Unconventional Techniques for Achieving Oustanding Stereo Imaging

[Elias]

This thread has been split off from the "What is the ideal Directivity Pattern for Stereo Speakers Thread"
This post seems to be the one that inspired the discussion of unconventional techniques so has been chosen for the start of the thread

To get back to the original point of "What is the ideal directivity pattern for stereo speakers?" and where Toole has already given his insight (quoted) in it's "naked ugliness" about stereophonic cross talk, which I consider one reason besides pinna localisation for the cause of another stereophonic artefact namely pin point imaging, or pin point imaging of the two tweeters. Just now once again I disappointedly listen to a stereo triangle in my living room, and instead of perceiving a high freq phantom image between I hear two tweeters making sound at +/-30 lateral degrees. Even I block the half frontal hemisphere with thick absorbing pillow i.e. only one ear sees one speaker, now I perceive one tweeter playing but image is less pin point.
There are two cures to this stereophonic artefact that I have observd succesfull:
1) block the direct sound (see my experiments on stereolith thread)
or
2) use a very wide directivity tweeter to spatially homogenise high freq cues.

So to answer the question:
What is the ideal directivity pattern for stereo speakers?

My answer is two folded:
Ideal directivity pattern above 1kHz is such that left speaker illuminates the left wall only, and vise versa for the right speaker. Direct sound should not be present. To maximise the illumination side firing center speaker could be used.
Below 1kHz, where most of the music signal information is located in terms of modulation, very high directivity is beneficial to maximise information transfer through the room. Dipole line array maybe.


- Elias

 

[Radugazon]

Unique system, certainly, but still far from any ideal.

this is the last setting :

A fast description without the many flaws (mainly in the polars):

1. it's a side firing set up, but with the particularities of the linear quadripole radiation, there is also a participation of the direct sound, but that's not a real omni. It's not a stereolith too.
2. the lateral reflections are not a malediction, they arrive after almost 20 ms, their spectrum is similar but slightly different, I have to integrate them in the time window to finish with a subjectively (!!!) correct voicing
3. the rear speakers group is running in opposite polarity and inversed sides, it's participation in the final result is very important
4. the rear speakers voicing is not linear but integrates peaks and dips that cheat our front vs back perception (Blauerts), and this works, even when running alone it's hard to say that the sound comes from the back.

5. the sub & bass uses IB and a pair of neo-cardioids, these last can be put anywhere in the room, I had nice results but timing problems with the cardioids each at 90° of the listener
6. as this system uses a lot of drivers (too much I know), someone could be concerned by alignment problems...no problem, with an impulse friendly filter I can get something satisfying

7. the final space rendering depends mainly of the filter Fc and of it's slopes, to make it short, a real first order is enjoyable at first but is not very faithful (too wide, too much echoes), I have better all around results with VI th order if the Fc Mids/HF is around 1600 Hz (with 800 Hz it gets too narrow).



An OT question could be about the supposed XTC efficiency. Just by hearing, I can say that's it's contribution is huge, but measuring is more interesting.

here, the CSD of the HF group front an back :

the huge side wall reflection is obvious (around 20 ms delay)


now compare with the mid group front and back :

the side wall reflection is very attenuated...

But, careful, this is a mono sweep R&L. In stereo it's very different ! Only the common part of the signal will be attenuated like this on both sides, the proper lateral info remains transmitted, but only to the good ear (supposedly).


Listening to this stuff is nice, center presence, progressive lateralizations till 135° on some records (never in classic of course), depth, speakers unlocalized, symphonic ability, very spectacular but not adding too much of it's own,

but everybody must have already guessed that the sweet spot is +/- 5 cm.

PS: Elias, nice to see you, no wavelets here because I'm short in time...

all this with these words in mind:

But how to achieve it in practise? Can an omni do it? Omni has the problem it illuminates the contralateral wall as much as the ipsilatelar one. Clearly directivity pattern higher than omni would be needed. 180 deg half omni perhaps, aimed sideways.

Ceiling and floor should be illuminated too, because it's the only way to keep the energy longer on one side of the room.

