Griesinger's teachings show up in Klippel, Linkwitz, Toole, and Geddes

[Duke LeJeune]

[

There is a much more interesting passage you did not quote:

"Image perception is dominated by the very earliest sound from the speaker, i.e. the direct sound (first arrival), and the sound that arrives in the first 5-10 milliseconds. The ear simply integrates this all into one lumped sum. This includes the speakers’ anechoic response along the listening axis, cabinet diffractions, and diffractions and reflections from nearby objects like equipment cabinets or televisions. Basically what one wants for good imaging is a pseudo point source response, i.e. a single direct sound free from any diffractions or reflections for at least 5, but hopefully a full 10 milliseconds. Let me call these Very Early Reflections VER (but they will also include the early diffractions as well)."

Yeah I just pulled out part of that passage - the whole thing is very informative but goes into topics I didn't want to get sidetracked into at this time.

As Earl describes it these VER would be from the same direction as the direct sound. In his AES paper 6888 (Audibility of Linear Distortion with Variations in Sound Pressure Level and Group Delay) he finds that sound (in this case linear distortion) delayed by 0-1 ms is more audible when delay increases. 0-1 ms is the range where cabinet edge diffraction might play a role. Ando finds that reflections arriving within 2-3 ms, such as from surfaces very close to loudspeakers, produce high interaural cross-correlation and are the least beneficial.

Thank you for digging that up! I was not aware of the specifics.

Earl goes to unusual lengths to minimize interaural cross-correlation. His aggressive toe-in results in the first significant sidewall reflection of the left speaker taking the long, across-the-room bounce and arriving at the right ear, and vice versa. This kills two birds with one stone: Long delay time in addition to low inter-aural cross-correlation. I do this too.

And in Earl's dedicated listening room, the left side wall has a hanging tapestry while the right side wall is made of fairly large, jagged rocks!

Why are late first reflections good and early first reflections bad, when both are above perception thresholds?

Early reflections have both beneficial and detrimental aspects, while in practice the detrimental aspects are relatively absent from the late reflections.

For instance, the early reflections are the ones which tell us we're in a small room ("Second Venue" cues), while the late reflections help to deliver the reverberant tail (hopefully) present on the recording, which tells us we're in the acoustic space of the recording ("First Venue" cues). (Sorry for all the parentheses.) (<- Oops I did it again.)

 

[Duke LeJeune]

Klaus, I'd like to offer a more complete response to something you posted.

The delay of reflections depends on locations of source and listener. You move one (or both) and the delays change. Does the perceived room size change as well?

To which I replied: "In my experience, it can!"

Let me elaborate on that a bit:

In addition to playing around with distance to the wall with dipole speakers, I have also built systems which have a fairly directional primary array of drivers aimed at the listening area and delivering the direct sound (along with some reverberant sound of course), and a secondary directional array aimed in a different direction such that it delivers only reverberant sound. My intention at first was richer timbre, but there seemed to be a spatial enhancement which I had not anticipated. I have been attributing it primarily to increasing the time delay before the strong onset of early reflections, but thanks to the rest of your post, I now realize that something else is going on, which MIGHT be even more significant:

What Kuhl stated in his 1978 paper apparently is not true, as I have pointed out with reference to more recent psychoacoustic research, where reverb. time was identified as main suspect.

The additional energy which the secondary array adds to the reverberant field would make the reverberation louder, therefore lasting longer before decaying into inaudibility.

In other words I have been focusing on manipulating the early reflections, and may have been inadvertently making a worthwhile improvement to the reverb time as well.

Thank you for posting so much information! You are teaching me about things I was unaware of.

 

[KlausR.]

Hello Duke,

Earl goes to unusual lengths to minimize interaural cross-correlation. His aggressive toe-in results in the first significant sidewall reflection of the left speaker taking the long, across-the-room bounce and arriving at the right ear, and vice versa. This kills two birds with one stone: Long delay time in addition to low inter-aural cross-correlation. I do this too.

And in Earl's dedicated listening room, the left side wall has a hanging tapestry while the right side wall is made of fairly large, jagged rocks!

Low interaural cross-correlation means high dissimilarity of right and left signals, which is what listeners prefer. If you look at the data of my SPL-in-3-rooms you see that the across-the-room reflection is delayed by 7 ms (room 1), 13 ms (room 2), 68 ms (room 3).

In rooms 1, 2 the reflection is still well above perception threshold, in room 3 it is close to that threshold, but at the same time above echo thresholds for most types of signals.

Perceiving a reflection means that you hear a difference (timbre or image) compared to when the reflection was below perception threshold. That is why I asked the question: what is the benefit of having late first reflections when they are still above perception threshold? Why should they be psychoacoustically different from early first reflections?

Early reflections have both beneficial and detrimental aspects, while in practice the detrimental aspects are relatively absent from the late reflections.

For instance, the early reflections are the ones which tell us we're in a small room ("Second Venue" cues), while the late reflections help to deliver the reverberant tail (hopefully) present on the recording, which tells us we're in the acoustic space of the recording ("First Venue" cues).

Re: Room size perception: the psychoacoustic literature I mentioned (Hameed, Yadav) states that reverb. time is the cue, not first reflections.

