry.

And a few speakers - like the Polk Stereo Dimensional Arrays - genuinely require stereo operation in order to begin to be fairly evaluated.

Further, I find that it would be more fruitfull if we used Harman's research as a starting point not as the definitive answer to all speaker-related questions. Because it isn't.
Harman's research has generated an enormous amount of extremely useful information, but I draw on other sources as well. And kudos to Dr. Toole for including data at a high enough resolution in his book (such as Figure 7.14) so that alternative conclusions can be contemplated.

The one thing I'm a bit skeptical of is this: Calculating an ironclad metric of merit based entirely on data which has been processed a certain way. Even if there is a strong correlation between controlled listening data and the metric, from what I can tell the database used in the derivation of the metric does not include adequate representation of polydirectional loudspeakers (bipoles, dipoles, omnis, Shahinians, etc.); nor does it adapt to different rooms (a particular type of room acoustic environment is assumed). Notably (in my opinion) absent from Harman's data is the Mirage M1, a bipolar speaker which spent many years in Toole's living room. And the M1 was a highly informed choice on his part.

What was it about the M1 that made it stand out from its competitors? And how much of a role did its bipolar radiation pattern play? Harman isn't really going to tell us, so we have to look elsewhere, assuming this is a question which matters. Imo it does... was the Mirage M1 an outstanding execution of a fundamentally flawed concept, or did the concept have merit then, and if so, does it (or some evolution thereof) have merit now?
 
Last edited:
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
27,652
Likes
68,176
Location
Seattle Area
Thread Starter #226
After analysing Fig. 7.14 I am convinced that, even though there are practical advantages to listening in mono (i.e. a stereo version of Harman speaker shuffler whould have been complicated to produce, as would one which positioned each single or pair of speakers optimally in the room)
Harman has a full multi-channel speaker shuffler:

Harman Shuffler Testing Room #2.jpg
 

Duke

Active Member
Manufacturer
Forum Donor
Joined
Apr 22, 2016
Messages
246
Likes
392
Location
Princeton, Texas
Harman has a full multi-channel speaker shuffler:

View attachment 71662
SWEEET!!

According to some numbers I found online, it looks like that room is about 30 feet long by 21.6 feet wide by 8.5 feet tall. Verrry interesting, in particular the width, which pushes the first sidewall reflections back in time to ballpark 8 milliseconds later than the direct sound, at the center sweet spot (assuming the speakers are 8 feet apart).

We can see the diffusors, but what are the walls like in between them? Reflective or absorptive or some of both?

I bet that room lets you clearly hear the acoustic space on the recording, without the playback room's signature intruding. Have you been there?
 
Last edited:

NTK

Senior Member
Forum Donor
Joined
Aug 11, 2019
Messages
392
Likes
772
Location
US East
SWEEET!!

According to some numbers I found online, it looks like that room is about 30 feet long by 21.6 feet wide by 8.5 feet tall. Verrry interesting, in particular the width, which pushes the first sidewall reflections back in time to ballpark 8 milliseconds later than the direct sound, at the center sweet spot (assuming the speakers are 8 feet apart).

We can see the diffusors, but what are the walls like in between them? Reflective or absorptive or some of both?

I bet that room lets you clearly hear the acoustic space on the recording, without the playback room's signature intruding. Have you been there?
You can find Dr. Olive's paper on the lab here:
https://www.semanticscholar.org/pap...stro/311598ba6306e0a55d35a078be7562e90f416c94

An excerpt on the construction from the paper:

1593723010364.png
 

tuga

Major Contributor
Forum Donor
Joined
Feb 5, 2020
Messages
1,855
Likes
1,384
Location
Oxford, England
Agreed. Listening in mono is much more practical and in most ways much more useful.

But one of the questions I have, as a speaker designer, is WHY did the Quads move up so much in perceived sound quality in stereo, even if their rank remained third out of the three? Was is because of improvements in spatial quality, and in turn was that because it was a dipole?

Spatial quality matters. How much does it matter?

In his book Toole refers to a study done by Wolfgang Klippel (yes, the same Klippel) in which he examined the relative contributions of spatial quality and sound quality to "naturalness" (how realistic the speakers sound), and "pleasantness" (how enjoyable they are).

He found that “naturalness” (realism and accuracy) was 30% related to sound quality (coloration, or the lack thereof); 20% related to tonal balance; and 50% related to the “feeling of space”.

“Pleasantness” (general satisfaction or preference) was 30% related to sound quality and 70% related to the “feeling of space”.

