The paper is excellent, I totally get the fact that, in a hall or very large room, there must be a benefit to go multi-channel. The paper proves the advantage of 3 channels (although large venues seemed to have moved to even more channels than that, sometimes much to my chagrin, because I like to hear stuff in front of me, which is what my eyes reveal), but then again it seems the paper (caveat: haven't finished reading it yet) also states the recording itself should happen with 3 microphones. Over-simplifying some here.
I would be very curious about a 3-channel setup, especially one that allows me to very easily dial in the volume of the center channel on the fly. However, the music I own is all stereo. How would I take advantage of a center channel? Even this paper states that the center channel may reduce stage width if inappropriately calibrated. Is Atmos the logical evolution of the setup the paper suggests?
I know multichannel music exists. MP3 has an extension to be multi-channel (not sure when it was introduced, I know originally it only supported mono and stereo). Can any system add artificial depth to a stereo recording? I find myself thinking whether my huge established preference for stereo (well, 2.1 these days: stereo + sub) is driven by intellectual laziness, force of habit, or the lack of media that let me test it out. :-D
Sorry for the stream of consciousness, my brain hasn't yet been able to process all the stuff I learned.
Thanks for a very eye-opening thread!
That would be "at least 3 microphones", but yes, you are getting the message.
"What do I do with 2 channel "stereo"?" is a very, very good question, indeed. The answer is "not as much as one would want", and yes, we're stuck with a great lot of 2-channel recordings (which nobody is going to throw out, well, we're not throwing out mono recordings either!).
As to specific multichannel systems, I am not going to comment on any particular system. A good "rendering" system could (and can) produce the best result from any given layout without more production, mixing, etc. That does exist, but it's not common (to say the least) in the modern world yet.
MP3 was, when we wrote the original system, only mono or stereo. Then "Backward compatible multichannel" was created (which was a bad idea for reasons related to multichannel content more than the MP3 algorithm), and then "NBC" standing for "Not Backward Compatible" MPEG_2 standard, which evolved after the "BC" systems for multichannel, and that was finalized as MPEG_2 AAC, which does from 1 channel to 16 channel pairs PLUS 16 mono channels PLUS 16 LFE channels PLUS 16 steerable channels. (yes, I wrote a great deal of that bitstream).
This was then further complexified by MPEG-4 AAC, adding "low rate" (i.e. bad qualiity if you want my opinion) options, more high rate options, some various options relating to other audio tools, etc, and to my thinking was complexified to a degree that just over did it all, and made subsets the only option for most uses and manufacturing.
BUT having a way to send compressed audio channels isn't the point. In a modern system we can stick to 24/48 or 24/96 without any real storage problems, and that's how I would personally like to see such content produced.
This brings us to channel layouts, again. First, let us consider 'quad'. Bzzzt. No center speaker. Absolutely the wrong thing to do, 3 front and one back is the way to go if you only have 4. That brings us to matrixed systems (of which there are a few dozen). Bzzzt, you can't get the time cues right. You can'get 5 independent data streams from 2 of the same rate and format per stream. Mathematics works, like it or not.
So, we move to independent multichannel (PLEASE!!!!). This does not solve your or my problem with current 2 channel recordings. So let's just let that be, although some subtle processing can improve things a bit, it is "improvement" rather than "accuracy".
What is required? First a CENTER SPEAKER. That is the most important position. +- 22 to 30 degrees come next, i.e. left and right.
For side and rear it's more interesting. For side, you must have a channel on each side (symmetric L/R positions) in front of the pinna shadow (the HF loss you get from a sound from the rear of your head) to get immersion at the sides of your body (even for a frontal source in a concert hall). You must also have 2 rear channels, again symmetric, behind the pinna shadow. Once you have that, you can create not only imaging (which is often silly if it's not in front of you, but not always) but also hall sensation that is actually something like the auditory cues you need to sound like you're in a particular venue (good, bad, whatever, any venue).
In a proper 3 channel setup, you would not want to vary the center channel independently of the rest, and the speakers must be effectively identical. For synthetic things, well, maybe yes, you do want a control, but the real question is "how do I get the center channel and not muck up the time domain cues". Strictly speaking, it's impossible, but you can do a "pretty good job" most of the time, more so if a source is panpotted, less so if it includes proper time cues in L and R. Such is life.
Subwoofers and elevation can be addressed elsewhere. Btw, subwoofers, NO unless there's at least 4 properly integrated.