Film vs. Music

As you all know, 5.1 surround sound calls for a center loudspeaker, placed dead ahead of us, midway between the left and right loudspeakers. Left and right are supposed to be in their traditional positions, 30 degrees off the center axis on either side of the median plane, just like in stereo.

As you also know, this setup evolved from cinema practice, where a mono center channel behind the screen provides dialog and on-screen FX, while left and right provide a stereo music score and off-screen FX, with both stereo and mono components. However, the integration of this practice into home theater and surround music systems has been troubled and uneven, for various reasons.

At the center of this is the desire (some might say need) for the home theater system to also play back music, or as viewed from the opposite end of things, be able to use a high-quality music listening system to reproduce TV sound. This is called convergence, which is the integration of various separate modalities into a single, all-encompassing modality. Convergence is at the center of digital system benefits, in that it reduces costs while increasing the capacity and versatility of a system.

In any case, music and film are quite different from a production standpoint, and a one-system-plays-all modality has to serve multiple sets of needs, often simultaneously.


The issue here has to do with the choice of a suitable center speaker to fit in the playback array. Because we have a television directly in front of us, a loudspeaker that is identical to the left and right ones, right in front of it, may not be suitable. Also, the TV has its own speakers. Why not use it then?

If all we're doing is watching video, that solution will be viable, if not great. As noted above, production convention drops mono dialogue into the center channel and leaves all the stereo stuff to left and right and the surround channels. However, that kind of screws up the idea of convergence... and here's where we get down to it: One of the conditions for any multichannel loudspeaker array that is going to generate phantom images (i.e., have stereo components) is that the speakers involved must be essentially identical in order to successfully generate those phantoms. Any pair (or more) of speakers involved in carrying stereophonic information need to be close to identical in order to be able to generate convincing phantom images and ambiences.


I've been working with a center channel since 1995, and until recently I've had no luck adapting it to the playback of stereo recordings. This is to say that I haven't been able to play back stereo recordings successfully with the center channel added. More about that in a minute.

However, when I make a surround music recording, and actually deploy a center-channel microphone in the recording space, the center becomes a compelling contributor to surround recording. I really like what it does. And I'm not alone in this. Several goodnesses accrue:

  • The frontal soundstage is much more stable with a center channel present;
  • The bass response is much moresolid and palpable;
  • The centered sounds, emitting from a real source, are much more compelling, solid and lifelike.

These things are generally true for all modalities, to the extent that stereophonic techniques are in use.


The basic problem is simple. If we try to add a center channel to any two-channel stereophonic mix, we have two possible techniques. First, we can take a mono sum of L & R and feed it to the center. Second, we can use a steering mechanism, wherein we use a detector and whenever L/R becomes strongly correlated, we feed it to the center, and when L & R are NOT strongly correlated, we leave the center off (this is, in a simple sense, what Dolby Pro Logic and other matrixing schemes do).

With the former, the stereo collapses. If the center speaker is audible, the stereo becomes about 30 degrees wide, in total (L and R collapse as phantoms to midway between L/Center and R/Center).

With the latter technique, the steering becomes quite audible and often annoying in complex stereo signals that have components that are highly correlated occurring simultaneously with components that aren't.

As a consequence, added to the uncertainty that end-users' playback systems will have identical center speakers (to say nothing of good center speakers), surround music producers tend to avoid the center speaker. It loses out by default.


I'd been treating this state of affairs as "the way it is" and simply not worried about it until I attended the AES Multichannel Conference in Banff this past June. There, after listening to Tom Holman's excellent 10.2 surround system, and keeping the capabilities of wide-dispersion loudspeakers in mind, I got to thinking; would it be possible, I wondered, to improve on the current surround topology (5.1) in a way that didn't involve changing our current production methods, that would work really well with stereo and would work for film as well or better as the present system?

Tom Holman's 10.2 array uses eight speakers in a horizontal array, with a center front, L/R at 30 degrees, L/R at 60 degrees, Ls/Rs at 110 degrees and center rear. One of the points he made was that the 60-degree L/R speakers really filled in on the sides, so that when we hear phantoms between them and the surrounds, they come pretty clearly from 90 degrees (or thereabouts). I wondered what would happen if we made Left and Right at default angles of +/- 60 degrees? Better yet, suppose we made the five-channel array a pure pentagon, with all of the speakers placed 72 degrees apart. What would happen?

Well, when I got home from Banff, I decided to reconfigure my main system just to see what would happen. Next month I'll tell you what I found. It's actually pretty interesting.

Thanks for listening.

Dave Moulton