Another option is to use Ambiophonics style room divider baffle Maybe that's one reason why Ambiophonics sound so good


- Elias

 

[LineArray]

There are two cures to this stereophonic artefact that I have observd succesfull:
1) block the direct sound (see my experiments on stereolith thread)
or
2) use a very wide directivity tweeter to spatially homogenise high freq cues.

3) Use decorellated radiators >2Khz, thereby cirumventing such pronounced
patterns arising from constructive and destructive interference in the
listening room.

Preliminaries are group delay being small throughout the used bandwidth
and lobes in angular dispersion being scattered and overlayed fine enough
(dense enough) over frequency.

This way there is no need to avoid direct sound. A bending wave transducer
with high enough modal overlap will do the job.

Control of the gross radiation pattern can be achieved by absorbers,
lenses, size, curvature, quasi dipole or quasi monopole configuration, just
the usual candidates.

 

[tnargs]

flooder speakers

diyAudio

What if (some ?) listener's preference of wide dispersion arises from
smoothing that (inter speaker) interferences by (also early) reflections ?

A speaker handling that problem in a different way, may not depend
on (early) reflections in the same way to statisfy even those listeners ...

Is this not worth thinking about ?

I don't see any connection between your quote from Toole and your post. Can you explain?

 

[LineArray]

I don't see any connection between your quote from Toole and your post. Can you explain?

In the discussion with graaf different estimations of usefulness/harmfulness of early reflections showed up.

Graaf posted that Toole citation in #153, so i just re-cited it to bring back into mind, that according to
Toole without reflections there is

"...acoustical crosstalk* that plagues stereo phantom images is present in its naked ugliness..."

* Own supplement: acoustical crosstalk and interspeaker interference

___

To me the question arose, whether smoothing interference patterns (by also early reflections) might be
a major reason for some listeners to prefer wide dispersion speakers, as the enjoyable listening zone might
be considerably widened, even if the zone of "proper imageing" will not be widened to same extent.

Preference for wide dispersion speakers would then as a consequence (at least partially) originate in a
shortcoming of the stereophonic system itself and not (solely) in (postulated) advantages in speaker/listening
room interaction.

If smoothing of interspeaker interference patterns can be achieved in a different way (not just using early
reflections introduced through wide dispersion), then also estimation of usefullness and enjoyability of more narrow
radiating speakers might change considerably even for that share of listeners preferring (conventional phase coherent)
wide dispersion speakers.

I suggested to partially overcome the interference by using radiators, which decorrelate phase between left and
right speaker above say 2KHz, if the listener is not in the median plane between the speakers.

That decorrelation can be achieved without exceeeding limits of group delay, which are accepted for high quality
reproduction.

 

[graaf]

first it was:

- excessive group delay

and

- early reflections

which was:

self evident from this brief description

then it became just:

group delay (depending on...

and in the end just:

possibly occurring with different delay due to side wall distance

so in a couple of posts You have travelled long way from "self-evidently excessive" to "possibly occurring" ...

and I responded on Yesterday, 02:15 PM with calm:

what is self-evident?

I am sorry but this is how it looked from my perspective because I am not used to rereading posts many times to find out whether author have possibly extensively edited them especially AFTER getting a response to them, because such ex post editing IS AT ODDS WITH GOOD MANNERS

and this was such a case because only at 02:22 PM You have edited Your post - instead of posting normal reply - and You added this:

If implemented perfectly, there would be no direct sound, so the early reflections...

and so on

ok! at least it is something I can discuss in contrast with:

on to the matter:

If implemented perfectly, there would be no direct sound, so the early reflections
in highs may not be considered "reflections" as such. OK, but also a bit "sci fi".

so blocking direct sound >1 kHz is sci-fi to You, really?

Delay of the highs relative to the lower spektrum would depend on room
and position (discontinuity in group delay).

so what kind of argument is that?
let me respond this way - a typical stereo pair of speakers is a complete waste of time because someone may place the speakers asymmetrically in a bathroom which inevitably leads to complete disaster blah blah blah

the quality of sound will IN ANY CASE depend on room and position of speakers in it, can't You understand it??!

