MID-SIDE vs SUM-DIFFERENCE

[groover]

I understand the mid-side mic'ing technique and have used it once, substituting 2 <3 SDCs for a fig-8.

My question has to do with processing devices and/or procedures that extract "mid-side" information from a stereo mix, also sometimes called Sum and Difference processing.

I understand how combining L + (-R) or vice versa gives you a mono track with sounds that are in the center being mostly cancelled out. So there's your difference, and your "side" information.

The quandary concerns the Sum, or "mid" information. It seems that in combining L + R into a mono stream you are not actually rejecting the "side" information, just diluting it if anything.

Am I missing something, or do these boxes and plug-ins do more than just sum-and-difference calculations to extract a true middle stream?

[groover]

To put it another way for clarity: If you were to mix 3 mono sources, A, B, and C, to a stereo track with A panned hard left, B panned center, and C panned hard right

A true mid-only stream would have no A or C component, but a summed mono mix would have A and C present but relatively 1/2 the level of B, because B would be self-reinforced.

So are people saying mid-side when they mean sum/difference? Like in a mastering context?

[vvv]

FWIW, and this is my own thoughts and experience, there is a difference between M/S as created by a plug-in, and mic'd M/S.

Mic'd M/S, for me, is a mono Mid, Left-Side is a figure-8 track and Right-Side is the figure-8 as flipped.

At the least this differs from a process-created M/S in that the Mid is from a different mic, placed (however carefully) in a different location; at the least it's aimed differently.

I suppose that if you used 3 identical non-figure-8 mic's (as you did, in an inverted "T" configuration) you might get a slightly different result (relative to cancellations) than I have experienced, but then that wouldn't be what is usually called M/S, as I understand it.

The fact that Right-Side is an exact but inverted copy of Left-Side is what makes it M/S, along with the Mid.

And even using two identical mic's, one in figure-8 and the other switched to cardoid or whatever, the Mid is still gonna be different because of placement.

All that said, the processed-type M/S, I recall reading, also may have some time-shifting involved, which of course changes the phase relationships.

This page has a prettyy good explanation, and a link to a freebie M/S pug-in at the bottom.

[ashcat_lt]

Once you've extracted the side-only information into a single mono track, it's a simple matter of subtracting that information from the mono L+R mix to leave you with the mid-only information.

[groover]

I don't see it a simple, because the side-only info is created by inverting one of the stereo channels, and then combining them to mono. Therefore either the left- or right-side information will be inverted. Simple subtraction from the mono mix won't work, because if you are subtracting the left-side info the right-side info will be additive, and vice-versa .

[groover]

This article on the UA website promulgates the above-described misunderstanding, by stating that Mid=(L+R)/2. (L+R)/2 is actually the SUM, and still contains all the stereo info, albeit combined into mono.

http://www.uaudio.com/webzine/2007/june/index7.html

[Scodiddly]

If you take a coincident-diaphragm LR stereo pair of cardioid microphones and sum them to mono, you end up with the same signal you'd get with a single cardioid microphone pointed straight - so that's how Mid = L + R works. Once you've done this you're mono with your "mid" signal.

The side signal is a bit more tricky - for one thing, a figure-8 microphone really is figure-8, with extremely deep (in theory, infinitely deep) side nulls since the "back" of the pattern has inverted polarity with respect to the front. So if you've got a figure-8 pointed to the sides, it shouldn't pick up anything from straight in front of the M/S array. So, it shouldn't have any of the mid microphone's signal. However, there of course will be some overlap of the cardioid pattern of the mid mic, so you can derive some stereo LR signal. L = M + S, R = M - S simply because the convention is for the side microphone to be pointed to the left. R = M - S is really "R = M + inverted S".

[vvv]

This article on the UA website promulgates the above-described misunderstanding, by stating that Mid=(L+R)/2. (L+R)/2 is actually the SUM, and still contains all the stereo info, albeit combined into mono.

http://www.uaudio.com/webzine/2007/june/index7.html

It's not M = (L+R)/2, it's "M = L+R/2". In the formula, then, Mid and the Sides are 1/2 the original stereo signal volume. (IOW, "(M = L+R)/2".)



I think that "L+R/2" is not to be taken, necessarily, as math, like you are adding popsicle sticks.

Rather, it is representative of the process of adding the two tracks together, and what results, more like adding, say, paint? Because there is more than just a count/i], you are dealing with frequency, and time (spatial cues and phase?), and volume, and their interactions

For example, I submit that if you add Left + Right and get Sum, and then invert, say Left and add (in effect, "subtract" through inversion) it to Sum so you have, LeftInvert + Sum, you will not get an un-altered Right thereafter.

Anyone more knowledgeable wanna tell me I'm wrong, or correct?

[vvv]

I don't see it a simple, because the side-only info is created by inverting one of the stereo channels, and then combining them to mono.

This is, I think, incorrect, or at least unclear.

Re a mic'd M/S group, the "side-only info" isn't like you have a mono track with only that locational info, and it isn't only one track (unless that track is stereo).

The side info comes from the use of the original side (traditionally Left) track, and an inverted copy of that track.

You bring that pair of tracks up or down against the Mid track to adjust the stereo spectrum, with the side tracks being panned out.


In the process-created (as opposed to actually mic'd) M/S stuff, they add the Left and Right sides of an already-existent Stereo track to create the mono Mid. The process then apparently subtracts that Mid information to create the Side track by inverting the Right side of the Stereo track and adding it to the original Left to create what I'll call the New Left. And again, you massage the stereo spread by adjusting the levels of the Mid against that New Left side (panned left), plus the New Right side (comprised of an inverted New Left) panned right.

