[Vapor Audio]

Arguably the most basic thing a speaker has to get right is the tonal balance, but because tonal balance is very much a room-interaction thing (since most of what we hear is reflected energy), right off the bat we're swimming in the "deep end" if we dive into the question of which sonic attributes matter the most.   So what looks like a tangent is just going deep into one of the basics, in my opinion anyway.

I just have to say, I couldn't disagree with this more.  IMO tonal balance is determined primarily by the first impulse, as is the majority of how we perceive a speakers sound.  When I take speakers (and we make models with very well controlled and even off-axis response) between my 3 different rooms, the overwhelming characteristics of how they sound remain intact.  What changes in how well they image and disappear into the room.

[*Scotty*]

Is a speaker more likely to sonically disappear in room that is more live or one that is better damped.
More specifically is a room that has minimal comb filter effects above the Schroeder frequency more likely to help a loudspeaker disappear when compared to a room that has not been treated so as to address these problems.
Scotty

[Duke]

I just have to say, I couldn't disagree with this more.

That's what makes this game so interesting!  If we agreed on everything, one of us would be redundant!

IMO tonal balance is determined primarily by the first impulse, as is the majority of how we perceive a speakers sound.  When I take speakers (and we make models with very well controlled and even off-axis response) between my 3 different rooms, the overwhelming characteristics of how they sound remain intact.  What changes in how well they image and disappear into the room.

This is what I think is going on:  It is precisely BECAUSE your speakers are getting the off-axis sound right that their characteristics remain intact from one room to the next!  Whatever the room bounces back, it is supporting the timbre of that first-arrival sound, whether the room bounces back a little energy or a lot.

On the other hand, if your speakers had poor off-axis behavior, the tonal balance would change audibly depending on how efficient the room was at returning that off-axis energy to you because it would be skewing the tonal balance according to how loud it was relative to the first-arrival sound. 

So the role that your speakers' off-axis behavior is playing is not intuitively obvious because they're doing it right, but it would be manifesting as an audible change in timbre from room to room if they had poor off-axis behavior.   

The audiophile's natural inclination is to "blame the room" if a speaker goes from sublime to suck when put into a different room, but it might well be the fault of the speaker's off-axis behavior.   Note that unamplified instruments don't go from sublime to suck when going from one room to another (though they can overpower too small a room). 

[G Georgopoulos]

Inside the speaker box the back of the cone produces exactly the same music the front of the cone does,it is like a speaker in a room ,think the room as a speaker box, then you'll see my point...

[JLM]

Inside the speaker box the back of the cone produces exactly the same music the front of the cone does,it is like a speaker in a room ,think the room as a speaker box, then you'll see my point...

Sorry, I don't see your point.  Could you elaborate please?

[JLM]

So Duke, with all this controlled directivity and ambience/reverberant field discussion, can you describe in your opinion what the ideal directivity would be?

The Princeton study shows many graphs and oblique numbers but offers no conclusions, it only serves to muddy the waters for me, especially since one of the lowest scoring speakers is one I own (previous version) and love.

[FullRangeMan]

Inside the speaker box the back of the cone produces exactly the same music the front of the cone does,it is like a speaker in a room ,think the room as a speaker box, then you'll see my point...

George are refering to the acoustic negative back side driver cone sound emissions.
The front driver sound emissions from the cone are in positive acoustic phase and negative at the back side of the driver.
On boxed lousdspeakers the back emissions are not utilized as music in the room, but they are muffled by the box.
Seems was this george was try to say.

[Duke]

3) Which one is more important to achieve, question 1 [pattern matching in the crossover region] or question 2 [overall controlled directivity]?

I'd choose generally well-behaved pattern control, wide or narrow, over pattern-matching in the crossover region, if I could only have one and not the other.  It's possible to "feather" dissimilar radiation patterns in the crossover region so that we don't end up with a big problem either on or off axis, but kinda spread the misery to both a bit.

4) How low of a frequency should one aim in controlling directivity? 1khz? 500Hz? 250Hz?

