Omnidirectional speakers

[Duke]

Some imo relevant quotes from acoustics and psychoacoustics expert David Griesinger, inventor of the original Lexicon processor, which (based on a conversation with an advocate of "tasteful upmixing" whose name has become controversial in this thread) was the best there ever was:

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.

In a small room there is almost never sufficient late reflected energy to contribute to the background [reverberation] perception.

Presence depends in the ability of the ear and brain to detect the direct sound as separate from the reflections.

When presence is lacking the earliest reflections are the most responsible.

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

Imo these statements offer insight into why omnis (and most speakers) produce better clarity and a better sense of acoustic space when their first-reflection path lengths are fairly long, with one of the statements offering insight into why one might want more late-onset reflection energy than conventional speakers normally deliver (and upmixing is one way to accomplish this).

It [optimal upmixer settings] is highly dependent on the recording and mixing technique. So the answer, IMHO, is a strong no. One of the first job of an upmixer is "directional decoding", which is to determine where the placement of the sound is.

For example, if you are adding a center channel to a 2 channel (stereo) mix, and there is a sound panned (using amplitude panning) to the right but not all the way. There you'll have the same sound in both the left and right channels, with the left channel weaker than the right. If you upmix by simply mixing the center channel with the sum of the left and right channels, the resultant image will be pulled toward the center. To "accurately" upmix it (not changing its perceived location vs the 2 channel playback), you'll need to reduce its strength in the left channel. Needless to say, this directional decoding process is far from simple. If the sound location info is encoded using time delay panning, as in recordings using spaced mics, it is almost hopeless. Therefore, a "successful" upmixing is highly depending how the source material is prepared. I believe you'll have a better chance of success for materials recorded/mixed for multichannel, downmixed to 2 channel (for distribution), and then upmixed again.

Thank you! That's what I suspected.

Ime good two-channel (which includes "getting the reflection field right") is "set it and forget it", though I concede that the performance potential is probably higher for upmixing.

In two-channel setups which enable the venue spatial cues on the recording to be perceptually dominant, I find the oft-unexpected and oft-dramatic variation in the sense of space from one recording to the next to be extremely enjoyable, in that it approximates listening to music performed in a corresponding variety of acoustic settings. Which leads to my next question:

Is there a great deal of variation in the sense of space from one recording to the next with upmixing?

 

[jim1274]

Adrian Acoustics moved their production to a much larger facility in Bridgeport, Connecticut, where they continued to produce cabinets for more high-end audio brands like McIntosh, Mark Levinson, and Polk. They still produce cabinets for Von Schweikert, VPI, and Ohm.

They make outdoor speakers not trash cans but planters.

View attachment 339852

View attachment 339853

Just looking at the photo, I don’t see how these could possibly be Omnis. Even if there are drivers on all fours sides, don’t think this a true Omni.

A quick perusal of their web site didn’t give any design details I could find.

 

[NTK]

Is there a great deal of variation in the sense of space from one recording to the next with upmixing?

I mostly approach this from a theoretical point of view, and have no actual experience. Judging from the abundance of posts regarding the "artificial" sound of some upmixers, my guess will be yes. But I think the technology is getting better.

 

[jim1274]

With up-mixing, it is done with algorithms designed and optimized by knowledgeable researchers. Having the sound coming from speakers all around you and when your room is nicely damped, the room effects will be overwhelmed by the surround speakers, and you get repeatable and controllable results. It is much more flexible and can do what you can't get from fixed room reflections, for example controlling the rate of reverberation decay. And if you don't like what you are getting, you can adjust the settings and there is always an off button. You can easily switch settings for different materials, including turning it off, according to your desire at the moment.

With speakers that spray sound into the room in random directions, if you want to adjust, you'll have to:
- Move the speaker around
- Move the listening position around
- Change the speakers
- Change the room
There is no off button, and you don't have the capability (or at the minimum highly compromised capability) to listen to native multichannel recordings (videos).

