We tend to gloss over some of the production difficulties we face as a consequence of our current broadcast distribution system. The simple fact of that glossing, as well as some of our well-established current practices, often makes matters even more troublesome.
In this particular case, I'm thinking about the relationship between monaural ("mono" to most of you), stereo and multichannel audio, and how they interact as a function of (a) production and (b) distribution.
Over the next several columns, I plan to examine in some detail, the basics of how the system works, and how things can go wrong for us as we try to make acceptable presentations for all of these formats. I got interested in this as a result of some conversations with Michael Nunan, post-sound supervisor for CTV in Toronto. I'll be citing his findings.
Let's start by considering what these terms mean, and how they apply to our work.
"Mono" refers to a single, unique audio signal trace. It also refers to a playback system that has a single audio channel and loudspeaker.
"Stereo" refers to a pair of signal traces, usually that have some elements in common and some within ca. 10 db and 40 µs. Stereo audio is used to convey a sense of spaciousness and verisimilitude with the audio. Stereo also refers to a playback system that has two audio channels and two approximately identical loudspeakers placed at different points in the same playback space.
Multichannel, or surround sound refers to a more complex system that has six signal traces, each with its own characteristics-four of the six channels behave approximately as stereo pairs. Multichannel is used to convey envelopment and significantly enhanced verisimilitude with the audio. Again, the term also refers to a playback system, one that has an array of five (hopefully) identical loudspeakers and a subwoofer.
You know all this, right? Good.
IN AND OUT
What we often forget to think about is this: each of these formats is both a production format and a playback system. And herein lies the problem. Take a look at Fig.1.
If only it were as simple as this looks!
(click thumbnail)Fig. 1
(click thumbnail)Fig. 2
If a mono source just went to a mono end-user and a multichannel source just went to a multichannel user, it'd all work fine. Unfortunately, that's not what happens. We need all formats to be playable on all systems, at least sort of. So we get Fig. 2.
Now we have the beginning of chaos. To make it even worse, all formats have to go through all transmission modes. I don't have the heart to even try to draw a flowchart of that madness, so you will just have to imagine it!
We went through something similar when we incorporated stereophony into broadcasting in the early 1960s. Our solution was simple and elegant. We created a pair of signals that were constituted of (a) the sum of the stereo channels and (b) the difference between the channels. The sum of the channels is, of course, the mono signal. The difference channel carries all of the stereo information. In each stereo receiver, a matrix reconstitutes left and right channels from the sum and difference signals. In mono receivers, the difference channel is simply not received.
What we learned from this was that we needed to take more than a little care in handling stereophony. This led to a practice of being concerned about mono compatibility and a craft that includes checking the mono as a basic part of stereo production. It is possible, for instance, to inadvertently generate a signal that is all difference; it will be out-of-polarity between left and right.
Such a signal will sound quite spacey and probably pleasant in stereo but won't sound at all on a mono receiver. Uh-oh. I've written about this before, and there is a substantial body of literature on the subject as well. Suffice it to say that interference effects can occur during summation that make summed mono sound much less nice than its stereo ancestor.
The main thing to keep in mind is that to make the signals in each format compatible with other formats, we need to make technical and aesthetic assumptions about the nature of those signals.
MIXING
When we wish to make more source channels fit into fewer playback channels, we indulge in downmixing. This occurs at the end-user's receiver or set-top box. The receiver simply mixes the extra channels together into fewer channels. Left and right are mixed to get mono, for instance. For surround-to-stereo downmixing, center and low frequency effects are equally distributed to left and right, and left surround is mixed to left while right surround is mixed to right. Usually, surround, center and LFE channels are attenuated by 3 dB. It depends on the receiver and how it is configured.
The ultimately degrading downmix occurs, of course, when the surround signal is downmixed to stereo and then summed to mono. Oh boy!
Upmixing is what occurs when we try to make fewer source channels fill out more playback channels. It is a little more troubling and difficult.
In the case of mono to stereo, however, it's cake. We simply send the mono signal to left and right. It works, and it sounds a little more entertaining than a single mono speaker.
Going from stereo to surround is a lot trickier. If we were to just send a mono sum of left and right to center, our stereo image would collapse, so that's out. Instead, we derive the center signal from left and right using a steering algorithm-how it works is beyond the scope of this column. Similarly, we like to derive the rear channels from the difference signal of left and right. I won't discuss LFE here.
The point here is that our derivations are based on assumptions about what should be in the center and rear channels. Those assumptions may or may not be true at any given moment with any given program.
What this means is, life ain't easy! Next month, I'll consider the transmission flow from production facility to end-user and how that affects these issues. Then we'll take a hard look at upmixing and downmixing as it's practiced today. Is there trouble in River City? I think so.
Thanks for listening.