Earl cited David Griesinger in suggesting 700 Hz as the target, and I certainly don't have any information to dispute that.   In practice, I think we can get away with controlled directivity down to an octave or so north of there, and if we cheat and feather it in carefully, maybe even a bit higher than that.  700 Hz calls for a ballpark 20-inch wide woofer, or (more likely) a short horn on a smaller woofer, and a correspondingly large waveguide.

5) How important is a flat DI versus a smoothly rising DI (from say 1khz to 20khz) as long as the DI is at least 6dB?

I'd rather have a fairly flat DI but when dealing with a rising DI, I start leaning towards compensating with a rear-firing tweeter.

 

6) Relative to controlling directivity in the horizontal axis, how important is it to control directivity in the vertical axis?

The horizontal matters more because a) listeners are more likely to be distributed horizontally rather than vertically and b) the lateral reflections are especially important.  But the vertical still contributes to the spectal balance of the reverberant field so its overall content matters, even if we don't end up with symmetry in most cases.   Recently I've started embracing narrower directivity in the vertical plane, which takes some wind out of the floor and ceiling bounce, and two woofers vertical helps to mitigate the floor-bounce notch because they will have different path lengths. 

7) How does all of the above impact room treatment at the 1st reflection points on the side wall, ceiling, floor, and front wall? What real world experience have you had (irrespective of measurements)? How does this affect image specificity vs sound staging and spaciousness?

In my opinion we get best imaging from a first arrival sound that is not quickly followed by strong, distinct ("specular") early reflections.  And our spaciousness cues come mainly from later-arrival (mostly after 10 milliseconds) reflections.   Normally we'd have to have a lot of early reflected energy if we still want to have a lot of later-arrival reverberant energy, but Earl's cross-firing setup and Jim Romeyn's Late Ceiling Splash concept give us ways to skew the reflections in favor of more late-arrival reflections without paying a penalty in too much early-arrival reflections.  In my opinion anyway.  (My various bipolars had the same goal in mind as Jim's LCS concept). 

I'm generally inclined towards diffusion rather than absorption at first reflection zones, as long as we don't have an obvious problem like slap-echo.  But having been in a professionally treated room (by Jeff Hedback), nowadays I really hesitate to armchair quarterback room treatment issues.   A professional not only knows where the goal posts really are, but he can take remote measurements and tell you what type of treatment and what values and where to put them... not unlike crossover design, where what type of component, what values, and where in the circuit all make a big difference in the final outcome.  And if he doesn't have to use room treatments to "fix" the speaker's off-axis response, we're already that much ahead of the game.

[Duke]

The Princeton study shows many graphs and oblique numbers but offers no conclusions, it only serves to muddy the waters for me, especially since one of the lowest scoring speakers is one I own (previous version) and love. 

Imo Toole was already way ahead of the Princeton study decades ago, as far as correlating measurements with subjective preference.  After all, we're designing for ears, not test equipment, so if we what we're measuring doesn't correlate with what we hear (acknowledging that we can be fooled by our biases), then we're either measuring the wrong thing or our measurements are incomplete or we're interpreting our measurements incorrectly. 

So Duke, with all this controlled directivity and ambience/reverberant field discussion, can you describe in your opinion what the ideal directivity would be?

Good question.  I'll have to think about it a bit. 

[JLM]

George are refering to the acoustic negative back side driver cone sound emissions.
The front driver sound emissions from the cone are in positive acoustic phase and negative at the back side of the driver.
On boxed lousdspeakers the back emissions are not utilized as music in the room, but they are muffled by the box.
Seems was this george was try to say.

Another nasty unspoken secret about speaker cabinets.  Keep in mind that while many manufacturer's gloat about how inert their cabinets are, the driver cone is extremely lightweight and is probably 1,000 times more acoustically transparent while the sound pressures inside the cabinet vastly exceed those we hear on the outside.  At the same time dipole/open baffle builders gleefully support the concept that we should hear this out of phase sound.  The key would be to keep the back-wave from reflecting back towards the driver.  Various materials/designs are used to reduce reflection have been introduced but perhaps the best is to slope the back wall relative to the front baffle as can be seen in pipes and transmission line speakers. 