Furthermore spektrum in highs would depend on the wall's frequency
dependent absorption coefficient with no direct sound establishing
"close tonality" in the reproduced sound.

"wall's frequency dependent absorption coefficient"? Have You got a persian carpet hanging on a side wall of Your listening room?

in 99% of cases of normal living rooms "wall's frequency dependent absorption coefficient" is pretty typical (and relatively low) and can be taken into account in the design -have You ever heard about HF level adjustment in speakers? it is especially popular solution in professional monitoring

yes, such configuration requires symmetry in that regard but any stereo setup requires symmetry, doesn't it?

The "indirect radiator" has to have a very constant directivity, to not
further introduce severely non flat group delay even in the highs itself.

All of these preliminaries and weak points are hard to be tackled to
make a "robust" reproduction system based on this concept.

yes it is really the same:

endless loop of repeated
arguments as associated with the ceiling flooder discussion some
time ago.

I say - give it a try and set it up as recommended! and only response I get is more and more blah blah blah

 

[LineArray]

so blocking direct sound >1 kHz is sci-fi to You, really?

Let me put it like this: "Blocking" direct sound from the speakers
in likely listening windows and having considerably wide dispersion
utilized for reflection (at side walls) at the same time at least calls
for compromise.

Maybe this formulation is more agreeable to you.

I did not bring up "sci fi" in the first place, it was about having
different properties at the same time.

Editing posts was not meant to displease those, who may
answer quickly.

Concerning the reflection coefficient: Even two symmetrically placed
"persian carpets" in the "reflection zones" of left and right speakers
impose a problem on the performance of the proposed "indirect only"
radiators, right ?

A "neutral" direct response is missing.

 

[graaf]

We are drifting well off from the topic.

have You read Elias' posts? Do You really consider them OT?

The other view of the objective is "what balance between direct and reflected?" That is, are we in the direct field, or at a distance where direct and reflected are equal or perhaps well into the reverberent field?

into the reverberant field because decorrelation occurs in the reverberant field

Even if we speculate that a certain reflection pattern is ideal we may not be able to achieve it within the dimensions of our room. If it turns out we can achieve it then likely a small variety of speaker room combinations might achieve it equally

both flooder and back-to-back configuration can achieve it in almost every room suitable for normal stereo, with one possible exception of acoustically overtreated audiophile room because it prevents the build up of adequate reverberant field

How can you discuss a speaker dispersion pattern without considering the total response it will create in the typical room? ...

total frequency response? can't it be equalized quite easily? on the other hand equalizing time response in room is more difficult

So what ETC (energy time curve) are we looking for?

I think David Moulton can be right:

My design philosophy for studios is: let's have a perfectly reflective space for 50 milliseconds and then let's have no reflections or reverb after that. So let's have all the early delays with as little frequency response change and as little amplitude loss as possible, and then nothing after that.

 

[graaf]

doubling the distance from listener to the CD horn system would give the same direct to reflected ratio as the wide dispersion system.

Who can proclaim that one is universally more desirable than the other, at least without talking about the room and listening distance?

but the listening distance is pretty typical and depends mainly on typical room sizes

 

[graaf]

My point is that there might be something else going on and the ETC (energy time curve) might not be the single defining factor. Furthermore, ETC is given in 2 dimensions and I think we dealing with 3 variables - time, intensity and frequency.

good points

 

[graaf]

Reflections will be there if you're in a room. The question is how can you work with that "mechanism", such that its damage is minimal, or in some cases an enhancement of the listening experience? (I knew I shouldn't have drank that coffee...)

yes, this is the question

 

[LineArray]

into the reverberant field because decorrelation occurs in the reverberant field

Side- or ceiling flooder approaches will not reliably provide a certain desired
decorrelation just by mainly utilizing a single reflection, which may be from
a flat and naked wall.

With such a device the "reflector" itself has to be part of the "scope of
delivery" to achieve somewhat reproducible results, as the "neutral direct
sound component" is completely missing in the passband of said
"transducer/reflector" arrangement.