Reflections play a role in determination of auditory distance:
Bronkhorst et al., “Auditory distance perception in rooms”, Nature 1999, vol. 397, S.517
Nielsen, “Auditory distance perception in different rooms”, J. of the Audio Engineering Society 1993, S.755
Michelsen et al., „Parameters of distance perception in stereo loudspeaker scenario”, Audio Engineering Society preprint 4472 (1997)

If you know of research that shows that first reflections play a role in room size perception, let me know. If you know of any research re: audibility of cabinet edge diffraction, please let me know as well.

Having said all this the main point is still: 1. there is, as far as I know, no hard/scientific evidence that first reflections have detrimental effects under all circumstances, and 2. IMO Griesinger’s demonstration cannot be used to prove that they have in small listening rooms. Maybe we are bound to disagree on the latter?

Klaus

 

[Duke LeJeune]

Thank you for your in-depth reply Klaus, and for extending the olive branch of "agreeing to disagree" at the end. I don't think we are quite as far apart as it may seem, but it is very nice to know that the "worst case scenario" still ends amicably.

I will reply as time permits, which may take me a few days.

 

[KlausR.]

Hi Duke,

One further point you might want to consider: I understand that Griesinger “matlabs” the recording for a seat he did not like: “I’m gonna talk about this one (apparently the lowermost of the seats indicated on the slide), which is a seat I know from going to concert performance I do not like. This one’s marginal, this is where I often sit, this one’s actually better. These seats are all pretty damn good, and this one is better than most halls.”

So to make his point Griesinger uses a seat that sounds bad (to him), fair enough. But what about that seat that is “better than most halls”? This seat also has first reflections, but still sounds great. If first reflections are detrimental as a matter of principle, then all seats should sound bad, shouldn’t they? But they don’t.

In the other thread you said: “We learn that early reflections are detrimental while late reflections are beneficial.” Yes, for a bad sounding seat in Boston Symphony, with the first reflections delayed by up to 80 ms, listening to the right channel of a binaural recording. How relevant can that be for 2-channel stereo in small domestic rooms. Maybe we don’t need the olive branch after all?

Klaus

 

[DaveC]

IME there's more than one way to achieve high quality immersive sound. I've heard different approaches work well, and believe there is some personal preference involved.

From my experimentation removing reflections results in an experience closer and closer to headphones, some like it and some do not. Some become accustomed to it and then like it, others may not become accustomed as much.

I've heard Duke's various iterations over the years at RMAF and in the context of a very directional speaker, adding a 2nd source in the way Duke describes results in bilpole-like reflections that many people enjoy, it reduces the perceived difference between a horn and a cone 'n dome, and gives people a listening experience more like a conventional speaker... but with the advantage of a less demanding setup. I'd also say the way Duke and James design the 2nd source is superior to most dipole/bipole as it's aimed to produce longer delays.

In the systems I've heard with conventional speakers that achieve good immersion, it takes some dedication to room design and setup to make sure the 1st reflections integrate with the direct sound properly so they don't interfere with one another, and I think this is where the importance of a smooth polar plot is important so the reflections are accepted by the brain as the same as the direct sound and they properly "merge". I'm not convinced it's quite as important with horns, as AGs generally have horrible polars but people get good results out of them anyways. Which brings into question the real importance of CD type horns, I don't believe they are in fact superior to a LeCleach... but that's another subject!

In any case, I'd say that room reflections are NOT required to produce a 3-D immersive soundstage, but may be preferred by some listeners in rooms that are large enough to allow them. I think many rooms are not, just like many rooms are really not large enough to give ideal performance from dipole speakers. In these smaller rooms I prefer a larger proportion of direct sound and no 2nd stream at all. If the room is large enough I still don't prefer them but I can see why some do and I believe it's possible to have them while maintaining immersion.

 

[DaveC]

Oh, yeah... I think this point is key:

If the system is not resolving enough to convincingly reproduce spatial cues in the recording then a higher ratio of direct to reflected sound can come off as dull and boring. You need a very high quality system without major flaws to achieve convincing immersion, and it takes a lot... clean AC power and most copper interconnect cables shouldn't apply at all, they will kill the immersive experience to a large degree.

This is one area where Toole and others have simply failed, they fail to setup a system that has this level of performance because they don't understand the importance of cables, AC power, controlling vibrations and using high quality electronics. Their test setups are only able to capture gross phenomenon, while IME achieving a highly immersive sound is dependent on things they completely ignore and believe can't matter.

 

[Duke LeJeune]

Hello Duke...

Thank you Klaus for letting me know that I still need to work on communicating clearly, and for giving me another chance to do so.

I'm going to start with some of the last things you said, if you don't mind:

The main point is still: 1. there is, as far as I know, no hard/scientific evidence that first reflections have detrimental effects under all circumstances...

I don't think I said that “first reflections have detrimental effects under all circumstances,” but obviously I have communicated poorly if that is what you heard.

Let me try again: Early reflections can have both beneficial and detrimental aspects. Late reflections in a home audio setting generally go not share the detrimental aspects of early reflections (assuming they are spectrally correct). So imo it makes sense to manipulate the early reflections to minimizes detrimental aspects, while encouraging beneficial late reflections. (I will reply to the relevant questions in your third paragraph another time.)

Regarding your call for “hard/scientific evidence”, most of my information comes from a few trusted individuals who have read the relevant research, who have done much of it themselves, and who I believe are qualified to speak without having to justify their every statement. Toole, Griesinger, and Geddes come to mind. I can quote things they have said, but I have not the time nor energy to dig up their sources and quote them.