[poseidonsvoice]

Duke,

Thank you for going through all my questions so thoroughly.

Can't wait to see you at RMAF!

Best,
Anand.

[G Georgopoulos]

George are refering to the acoustic negative back side driver cone sound emissions.
The front driver sound emissions from the cone are in positive acoustic phase and negative at the back side of the driver.
On boxed lousdspeakers the back emissions are not utilized as music in the room, but they are muffled by the box.
Seems was this george was try to say.

Exactly FRM

JLM,if we were to be placed inside the speaker box and listen to the back of the driver,all those problems going on there are similar to listening the front of the driver inside the room (room acts as a speaker box too)

All speakers are effected by room acoustics,that's what I was trying to say.

[Duke]

So Duke, with all this controlled directivity and ambience/reverberant field discussion, can you describe in your opinion what the ideal directivity would be?

Okay first of all, the direct sound needs to stand apart from the reflections, because the direct sound is what we get most of our cues from.  According to David Griesinger, "The earlier a reflection arrives, the more it contributes to masking the direct sound."  A highly directional speaker will put less energy into those early reflections, and can be crossfired to delay and de-correlate the early sidewall reflections (in this context, by "de-correlate" I'm referring to the left speaker's first sidewall reflection arriving at the right ear and vice versa).   And the more directional we can make the speaker in the vertical plane, the less energy we're putting into the floor and ceiling bounces.  I think the 90 degree pattern is ideal in the horizontal plane, but something more narrow than that may be better in the vertical plane, maybe 45 degrees or so. 

Now here's another piece of the puzzle, again in Griesinger's words:  "Envelopment is perceived when the ear and brain can detect TWO separate streams:  A foreground stream of direct sound.  And a background stream of reverberant sound.  Both must be present if sound is to be perceived as enveloping." 

So what I try to do is introduce a time lag in between those two streams by a) using a directional main array aimed to minimize early reflections; and b) using a secondary directional array likewise aimed (in another direction) to differentiate its later-arriving reverberant field contribution from the direct sound and ensuing early reflections from the main array.   We don't consciously hear these events as separate, but according to Geddes we'd like a roughly 10 millisecond "lull in the action" in between the first-arrival sound and the onset of reflections.   (Where Earl and I differ is in how much energy we think ought to be arriving after that 10-millisecond interval.)

I haven't seen a whole lot written about the reverberant field enhancing timbre, but I'm quite confident that it does so, as long as several criteria are met:  First, the reflections should approximate the same spectral balance as the first-arrival sound, and it's okay if the room's natural acoustic signature is superimposed on top of that because it will still sound natural.  (For an illustration of this, walk from room to room in your house talking out loud, preferably when no one is around to question your behavior.   If you focus on the sound of your voice, you'll notice that the timbre changes.  But your ears accept the room's natural acoustic signature on top of your voice's reflections, so your voice never sounds wrong even though you can tell that it sounds different if you pay attention.)

Second, the early reflections should come from a different direction than the direct sound.  In tests where they use an electronically generated "early reflection" coming from the same speaker as the direct sound, it just sounded like coloration.   But if we can get a long enough time delay, it's okay for the reflections to come from nearly the same direction (like the backwave bounce of Maggeis).   However the ideal directions for reflections to arrive from are about 10 o'clock and 2 o'clock, this from Toole. 

So we might well ask the question - Which matters most: Having the reflections come in from the ideal direction, or having a longer time delay?  The only person I know of who has directly investigated that question is Jim Romeyn, and to my surprise he found that getting a longer time delay is more beneficial, within the practical limits imposed by our room sizes.  So I'm fortunate that he's letting me use his ideas.   