I have done a lot of playing around with various DSP schemes to enhance stereo to create “surround sound” on my 11.1 system but never really liked the effect. Is that “cheating” since my defense of Omnis is only compared to box speakers (mainly) or other designs using only 2 speakers in a stereo pair?

 

[jim1274]

In my opinion, the good idea is that you have two speakers system, one optimized for stereo recordings and one for native multichannel recordings.

YES! Amen, brother. I could not agree more on that statement.

I never really listened to much 2 channel stereo on my 11.1 ATOMOS capable rig even after playing around with stereo source “enhancement” using the added surround channels . Discrete multichannel recordings on a surround system is a completely different thing.

So far, the “optimized for stereo recordings” is a pair of Omnis in my room, at least for me.

 

[NTK]

I have done a lot of playing around with various DSP schemes to enhance stereo to create “surround sound” on my 11.1 system but never really liked the effect. Is that “cheating” since my defense of Omnis is only compared to box speakers (mainly) or other designs using only 2 speakers in a stereo pair?

They are all "cheating". I will quote this Dr Toole's post. And I'll include the use of onmi/dipole/multi-directional speakers into it.

BACCH4Mac Pro Edition: a report

Many thanks to Dr. Toole for joining the forum and providing some context on the development of crosstalk cancellation. In my ignorance, I had not known of Cooper-Bauck. Perhaps I need to hear a state-of-the-art multichannel system before I make more statements about the relative merits of...

www.audiosciencereview.com

...​

I think I can end by quoting myself from the earlier post:" Ideally we want an encode-decode system, so that results are predictable from the creative artists through to the listeners." This necessarily includes neutral loudspeakers throughout, and that too is a huge problem. Binaurally post processing existing recordings mixed and mastered for loudspeaker reproduction is not that. Neither is upmixing stereo to a multichannel system. Both can enhance a basic stereo playback, which is why both of us have found our ways to do it. We need a different approach, but from where I sit, I cannot imagine the global audio industry changing from its pathetically obsolete two-channel habits. One can only hope.​

​

Good luck ​

 

[Sal1950]

Given that you prefer to "take the room out of the equation as much as possible", and apparently you prefer upmixing over straight 2-channel, I get the impression that you DO NOT like the reflection field stereo speakers generate in a room, but you DO like the reflection field generated by upmixing 2-channel to multi-channel. Is this correct?

You seem to be mistaking the idea of upsampling as simply to add a bunch of concerthall like reverberation to the surround channels. That is what Floyd Toole loved about the old Logic 7 system of which it is still one of the best at that. But that's more for the classical music crowd.
OTOH modern systems like the latest Dolby Surround, DTS-X, and Auro 2&3D attempt to do so much more by actually trying to separate individual parts of the source into the surround channels. Their effectiveness at that varies from with each codec and musical source.
But to get back to the real debate here, I don't know of one recording studio that uses omni's to do their mixing and if your goal is High Fidelity Reproduction of the source, how you can believe spraying sound all over the room is going to get you there doesn't add up to me. I've listened to them many times over the years and all I ever got from the experience was a soundstage that was a jumbled mess.
YMMV

 

[jim1274]

Some imo relevant quotes from acoustics and psychoacoustics expert David Griesinger, inventor of the original Lexicon processor, which (based on a conversation with an advocate of "tasteful upmixing" whose name has become controversial in this thread) was the best there ever was:

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.

I mentioned envelopment with respect to the Omni “sound”, but the initial more noticeable characteristic was the soundstage expansion in width and depth.

In a small room there is almost never sufficient late reflected energy to contribute to the background [stream of reverberation] perception.

Presence depends in the ability of the ear and brain to detect the direct sound as separate from the reflections.

When presence is lacking the earliest reflections are the most responsible.

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

Imo these statements offer insight into why omnis (and most speakers) produce better clarity and a better sense of acoustic space when their first-reflection path lengths are fairly long, with one of the statements offering insight into why one might want more late-onset reflection energy than conventional speakers normally deliver (and upmixing is one way to accomplish this).

I just now moved the Omnis further out into the room, increasing distance from the rear wall, and first reaction was more envelopment—that’s what would be expected if I understood correctly.