Transducer and reflector have to make up a system, with the reflector to be mounted
in a defined way (at least a defined range due to distance and orientation) and the
group delay compensated in a tunable network.

Shape and material of the reflector would provide the desired amount
of decorrelation (diffusivity) and together with the transducer a certain
desired radiation pattern will build up.

Now a pair of speakers has to be placed in the room and a pair of reflectors
additionally ...

A suitably designed bending wave loudspaker optionally equipped with
means for dispersion control (narrowing)

- yields the desired dispersion and decorrelation
- is one single unit
- performs far more independent from idiosyncrasies of listening room and placement

- can be made without excessive group delay

(which e.g. is present in a transducer/reflector arrangement).

____

But - different approaches aside - i think Elias' post (including provided link)
is in no way off topic.

 

[graaf]

Side- or ceiling flooder approaches will not reliably provide decorrelation
just because they utilize a single reflection (which may be from a flat and naked wall).

in particular in case of a ceiling flooder You can have at least four ipsilateral first-order reflections - side-wall, front-wall, back-wall and - of course - ceiling reflection, no floor reflection

A suitable designed bending wave loudspaker

- yields the desired dispersion and decorrelation
- is one single unit
- performs far more independently from the properties of the listening
room and placement

- can be made without the excessive group delay

(of a transducer / reflector arrangement).

but is it real life option? can we diy such a bending wave loudspaker?

 

[speaker dave]

Originally Posted by speaker dave
The other view of the objective is "what balance between direct and reflected?" That is, are we in the direct field, or at a distance where direct and reflected are equal or perhaps well into the reverberent field?

(Graaf) into the reverberant field because decorrelation occurs in the reverberant field

You need to explain the term "decorrelation' that you constantly use. The reverberent field is a consequence of the direct sound so it is correlated with it. Not sure how a field (or how the treble of a speaker) can be "decorrelated" in a proper sense.

Originally Posted by speaker dave
"Even if we speculate that a certain reflection pattern is ideal we may not be able to achieve it within the dimensions of our room. If it turns out we can achieve it then likely a small variety of speaker room combinations might achieve it equally."

(Graaf) both flooder and back-to-back configuration can achieve it in almost every room suitable for normal stereo, with one possible exception of acoustically overtreated audiophile room because it prevents the build up of adequate reverberant field

If the desired goal is low reflection level in the first 20 msec, then a wide dispersion speaker in a small live room will not be able to achieve it. The goal may not be how to achieve more reflections but how to achieve less, at least in the early period.

total frequency response? can't it be equalized quite easily? on the other hand equalizing time response in room is more difficult

Definitely talking about achieving a particular time response. My assumption is that the direct response is flat. I haven't seen any evidence that the delayed reflections should have any particular frequency response but they should have a particular leval and delay. Frequency EQ is not the solution to a time based problem. This is what most of the DSP room EQ systems get wrong.

(Graaf) I think David Moulton can be right: "My design philosophy for studios is: let's have a perfectly reflective space for 50 milliseconds and then let's have no reflections or reverb after that. So let's have all the early delays with as little frequency response change and as little amplitude loss as possible, and then nothing after that."

Really?, that is directly opposite what the research is saying, in terms of suppresion of the early reflections to reduce timbral effects. And how do we get a studio to be "perfectly reflective for 50 milleseconds and then no reverb after that"? Acoustic noise gates?

He has invented a speaker that gives wide horizontal dispersion and now he needs to justify it. This is a popular notion that I see running through this thread and throughout the forum: "If reflections are just the same response as the direct sound then everything is okay." I've never seen any research to support this. If we have direct response and a single delayed reflection making the reflection flat in spectrum doesn't change the comb filtered mess of the combined response. It just determines the extent of the combing. Bech shows that an angular related hole in the reflection might just depress it below audibility. It is frequency selective directivity and in this case directivity was a good thing.

 

[graaf]

As you say, 3ms can give a satisfactory result, if overall reflectivity and reverberation are kept down to a manageable level.

yes, bathroom is no good place for a listening room
I believe that if overall reflectivity and reverberation are kept down to a manageable level then just anything above say 1.5 ms can bring satisfactory results

Moulton:

let's have a perfectly reflective space for 50 milliseconds and then let's have no reflections or reverb after that. So let's have all the early delays with as little frequency response change and as little amplitude loss as possible, and then nothing after that.