Anyway I think the ideal would be for both arrays to be directional to the tune of 90 degree patterns down to about 700 Hz in the horizontal plane, which means they'd be huge, especially if I also wanted a 45 degree pattern in the vertical plane from the main array.   Their relative levels would be adjustable to deal with differing room acoustic situations - there's a "Goldilocks zone" for how loud the direct and reverberant sounds should be relative to one another.   

Several decades ago Ken Kantor designed a speaker system for Acoustic Research that did just about everything I'm talking about here.  It was called the "Magic Speaker".  It was a commercial failure and I never got to hear a pair, but I think he was aiming at pretty much the same goal posts that I am.   There's some good information in the brochure pages at the link. 

[FullRangeMan]

Another nasty unspoken secret about speaker cabinets.  Keep in mind that while many manufacturer's gloat about how inert their cabinets are, the driver cone is extremely lightweight and is probably 1,000 times more acoustically transparent while the sound pressures inside the cabinet vastly exceed those we hear on the outside.  At the same time dipole/open baffle builders gleefully support the concept that we should hear this out of phase sound.  The key would be to keep the back-wave from reflecting back towards the driver.  Various materials/designs are used to reduce reflection have been introduced but perhaps the best is to slope the back wall relative to the front baffle as can be seen in pipes and transmission line speakers.

At the same time dipole/open baffle builders gleefully support the concept that we should hear this out of phase sound.
Most or all musical instruments emit phase - and + in various patterns and different intensities.
Phase - arent out of phase, its the half pair from the recording event, not - and + phases are need to for a real, natural soundstahe.

The key would be to keep the back-wave from reflecting back towards the driver.
If it was true sealed box would be the best enclousure.
Dipole and OB use the neg sound phase to made the a 3D and bigger sound stage than monopole speakers.
Monopole speakers have the soundstage and instrumental image harmonically poor, they lack 50% of the sound to complete the real musical event.
Unfortunately Dipoles and OB speakers lost 4 to 5dB at sensitivity in attenuation by use both sound phases.

[poseidonsvoice]

Duke,

Thanks again for the detailed explanation. We'll talk more about LCS etc...at RMAF. Since it's a commercial product, we should leave it out of this thread but the theoretical details are all valid here.

Several decades ago Ken Kantor designed a speaker system for Acoustic Research that did just about everything I'm talking about here.  It was called the "Magic Speaker".  It was a commercial failure and I never got to hear a pair, but I think he was aiming at pretty much the same goal posts that I am.   There's some good information in the brochure pages at the link.

It is an absolute TRAVESTY that, that speaker was a commercial failure! Way ahead of its time and if done right, would have answered much and advanced the state of the art. I would recommend everybody to thoroughly read the brochure from pages 1-5 (there is a picture on page 2), and seriously concentrate on the "acoustic" answers to the fundamental problems that plague loudspeaker designs today. Moreover, this talk by Linkwitz goes over that as well: http://www.linkwitzlab.com/ALMA'14/Sound_quality.htm

Best,
Anand.

[JLM]

At the same time dipole/open baffle builders gleefully support the concept that we should hear this out of phase sound.
Most or all musical instruments emit phase - and + in various patterns and different intensities.
Phase - arent out of phase, its the half pair from the recording event, not - and + phases are need to for a real, natural soundstahe.

The key would be to keep the back-wave from reflecting back towards the driver.
If it was true sealed box would be the best enclousure.
Dipole and OB use the neg sound phase to made the a 3D and bigger sound stage than monopole speakers.
Monopole speakers have the soundstage and instrumental image harmonically poor, they lack 50% of the sound to complete the real musical event.
Unfortunately Dipoles and OB speakers lost 4 to 5dB at sensitivity in attenuation by use both sound phases.

Dipole (including open baffle) fans should be supporting their circle, instead of deriding enclosed speakers here.  Regardless:

Except for large hung gongs I'm not aware of any musical instruments that act as dipoles in a vertical plane, but what does that have to do with how signals are captured by one or more microphones, then processed by two enclosed speakers in the recording, mixing, and mastering studios?