...and 2. IMO Griesinger’s demonstration cannot be used to prove that [early reflections] have [detrimental effects] in small listening rooms.

I don't think I indicated that “Griesingers' demonstrations prove that early reflections have detrimental effects in small listening rooms.” And I apologize if that is the impression I have given.

I think Griesinger's clips are an example from a concert hall which rather vividly illustrates principles he teaches.

But I'm okay with excluding Griesinger's clips since they were not made in a small room. And I concede that they do not “prove” anything about small rooms, whether or not I said or implied that they did.

So, here is my fall-back position:

“The earlier a reflection arrives the more it contributes to masking the direct sound.”

“Envelopment is perceived when the ear and brain can detect TWO separate streams: A foreground stream of direct sound, and a background stream of reverberation. Both streams must be present if sound is perceived as enveloping.” (Implied is that there must be a time gap in between the two streams to differentiate them.)

“Envelopment is the holy grail of concert hall design. When reproducing sound in small spaces [such as home listening rooms], envelopment is often absent.”

I will reply to some of your other points in the same post when I have some time again, and then will move on to your more recent post.

 

[Duke LeJeune]

Wwhat is the benefit of having late first reflections when they are still above perception threshold? Why should they be psychoacoustically different from early first reflections?

Perception thresholds do not inform us about preference. If reflections were all undesirable, the holy grail would be the anechoic chamber. So whether or not a reflection falls within perception thresholds does not tell us whether it is beneficial or detrimental or maybe even some of both.

Here are four ways in which early reflections may be detrimental:

1. It is the earliest reflections which are most likely to degrade clarity (Griesinger).

2. Coloration from reflections peaks between 2 and 3 milliseconds as I recall (James M. Kates).

3. “The delay time of the early reflections determines the impression of the size of the room” (Kuhl). I know that you dismiss Kuhl because of subsequent research indicating that reverberation determines the impression of distance (and I understand reverberation does that). But please note that room size and distance to the sound source are NOT the same thing. Another researcher came to the conclusion that it is the “center of gravity” of the reflections which determines the impression of room size, but I do not recall who that was.

4. Early lateral reflections do contribute to enlarging the Apparent Source Width (ASW) according to multiple researchers. The desirability of increasing the Apparent Source Width via early lateral reflections is not agreed upon; Toole finds it to be desirable while Griesinger and Geddes do not. But note that earlier in this thread I pointed how how two out of three of Toole's personal home stereo setups have weak or even non-existent early lateral reflections.

5. Reflections arriving before 10 milliseconds can corrupt transients (Griesinger). [This item was added several days after I originally made this post; see Post Number 38 below.]

The benefit of late reflections is, they contribute to timbre and spaciousness without the downsides listed above. If we are speaking in terms of Two Venues, late Second Venue reflections are desirable because they deliver the First Venue reverberation which informs our perception of First Venue distances and contributes to our perception the First Venue room size.

Here is a short clip of Earl Geddes speaking on the subject:

Re: Room size perception: the psychoacoustic literature I mentioned (Hameed, Yadav) states that reverb. time is the cue, not first reflections.

Does that really paper say first reflections play zero role in room size perception? Because if so, that has not been my experience, at least as it relates to Second Venue cues in a home audio setting, or listening in a professionally designed recording studio. Not that I expect my experiences to mean anything to anyone other than myself.

I suspect that both the first reflections and the reberberation play a role, and that room size perception corresponds to some weighted average.

 

[Duke LeJeune]

If you know of research that shows that first reflections play a role in room size perception, let me know. If you know of any research re: audibility of cabinet edge diffraction, please let me know as well.

I get the impression that you hold early reflections and diffraction to be "innocent until proven guilty", with the burden of proof being upon me to prove them guilty. I do not have the time & energy to dig up the research that has informed people like Geddes and Griesinger on the subject.

So instead, I'll copy-n-paste an imo relevant passage from one of Earl Geddes' papers:

"There are a couple of characteristics of the ear that need to be understood in order to understand what is important in small room acoustics at mid to high frequencies. These are mostly related to how the human ear perceives reflections and diffraction. The perception of reflections and diffraction is a highly complex topic, but there are a few principles that are most important here:

"The earlier and the greater in level the first room reflections are, the worse they are. This aspect of sound perception is controversial. Some believe that all reflections are good because they increase the listener's feeling of space – they increase the spaciousness of the sound. While it is certainly true that all reflections add to spaciousness, the very early ones (< 10 ms.) do so at the sake of imaging and coloration. There is no contention that reflections > 20 ms are positive and perceived as early reverberation and acoustic spaciousness within the space. In small rooms, the first reflections from an arbitrary source, mainly omni-directional, will never occur later than 10-20 ms (basically this is the definition of a small room), hence the first reflections in small rooms must be thought of as a serious problem that causes coloration and image blurring. These reflections must be considered in the design and should be also be considered in the room as well.

"Reflections become less of a problem as coloration and image shift at lower frequencies. Below about 500 Hz. early reflections are not as much of an issue. The ear has a longer integration time at lower frequencies and it has a poorer ability to localize resulting in a lower sensitivity to early reflections. Image localization is strongly weighted towards the higher frequencies.