 

[speaker dave]

Originally Posted by speaker dave "doubling the distance from listener to the CD horn system would give the same direct to reflected ratio as the wide dispersion system. Who can proclaim that one is universally more desirable than the other, at least without talking about the room and listening distance?"

but the listening distance is pretty typical and depends mainly on typical room sizes

The point was that even with speakers of widely different directivity a reasonable seating distance difference could equalize their direct to reflected ratio.

 

[graaf]

You need to explain the term "decorrelation' that you constantly use.

less IACC and less IACC = spaciousness and envelopment

The goal may not be how to achieve more reflections but how to achieve less, at least in the early period.

why may not be?? who said so? Roman Pontiff?

Really?, that is directly opposite what the research is saying,

oh yeah! with minor exception of Dr Floyd Toole for example

how do we get a studio to be "perfectly reflective for 50 milleseconds and then no reverb after that"? Acoustic noise gates?

this is ideal of course, a goal to be pursued (just like an anechoic chamber is such a goal/ideal for many audiophiles) and here is about what can be done in real life:
Moulton Laboratories :: The Real World of Project Control Room Monitoring

here is about how it sounds in opinion of Bob Olhsson (from gearslutz forum):

Dave Moulton did a simple demo for me in a bare room that turns most of what we thought we knew about acoustic treatment and imaging right on its ear.

He had designed some speakers that deliver a flat response across 180 degrees. The imaging in the bare room was holographic, among the best I've ever heard. His conclusion is that early reflections aren't any problem at all but early reflections that don't have a flat frequency response are a big problem because they change the perceived tonality of the speaker.

Something that led him to this was the fact that many of the best translating control rooms a lot of us "old timers" ever encountered did not have any early reflection treatment while some of the worst have been certified LEDE rooms.

The problem with room tuning and early reflection absorption is that they are oversimplifications that have little to do with how we actually hear. Dave actually has serious academic credentials in addition to having been an engineer at Columbia Records. He also, last I heard, wasn't in the room design business.

neither is Mr Moulton in the speaker design business, so it is not true in his case that:

He has invented a speaker that gives wide horizontal dispersion and now he needs to justify it.

Mr Moulton's business is "music production, orchestral recording in stereo & surround, mixing and mastering for a variety of clients." and also "educational and technical writing, multichannel music composition and synthesis, acoustical and psychoacoustical research and consultation, and education"

Moulton's Takes
KIQ Productions

we have direct response and a single delayed reflection making the reflection flat in spectrum doesn't change the comb filtered mess of the combined response. It just determines the extent of the combing. Bech shows that an angular related hole in the reflection might just depress it below audibility.

and Dr Toole says that comb filtering is a measurement artifact, not audible at all
so, who knows? ...without trying...

 

[graaf]

The point was that even with speakers of widely different directivity a reasonable seating distance difference could equalize their direct to reflected ratio.

yes, but the point is that usually it is not an option in practice, one just cannot move one's listening chair from three meters distance from the speakers to six meters, and so one cannot equalize anything in such a way

 

[graaf]

10ms to first reflection is a complete fantasy for anyone living in Europe, even 6ms is impractical.

Which is why you want speaker directivity or room treatment to help to deal with those early reflections.

and Moulton says no need to worry because

it should get better in small rooms

because

if you take a look at what's really going on in recordings, playback rooms are generally small and the early reflections happen very quickly-whereas in a recording space (or simulation of a recording space that we do with artificial reverb), those reflections are much, much later in time.

What happens is that the early reflections of the playback room carry information about the recording room

 

[graaf]

ps.

10ms to first reflection is a complete fantasy for anyone living in Europe, even 6ms is impractical.

in fact, if You are a RFZ-believer and You have room size constraints making this >6 ms impossible to obtain, then a flooder in Beveridge placement is really the way to go for You

with a flooder in Beveridge placement You can have all early reflections delayed more than 6 ms easily