Omnidirectional Loudspeakers : Soundstage, Imaging and Placement

Now that I've been listening to omnidirectional speakers for several months, I've come to a few conclusions about their characteristics, pro & con.

— initial remarks from 2008 — update follows

The number one disadvantage of omnidirectional speakers is also the root of their greatest attribute. Omnidirectional dispersion maximally interacts with the room. Room interaction and the subsequent reflections are the source of peaks and nulls in listening position frequency response, and time delayed sound which can be viewed as a form of distortion; that which does not exist in the original recording is distortion. How do conventional direct radiating speakers differ from omnis in this regard? As mentioned elsewhere in this site, direct radiators are omnidirectional below a few hundred hertz and become progressively more directional with increasing frequency. Obviously, direct radiators maximally interact with the room only in the lower part of their bandwidth, then progressively less in the upper registers. Yet, the most detrimental interactions are in the lower frequencies where room modes create strong peaks and deep nulls. In the upper mid and high frequencies summing and canceling level out. This is true with any dispersion pattern. But with direct radiators, the time delayed reflections of higher frequencies are reduced due to the narrowing dispersion. So shouldn't this be a good thing? Not really. It leaves most of the reverberant field low passed and lacking in treble energy, i.e., a distorted reverberant field. The consequence of a low passed soundfield is a more boxy, stuffy, out of balance sound and a diminished perception of the ambience in a recording. Omnis avoid this by delivering full spectrum sound to the whole room, evenly dispersing the entire audible bandwidth and maximally interacting with the room. There is no way around room related issues except an anechoic chamber.

If we were to make our listening rooms anechoic, the dispersion pattern and off-axis response of a speaker would be irrelevant, at least to the extent that it's wide enough to cover the listening area. For a single listener, a speaker could beam to the perfect spot. No problem. But anechoic chambers sound odd and artificial. We are accustomed to hearing the acoustics of the room and spacial cues coming from many directions. Although a recording already contains the acoustics of the concert hall, during playback those spacial cues are not coming from the original direction—they are all coming from the two speakers in front of us—very artificial. It is a crude and unnatural way to simulate an acoustic environment. We need to hear those spacial cues coming from all around us. In an anechoic chamber they can't. In contrast, I suspect the reason stereo works as well as it does in our homes is because of room acoustics. In a way, the room reflections are substitutes for the ones we would get at a live event. The reverberant field in our home listening room surrounds us with sound, not as a simulacrum of the actual location, rather as a substitute. Those cues in the recording can then be interpreted as if coming from their original locations. This partially explains why headphone listening isn't as satisfying at achieving the effect of “they are here” or “I am there.” We need to be enveloped with sound, not injected with it. (This also argues strongly for multichannel recordings, which for music hasn't caught on well. And judging from the few multichannel music recordings I have, it's no wonder. They are very poorly executed. The recording industry has yet to develop a robust set of multichannel techniques.)

In the real world, most of us don't have the resources to build an anechoic chamber, nor would we want to. (See (II) UPDATE 21-01-10 below.) So we have to deal with room interactions and time delayed reflected sound. Perhaps the room isn't the diabolical agent of fidelity destruction it's assumed to be. Nevertheless, we may wish to have some control over the quality and quantity of the reflections. One way is through room treatment. Usually room treatment is just a band-aid approach and always sorely insufficient in the bass. Decimeters, not centimeters, of material with sufficient mass are needed to effectively manage frequencies below 100 Hz. Room treatment may help a portion of the room's contribution to the problem, but it cannot do anything for the speakers' contribution : its radiation pattern. One way to control room-speaker interaction is with a large, floor to ceiling line source—wide horizontal dispersion, limited vertical. It significantly reduces floor and ceiling reflections (except in the lowest frequencies), yet still succumbs to the low passed reverberant field problem by beaming the higher frequencies. Plus, you're getting the interference patterns created by the line source in addition to those created by the room. A similar situation occurs with large panel speakers which reduce both vertical and horizontal dispersion, causing the beaming to start at lower frequencies. Another possibility for controlling dispersion is a dipole radiation pattern. Full frequency dipoles cancel side radiation over the entire bandwidth. This leaves the reverberant field balanced and increases the ratio of direct to reflected sound, thereby reducing the reverberant 'distortion.' Dipoles are most likely the ideal for home audio reproduction and for minimizing room interaction. (Horns are another means of directivity control. They open a whole new can of worms. But no matter, the low frequencies are still omnidirectional—same irregular dispersion pattern, only worse.)

Where does this place omnidirectional speakers? It puts them in a compromised category. More compromised than dipole, but much less than conventional direct radiators or beaming dispersion types like panels or line sources. So, why accept this compromise?