Since the room is more long and narrow shape, maybe moving them closer together would add even more by reducing early side reflections? Obviously, there would be a trade-off in spatial imaging as left and right channels got closer together. I’ll try that next. I have been lax in trying more positions within the room.

Thank you! That's what I suspected.

Ime good two-channel (which includes "getting the reflection field right") is "set it and forget it", though I concede that the performance potential is probably higher for upmixing.

In two-channel setups which enable the venue spatial cues on the recording to be perceptually dominant, I find the oft-unexpected and oft-dramatic variation in the sense of space from one recording to the next to be extremely enjoyable, in that it approximates listening to music performed in a corresponding variety of acoustic settings. Which leads to my next question:

Is there a great deal of variation in the sense of space from one recording to the next with upmixing?

 

[jim1274]

They are all "cheating". I will quote this Dr Toole's post. And I'll include the use of onmi/dipole/multi-directional speakers into it.

BACCH4Mac Pro Edition: a report

Many thanks to Dr. Toole for joining the forum and providing some context on the development of crosstalk cancellation. In my ignorance, I had not known of Cooper-Bauck. Perhaps I need to hear a state-of-the-art multichannel system before I make more statements about the relative merits of...

www.audiosciencereview.com

...​

I think I can end by quoting myself from the earlier post:" Ideally we want an encode-decode system, so that results are predictable from the creative artists through to the listeners." This necessarily includes neutral loudspeakers throughout, and that too is a huge problem. Binaurally post processing existing recordings mixed and mastered for loudspeaker reproduction is not that. Neither is upmixing stereo to a multichannel system. Both can enhance a basic stereo playback, which is why both of us have found our ways to do it. We need a different approach, but from where I sit, I cannot imagine the global audio industry changing from its pathetically obsolete two-channel habits. One can only hope.​

​

Good luck ​

That really hit home after reading the linked Dr. Poole post. I did some live recording some years back, mostly local bands with a pair of stage mics and a second pair for rear “ambience”, but some with a larger number of channels, but it all was ultimately mixed down to stereo. When recording a friend’s live performances, usually with only percussion accomaniement using congas, that was 8 channels that got mixed down to two. Only did one recording session with my 24 channel mobile rig. The mixing/mastering was farmed out to another friend who had experience. If I ever get ambitious, I’ll maybe try making some discrete multichannel mixes of those old sessions, or more likely, paying my friend to do it. The learning curve is steep for this—learned that quickly when opening ProTools the first time.

That was probably off topic…sorry..

 

[jim1274]

You seem to be mistaking the idea of upsampling as simply to add a bunch of concerthall like reverberation to the surround channels. That is what Floyd Toole loved about the old Logic 7 system of which it is still one of the best at that. But that's more for the classical music crowd.
OTOH modern systems like the latest Dolby Surround, DTS-X, and Auro 2&3D attempt to do so much more by actually trying to separate individual parts of the source into the surround channels. Their effectiveness at that varies from with each codec and musical source.

I think many older recordings fall short when re-mixed into discrete channel surround because the original multi-channel recordings were never done with any intention for surround sound. Just moving instruments around the room in dramatic ways like in many quad recordings from back in the day or more modern remasters of older recording (most only max of 8 channel source tracks), has significant compromises.

But to get back to the real debate here, I don't know of one recording studio that uses omni's to do their mixing and if your goal is High Fidelity Reproduction of the source, how you can believe spraying sound all over the room is going to get you there doesn't add up to me. I've listened to them many times over the years and all I ever got from the experience was a soundstage that was a jumbled mess.
YMMV

I can’t even imagine using Omnis for mixing/mastering. Now Omni mics for live recording is another thing if used skillfully and properly. The whole choice of mic types and positioning is a “black art”, some folks holding their methods tighter than a secret family recipe.

 

[Flaesh]

I meant currently

- it's much easier than Duevel. Make omnis great again! Karlson's slots aren't [absolutely] necessary. A pair of used decent cinema\church speakers may be obtained in the US for cheap. You can call this Project JBueveL.

