omnidirectional loudspeakers = best design available

[q3cpma]

Just posting something that wasn't mentioned once in the thread (I hope): omni proponents seem to rely on the implicit assumption that room absorption is perfectly smooth and broadband, which of course is rarely true. Thus, a CD design will have the same advantages as a smooth omni but probably even better/more reliable, as the room dependency is lessened.

Really, my view is that omnis' goal is to create a new live event in your own room based on the recording (not even mentioning that it'd require a special kind of production/mixing to match) while monopoles can at least try to reproduce the recording itself, and maybe hint at the performance if the studio did a good job.

 

[oivavoi]

My interpretation of the above (and other Toole writings on loudspeaker directivity) is that he believes that omnidirectional loudspeakers are an excellent choice for two-speaker stereo reproduction (as they enhance perceived spaciousness) but a poor choice for multichannel reproduction (as the spatial cues already contained in multichannel recordings are actually diminished by too many/too strong reflections).

Yap. If I ever get around to setting up a Auro3D system, or at the very minimum a 5-channel system with trusty old Logic 7, I'll most certainly not use omni speakers for that. Neumann KH80 with its controlled and limited directivity seem like a good choice for that kind of application.

 

[Blumlein 88]

Since we all love quoting Toole so much, here are some quotes of things he posted last year. Take from them what you will.:


In response to a question regarding "the reconstruction of the original musical event.":
"Which, of course, is quite impossible with only two channels! What we hear is only hints of what might have been, as tweaked by recording and mastering engineers. Nothing approaching the original sound field is captured, stored or reproduced. An active imagination is required, and that very likely is strongly individualistic. The problem is the "system" (stereo), not the loudspeakers. This does not mean that music cannot be extremely pleasurable, but the expanded "circle of confusion", if reality is the goal, is not a circle at all. It is a dead end, so long as we stick with two channels. Let the flames begin . . ."


Discussing his Mirage M1s:
"I have not only subjectively and objectively evaluated multidirectional designs, but I owned a pair of Mirage M1s. My comments on them as well as in-room and anechoic measurements are in Chapter 7 in my book. They are absolutely amenable to conventional evaluation. The added room reflections soften hard L & R panned images and slightly blur the panned images. This is most advantageous for large scale classical music in my opinion, but not offensive with any musical genre. I'm sure some others will disagree."

"The closest I have experienced are the Mirage M1, a bipole design that was crudely omni in the horizontal plane, and an Ohm Walsh 2 which was horizontally omni up to about 2 kHz. Both acquitted themselves very well in double-blind listening tests compared to good forward firing designs. The most obvious difference was in the spatial domain, which is to be expected, but even that was more subtle than many of us expected in direct-comparison blind tests."


Specifically discussing Omnidirectional speakers:
"The dominant sound energy and perceptions are attributable to the direct sound and the first reflections from walls, floor and ceiling (horizontal and vertical planes) - all other reflections travel much farther and encounter multiple reflecting surfaces. So, it is not necessary to have "true" omnidirectIonality, a point source, even though it is a popular theoretical, academic, concept. The question is "how close to the direct sound must the off-axis sounds be?"

"A truly omnidirectional speaker radiating a flat direct sound would exhibit flat sound power and a "flattish" room curve. There would inevitably be a small downward tilt because of air absorption, absorption at room boundaries and furnishings that tend to be a higher frequencies."

"Harman sells monitor loudspeakers to music and movie studios, and to consumers ... you will see that in order to minimize the influence of the "circle of confusion" and thereby have any hope of delivering the "art" as it was created, one needs similar loudspeakers everywhere. I know of no recording facility that uses multidirectional loudspeakers."

"They exist in homes, I believe, mainly in an attempt to improve on the spatial limitations of stereo. They are contraindicated for multichannel installations."

"However, I long ago decided that multichannel upmixing was a more rewarding way to embellish stereo. It is adjustable, and it can be turned off. A permanent form of embellishment, as in a loudspeaker design, cannot work for all recordings."