"A reflected signal that arrives at the opposite ear from the direct sound is less perceptible as coloration and image shift than if both signals arrive at the same ear. This is because of head shadowing above about 500 Hz and the fact that our ears can process signals between them. When the two signals arrive at the same ear, the signals are physically merged in space even before they enter the ear and no amount of auditory processing can separate them. When these signals arrive at different ears, the auditory processing system can diminish the adverse effects of these early reflections through cognitive processing between the ears.

"Very early reflection-like signals can be quite audible due to non constant group delay. Diffraction acts much like reflection, but is usually frequency dependent – far more so than a typical reflection. Diffraction can occur from any nearby discontinuity be it on the cabinet or the room itself. These very early reflections/diffractions can have a pronounced audible effect far greater than what their frequency response effect would indicate. This is because of temporal masking effects in the ear. This audibility can also be level dependent making it a non-linear distortion-like effect. Diffraction cannot be electronically corrected except at a single point in space. Only acoustical corrections to diffraction (like not generating any) can have a global effect.

"From an acoustics reproduction standpoint then, the loudspeaker system design must help to provide as much delay as possible in the early reflections and allow for speaker placement and orientation such that the earliest reflections occur at opposite ears rather than the same ear. This needs to be done above about 500 Hz. Below 500 Hz other factors, such as room characteristics and our hearing mechanism, may dictate an entirely different approach. The cabinet and nearby room geometry must be such as to minimize the generation of diffraction."

The occasional emphasis in the foregoing is mine. Imo there is an enormous amount of valuable information packed into those six paragraphs. Earl's signal-to-noise ratio is very high.

Of course you are free to disagree with him, and I would be surprised if you do not, but I'm not making things up out of thin air - I'm accepting the views of experts in the field who you happen to disagree with. Just for the record, my acceptance of Geddes and Griesinger on this subject is not arbitrary, it is based on experiences which constitute "evidence" to me but would of course be "anecdotal at best" to anyone else.

Maybe we are bound to disagree...?

Maybe so. And that would be okay with me.

 

[Duke LeJeune]

Finally moving on to KlausR.'s subsequent post:

So to make his point Griesinger uses a seat that sounds bad (to him), fair enough. But what about that seat that is “better than most halls”? This seat also has first reflections, but still sounds great. If first reflections are detrimental as a matter of principle, then all seats should sound bad, shouldn’t they? But they don’t.

Quoting Geddes: "The earlier and the greater in level the first room reflections are, the worse they are." And of course first reflections are not the only thing happening at any given seat in a concert hall... multiple things are happening.

Briefly, Griesinger finds the difference between a good seat and a bad seat in a concert hall largely boils down to whether or not the direct sound is distinctly audible. I believe this principle is applicable to home audio as well, hence my emphasis on the desirability of a "time gap" in between the first-arrival sound and the strong onset of reflections. In other words, unless the early reflections "get out of the way" such that the direct sound is distinctly audible, clarity is degraded. If you go back and re-read Mike Lavigne's experience with his dedicated room, Post Number 11, you'll be reading about a small domestic room application of this principle.

In the other thread you said: “We learn that early reflections are detrimental while late reflections are beneficial". Yes, for a bad sounding seat in Boston Symphony, with the first reflections delayed by up to 80 ms, listening to the right channel of a binaural recording. How relevant can that be for 2-channel stereo in small domestic rooms.

I have posted a fair amount of information from respected sources which is relevant for 2-channel stereo in small domestic rooms and which has nothing to do with Griesinger's clips. So if you are going to dispute my statement that "early reflections are detrimental while late reflections are beneficial" (which in retrospect was painting with an overly broad brush), you have to dispute a lot more than the relevance of those Griesinger clips.

At any rate, Klaus I thank you for motivating me to come up with information supporting my statements, as I'm sure you were not the only person wondering about them.

 

[Duke LeJeune]

IME there's more than one way to achieve high quality immersive sound. I've heard different approaches work well, and believe there is some personal preference involved.

Agreed!

I've heard Duke's various iterations over the years at RMAF and in the context of a very directional speaker, adding a 2nd source in the way Duke describes results in bipole-like reflections that many people enjoy, it reduces the perceived difference between a horn and a cone 'n dome, and gives people a listening experience more like a conventional speaker... but with the advantage of a less demanding setup. I'd also say the way Duke and James design the 2nd source is superior to most dipole/bipole as it's aimed to produce longer delays.

Thank you very much. Yes we are trying to do something sufficiently room-friendly that extensive room treatments are not required for good results, though they are still quite beneficial. I suppose that on a forum called "What's Best" the question may arise of how our approach + room treatment competes with the best conventional speakers + room treatment, and I'm cautiously optimistic.

In the systems I've heard with conventional speakers that achieve good immersion, it takes some dedication to room design and setup to make sure the 1st reflections integrate with the direct sound properly... I'm not convinced it's quite as important with horns, as AGs generally have horrible polars but people get good results out of them anyways. Which brings into question the real importance of CD type horns, I don't believe they are in fact superior to a LeCleach... but that's another subject!

I have not worked with LeCleach horns. I have worked with exponential horns and Tractrix horns and a few other non-CD types, and they do some things better than constant-directivity horns and some things not as well. It would be very interesting to work with a LeCleach horn some day, as from my reading thus far it seems to be the most innovative competing profile. Let's just say that I do not necessarily rule out non-CD types for some applications.