For maximum dispersion and the widest listening area, for group listening or listening outside the prime center position, no other dispersion pattern will provide better spectral balance and channel to channel balance throughout a room than omnidirectional speakers. I have also found this room filling quality the most capable of performing the speaker 'disappearing act.' The musicians appear to float in an imaginary space unconstrained by the speakers. As a result of the uniform dispersion of full spectrum energy, the musicians and the acoustics of the recording seem to fuse together with the listening room. This is partially a result of a lower direct to reflected sound ratio, and it can be argued, higher room related distortion, yet because of this, the as if quality is maximally presented.

Speaker placement has an effect also. At only 75cm (2.5') from the front wall, omnis perform very well. Move them out further and the soundstage appears to deepen, adding greater dimension and specificity to the imaging. This effect increases up to about 1.5m (5') out from the front wall. The benefits are both theoretical and practical when one keeps the listening position fixed. The increased distance (more from the front wall, less to the listening position) accomplishes three things. First, it increases the time delay of the front wall reflections. The longer delay helps the ear/brain separate the direct from reflected sound more easily which makes for better clarity. Second, it increases the direct to reflected SPL ratio, so that not only is the time delay greater, the relative volume of the direct sound is greater, which again improves clarity and detail perception. And third, it makes a noticeable and measurable improvement in the transients, preserving the music's original impact and presence. The combined effect makes the speakers vanish into the soundscape of the recording. It sounds as if there's a field of musicians spanning across the space in front of me and the speakers are sitting there silent.

This seems perhaps a little too perfect? I'm going to do more listening and comparing between dipoles and omnis. At this point it's a little difficult because my dipole speakers are in a room with very different characteristics and I can't be certain which part of what I'm hearing is due to the room's contribution and what is due to the radiation pattern. I can say that the dipoles, which are in a quieter, more damped room, sound marginally better to me in the sweetspot—slightly drier, and a trifle more definition in the bass. Some of the dryness may be a result of the reduced reflections, downplaying the recording's ambience. Or should it be said, the omnis retain the ambience by putting more of its energy into the room?

But is there any advantage to omnidirectional speakers over dipoles? . . . I'm still out with the jury. . .

. . .

. . . . . .

. . . . . . . . .

(II) UPDATE 24-01-11 : . . . . the jury is in—after nearly two and a half years of deliberation. With the Parallel system in the new media room, a very different space than before, and having more experience with both dispersion types in various size rooms, I have come to a few cautious conclusions. I find myself hearing the room more than I use to. Larger rooms in general are easier to deal with, yet smaller rooms are still workable. (Not too small. Rooms less than 4m x 5m [12'x15'] are very limited.) Smaller rooms seem to benefit from more damping. Lower ceilings seem to give a closer perspective of the soundstage. Floors on joists transmit the physical feeling of the bass, nice, but also reradiate spurious bass, not nice. All speakers benefit from 'breathing room.' Normal untreated lateral room reflections are good, but they need adequate delay. I can confirm my earlier observation that front wall reflections deepen the soundstage especially when the speakers are about 1 to 1.5 meters (3-5 feet) from the front wall. Contrary to popular opinion, omnis image very well. The excellent imaging may actually be a function of front and side wall reflections giving our ears another copy of the directional cues from the recording. I'm not noticing the ambience difference anymore, but make note of how the rooms have changed. The new room with the omnis is now the more quiet and damped space. The dipoles' new location is a slightly more live space of about the same volume, but a bit smaller floor area. I continue to prefer a more damped room. It makes it easier to hear 'deeper' into the recording.

So, what about the dispersion patterns? Well, the dipole bass is not saving itself from the effects of a smaller room. Would it be worse for the sealed PX30s? It hasn't proven to be since they are also in a smaller room than before. I think whatever benefits there are with dipole dispersion, they are fairly minimal. Small room acoustics dominate in the low frequencies no matter what the dispersion pattern. Once again, here is where the off-axis behavior of the speaker plays a vital role. Since lateral reflections are helpful in creating a sense of space, the spectral balance of those reflections effect the overall tonal balance and clarity. The reduction of those reflections with dipole radiation may be beneficial in rooms with less furnishing and more hard surfaces, that is, livelier rooms, but as the room size increases the benefits may diminish. It may boil down to, how much sense of space do you want to create? More room interaction helps to broaden and deepen the soundstage. It also helps to 'hide' the speakers' location. The really critical factor, which to my ears keeps coming back again and again, is spectral balance. The linearity of the speakers' on and off axis behavior is more important than the dispersion pattern.

Tips for Auditioning Loudspeakers and views of the Media Room.