 

[jim1274]

Hmmmmm…then maybe LXMini is back to barking up the wrong tree for me…

This is late to the game here, but stumbled on Erin’s LXMini review by chance, and his comments on soundstage early in the review were interesting in the context of comparison to at least conventional speakers. I now see why they were suggested as something that had some similarity to Omnis in that department. I didn’t notice any reviews he did on true Omnis, but didn’t fully peruse his list, so will check. Might be interesting if he did, as while he does a lot of measurement based evaluation, he also makes subjective impression comments like soundstage of the LXMini. I found the shortcomings of the LXMini enough to not make me want to chase a pair, but they are interesting and different.

 

[Duke]

You seem to be mistaking the idea of upsampling as simply to add a bunch of concerthall like reverberation to the surround channels. That is what Floyd Toole loved about the old Logic 7 system of which it is still one of the best at that. But that's more for the classical music crowd.

Thank you for replying to my question, I appreciate it very much.

My assumption was that upmixing involves extracting and/or deriving ambience information from the recording so that it can be separately amplified and delivered from directions and with time delays that support the recording's spatial and ambience qualities. BUT...

OTOH modern systems like the latest Dolby Surround, DTS-X, and Auro 2&3D attempt to do so much more by actually trying to separate individual parts of the source into the surround channels.

... it seems I was badly mistaken, and that the modern intention of upmixing is to re-position sound images to locations made possible by having multiple speakers arrayed around the listening room. Obviously 2-channel cannot compete with that, outside of a few specialty recordings (like "Amused to Death").

Their effectiveness at that varies from with each codec and musical source.

So it sounds like the upmixing experience is either inconsistent or often calls for different settings for different recordings. Assuming this just means selecting stored pre-sets, probably not a big sacrifice from a convenience standpoint.

But to get back to the real debate here, I don't know of one recording studio that uses omni's to do their mixing

And I don't know of any recording studio that uses upmixing to do their 2-channel mixes.

and if your goal is High Fidelity Reproduction of the source, how you can believe spraying sound all over the room is going to get you there doesn't add up to me.

"Spraying sound all over the room" is what upmixing does too! Except that upmixing is spraying a deliberately-altered signal all over the room (and for the record, I have no problem with upmixing doing that).

"High Fidelity reproduction of the source" could mean "replicating what the mixing and mastering engineers heard in the studio control room." In this case, upmixing and omnis would both be mistakes.

Alternatively, "High Fidelity reproduction of the source" could mean "creating the perception of hearing the original live performance, or of hearing what that performance would have sounded like if it had actually taken place." In this case, upmixing is one way of doing so. And making good use of the spatial and ambience information already on the recording is another.

Making good use of the spatial and ambience information already on the recording is something good omnis can do well, with set-up geometry playing a role. Apparent instrument width is an area where omnis can exaggerate things. Clarity and image precision can be compromised if too much reflection energy arrives too early. Timbre tends to be very good with good omnis because the reflections reinforce the timbre of the first-arrival sound, but even here excessive early reflections are undesirable. All of this assumes fairly neutral and NOT overdamped room acoustics. With omnis in general, the longer the in-room reflection paths, the better.

In my opinion.

Arguably most "well-treated rooms" would be poor environments for omnis, if "well-treated" means "heavily damped". We'd end up with an abundance of spectrally-incorrect reflections, the shorter wavelengths having been attenuated moreso than the longer ones.

 

[jim1274]

I mentioned envelopment with respect to the Omni “sound”, but the initial more noticeable characteristic was the soundstage expansion in width and depth.



I just now moved the Omnis further out into the room, increasing distance from the rear wall, and first reaction was more envelopment—that’s what would be expected if I understood correctly.

I moved them again, even further from the back walls, getting even more envelopment, plus a little better imaging in terms of spatial relationship of instruments from a front center listening position, getting closer to an equilateral triangle of speakers and listening. I move my listening position up a bit to the equilateral position, and spacing and more distinct positioning of instruments improved.