In other words, according to Klippel, the "feeling of space" was 50% of what made speakers sound realistic, and 70% of what made speakers enjoyable! Given that stereo plays a major role in "the feeling of space", imo evaluations which leave it out are incomplete. But they are NOT useless!! They are still extremely valuable, just not the complete picture. (I am not arguing that these percentages are absolutes; only that spatial quality matters enough that it should be fully included in any subjective evaluation which claims to be complete.)

Also, Geddes and I design speakers that are supposed to be aggressively toed-in, like at about 45 degrees. This results in the first significant sidewall reflection being the long bounce off the far side wall. That reflection arrives at the opposite ear, which decreases the interaural cross-correlation, which in turn enhances the perception of spaciousness. Any perceptual benefit that might arise won't show up unless the speakers are evaluated in stereo with the recommended setup geometry.

And a few speakers - like the Polk Stereo Dimensional Arrays - genuinely require stereo operation in order to begin to be fairly evaluated.
Thank you for mentioning Klippel's findings, they are very interesting.
I guess that wide vs. narrow dispersion is probably cause for the greatest divide when it comes to loudspeaker design.

In regard to the "feeling of space", even though it appears that most listeners enjoy the effects of room "interaction" (increased envelopment and wider soundstage) some (like myself) find the trade-offs (blurrier or stretched phantom images, poor reproduction of recorded room acoustics) too taxing.
I used to think that it would be logical for classical music listeners to favour less room "interaction" as this would produce a more palpable illusion of listening to music being played in a music hall (crisper phantom images, better "reproduction" of the recorded acoustics) but audiophiles in this and other forums have proved me wrong. I would venture that even though preference for more or less room "interaction" is a matter of taste it does not appear to be driven by preferred musical genre.
It is also interesting to note that whilst the spatial illusion of 2-channel stereo appears to be enough for a large number of listeners, some are unsatisfied with the lack of envelopment and prefer multi-channel, upmixing or even some form of digitally processed 3D-ness.
This makes me wonder whether for some people the experience of sound may be as important as the experience of music.
I think we can safely conclude that there isn't a universal solution.

I've never listener to any Geddes' speakers unfortunately, they're not very common in Europe.
 
Joined
May 28, 2020
Messages
80
Likes
93
If reflections are not good than is it safe to conclude that if the room is an acoustic nightmare with lot of hard reflecting surfaces than its better to get speakers with narrow dispersion window and point them to MLP?
Sadly no one has bothered to answer this question. I was wondering the same thing. I believe this is the approach that Gradient of Finland advocate to “get rid of the room”, although they don’t explicitly refer to early reflections.
 

Duke

Active Member
Manufacturer
Forum Donor
Joined
Apr 22, 2016
Messages
246
Likes
392
Location
Princeton, Texas
Thank you very much, NTK!!

Incredible room construction:

"The inner walls and ceiling of the double-wall IAC shell are made of heavy gauge steel panels separated 10 cm and filled with fiberglass. The inner surfaces are perforated with 2.34-mm openings to provide substantial sound absorption inside the room. The inner walls are entirely floated and separated from the outer wall of the shell by a 10 cm space to minimize mechanical and acoustic transmission of noise."

My (amateur level) understanding of those fiberglass-filled steel walls with small openings is that they would provide very even and controlled broadband absorption, unlike slabs of foam which completely absorb short wavelengths but have diminished effectiveness as the wavelengths increase (thus skewing the spectral balance of reflections). I think the technique Harman used would weaken reflections a little bit, but not too much, without significantly altering their spectral content.

Another excerpt from the paper:

"...strong early reflections are known to influence the perceived spatial and timbral qualities of reproduced sound...

"The EBU standard recommends that all reflections within the first 15 ms after the arrival of sound be no greater than 10 dB in level relative to the direct sound from each sound source."

The word "influence" doesn't tell us whether the effect of strong early reflections on the "perceived spatial and timbral qualities of reproduced sound" is desirable or not.

The implication of the EBU standers is that it's desirable to weaken early reflections (those arriving within 15 milliseconds) below a certain threshold (-10 dB). Very interesting! This is not exactly what I do (my constraints being more severe than Harman's), but I LIKE it!! And it (arguably) tells me what I SHOULD be doing.
 
Last edited:

Duke

Active Member
Manufacturer
Forum Donor
Joined
Apr 22, 2016
Messages
246
Likes
392
Location
Princeton, Texas
Interesting room in that it's quite wide and treated, I wonder if wide dispersion would produce enough sense of envelopment.
I'm also interested to hear from anyone who has listened in this room. I THINK the absorptive properties of the walls, in between the diffusion panels, is gentle enough that envelopment can happen with a good recording.
 

