Any Reason not to Mid/Side encode to L/R in recording?!

[Prime Mover]

Maybe this is old news to many of you, but it just hit me that if something is recorded in a Mid/Side array and encoded to left/right, by the equations I'm seeing, you can reproduce the original mid/side signals perfectly by inverting the function:

L = (M+S) / Sqrt(2)
R = (M-S) / Sqrt(2)

M = (L+R) / Sqrt(2)
S = (L-R) / Sqrt(2)

I've been banging my head trying to figure out how to monitor Mid/Side for future projects, but maybe I should just reverse the process: encode to Left/Right right in my field recording rig (so it's a normal L/R file). Then in post: Decode, tweak, reencode.

Is there any downside to this? Will repeated Encoding/Decoding cause any drift or signal degradation? I guess there might be some SLIGHT floating point discrepancy over generation, but would it be at all perceptible within a 24bit raster?

[FMiguelez]

This operation is 100% transparent.
You can convert from LR to MS or viceversa as many times as you like. There's no loss or distortion at all, and if you follow the equations with mixer faders (best and most useful way) or a commercial plugin, there won't be any gain changes in the signal.
I also love MS recording and processing. I see no downside about using it and converting it as needed.

If you ever notice any discrepancy in the signal level at all, it's most likely due to DP's old bug of NOT summing to mono correctly.

To encode to MS:
M = (L + R) / 2
S = (L - R) / 2

To decode to LR:
L = (M + S) / 2
R = (M - S) / 2

[stubbsonic]

I've not tested this theory (which is more like "mental garbage") but I have a suspicion.

If you have the track in M/S format, and then you are matrixed to stereo, then you can apply processing to the resulting stereo just as you would a stereo format.

However, if you try to apply processing to either the Mid channel or the Side channel before matrixing, my working theory is that you would change the stereo width of whatever you change. Because M/S requires a delicate relationship between the phase of the Mid (sum) and the Side (difference) channels, things you remove or add to either will affect the stereo width.

A while back, I did a little test with one of DP's plugins and confirmed that changes to mid or side reduced width, but I don't recall the specifics. As FMiguelez mentioned, that might have had to do with the way the plugin was working. I may be wrong about all this.

[FMiguelez]

However, if you try to apply processing to either the Mid channel or the Side channel before matrixing, my working theory is that you would change the stereo width of whatever you change. Because M/S requires a delicate relationship between the phase of the Mid (sum) and the Side (difference) channels, things you remove or add to either will affect the stereo width.

I totally agree.
It would enter the matrix already affected or distorted (as in not identical to the original). It makes more sense to process the Side or Mid inside the matrix.

BTW, what I mentioned above about the transparency of the conversions between these 2 stereo formats is exclusively meant for unprocessed conversions, of course (and gain compensated, according to the equations).

A while back, I did a little test with one of DP's plugins and confirmed that changes to mid or side reduced width, but I don't recall the specifics. As FMiguelez mentioned, that might have had to do with the way the plugin was working. I may be wrong about all this.

It may definitely be related, Stubbs.

According to my last DP9 test, the ONLY way to get DP to sum to mono correctly is by using the Spatial Maximizer plugin and solo its mono component WITH NO FILTERS.

Anything else native to DP, including the Trim plugin, duplicating/panning, summing via sends, etc., will be incorrect and will NOT cancel out with a properly summed mono signal from any other commercial monoing application/plugin.

BTW, I discovered, by pure accident, a little trick to make DP's mono-summing around 35% "less incorrect" (and probably the very exact cause of the problem)...
It gets much better, but still not as it should be. If anyone's interested, I can spill it out...

MOTU?

[mikehalloran]

I never encode the tracks while recording. The the raw tracks are recorded to separate channels. Every once in awhile, I use an M+S 3-mic setup using cardioids and record each mic to a separate track.

Although matrixing is easy in DP, it’s not necessary. Copy the S to another track and invert. Pan both S channels hard R & L. Mix the resulting three tracks.

You can get interesting results by summing the two S channels to mono and eliminating the M track. I did this 20 years ago to eliminate the sound of a fireworks show outside the church where I was recording a chamber concert (SF Bay Area folks may remember the annual KFOG Kaboom! This church was about 200 yards away from the barge). My choice was between summed mono with a tiny puff of explosion noise in headphones or ‘all hell breaking loose’ stereo. When I was remastering those tracks a couple of months ago, I was able to use modern tools, mostly RX 7 Advanced, on those 20 year old tracks to bring back the stereo. The composer didn’t recall the explosions that shook the church — thought I was making it up till I played him the raw tracks. Imagine the 1812 Overture cannons accompanying a violin/cello duet—but worse.

That was a concert I’d recorded in a 3-mic cardioid M+S array instead of a cardioid + figure 8. Two of the cardioids were placed 180° facing the sides while the third faced the ensemble. No matrix required as the two side mics are out of polarity. For a small ensemble with the mic array fairly close, I find that it gives a more natural sound than the narrow lobes of a figure 8 ribbon.