When I then moved the speakers even closer together to about 3’ seperation, got a little more envelopment from reduced early side reflections. This positioning definitely needs to be tested further. I’m not sure if closer together and well away from rear wall and only a little less from the sides isn’t the “best” soundfield yet.

The wife came out a bit ago and said, “that really sounds good”. I’ve been with this woman for 48 years and she has NEVER, unprompted for her opinion, made a comment about sound other than saying it’s too loud, that more times than I could count and then some. I know eyes are rolling now, but that got my attention.

Since the room is more long and narrow shape, maybe moving them closer together would add even more by reducing early side reflections? Obviously, there would be a trade-off in spatial imaging as left and right channels got closer together. I’ll try that next. I have been lax in trying more positions within the room.

 

[MarkS]

I don't know of one recording studio that uses omni's to do their mixing and if your goal is High Fidelity Reproduction of the source, how you can believe spraying sound all over the room is going to get you there doesn't add up to me.

Well here again the music matters. I listen mostly to classical, and the very best recordings IMO are Telarc's minimally miked ones. And I preferred the sound of those recordings on bipolar DefTechs compared to conventional forward-firers when I went to buy some new speakers 20 years ago. (I went to hear the well-measuring Paradigm Studio 60 v3, and was underwhelmed.) I also liked their sound on arena rock, which in real life you hear over giant banks of PA speakers.

However bipolars and omnis have fallen out of popular favor. Mirage (makers of the bipolar M1 owned by Floyd Toole back in the day) is out of business, DefTech's founders left and the new owners dialed down the rear arrays by 6dB. The only bipolar speakers left that I know of are the Axiom LFR series (Axiom calls them omnis but they're really bipolar). True omnis, as noted above, are very rare.

So when one of my DefTechs went wonky three years ago, I ended up with forward-firing replacements, GoldenEar Triton 7s. (I hadn't yet found ASR; if I had, I might have Revel F35s instead.) I'm happy enough with them, though I still sometimes miss the solidity of the sound field that I experienced when moving around the room with the DefTechs.

 

[jim1274]

Have I mentioned lately how great this forum is?

I had to take a break from playing with Omni positioning tests, but did notice with them close together, the off-axis listening was enhanced, at least to 45 degree point I tried. If anything, trying to optimize the positioning within my room made the difference in soundstage compared to box speakers even more pronounced, in a good way.

There was still something “left on the table” not that they didn’t impress even poorly placed.

 

[Duke]

I moved them again, even further from the back walls, getting even more envelopment, plus a little better imaging in terms of spatial relationship of instruments from a front center listening position, getting closer to an equilateral triangle of speakers and listening. I move my listening position up a bit to the equilateral position, and spacing and more distinct positioning of instruments improved.

When I then moved the speakers even closer together to about 3’ seperation, got a little more envelopment from reduced early side reflections. This positioning definitely needs to be tested further. I’m not sure if closer together and well away from rear wall and only a little less from the sides isn’t the “best” soundfield yet.

The improvements you report make sense to me.

Increasing the time gap between the direct sound and the strong onset of reflections is virtually always desirable in home audio (assuming the spectral balance of the reflections is preserved). This is implied by the David Griesinger quotes in post number 201.

Imo omnis tend to produce a lower in-room direct-to-reflected sound ratio than is ideal, but when you increased their reflection path lengths by moving them away from the walls (and closer to the listening position), you pushed the direct-to-reflected sound ratio significantly higher at the listening position.

Finally, restoring the angular relationship between you and the speakers by moving your listening position forward should largely restore the soundstage width. You might even schooch your chair up a bit more, further widening the angle, for the following reason:

You might have lost a little bit of soundstage width in the process of moving your speakiers away from the side walls. You may now have less "apparent source width", having reduced the intensity and increased the time-delay of the first same-side-wall reflections. My understanding is that most listeners really like having sound images appear wide to the sides, beyond the speakers, but the fairly strong and early same-side-wall reflections which bestow this characteristic also degrade the image precision, soundstage depth, and perception of immersion. So there is some juggling of tradeoffs involved.