. . . and A Different View of Dispersion

(II) UPDATE 15-09-08 : Recently install a pair of SP18M with a pair of PX30 in an Interesting open loft-style space with exposed brick on the front wall, brick and windows on the right wall, drywall to the rear and left, open joists 12' ceiling, and floating hardwood floors over concrete slab. Some adjustments to the balance had to be made to compensate for the left/right asymmetry. Easy. But when orchestral music crescendoed to fortissimo, yikes! First thought was too much brick and glass, yet there was obviously also a great amount of energy being transfered to the floor. Returned a couple of weeks later, after a large area rug and L-shaped sofa were in place. Wow, what a difference! The congestion was gone. Nothing else changed. Clearly, the floor was the culprit. Damping the secondary radiation from the floor with the rug & sofa, and isolating the SP18Ms with felt pads, solved the problem.

(II) UPDATE 22-09-08 : I've been reminded once again of how resonances, high Qs, and low powered amps can spoil a system. Yesterday I heard a pair of quasi-omnidirectional speakers. I knew of these speakers and long been curious to give them a listen. Couldn't remember if they were ported or sealed. The first cut I heard on them started out quietly with a solo female voice and guitar, sounded very nice. As soon as the bass and drums and a couple other instruments came in I was struck by the lack of definition of both the midrange and bass—immediately thought to myself, "must be ported." Sure enough, they are. The size of the enclosure is small for a 6" or 7" driver, indicating the Q is likely high. With the ports and the high Q, there is just too much resonance in the midbass and well into the midrange resulting in a stuffy, muddy sound. Not too bad on small group acoustic, but when complex orchestral music was played everything got boggy, boomy, and intolerable. This was heard even on a relatively tame baroque concerto grosso played on period instruments. What would these sound like with a big modern orchestra playing Tchaikovsky or Shostakovich? (Don't wanna know.) When a wonderful jazz recording of a bass and piano duo was played, there was gross distortion in a narrow range of pitches played by the bassist. I can't say exactly what the cause was without testing, but I could speculate a few possibilities, a) amp clipping, b) woofer driven beyond excursion limits, c) underdamped driver at its resonance frequency/impedance peak, d) cartridge distortion, or e) any combination. It doesn't matter. These speakers are inferior. They have a classic 'smiley face' response—a midbass peak, midrange dip, mid-treble peak, just like a smile—and a smile is how most inexperienced ears respond to this common type of voicing. Anyway, their near omnidirectional dispersion is quite nice, too bad nothing else about them shines.

(II) UPDATE 21-01-10 : An anechoic chamber may be disturbing and unpleasant for playing music, but some claim it's great for listening to recorded music. There will always be differing opinions and differing preferences. It is true that all the ambient cues, spacial information and reverb is already in the recording; that there is no need to add more to it; that room reflections add to it and are therefore distortion; and that careless, uneducated room treatment is worse than none. What I'm not certain about is whether or not the speaker/room fusion hypothesis is completely off base? Without a first hand anechoic chamber listening experience, I won't know except through others' reports. Is it best to eliminate room reflections as much as possible, or are some actually necessary? Listening levels are a factor also. At low to moderate levels the room acoustics may be of less concern, but realistic playback levels reveal problems not experienced at lower levels.

Here are three links on the topic—first two reluctantly included because they are commercial sources—a third from a recording studio.

(II) UPDATE 02-03-10 : Attention streams. Listening to music and evaluating an audio system requires listening to multitude aspects simultaneously. Music is multilayered, and music reproduction involves many elements. Our minds can focus on one layer or one element at a time with a single attention stream. Each listening allows us to focus on a different area. An example of this is the latest remastered collection of The Beatles. The praise for these new remasters was glowing with nearly unanimous kudos. Peopled stated, "Truly jaw dropping," "greater clarity and better dynamic contrasts," "kick drum and bass line that I never heard," "Ringo is a better drummer. . , Paul is a much better bass player than I thought," "voices sound like they're in the room," "eleven on a scale of ten," and other similar raves. So, after I plunked down $200 to get this "must have" reissue, I did an A/B with my previous recordings (pirated digital from pristine condition original US release LPs). Many of the cuts that I compared show the new remasters are slightly drier (voices in the room) and a bit less treble, smoother and cleaner sounding (greater clarity), but not all. The differences on some cuts are negligible, the low bass is mostly identical, yet the upper bass is lifted on many cuts (kick drum and bass line that I never heard), that also imparts a fuller, richer sound. Truly jaw dropping? No. Improved? Yes. I unreservedly prefer the new remasters. Unlike many contemporary remasters, they haven't been compressed. These can be properly called remasters. They make some real gains from the originals, although mostly only marginal. None of the improvements are enough to excuse the hyperbole. These new releases brought new attention to The Beatles' recordings. People were listening with renewed focus and great expectations. Of course, they heard things they never heard before—their attention stream was redirected.

back to the Observations main page

(|| ) Audio Home
© 2008-2024 Parallel Audio all rights reserved