I intended to get my editing computer back in operation and get down to mixing past recordings over the past year.. but, well same old story.
Anyway, I found an old installation of Audacity on my work computer last week and pulled up a classical recording I'd made prior to shutdown, for a much needed listen and a grin. I used no EQ at all, which is clunky in Audacity (this version at least) because it is not real-time adjustable. So I decided to use simple channel mixing only and see how good I could get it.
This recording was made using my simpler four channel OMT rig using 4 x DPA 4060s. The raw recording is in a LRCB (B = back) format using one microphone mounted centrally on each side of an acoustically absorbent rectangular baffle measuring approximately ~20" wide x ~4"-7" thick. You might think of the arrangement as being somewhat similar to mixing a pair of spaced omnis with a front/back oriented pair of directional mics in the center.
Anyway, I brought in the L/R pair and level balanced those two channels.
Brought in the C/B pair and split it into two mono tracks, panned the Center channel mono track to center and level balanced that against the L/R pair.
Good.
I then panned the mono back channel to center and brought it up to taste.
As usual, adding some rear facing (B) channel improved on the LCR mix alone.
Taking it a step further.. and the reason for this post-
I then copied the Back channel to a second track, inverted polarity on the copy, and combined straight and polarity inverted channels to a stereo track again.
This was essentially treating the rear-facing channel like the Side channel of a M/S pair, except instead of a side-facing fig-8 coincident with C as Mid, it is a rear facing baffled omni.
I check it by soloing B+/B-.. and yep, it sounds much like a polarity inverted pair or a Side fig-8.
I un-mute the other 3 channels, bring the +B/-B fader back up and wow. Such ambient width and dimension. So much better than mono B panned center.
Really good.
I mute the L/R pair, soloing the essentially M/S'ed C/B pair alone.
Not bad at all. Not as much missing as I expected. Great clarity, depth and dimension. No discrete L/R imaging, but it takes some listening to hear that the violins are no longer positioned left and cello and bass positioned right behind the up front piano dominating the center, instead the instruments that are further away become more diffuse, reverberant and enveloping, with a lush width complementing the clear center, yet without discrete imaging.
I do the opposite and solo the L/R pair.
I hear L/R imaging again, but if I had to choose one or the other I think I'd prefer the C/B alone. It lacks directional imaging but has better clarity with a more engaging reverberant width and depth at the same time.
I go back to the full mix.
Yes, this is the full ticket, the complete package.
I play a bit with fine tuning the balance trade off between L/R and the now stereoized B (stereoized only in combination with the other channels).
It's best getting L/C/R balanced first for imaging, then bringing up +B/-B for depth, reverberant ambience, and naturally enveloping audience applause.
The caveat? Bringing up +B/-B begins to cancel out some low-end on the polarity inverted side, skewing balance in the low end.
I make a slight adjustment and bring up R a couple dB to compensate. Not perfect, but OK
The better fix? Instead of polarity inverting one copy of B, use the Voxengo PHA-979 stereo phase rotation VST to modify phase of one copy of B by +90° and the other by -90°. This will keep low-frequency cancellation symmetrical. Might also try some slight time-shifting of B, either using the delaying function in PHA-979 or via the multi-track editor itself. PHA-979 is typically used for distance and phase alignment of drumkit mics, and making low frequency phase corrections. There may be others, but I know of no other plugin with the same direct functionality.
Additional potential benefit? This kind of symmetrical phase rotation is the basic operation a matrix surround encoder does with surround channel information when creating a 2ch LtRt output. When played back through a matrix decoder, the B channel information will be routed to the surrounds. Would need to play with it to make sure the levels end up right played back that way as well as in 2channel, and sure, matrix surround is no longer of much interest to many these days, but this is a cool additional benefit with no additional cost or detriment. Although it has long been interesting to me I don't really wish to emphasize this last point, as I am mostly just enamored with how well this technique works as a mixing technique for OMT recordings intended for traditional stereo playback. Yes, I've speculated about this technique previously in these threads, but its time to bring it back to focus again because it's just plain better IMO. Rocksuitcase, what enlightenment might you provide from your past experience with the surround encoder/decoder you once were involved with developing and selling decades back?