 

[Duke]

Well here again the music matters. I listen mostly to classical, and the very best recordings IMO are Telarc's minimally miked ones. And I preferred the sound of those recordings on bipolar DefTechs compared to conventional forward-firers when I went to buy some new speakers 20 years ago. (I went to hear the well-measuring Paradigm Studio 60 v3, and was underwhelmed.) I also liked their sound on arena rock, which in real life you hear over giant banks of PA speakers.

Based on my experience with SoundLab's 90-degree-pattern (front and back) fullrange electrostats, I manufactured a 90-degree-pattern (front and back) bipolar speaker for several years starting in 2008. I stopped making it when the most critical components were discontinued.

However bipolars and omnis have fallen out of popular favor. Mirage (makers of the bipolar M1 owned by Floyd Toole back in the day) is out of business,

I've mulled over doing a revised version of my controlled-pattern bipolar. Marketing would probably be an uphill battle because, as you say, they have fallen out of favor.

I continued doing multidirectional speakers after discontinuing my bipolars, and they haven't exactly set the world on fire. Ironically most of the interest in my multidirectional approach has come from bass guitar players; I make a couple of multidirectionals for that market.

DefTech's founders left and the new owners dialed down the rear arrays by 6dB.

They probably found most people were placing them too close to the wall, which tends to degrade clarity with multidirectional speakers having a lot of rear-firing energy. One way to offset that is to reduce the SPL of the rear-firing energy.

I still sometimes miss the solidity of the sound field that I experienced when moving around the room with the DefTechs.

I understand.

If you have any interest in DIY, you might try adding an "ambience tweeter". Eyeballing the far off-axis curves in SoundStage's measurements, it looks to me like there is a "window of opportunity" to add some rear-firing energy in the region from about 1.5 kHz to about 10 kHz. The fairly hot on-axis response north of 10 kHz inclines me against adding much more off-axis energy that high up.

Place the ambience tweeters on the floor behind the Triton 7's, facing upwards (assuming your ceiling is reflective rather than absorptive), and "shielded" from the listening position by the speaker enclosures. The idea is to add some relatively late-onset reflection energy to the reflection field, complementing the lower-range later-arriving reflection energy already present due to the speaker's wide radiation pattern south of 1 kHz or so. Optimize the response of the ambience tweeters for good power response, ignoring their on-axis response. You can adjust the loudness of the ambience tweeters by ear; imo the ideal setting is just below the threshold where they just begin to degrade the system's clarity, when auditioned from the main listening position.

This probably won't give you everything the Def Tech's once did, but I think it will be a net improvement in spatial quality and in timbre.

 

[Newman]

Does Toole also point out that essentially no recordings are mastered with upmixing to multi-channel (his preferred method), so what you are hearing with upmixing is not what they produced, in almost every way?

I don't think he says it in as many words, but I am sure he would agree with you. That is why he usually doesn't like what most of them do, and pointedly says so.

Upmixing to multichannel is also correctly described as a sound effect generator.

Absolutely! And so is allowing first lateral reflections from conventional speakers in 2-ch, yet that is positively perceived.

And this is the main point: the shortcomings of 2-channel audio are so undesirable that sometimes the harm done by just the right amount of sound effect is outweighed by the good. For speaker reflections, experiments showed that 'the right amount' is the first lateral reflection. But go too far, and preference falls off a cliff.

Same for upmixing. It's an undesirable kludge, but in small, sensitively calibrated amounts, it can be worth it. Even then, probably not for all recordings. It's analogous to allowing that first lateral reflection.

But omni speakers are way OTT, as a sound effect generator. Sort of like upmixing by sending the full L channel to every L speaker and the full R channel to every R speaker.

cheers

 

[Newman]

Some people confuse his book with the Bible.

That back-handed pot-shot of a comment shows a remarkable ignorance about how science works, and how learning from scientists works. Thanks for sharing: it's very helpful to know. I shall also note who gives it a Like, which also yields quite helpful information.

Let us know when you are ready to move on from the starting point of "everything is relative, and everyone's preferences are different and unique". Some people just can't get past it.