In any case, I'd say that room reflections are NOT required to produce a 3-D immersive soundstage, but may be preferred by some listeners in rooms that are large enough to allow them. I think many rooms are not, just like many rooms are really not large enough to give ideal performance from dipole speakers. In these smaller rooms I prefer a larger proportion of direct sound and no 2nd stream at all. If the room is large enough I still don't prefer them but I can see why some do and I believe it's possible to have them while maintaining immersion.

All that is required for that Second Stream of reverberant sound is, a time-gap in between the first arrival sound and the strong onset of reflections, along with a decent amount of reflected sound. A second sound source (like in my polydirectionals, or like what a dipole does) is NOT required. So you could have narrow-pattern LeCleach horns which inherently do not interact with the sidewalls, and then assuming you aren't sitting with your back against the wall, you should have a decent time delay before the first significant reflections start to arrive. Those reflections would then constitute our Second Stream.

 

[DaveC]

All that is required for that Second Stream of reverberant sound is, a time-gap in between the first arrival sound and the strong onset of reflections, along with a decent amount of reflected sound. A second sound source (like in my polydirectionals, or like what a dipole does) is NOT required. So you could have narrow-pattern LeCleach horns which inherently do not interact with the sidewalls, and then assuming you aren't sitting with your back against the wall, you should have a decent time delay before the first significant reflections start to arrive. Those reflections would then constitute our Second Stream.

Sorry I wasn't clear, I meant no ADDED 2nd stream in the form of bipole/dipole, rear firing tweeter, etc.

I experimented with using the midrange horn driver with an open-backed box and stuffed it with varying thickness of reticulated foam to adjust how much rear output I go,t and in my room, with the driver about 3 ft from the front wall, I think the box was best completely sealed, as long as the backwave was properly absorbed. However, I can see in a larger room it's possible I'd like the rear output more.

 

[Duke LeJeune]

Sorry I wasn't clear, I meant no ADDED 2nd stream in the form of bipole/dipole, rear firing tweeter, etc.

That's makes sense, thanks for clarifying.

I experimented with using the midrange horn driver with an open-backed box and stuffed it with varying thickness of reticulated foam to adjust how much rear output I go,t and in my room, with the driver about 3 ft from the front wall, I think the box was best completely sealed, as long as the backwave was properly absorbed. However, I can see in a larger room it's possible I'd like the rear output more.

I'm trying to picture this - was it a cone midrange driver in an open-backed box? Or a midrange compression driver with the back removed, which was then mounted in an open-backed box?

Either way, this is pretty much consistent with my experiences. Unless there is sufficient path length for the backwave, ime it can end up doing more harm than good.

 

[DaveC]

That's makes sense, thanks for clarifying.



I'm trying to picture this - was it a cone midrange driver in an open-backed box? Or a midrange compression driver with the back removed, which was then mounted in an open-backed box?

Either way, this is pretty much consistent with my experiences. Unless there is sufficient path length for the backwave, ime it can end up doing more harm than good.

It's a 4.5" cone driver, Louis at Omega helped me develop it for use in the LeCleach horn. It has a rising response that the horn levels out. It also has the EnABL treatment which keeps the driver sounding the same at all SPLs, without it you get a bit of "shout" at higher SPLS. I'm a big fan, but I have not been able to apply it myself yet, it takes some practice!

The box was simply a 7-8" cube with a back that could be removed, then I had layers of foam that could be stuffed into the box behind the driver to attenuate the backwave. This definitely drove home how important it is to kill the backwave in a sealed box, it sounds really bad if you simply put the back panel on. Currently I have a sealed box with more room for absorption behind the driver but I can see how a tapered back would be great, like B&W and Vivid's snail-shell design.

 

[Duke LeJeune]

It's a 4.5" cone driver, Louis at Omega helped me develop it for use in the LeCleach horn. It has a rising response that the horn levels out.

That sounds absolutely brilliant to me. I was a dealer for Louis many years ago.

The box was simply a 7-8" cube with a back that could be removed, then I had layers of foam that could be stuffed into the box behind the driver to attenuate the backwave. This definitely drove home how important it is to kill the backwave in a sealed box, it sounds really bad if you simply put the back panel on. Currently I have a sealed box with more room for absorption behind the driver but I can see how a tapered back would be great, like B&W and Vivid's snail-shell design.

You might find a stuffed transmission line to work well for absorbing that backwave energy, ideally with the line length being at least twice the wavelength at the lower crossover frequency. The idea would not be to get any sort of reinforcement from the transmission line, but rather to minimize any reflections back into the cone, nor would very much energy be escaping from the end of the line either.

 

[KlausR.]

Hello Duke,

I don't think I said that “first reflections have detrimental effects under all circumstances,” but obviously I have communicated poorly if that is what you heard.

Indeed, the combination of different statements in the other thread led the reader to believe that first reflections are bad, because Griesinger’s clip demonstrated just that:

“Done wrong in a home audio setting, the reverberant field degrades timbre and clarity...”

“So what do we learn from this [Griesinger’s clip]? We learn that early reflections are detrimental ...”