And keep in mind, there can be more fun as you adjust the panning of the Side tracks because of overlap of their signals (works best when the side tracks are not linked, IOW, are dual mono.)

Bonus: try to vary the processing you apply (either type or amount, if any) on the various tracks.

[vvv]

So, for me as much as anyone:

A mic'd M/S group might be:

Start with a center mono mic, and a figure-8 with the null centered.

1. Left = figure-8 (mono)

2. Mid = center mic (mono)

3. Right = #1 inverted (mono)


A processed M/S group might be:

Start with a Stereo track.

1. Left = original Left + inverted Right of original Stereo track (mono)

2. Mid = Left + Right of original Stereo track (mono)

3. Right = #1 inverted (mono)


Alternatively, a processed M/S group might have #2 above, and numbers 1 & 3 as a Stereo track.

[groover]

1)I wrote the formula MID=(L+R)/2 because the summed signal is reduced in volume by 1/2. The parenthesis indicate that the division by 2 operates on the sum of L+R, not just on R. If you follow normal algebraic convention, MID=L+R/2 indicates that only the Right channel is reduced by 1/2.

b)That notwithstanding, I take exception to referring to the combination of L+R as MID, although it borders on being a semantic distinction. If you add either L+(-R) or (-L)+R you end up cancelling out all the information that is equally present in both sides of the original stereo track. However if you add L+R together you haven't cancelled out any information, just combined it to mono. So IMHO it is not really accurate to call it "MID".

3)Digital audio is all math, albeit extremely complicated math being calculated very quickly!

[groover]

If you take a coincident-diaphragm LR stereo pair of cardioid microphones and sum them to mono, you end up with the same signal you'd get with a single cardioid microphone pointed straight - so that's how Mid = L + R works.

Not exactly, wouldn't the combined coincident pair give you a wider pickup pattern that the single cardoid mic (assuming all the "cardiod" patterns are the same)? Unless the two were both pointed straight ahead of course.

[vvv]

1)I wrote the formula MID=(L+R)/2 because the summed signal is reduced in volume by 1/2. The parenthesis indicate that the division by 2 operates on the sum of L+R, not just on R. If you follow normal algebraic convention, MID=L+R/2 indicates that only the Right channel is reduced by 1/2.

** I hear ya! However, I think your way arguably states that "Mid" is the same as "L+R" divided by two. It's not, as "Mid" equals "L+R". The division by 2 refers to its level relative to the original stereo track. That's why I explained it as "(Mid=L+R)/2", which is itself still unclear, unless the reader knows of the comparison to the original stereo track. To be 100% accurate, mebbe I shoulda wrote, "Level of Mid (Mid=L+R) = 2x original Stereo level". I love me some semantics!
-------------------------------------------------------------------------------------------------------

b)That notwithstanding, I take exception to referring to the combination of L+R as MID, although it borders on being a semantic distinction. If you add either L+(-R) or (-L)+R you end up cancelling out all the information that is equally present in both sides of the original stereo track. However if you add L+R together you haven't cancelled out any information, just combined it to mono. So IMHO it is not really accurate to call it "MID".

**Like stereo itself, it's all a matter of representational relationships; "stereo" is artificial, and the "Mid" you speak of is processed from that "stereo" to represent a "true" mid. Further, if part of the left side of the stereo wave-form is negative and part of the right is exactly opposite, won't it actually cancel when combined? These are not sine waves, and the two sides have varying degrees of phase which should cause both some cancellation and some reinforcement. (See the "/2" above, and even consider the concepts of "masking", linear EQ, etc.)
-------------------------------------------------------------------------------------------------------

3)Digital audio is all math, albeit extremely complicated math being calculated very quickly!

** Ah, but sound is, I submit, different than digital audio!

[vvv]

If you take a coincident-diaphragm LR stereo pair of cardioid microphones and sum them to mono, you end up with the same signal you'd get with a single cardioid microphone pointed straight - so that's how Mid = L + R works.

Not exactly, wouldn't the combined coincident pair give you a wider pickup pattern that the single cardoid mic (assuming all the "cardiod" patterns are the same)? Unless the two were both pointed straight ahead of course.

I think this illustrates an important point I was tryna make about processed M/S.

I think in Sco's example the difference in the result of a true mono mid, versus the combined tracks summed to mono, would be indiscernible without a very close and meticulous comparison of such tracks.

But I also think that processed M/S can sometimes sound, well, processed. I mean by that, that processed M/S sometimes sounds different than true M/S, and I suspect you would get different results if you set out to record, say, a guitar as true M/S, or processed a stereo recording of the same performance to get M/S tracks. Or mebbe not (he waffles)?

Me, I think of true M/S is a method of recording that you undertake to get a sound that is very adjustable to stereo (or whatever) in mixdown; processed M/S is a method of adjusting an already-recorded stereo track.

Have I beaten this to death, yet?

[groover]

When I started this thread I was concerned with processed "MID-SIDE" examples, not true Mid-Side recording with a center mic and a figure-8 side mic.

Part of my question remains unanswered - Do these processors, such as the DrMS plug-in and the Avenson Mid-Side box, go through extra "steps" to remove the information contained in the "side" signal from the "mid" signal (which I figure can be done with a few more stages of combining, panning, and polarity manipulation) or are they really SUM - DIFFERENCE processors?