Features

It's rather like height information: Some systems sound as if they reproduce it, even though there's no way two mikes can capture it or two speakers can reproduce it. The up-and-across imaging of the Chesky Test CD LEDR demo is ersatz, too, because it relies entirely on the ear's interpretation of a progressive change in the spectral distribution of the signal. If the CD is paused when the "image" is at its highest point, then restarted for a fraction of a second, the sound will be heard to image at the height of the speakers, not above them. We have to hear it changing in order to interpret its equalization as height information.

Reflections from the side and rear walls of the listening room are necessary, of course, because without them, our ears would receive no 360 degrees reflections at all, and their lack while listening to music would be oppressive. Yet even when a listening-room's acoustics are optimized, no one will argue for a moment that the "sound" of that room bears any resemblance to the reverberant space of a large hall.

Surround reproduction allows the reverberant energy in the recording itself to reach us from the sides and back as well as from the front, providing a sense of envelopment by and involvement with the music that's impossible with two-channel stereo. And because the "room" sound is coming from the speakers and not from the original recording space, it sounds like a hall rather than a small room. Indeed, true ambient surround-sound works best in a very dead listening room, in which the only spatial cues returned to the listener are those of the original recording venue. This can seem to put you into the concert hall, instead of making you listen to it though a picture window. The difference must be heard to be appreciated.

The ambient surround field extends all the way around the listening seat, giving a much more palpable impression of the size and acoustical qualities of the performing space. Off-stage sounds will frequently image all the way out to the sidewalls, and the images are firm; ie, you can turn to face them and they'll stay put, unlike similar beyond-the-speakers effects that are sometimes heard from two-channel systems.

And there are other benefits. Surround reproduction makes a system sound effortlessly louder during fortissimos, as though its dynamic range has increased. The difference sounds like about 3dB---the equivalent of doubling the available amplifier power, even though a sound-pressure meter may show no increase in level. And surround reproduction can yield more realistic instrumental timbres than you may have ever heard, because it's no longer necessary to compromise the sound of the loudspeakers in order to make two of them sound as good as four.

Or weren't you aware that those "state-of-the-art" loudspeakers in your cherished audio system were purposely designed to distort the sound? They were, because literally accurate loudspeakers don't sound as realistic in two-channel stereo as they would in four.

The two-channel difference
If you're skeptical, then try this simple experiment: Dig out a stereo recording of any large-scale musical work with an active bass line---symphonic, operatic, cathedral pipe organ, what have you---and fire it up. Then switch your preamp to mono A+B mode. (If your preamp lacks a mono mode, see fig.1.) The first thing you'll notice, of course, is that the soundstage has collapsed. The second thing you'll notice is how much thinner the sound has become. Much of the upper bass/lower-midrange warmth and richness are gone (footnote 3), and the sound seems almost pinched in comparison to the stereo reproduction.

Contrary to what your ears tell you, you're not hearing a difference in frequency response. But one characteristic common to large-scale recordings is that they have lots of big-hall ambience on them, and that's what accounts for the difference. Our ears, it would seem, respond to hall ambience quite differently in stereo from the way they do in mono.

Fig.1 A mono blend hookup using Y-adaptors. Connecting A and B combines the stereo outputs for A+B operation.

Psychoacousticians seem unable to explain why this is so. The consensus appears to be that the phenomenon is related to auditory masking---the tendency for sounds to obscure softer sounds of similar frequency coming from the same perceived direction. In mono, all sounds come from the same direction, so the hall reverb, being softer than the direct sounds, is largely masked by them. In stereo, the instruments producing those reverberations are heard to image at specific locations across the soundstage, while the reverberant energy comes from two spatially separated sources, offsetting some of the masking effects of the direct sounds. The result is that the reverberant energy becomes more audible---it sounds louder, even though, as mentioned previously, frequency-response measurements will reveal no differences that could begin to account for the extent of the perceived difference.

Fig.2 Averaged response of a flat-response onstage loudspeaker, as measured from a mid-hall listening seat. The measured HF rolloff is due to absorption by the air and the upholstered seats. Listeners will perceive a much flatter response (see text).

In a large hall, the upholstered seats and the air itself absorb a lot of treble energy, so the reverberation consists predominantly of midrange and midbass. If we place a linear-response loudspeaker on the stage and measure its frequency response from an audience seat, we'll get something that looks like fig.2 (footnote 4). Because it's coming from all directions, the reverb is equally loud throughout the hall, but the direct sound waves from the stage become increasingly weak as they travel away from the source. All listeners farther than a certain distance from the stage (the so-called "far field") will hear more of the reverberant energy than direct sound, but for those in the near field (where direct sounds predominate), the reverb is quieter than the direct sounds and becomes susceptible to masking by them.

---

Footnote 3: This is partly because much of the low-frequency information in digital orchestral recordings is not correlated between the channels, as JGH explains later. Played in mono, this bass information will therefore tend to cancel. With LP recordings, however, the mastering engineer would often sum the channels in the bass, to avoid large vertical excursions of the cutting stylus that would ruin the master by breaking through the lacquer to the aluminum below. Would using LP rather than CD as the source affect the outcome of JGH's experiment?---JA

Footnote 4: This is not the response that audience members hear, because our ears judge upper-frequency response on the basis of first-arrival sounds: What we "hear" is essentially the actual high-frequency response of the sound source itself, rather than that of its reverberation.