Duke

Active Member
Manufacturer
Forum Donor
Joined
Apr 22, 2016
Messages
246
Likes
392
Location
Princeton, Texas
I guess that wide vs. narrow dispersion is probably cause for the greatest divide when it comes to loudspeaker design.
Radiation pattern is definitely the issue I'm fixated on, BUT to put things in perspective, Toole finds that freedom from resonances is WHAT MATTERS MOST. And I agree with him! There is not much point in having the perfect radiation pattern if what we're radiating sucks.

In regard to the "feeling of space", even though it appears that most listeners enjoy the effects of room "interaction" (increased envelopment and wider soundstage) some (like myself) find the trade-offs (blurrier or stretched phantom images, poor reproduction of recorded room acoustics) too taxing.
Everything's a tradeoff. I think it's possible to get good envelopment and good clarity at the same time, but I don't think it's possible to maximize envelopment, clarity, AND soundstage width all at the same time, at least not with two-channel sound in our typically atypical home listening rooms. Imo the early sidewall reflections which enhance the Apparent Source Width (ASW) can be detrimental to clarity, soundstage depth, and envelopment. The above-mentioned EBU standard calls for reducing the level of early reflections below a certain SPL threshold, which implies that strong early reflections (which includes the first same-side-wall reflections) are undesirable.

I've never listener to any Geddes' speakers unfortunately, they're not very common in Europe.
The Dutch & Dutch 8c looks to me like it embodies many of Geddes' principles, with a little Jorma Salmi mixed in (cardioid midbass), all in an incredibly elegant and well thought-out compact package.
 
Last edited:
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
27,652
Likes
68,176
Location
Seattle Area
Thread Starter #235
SWEEET!!

According to some numbers I found online, it looks like that room is about 30 feet long by 21.6 feet wide by 8.5 feet tall. Verrry interesting, in particular the width, which pushes the first sidewall reflections back in time to ballpark 8 milliseconds later than the direct sound, at the center sweet spot (assuming the speakers are 8 feet apart).

We can see the diffusors, but what are the walls like in between them? Reflective or absorptive or some of both?

I bet that room lets you clearly hear the acoustic space on the recording, without the playback room's signature intruding. Have you been there?
I have not. But there is an AES paper on it. http://www.aes.org/e-lib/browse.cfm?elib=8338

I can't tell if there is anything on the walls between them. If I had to guess I would say reflective.
 

Duke

Active Member
Manufacturer
Forum Donor
Joined
Apr 22, 2016
Messages
246
Likes
392
Location
Princeton, Texas
If reflections are not good than is it safe to conclude that if the room is an acoustic nightmare with lot of hard reflecting surfaces than its better to get speakers with narrow dispersion window and point them to MLP?
Ttimer quoted this and added:

Sadly no one has bothered to answer this question. I was wondering the same thing. I believe this is the approach that Gradient of Finland advocate to “get rid of the room”, although they don’t explicitly refer to early reflections.
This is my opinion:

First, the room might NOT be an acoustic nightmare. Does it have slap-echo? If so, then yes it's an acoustic nightmare, but one that can be solved.

When you walk from room to room talking out loud and listening to the timbre of your voice, is it particularly awful in that room? If so, then maybe it is an acoustic nightmare which needs professional help.

However if neither of the above situations apply, it is POSSIBLE that the speaker's off-axis response is causing the acoustic nightmare. Let me give an example: Suppose the speaker is a 6" two-way with a 1" dome tweeter, crossed over at 2 kHz ballpark. Below the crossover frequency, the midwoofer is starting to beam. But above the crossover frequency the tweeter's radiation pattern is extremely wide, like virtually full power at 90 degrees off-axis. An all of this excess off-axis energy is right smack in the region where the ear is most sensitive, 2-4 kHz or so. The net result can include harshness and listening fatigue. It may seem like the room is to blame, when it's really the speaker's off-axis response. Unfortunately it is very difficult to selectively absorb the bottom end of the tweeter's range with room treatments. Presumably there are professionals who know how, but I don't.

But now back to what to do if the room really IS an acoustic nightmare (aside from investing in headphones!):

Imo what you want to do is, have the direct sound from the speakers dominate over the reflected sound. This can be done with very directional speakers, with nearfield listening, or both. If you opt for nearfield or both, you might want to consider speakers with a coaxial mid/tweet driver, so that you can listen from very close without hearing a vertical discontinuity. For nearfield listening that KEF R3 speaker Amir reviewed looks very promising to me. It uses a dedicated woofer in addition to the coaxial unit, which imo is a major advantage over the coaxial-only LS50, in addition to the fine performance Amir's measurements indicate.

If you do not want to listen nearfield, then highly directional main speakers aimed at the sweet spot would imo be a good strategy. If you want a sweet spot wide enough for two or three people, and if using horn or waveguide speakers, you might try using extreme toe-in such that the speakers' axes criss-cross in front of the center of the sweet spot. Note that "highly directional" can include dipole speakers, though some placement restrictions apply. The Sanders Sound Model 10 is imo a very good single-listener speaker.

