Space...the Final Frontier Letters

While I must agree with JGH on surround-sound ("Space...the Final Frontier," Vol.17 No.3, p.60), I must side with the editor regarding the relationship between the distance of the speaker from the rear wall and apparent image depth. I have used a Yamaha DSP-1 Digital Sound Field Processor for about two years, and now find the image unbearably flat without the Yamaha's added ambience. This is despite the great image depth afforded by the bipolar speakers I built: my three-way revision of the Delac S10 ("one of the weirdest speakers I have ever laid eyes on," said JA in Stereophile, April 1989). I use 10" Focal woofers residing in separate cabinets to cover the frequency range up to 320Hz, with the front- and rear-facing 4" SEAS mid-woofers taking over from there and carrying it on up to 3.2kHz. Therefore, there is plenty of information being sent aft.

Until recently, I had these placed along the short wall of my 138" by 235" listening room. However, after devising some new positions with The Listening Room software, I decided to try placing them along the long wall. I was trying to get the speakers farther from the side walls and closer to my listening position to eliminate early reflections. (I have wondered if the improvement in sound in GL's room [Stereophile, March '92] was due more to the nearfield placement of the speakers than to the RoomTunes.) This now placed the speakers 28" from the rear wall and 55" from the listening plane, compared to 46" and 89", respectively, in the former setup.

Although there seemed to be fewer colorations in this new position, I had lost the sense of great sound-space depth that I had formerly enjoyed. I have had to compromise, bringing the speakers out 46" into the room, situating myself 64" from the speakers and 28" from the back wall. There are a few more peaks and dips, but image depth is restored. I might have agreed that placing the speakers nearer the rear wall, and the consequent strengthening of the rear waves, would increase sense of depth. However, this experience refutes that notion.
---Robert M. Godwin, Marietta, GA

Height information, please
J. Gordon Holt is right in saying that first-rate multichannel surround-sound will always sound more realistic than first-rate, two-channel conventional stereo (March, pp.60-75). But he is mistaken to say that 3-D space and imaging can be reproduced by an array of horizontal channels. That would still allow us only to hear horizontal surround-sound with unprocessed conventional stereo recordings.

Stereo is distorted 2-D sound; horizontal surround-sound is more realistic, but it is still 2-D sound. To realize a realistic 3-D soundfield with the height dimension, a full Ambisonics system with floor and ceiling speakers is required. It's time reviewers wake up to this fact. Too much redundant praise has been heaped upon two-channel stereo's ability to generate 3-D space. It's a pity that no one is interested---not Nimbus, not Minim, not Sansui; UHJ-encoded [horizontal-surround] sound is all they want us to have.

To date, the most realistic soundfield re-creation I've heard is achieved through replaying Gordon Hempton's dummy-head recordings on two pairs of Spendor SP1/2s via a Minim UHJ decoder: There's height info aplenty; I can tilt my head and practically see the thunder sound coming from the ceiling. I am lucky that the ears of Hempton's dummy-head are very likely to have a configuration close to mine; otherwise, the height impression may not be possible, as it is well-known that spectral shaping by the pinnae can create height information.
---Yip Mang Meng, Singapore

Ambisonics heard from
Gordon Holt's "Space...the Final Frontier" article in March has us reconsider something that's been around and around for a long, long time: surround-sound. Gordon's is an admirably objective treatment of a subject that touches on some tender issues, such as our audio goals and expectations, and how we realize---or convince ourselves we realize---them.

If the reawakening of interest in surround-sound has been stimulated by Home Theater and motion pictures, one should realize that this approach to surround has essential differences from the needs of an audiophile application. First, motion-picture surround starts and ends as an ancillary to "picture." It is usually more or less a concoction of wild sound, mixed mono, processed ambience, and effects that add up to something that's anything but subtle. It is mixed or balanced for a true theater venue vs a Home Theater, but works to a degree at home due to its hotchpotch nature.

Motion-picture surround is not really any different from the old proprietary quad systems: SQ, QS, CD-4, and the like. In all cases, hardly anyone was---or is---looking at how to record surround as much as at, say, how to deliver it.

Notwithstanding its rather '70s-ish flip-sounding name, Ambisonics is a system that does start with the recording process. It has been described as 3-D Blumlein in that it is an intensity system that allows phase and gain to serve as cues for directional information. I disagree with editor Atkinson's assertion (and connection to Blumlein) on p.65 that ambient information folded into conventional stereo can serve sufficiently to provide an ambience effect in playback any more than single-channel mono can be expected to be processed by our brains as stereo. If either of these work for some people, I submit that it is basic autosuggestion operating here, not some kind of exotic or unexplainable recording technique. In both cases the music is all there; it's just "whether," and if so, what, can be done to provide accurate and sufficient cues for directional information?

Ambisonics is, in fact, the logical extension of Blumlein. The way it works is to distill out and save separately all the directional information, including height, for a given soundfield. This information is then recombined in the playback with respect to the playback-speaker configuration and dimensions of the playback setup. Thus, it's part what Ambisonics gives you and part what it doesn't; Ambisonics eliminates the spatial timing errors between recording and reproducing a continuous soundfield. In other words, on playback you are not presented ambient information recorded with some main signal picked up with distant microphones to be reproduced with relatively close speaker(s). Since the speed of sound is a constant, you simply cannot reduce required distances between main and ambient sources without big "time" problems.

In my opinion, this is why conventional recording of smaller soundfields often sounds better; that is, more natural---not necessarily more impressive---than larger-group recordings. In Ambisonics, these spatial time cues are simply left out so as to not provide "impossible" time-space-directional cues. Less is more.