“So, we want to minimize the early reflections but encourage the later ones. The time scale will be of course different in our smaller rooms, but the general principle seems to be valid”

Let me try again: Early reflections can have both beneficial and detrimental aspects. Late reflections in a home audio setting generally go not share the detrimental aspects of early reflections (assuming they are spectrally correct). So imo it makes sense to manipulate the early reflections to minimizes detrimental aspects, while encouraging beneficial late reflections.

I assume that by early reflections you mean first, and by late reflections you mean reverberation? Since there are perception thresholds, which mean that the reflection somehow changes the sound or image, the reflection has an effect. Whether that effect is considered as detrimental or not , is the listener’s decision. In his 2006 AES paper Toole wrote: “The inevitable conclusion is that, in natural listening, room reflections are not problems.”

Regarding your call for “hard/scientific evidence”, most of my information comes from a few trusted individuals who have read the relevant research, who have done much of it themselves, and who I believe are qualified to speak without having to justify their every statement. Toole, Griesinger, and Geddes come to mind.

To the best of my knowledge of those three only Toole has done (published) research on first reflections in small rooms himself. Toole again, this time from his book:

“When listening to music, all preferred levels are far above the natural reflections
provided by small rooms. . . . They are simply not consequential factors in
this matter. . .”

“There appears to be no evidence in the now substantial literature that these
first-order lateral reflections are problems in normally furnished or the equivalent
moderately treated rooms.”

“The ritual had its origins in recording control rooms—listening in stereo—encouraged by alarmist cautions about comb filtering (see Chapter 9) or degraded speech intelligibility (see Chapter 10) or masking of other reflections within recordings (Olive and Toole, 1989). When examined, none of these turn out to be problems.”

So, here is my fall-back position:

“The earlier a reflection arrives the more it contributes to masking the direct sound.”

I understood from one of your posts that masking should be understood as blurring or smearing. This statement of Griesinger is probably based on concert hall acoustics.

“Envelopment is perceived when the ear and brain can detect TWO separate streams: A foreground stream of direct sound, and a background stream of reverberation. Both streams must be present if sound is perceived as enveloping.” (Implied is that there must be a time gap in between the two streams to differentiate them.)

In small rooms only the first 10 or 20 dB can be perceived, the rest is masked by the continuing music:

Blesser (2001), “An interdisciplinary synthesis of reverberation viewpoints”, J. of the Audio Engineering Society 2001, p.867
Toole (2008), „Sound reproduction - Loudspeakers and rooms”, Focal Press

Toole: “Considering the distances at which we listen in our entertainment spaces and control rooms, it is clear that we are in the transitional region, where the direct and early
reflected sounds dominate, and late reflected sounds are subdued, and progressively attenuated with distance.”
Looks like there is only foreground but no background stream.

Here are four ways in which early reflections may be detrimental:
It is the earliest reflections which are most likely to degrade clarity (Griesinger).

Griesinger’s statement is based on what? Concert halls, or stereo in small rooms.

2. Coloration from reflections peaks between 2 and 3 milliseconds as I recall (James M. Kates).

In his 1979 paper on loudspeaker cabinet reflections Kates describes listening tests, but without giving any details relating to setup and test signals. As jj so nicely said: tell me what you want to hear and I’ll create you a test signal that allows to hear it.

The 1978 Kuhl paper, first reflections - room size

In that paper Kuhl refers to experiments published in a paper in 1956. In these experiments artificial reverberation with different delays was added to recordings of speech and music, which delays provided the impression, when listened to in a heavily damped studio control, of the recording having been made in a larger room. The listening tests were repeated in a normal domestic room and the early reflections of that room did not change that impression. In further experiment a single 35 ms reflection was added to the recording between direct sound and delayed (50 ms) reverberation. This reflection increased the impression of the larger room only slightly.

Earl Geddes' clip on first reflections

He mentions the work of Griesinger and his theory of hearing, based on large room acoustics. I had a quick look at the Griesinger papers in my archive and found the following which might be the basis for the 10 ms Geddes advocates: “We localize sounds at low frequencies through the HD. Can we localize low frequency sounds in a room? The answer is yes. The human localization mechanism is highly dominated by the ITD of a transient, if transients exist in the source material. Transients are not corrupted by reflections if the room is large enough - and 10ms of reflection flee time is enough. Thus we can localize a drum hit even if it has been sharply band limited to frequencies below200Hz (or even 60Hz).” From “Spatial Impression and Envelopment in Small Rooms”, AES paper 4638.
Nothing in those papers about first reflections in small rooms with stereo systems.

Does that paper [Hameed, Yadav] really say first reflections play zero role in room size perception?

Hameed: However, since it was established in [3] that humans are incapable of discriminating between the exact amplitude, timing and direction of early reflections, it was hypothesized that precise details of early reflections could not contribute significantly to room size perception, and early reflections was, hence, left out of this study.

Yadav: Even though reverberation time was the strongest global predictor for room size judgements in this study, the finegrained analyses suggest the likelihood of relatively smaller volume rooms being judged with a different psychoacoustic mechanism than larger rooms. Rooms were recording booth, music practice rooms, lecture theatre, performance space, music recital hall.
But this issue is a bit off-topic IMO.

Geddes' White Paper “Loudspeaker System Design Philosophy”: The earlier and the greater in level the first room reflections are, the worse they are.

The earliest and strongest reflections in a room generally are the vertical ones. From his clip: “He [Toole] also believes, as I do, that the vertical reflections are not a serious matter.” Is Geddes contradicting himself?