Toole likes upmixed stereo, and I've never heard upmixed stereo that I liked at all. I like mch when the source is. Upmixing no.

 

[Blumlein 88]

My interpretation of the above (and other Toole writings on loudspeaker directivity) is that he believes that omnidirectional loudspeakers are an excellent choice for two-speaker stereo reproduction (as they enhance perceived spaciousness) but a poor choice for multichannel reproduction (as the spatial cues already contained in multichannel recordings are actually diminished by too many/too strong reflections).

Having heard M1 and M3 speakers and noting his comments, I think Toole prefers a more diffused sound with softer imaging than I do.

 

[andreasmaaan]

Having heard M1 and M3 speakers and noting his comments, I think Toole prefers a more diffused sound with softer imaging than I do.

Yeh, I think that's right. Spaciousness ranks high on Toole's list of desirable attributes (hence also his strong preference for wide-directivity loudspeakers in general).

 

[BenB]

Toole likes upmixed stereo, and I've never heard upmixed stereo that I liked at all. I like mch when the source is. Upmixing no.

I've also never heard an upmixed stereo that I liked. I'm can't even claim to like multi-channel when the source is. I have SACD (and even DVD-audio) discs that I'll listen to in stereo (sometimes they include a high-res stereo track, sometimes I'm down-mixing). I suspect I would like the PSR techniques if I had a chance to hear them, based on the level of attention paid to capturing and reproducing something authentically.

 

[aarons915]

Toole likes upmixed stereo, and I've never heard upmixed stereo that I liked at all. I like mch when the source is. Upmixing no.

Have you ever tried DTS Neo 6 music? It's the only upmixing I've liked as well but I only use the front 3 channels, surrounds have always sounded a bit odd with music. Since modern Denons don't have Neo 6 I've been using Equalizer APO to Upmix by sending the L and R channels at -11db each to the center and it gets you pretty close as well. I think it takes away some of the stereo imaging and sounds more like live music.

 

[krabapple]

Can you please provide a specific reference or quotation of what you say that Toole said? In your words, "the universal rules that I use" and "quote marks for Toole’s words, otherwise they are mine,"

Toole/Olive/Harman settled on using single loudspeakers (mono) as the most sensitive test of their subjective performance, based on research findings that sensitivity decreased as delivery channels increased In fact, some loudspeaker differences heard reliably in mono already start 'disappearing' in stereo.

3.4 How Many Channels

More recently, tests were done that included mono, stereo and multichannel presentations (Olive et al., 2008). The results, shown in Figure 3.10, indicate that when there was a perceived preference among the optional sound qualities, it was most clearly revealed in monophonic listening. As active channels were added to the presentations, the ability to distinguish between sounds of different timbres appeared to deteriorate.

4.6.3 Finding and FIxing Resonances

Another feature of these tests, and one of the motivations, was to compare subjective rating performances when listening in mono, stereo and 5-channel multichannel modes. Figure 4.14 shows the result, and it is very clear, as discussed in Section 3.4 and 7.4.2, that adding more active channels to a listening experience reduces one’s ability to discern details of loudspeaker performance.

-- Toole, Floyd E.. Sound Reproduction ,3rd Ed

Toole has also said as much in presentations/online.

 

[krabapple]

Toole likes upmixed stereo, and I've never heard upmixed stereo that I liked at all. I like mch when the source is. Upmixing no.

I'm in Camp Toole: love it, live with it, would never go back to 'stereo'. And I'm using 'just' Dolby Pro Logic II (music mode), or more recently, Dolby Surround....

 

[krabapple]

Yeh, I think that's right. Spaciousness ranks high on Toole's list of desirable attributes (hence also his strong preference for wide-directivity loudspeakers in general).

'envelopment'

 

[andreasmaaan]

'envelopment'

I chose "spaciousness" as it (at least by Toole's definition) encompasses both "envelopmemt" and "apparent source width" - both of which are enhanced by wide-directivity (and especially omnidirectional) stereo loudspeakers.