Yes, perhaps the ideal sealed speaker would look like the B&W Nautilus, with severely tapered shapes to redirect the back wave away from the driver (that actually acts like an infinite transmission line).

As home audio is all about enjoyment, to each their own, but a large blurred soundstage isn't necessarily better than a smaller more highly resolved one.

What do you mean by "instrumental image harmonically poor"?


The AR Magic Speaker is an interesting beast.


Regarding Linkwitz's presentation I agree with much of what he says (too many microphones/signal compression, subjectivity of audio cables, & passive speaker design), but the farther I read the worse it got:

- It is frustrating that production goals aim towards "good enough" but without newer/better gear we might have remained "blissfully ignorant" of it
- Not so sure tube amplification is obsolete
- He apparently lumps all "vented" enclosures together without consideration of various types (pipes, horns, transmission lines, etc.)
- No justification given for the figure 8 dispersion pattern as being ideal
- Claims that vented designs provide "unnatural bass enhancement" (and open drivers are more "natural"?)

[poseidonsvoice]

JLM,

I agree! I don't take Linkwitz' views on bass too seriously...I have heard wonderful examples of both multiple subs (regardless of LF alignment) and dipole bass (but without a multisub application/utilization). I do agree a lot with his views above 300 Hz, although the devil is more in the details regarding directivity and dispersion which Duke explained well. Edit: I admit, I haven't thoroughly read all of Linkwitz' site so I bet some of the answers/details are buried deep within! An absolute jewel of a website though!

Best,
Anand.

[DaveC113]

I'd post something but I pretty much agree with Duke. Toole/Harmon's preference testing indicates an extremely strong preference for speakers with smooth off axis response, but also for the presence of 1st reflections which is an area I don't necessarily agree with. As far as JBL products, I'd take a K2 over an M2... 

In my own not nearly as scientific testing I've found people like what they are used to, but after experiencing the benefits of controlled dispersion resulting in greatly reduced or eliminated 1st reflections and a much higher direct/reflected sound ratio they will prefer it. But at first it sounds a bit odd, almost a headphone-like experience. After you listen and your brain adjusts you notice a far more enveloping 3-D soundstage, the room boundaries acoustically disappear and fine detail is less obscured. Once you hear this you can't go back. I have heard a couple of conventional speaker systems achieve this, or something close to it, but it's dependent on room acoustics to a far larger degree. 

As far as frequency range for maintaining directivity, the lower the better but I don't think dipoles are the solution.

[Duke]

Toole/Harmon's preference testing indicates an extremely strong preference for speakers with smooth off axis response, but also for the presence of 1st reflections which is an area I don't necessarily agree with.

I've read seemingly divergent ideas about the desirability of early reflections from various authorities in the field.   One of the problems may be that the definition of "early reflections" isn't consistent - some people mean reflections that arrive after only one bounce ("first order reflections"), and some people are going by the time delay between the first-arrival sound and the reflection.  The former is looking more through the lens of acoustics, the latter more through the lens of psychoacoustics, and I look through the latter lens.  But that still doesn't clear up the discrepancies! 

When in doubt, I often go back to asking myself this question:  "How does what happens in a good recital hall differ from what happens in my living room, and how can I take a step in the direction of the recital hall?" 

Now many people have a philosophical objection to what seems to be imposing more of the listening room's signature on the sound via increased later-than-10-millisecond reflections.   When I've done informal demonstrations at audio shows and asked listeners whether they felt like they were hearing "more of the recording" or "more hotel room" when the additional reverberant energy was in play, the responses heavily favored "more of the recording".  Toole suggests that the ear/brain system can better decipher complex acoustic events if it gets multiple "looks" via reflections, and I think this is true, as long as we don't overdo it.

[poseidonsvoice]

I tracked down Ken Kantor's paper on the Magic Speaker: http://www.kenkantor.com/publications/magic_speaker/magic_speaker.pdf

Best,
Anand.