Of course you are free to disagree with him, and I would be surprised if you do not, but I'm not making things up out of thin air - I'm accepting the views of experts in the field who you happen to disagree with.

When I read Toole’s 2006 AES paper, which basically is an extremely short version of his book, I really was surprised, so I got copies of most, if not all, of his source papers. And I checked the most relevant technical journals for more: JAES, JASA, Applied acoustics, J. of sound and vibration, J. of the Acoustical Society of Japan, Rundfunktechnische Mitteilungen, Wireless World, IEEE, J. of Building Acoustics, BBC Research Disclosure, Acustica, Acustica - acta acustica. Having read all the papers relating to room acoustics and related psychoacoustics I could find, I agree with Toole simply because of the available (lack of) evidence. Thin air is not my piece of cake either. If Geddes has read all the relevant stuff and comes to a different conclusion, that’s fine.

I have posted a fair amount of information from respected sources on that topic which is relevant for 2-channel stereo in small domestic rooms and which has nothing to do with Griesinger's clips.

In the first instance I did not rely on a respected source, i.e. Floyd Toole, but having read what he has, and more, my conclusions are about the same. What concerns disputing more than Griesinger’s clip, all available facts are in my write-up (which took me about 2 years to produce). I have all those papers mentioned there, and sending the PDFs to anyone interested is no problem.
Klaus

Btw., I found a paper which probably (didn't read it yet) describes how Griesinger’s Matlab clips were produced
http://www.aes.org/e-lib/browse.cfm?elib=18569

It’s not open access, so I probably shouldn’t attach it to a forum reply. If interested, drop me a forum-mail and I can send the PDF.

Klaus

 

[Duke LeJeune]

Thank you Klaus for your in-depth, knowledgeable, and courteous reply.

Hello Duke, ... the combination of different statements in the other thread led the reader to believe that first reflections are bad...

Yes, I should have been more careful with my wording... I didn't use the wording "all circumstance", but can see how what I said could come across that way.

I assume that by early reflections you mean first, and by late reflections you mean reverberation?

I tend to think in terms of "before 10 milliseconds = early" and "after 10 milliseconds = late", within the context of a small listening room, like what we typically have at home. There could be first reflections which arrive after 10 milliseconds, and there could be secondary reflections which arrive before 10 milliseconds.

Whether that effect is considered as detrimental or not , is the listener’s decision.

Agreed.

In his 2006 AES paper Toole wrote: “The inevitable conclusion is that, in natural listening, room reflections are not problems.”

Do some rooms sound better than others? If so, then the effects of reflections are in play. I submit that it is a semantic choice whether or not one uses the word "problems" to describe the effects of reflections which result in some rooms not sounding as good.

Toole again, this time from his book:

“When listening to music, all preferred levels are far above the natural reflections provided by small rooms.”

This is one of the reasons why I suggest increasing the amount of reflections within a small room by adding relatively late-onset reverberant energy, such as that contributed by the backwave of a properly positioned dipole loudspeaker.

Toole: “Considering the distances at which we listen in our entertainment spaces and control rooms, it is clear that we are in the transitional region, where the direct and early reflected sounds dominate, and late reflected sounds are subdued, and progressively attenuated with distance.”

Looks like there is only foreground but no background stream.

In most home audio set-ups, that's probably true... there isn't enough of a time gap in between the direct sound and the early reflections for the direct sound to be distinctly audible.

However in really outstanding home audio set-ups, like in Mike Lavigne's room, I think we are far more likely to find a foreground stream of direct sound followed by a background stream of reverberant sound.

Griesinger’s statement is based on what? Concert halls, or stereo in small rooms.

He doesn't specify whether his statement is generally true (which is what I think), or only true for large rooms like concert halls (which is what I think you think). Three days ago I wrote to him asking for clarification on exactly that point. I haven't heard back from him yet.

[quoting Griesinger]The human localization mechanism is highly dominated by the ITD of a transient, if transients exist in the source material. Transients are not corrupted by reflections if the room is large enough - and 10ms of reflection flee time is enough [emphasis Duke's]. Thus we can localize a drum hit even if it has been sharply band limited to frequencies below200Hz (or even 60Hz).”

Nothing in those papers about first reflections in small rooms with stereo systems.

Thank you for digging that up. If "Transients are not corrupted by reflections if the room is large enough - and 10ms of reflections free time is enough", THEN some amount less than 10ms of reflections free time is NOT enough, and transients ARE corrupted.

Imo that quote tells us A LOT about early reflections in small rooms - that they can corrupt transients if they arrive before 10 milliseconds (exactly when isn't specified). THEREFORE, it would be beneficial to push those early reflections back at least 10 milliseconds where feasible.

So while there may not be anything explicitly about first reflections in small rooms in stereo systems in those papers, the information therein tells us that early reflections can degrade transients. So I'd like to add this to my list in Post Number 29 above:

5. Reflections arriving before 10 milliseconds can corrupt transients.

Hameed: However, since it was established in [3] that humans are incapable of discriminating between the exact amplitude, timing and direction of early reflections, it was hypothesized that precise details of early reflections could not contribute significantly to room size perception, and early reflections was, hence, left out of this study.

Yadav: Even though reverberation time was the strongest global predictor for room size judgements in this study, the finegrained analyses suggest the likelihood of relatively smaller volume rooms being judged with a different psychoacoustic mechanism than larger rooms. Rooms were recording booth, music practice rooms, lecture theatre, performance space, music recital hall.