 

[andreasmaaan]

Toole likes upmixed stereo, and I've never heard upmixed stereo that I liked at all.

I've also never heard an upmixed stereo that I liked.

I'm not sure if it fits the usual definition of "upmixing", but I'd highly recommend using three speakers for stereo, with the mid channel sent to C and the side channels sent to L and R.

I'm pretty sure even strong advocates of true stereo over upmixing would tend to prefer this three-channel approach, which essentially retains the soundstage of two-channel stereo, with the added benefits of a true centre channel (i.e. greater image solidity and greater freedom of movement for the listener).

 

[Duke]

Putting this in perspective using the 6 millisecond threshold, if someone sits 10 feet back from speakers that are separated by 10 feet, then they will need to have the speakers 5 feet from the sidewalls, and 3.5 feet out from the wall behind them. This implies a room that's at least 20 feet wide and presumably 17 feet deep (assuming the listener needs to be 3.5 away from the back wall).

The required room size for sufficient reflection path lengths can be reduced significantly by using fairly directional loudspeakers and aiming them appropriately. Ten milliseconds delay on all horizontal-plane early reflections except for the wall behind the listener is actually feasible in most rooms, but obviously not with omnis. I do this with controlled-pattern polydirectionals.

I don't think we'd want all the reflections arriving together at the listening position, so the room would have to be a few feet bigger than that to allow for diversity in arrival time. Based on prior research and personal experience, I have estimated that most people would probably prefer any sidewall reflection from closer than 3 feet be treated, and that any sidewall reflection from 5 feet or more away should be maintained.

Ime having that inrush of reflections at approximately the same time is not an issue as they will be arriving from different directions. This is the sort of thing one microphone cannot sort out, but two ears easily can.

I chose "spaciousness" as it (at least by Toole's definition) encompasses both "envelopmemt" and "apparent source width" - both of which are enhanced by wide-directivity (and especially omnidirectional) stereo loudspeakers.

My understanding is that "ASW" is conveyed primarily by the early sidewall reflections, while "envelopment" is conveyed primarily by the later-onset reflections which arrive from all around. And my understanding is that significant early reflections tend to work against envelopment, which implies a tradeoff relationship between the two. And that has been my experience, but I do not claim my experience to be all-encompassing or derived under controlled blind testing conditions.

I'm not sure if it fits the usual definition of "upmixing", but I'd highly recommend using three speakers for stereo, with the mid channel sent to C and the side channels sent to L and R.

I'm pretty sure even strong advocates of true stereo over upmixing would tend to prefer this three-channel approach, which essentially retains the soundstage of two-channel stereo, with the added benefits of a true centre channel (i.e. greater image solidity and greater freedom of movement for the listener).

In my experience, the center channel stereo setup - like many things in home audio - involves tradeoffs. I don't dispute the advantages you mention, but the downside can be a loss of depth and sometimes image-attached-to-a-box in the middle of the soundstage.

 

[MattHooper]

I'm in Camp Toole: love it, live with it, would never go back to 'stereo'. And I'm using 'just' Dolby Pro Logic II (music mode), or more recently, Dolby Surround....

Wait! Those poor engineers went to all those lengths to produce the stereo mix just as they and the artist intended, and now you are choosing to add COLORATION and inaccuracy!??

Please turn in your High Fidelity Club card, sir!

 

[Blumlein 88]

Have you ever tried DTS Neo 6 music? It's the only upmixing I've liked as well but I only use the front 3 channels, surrounds have always sounded a bit odd with music. Since modern Denons don't have Neo 6 I've been using Equalizer APO to Upmix by sending the L and R channels at -11db each to the center and it gets you pretty close as well. I think it takes away some of the stereo imaging and sounds more like live music.

I've got DTS NEOS 6.0 Cinema as an upmix choice. And another NEOS that supposedly adds height artificially. I've not cared for it either though I did use 5 channels when I listened to it. Lots of people recommended Logic 7 for years. I suppose it was less bad than most of the others.