You mentioned the Gradients - imo they are excellent speakers for problematic rooms. While they are not what I'd call "narrow pattern", their off-axis response tracks their on-axis response very well. So whether the room reflects back a little or a lot, chances are it'll be spectrally correct. If the room is really bad and skews the spectral balance of the reflections (like if it was damped with a bunch of inexpensive foam squares) then the Gradients can't fix that, but ime they are considerably more room-friendly than most. I was a Gradient dealer for many years and only stopped because I wanted to concentrate on my own designs.

“A speaker that has controlled dispersion does basically the same thing you'd expect an acoustic panel to do, but it does a better job. And it allows you to get away with no panels on the wall." - Matt Poes, acoustical consultant (in other words, Matt makes money off of acoustic panel installations, NOT off of speaker sales)
 
Last edited:

Kal Rubinson

Major Contributor
Industry Insider
Joined
Mar 23, 2016
Messages
2,086
Likes
2,605
Location
NYC/CT
It is also interesting to note that whilst the spatial illusion of 2-channel stereo appears to be enough for a large number of listeners, some are unsatisfied with the lack of envelopment and prefer multi-channel, upmixing or even some form of digitally processed 3D-ness.
I think that argument is not fairly presented. Stereo is only an intermediary in a continuum from monophonic to a full and true recreation of the performance experience. While stereo is "enough for a large number of listeners," many others are happy with a single speaker and, for others, it can be improved on. The only thing that is unique about stereo is its current wide acceptance.
This makes me wonder whether for some people the experience of sound may be as important as the experience of music.
What do you think stereo is about? Are you saying that it is the "perfect" medium for music reproduction? Do you really think it can reproduce all the audible content of a musical event?
 

eliash

Active Member
Joined
May 29, 2019
Messages
287
Likes
116
Location
Bavaria, near lake Ammersee
You can find Dr. Olive's paper on the lab here:
https://www.semanticscholar.org/pap...stro/311598ba6306e0a55d35a078be7562e90f416c94

An excerpt on the construction from the paper:

View attachment 71675
...Transmission of low bass into such a floor construction means that the speaker is (counter-) moving, driven by coil force and membrane mass, since the floor is elastic - this causes poor bass attack, according to own experience. At least true for slim underweight tower speakers (here Dynaudio Focus 220´s), which don´t have enough mass in the moving axis of the membane to attenuate such movement...hard ground coupling solves part of that problem...
 

tuga

Major Contributor
Forum Donor
Joined
Feb 5, 2020
Messages
1,855
Likes
1,384
Location
Oxford, England
I think that argument is not fairly presented. Stereo is only an intermediary in a continuum from monophonic to a full and true recreation of the performance experience. While stereo is "enough for a large number of listeners," many others are happy with a single speaker and, for others, it can be improved on. The only thing that is unique about stereo is its current wide acceptance.
What do you think stereo is about? Are you saying that it is the "perfect" medium for music reproduction? Do you really think it can reproduce all the audible content of a musical event?
I think we've discussed 2- vs multi-channel to exhaustion already in the appropriate thread.

How do you feel about 3D video, do you find it necessary, an improvement, a gimmick?
What about stereoscopic photography, are you pursuing it? Is it indispensable?

The domestic reproduction of recorded music will always be about creating an illusion, a second nature, and ultimately one has to draw a line on how little abstraction can be required from the listener. I am perfectly able to enjoy music in the car or even through my monaural Tivoli Audio tabletop radio. The latter struggles with instrument separation in complex passages and more importantly for you it cannot do the space thing...
You give more importance to space than I do and might only be satisfied with something like this:

 
Last edited:

tuga

Major Contributor
Forum Donor
Joined
Feb 5, 2020
Messages
1,855
Likes
1,384
Location
Oxford, England
Radiation pattern is definitely the issue I'm fixated on, BUT to put things in perspective, Toole finds that freedom from resonances is WHAT MATTERS MOST. And I agree with him! There is not much point in having the perfect radiation pattern if what we're radiating sucks.
What is your view in regard to resonance Q-factor (frequency vs. time)?
Toole's followers here generally dismiss the usefulness of CSD measurements for identifying audible problems, claiming that frequency response alone is enough.

Toole's paper makes a case for low-Q (wide-band, shorter decay) resonances being more audible yet my anecdotal experience has been that high-Q resonances are more taxing on the ear (irritating, fatigue), f.e. when reproducing violins. I have, perhaps mistakenly, attributed this to hard cone mid and mid-woofer break up (finding most probable causes for audible shortcomings is an extremely difficult task).
 
Top Bottom