Again, thank you for including this information.

Based on those quotes, I do not think either study conclusively proves that, for small rooms, "reverb time is the cue, not first reflections [emphasis Duke's]." I think they BOTH contribute.

The earliest and strongest reflections in a room generally are the vertical ones. From his clip: “He [Toole] also believes, as I do, that the vertical reflections are not a serious matter.” Is Geddes contradicting himself?

I don't think so. My understanding is that Geddes and Toole both believe early reflections in the vertical plane are perceptually relatively benign. (And I think Earl would say that a strong early vertical reflection is less benign than a weak one.)

Having read all the papers relating to room acoustics and related psychoacoustics I could find, I agree with Toole simply because of the available (lack of) evidence. Thin air is not my piece of cake either. If Geddes has read all the relevant stuff and comes to a different conclusion, that’s fine.

And I am willing to concede that the majority opinion seems to be consistent with Toole's position that "room reflections are not problems."

I side with Geddes and his minority opinion that "the earlier and the greater in level the first room reflections are, the worse they are."

Klaus, it looks to me like you and I are in the position of dueling via quotes by experts who disagree with each other on this topic, and unless one of us is ready to concede that the other's experts are right and ours are wrong, I think the time may have come for you and I to consider "agreeing to disagree".

And thank you for the offer to provide me with a copy of the paper that would explain Griesinger's Matlab processing protocol, but I probably wouldn't understand it.

 

[Duke LeJeune]

For anyone who has followed the exchange between Klaus and me on this topic, I'd like to point out that I'm probably coming from a different angle than he is. I THINK Klaus approaches the topic of reflections in a small room from the perspective of a researcher studying peer-reviewed research. I approach the topic from the perspective of a loudspeaker designer hoping to compete on the basis of sound quality.

Missing from my conversation with Klaus is the in-house research conducted by myself and Jim Romeyn, because non-peer-reviewed research would have no credibility in the context of our conversation, whether or not our listening tests were properly controlled to eliminate bias.

So I have to argue my position based on the works and words of others who do have credibility.

Out there in the world of published research there is insufficient information to answer all questions which may arise. For instance, in his book Toole tells us what the optimum direction is for reflections to arrive from. And Geddes suggests a target time delay for early reflections. Which matters the most? If we can choose between reflections arriving from the ideal direction but at a non-ideal time, and reflections arriving at a more ideal time but from a less ideal direction, which should we choose? There is no study out there which answers that question.

In situations like this, I think in-house experimentation is the quickest way to find out. But our in-house findings would not contribute anything to a discussion such as the one Klaus and I have been having.

So in this thread Klaus and I have been debating whether or not early reflections have detrimental attributes, but in-house AudioKinesis is working on topics much further down the road than that.

 

[KlausR.]

Hello Duke,

Do some rooms sound better than others? If so, then the effects of reflections are in play. I submit that it is a semantic choice whether or not one uses the word "problems" to describe the effects of reflections which result in some rooms not sounding as good.

When a room sounds bad, it is generally reverb time which is too high. Speech intelligibility is a good indicator. As Toole said/wrote, what we are listening too above the Schroeder-frequency is direct sound + interaction of the speakers’ off-axis response and room boundaries. If the off-axis response is rollercoaster an otherwise good room might not do the trick all on its own. If the off-axis response is well shaped and you have a wall on the left and an opening on the right it might sound strange.

He [Griesinger] doesn't specify whether his statement [reflections are bad] is generally true (which is what I think), or only true for large rooms like concert halls (which is what I think you think). Three days ago I wrote to him asking for clarification on exactly that point. I haven't heard back from him yet.

I looked up the seat chart of Boston Symphony to see where seat DD11 is which Griesinger used for his lecture demo, and am not surprised about the results: that seat is acoustically very asymmetric, no wonder its sounds bad. One certainly cannot generalize from there.
https://www.bso.org/brands/bso/plan-your-visit/attending/building-maps-and-seating-charts.aspx

Thank you for digging that up. If "Transients are not corrupted by reflections if the room is large enough - and 10ms of reflections free time is enough", THEN some amount less than 10ms of reflections free time is NOT enough, and transients ARE corrupted.

I suppose that Rakerd et al., “Localization of sound in rooms, II: The effects of a single reflecting surface”, JASA 1985, p.524 is the basis for that statement. That stuff is not easy to digest, but the issue seems to be the precedence effect and its failure under certain circumstances, and one would have to dig much deeper into the scientific literature to get a good grasp, so I’ll leave it at that. Rakerd’s experiments, however, were made in an anechoic chamber with a single speaker as sound source and a single reflection. What happens when there are reflections from left AND right sides in a reflective room with stereo loudspeakers?

Klaus, it looks to me like you and I are in the position of dueling via quotes by experts who disagree with each other on this topic, and unless one of us is ready to concede that the other's experts are right and ours are wrong, I think the time may have come for you and I to consider "agreeing to disagree".

Well, I’m siding with Toole, not because it’s Toole, but because the relevant literature supports his point of view. That’s why I would like to see the literature Geddes is basing his differing opinion upon. Based on what I’ve read so far I disagree with him. If you happen to have good relation with Earl, maybe you could ask him.

Klaus