I can find the odd stereo recording some of the upmixes are okay on. None seem like something I want to switch on and leave on. And even when okay they don't sound better to me just different. So I can't hear the advantage.

 

[Blumlein 88]

I'm not sure if it fits the usual definition of "upmixing", but I'd highly recommend using three speakers for stereo, with the mid channel sent to C and the side channels sent to L and R.

I'm pretty sure even strong advocates of true stereo over upmixing would tend to prefer this three-channel approach, which essentially retains the soundstage of two-channel stereo, with the added benefits of a true centre channel (i.e. greater image solidity and greater freedom of movement for the listener).

Now that has been tested for accuracy. I don't know if I still have a copy of it. At least with a centered listening position the center channel is a big boon for 3 channel recordings. For two and three channel recordings listeners were tested by drawing a diagram of where they heard instruments in simple unprocessed recordings. 3 channel over 3 speakers was best by far. Stereo over stereo speakers was better than stereo over 3 speakers with the center channel made up of a mix of L+R and the level reduced. The depth information was reduced in accuracy if the center was used this way. Using 3 channel sources with the center split right and left didn't work well either.

Now that doesn't mean some may not prefer a center channel on stereo, especially if they gain freedom of the LP without a collapse of the soundstage.

 

[Duke]

Think about how to better quantify the difference when I have my front / rear firing speakers playing together vs just my front firing speakers, and to measure the 'more enveloping sound' of a dipole setup.

Question:
In a nearfield desktop scenario (ears ~2-3 feet from speakers), I can play both
1. front/rear firing speakers together, or
2. front firing speakers only, or
3. front firing speakers only, but at higher volume (to contrast against 1).

Which are the measurements I should expect to see differences in, group delays? Other than cancellations in some frequencies of course.

And measurements should be on axis AND off axis as well? I suppose this is where the klippel is amazing at where it can break down direct / reflected sound.

One of the challenges of measuring the combined output of a polydirectional loudspeaker system is this: A microphone does not combine the direct and reflected sound the same way that two ears and a brain do. The in-room measurement will include reinforcement peaks and cancellation dips which are non-existent to the ear/brain system. Actually this is true of most in-room measurements, but deliberately adding more reverberant energy via a polydirectional system is especially likely to result in appallingly (and misleadingly) poor measured response.

 

[andreasmaaan]

Now that has been tested for accuracy. I don't know if I still have a copy of it. At least with a centered listening position the center channel is a big boon for 3 channel recordings. For two and three channel recordings listeners were tested by drawing a diagram of where they heard instruments in simple unprocessed recordings. 3 channel over 3 speakers was best by far. Stereo over stereo speakers was better than stereo over 3 speakers with the center channel made up of a mix of L+R and the level reduced. The depth information was reduced in accuracy if the center was used this way. Using 3 channel sources with the center split right and left didn't work well either.

Now that doesn't mean some may not prefer a center channel on stereo, especially if they gain freedom of the LP without a collapse of the soundstage.

That's interesting, do you have a link to the research?

It seems to me that "stereo over 3 speakers with the center channel made up of a mix of L+R and the level reduced" is not the same as the setup I described, in which the signal is split into mid/side channels and then only the mid channel sent to the centre speaker...

 

[Duke]

Now that doesn't mean some may not prefer a center channel on stereo, especially if they gain freedom of the LP without a collapse of the soundstage.

The instances where I've had customers and colleagues go from three-channel stereo back to two-channel stereo have all involved switching to highly directional main speakers strongly toed-in, such that their axes criss-cross in front of the center sweet spot. Credit to Earl Geddes for this idea, which (somewhat counter-intuirively) results in good soundstaging across a wide listening area.

 

[Blumlein 88]

That's interesting, do you have a link to the research?

It seems to me that "stereo over 3 speakers with the center channel made up of a mix of L+R and the level reduced" is not the same as the setup I described, in which the signal is split into mid/side channels and then only the mid channel sent to the centre speaker...

Yes the sides cancel out and you have the Mid left. Should be the same thing.