Ambisonics does not require a compromise (read: lowest common denominator) for widely varying playback-room dimensions or huge variations in consumer playback equipment as a part of the recording considerations. Much of the so-called art of recording is really putting round pegs in square holes, and this is eliminated in Ambisonics. The result is not as "dramatic" as it is "faithful," bringing up, again, the issue of what we think we want to hear vs what "sounds" best.

Listening to Ambisonics, you come to realize a notably smooth, continuous, and natural-sounding imaging; less exact-sounding than your latest or favorite demo discs, but finer, if you will, more pure, and with more of a sense of occasion and ultimate "realism." As such, the recording venue plays a much more fundamental part in the production of the overall sound, making it less of a controllable variable. Curiously, Ambisonics is thought to be "tricky" to record, perhaps because there are fewer microphone placements and general recording tricks of the trade that can be deployed in the name of its art.

So what goes around comes around. Have you listened to Ambisonics lately?
---David S. Monett, Atlanta, GA

Binaural heard from
In Stereophile, Vol.17 No.3, J. Gordon Holt proclaimed surround-sound to be the way of the future...feeling that "the High End should abandon two-channel stereo." Why was the binaural system not considered as a surround-sound option? The binaural system produces remarkably realistic recordings which surround the listener with the original acoustic space. Additionally, the binaural system requires only two channels, as it is meant to be reproduced via headphones.

I see no reason why we should abandon stereophonic sound reproduction, as all systems could coexist simultaneously. Stereophonic, binaural, and encoded Ambisonic recordings could be played back on two channels, while storage formats which deliver more than two channels could play back full Ambisonic recordings and all the others previously mentioned. Many Ambisonic recordings are currently available. Why, then, is coverage of the Ambisonic system diminishing?
---Hrvoje Hrvojic, Toronto, Ontario, Canada

Stereophile's coverage of Ambisonics has diminished almost to nothing because of the system's commercial failure. Nimbus continues to record all of their CDs Ambisonically, but other than some Onkyo receivers, no decoders currently available play these encoded two-channel recordings. Not that it matters, in my opinion, because I have found these to be compromised in channel separation and image localization compared with a full discrete Ambisonics record/playback system. There is no source of commercial recordings in the latter format, however, nor is there any hardware available to consumers to play them on, even if they did exist. All that is left of the Ambisonics work is a remarkable microphone, the SoundField, which can be used as a versatile stereo mike. (I used it to make my piano and choral/orchestral recordings on the two Stereophile Test CDs.)

The real mystery is why the devotees of a commercially nonexistent format are so devoted to it. I guess the answer lies in the philosophically thought-through nature of Ambisonics, as revealed in the next letter.

Ambisonics redux
In Peter Mitchell's discussion of Dolby Surround formats in the April issue (Vol.17 No.4, p.49), he says, "Surround-sound is an evolving medium, and there is still much to be learned about what works best." On the contrary, surround-sound is a fully developed medium, about which there is little that is not known. JVC's Quad-Biphonics and NRDC's Ambisonics are fully realized surround-sound systems that give the recording engineer total control over all surround effects and allow the listener to hear exactly what the producer intended. All other systems are based on wishful thinking rather than psychoacoustics. None of them provide the control and accuracy of Quad-Biphonics or Ambisonics.

"...[T]here will be vigorous debates during the next few years about whether the diffusion provided by the bipolar radiation pattern of THX surround speakers will still be desirable when the mono surround signal of Dolby Surround is replaced by the discrete stereo surrounds of DSD," also wrote Peter in April.

There may be a lot of debate, but the point is not debatable. You can get a sharp, clear image only from speakers that have some directionality. The Bose 901 is a classic example of how not to design a stereo speaker. [Its idiosyncratic radiation pattern] destroys most of the directional cues that are already in the recording, substituting its own ersatz ambience. You cannot create a rear spread of images with speakers that diffuse the sound.

On the other hand, Peter's remarks about "envelopment" are well taken. One major cause of this problem is the fact that most surround systems are incapable of creating stable side images. Since the sound of any "performance space" is heavily dependent on lateral reverberation, it follows that any surround system that cannot image to the sides (and this includes so-called "discrete" systems) is, ipso facto, incapable of accurately reproducing hall acoustics.

Having made my own "discrete quad" recordings, I can state this as fact. Even when you place four cardioid mikes in a near-coincident square, facing outward, you still don't get a sense of envelopment. The ambience coming from the sides is encoded as amplitude panning between the front and rear mikes. Unfortunately, that isn't the way the brain images side sounds. The result is what I call "empty sides" and a lack of envelopment, even from a "discrete" system (footnote 1).

The various ambience-synthesis systems (such as those from Lexicon, JVC, and Yamaha) get around this problem (to some extent) by using four speakers. The two front speakers are usually placed beside the main speakers; Lexicon even recommends placing them at the sides of the listening area. Properly adjusted, these systems provide an excellent sense of envelopment. The speakers completely disappear, and you seem to be sitting in a "real" hall.

Surround-sound cannot turn a poor system into a good one, and I always recommend that listeners with x dollars to spend put it into stereo first. This rationale, however, creates a vicious cycle: When it's time to upgrade, should I upgrade the stereo system I already have, or add surround components? I have always resolved the problem by buying four channels' worth of expensive components. But then, I'm a bachelor.

This dilemma---better stereo vs the benefits of surround---will plague most listeners. My suggestion is that, once you have a system that satisfies you (either in terms of accuracy or "musicality"), you should seriously think about adding an ambience-synthesis or -extraction system. Once you have it, you will wonder how you ever enjoyed listening to reproduced music without it!
---William Sommerwerck, Bellevue, WA

Footnote 1: The quality of a surround system is not determined by how many channels it has, or how much "separation" there is among them, but whether the system can reproduce the original (or intended) directional effects. "Discrete" systems cannot do this. They are fundamentally and uncorrectably flawed (as is conventional stereo, and for much the same reasons).