Topic: MID-SIDE thoughts

[Nick's Picks]

i've been wondering...

now...if you run mid side and keep the gain of each capsule to be equal (and matrix on the fly)..., then the output is equal to just running XY at 90.

if you run crossed figure of 8 in m-s mode, the output is blumlein (again, if the gain is equal on each side).

Yet...it does not sound the same as running in those configs to begin with.  My thought...
in MS mode, one capsule is directly on axis w/the sound source.  not much in the way of other stereo recording techniques allow a microphone to be placed directly on axis, which will certainly produce a better sounding recording than a mic pointing off axis (think a single mic, for example).

so w/that in mind.  you have one capsule on axis that sounds (arguably) as good as it can ever sound.  When you do the matrix you have different sounding characteristics then if the mics were set up w/an off axis stereo technique.

follow ?
is this all in my head?

[stirinthesauce]

i've been wondering...

now...if you run mid side and keep the gain of each capsule to be equal (and matrix on the fly)..., then the output is equal to just running XY at 90.

if you run crossed figure of 8 in m-s mode, the output is blumlein (again, if the gain is equal on each side).

Yet...it does not sound the same as running in those configs to begin with.  My thought...
in MS mode, one capsule is directly on axis w/the sound source.  not much in the way of other stereo recording techniques allow a microphone to be placed directly on axis, which will certainly produce a better sounding recording than a mic pointing off axis (think a single mic, for example).

so w/that in mind.  you have one capsule on axis that sounds (arguably) as good as it can ever sound.  When you do the matrix you have different sounding characteristics then if the mics were set up w/an off axis stereo technique.

follow ?
is this all in my head?

I think, according to my understanding of DSatz's post in another thread, is if you ran an omni as your center mic, then yes, essentially what you get is xy cardoid. 

Re-read his posts, great insight in those.

[Nick's Picks]

pretty sure a card as mid outputs XY.  I've got a great set of pictures showing all this ...from the new stereo soundbook.  I think I scanned/posted this info once before.

in any case, the point was really about the sonic characteristic of having a mic directly on axis vs. both mics pointing off axis.

[stirinthesauce]

pretty sure a card as mid outputs XY.  I've got a great set of pictures showing all this ...from the new stereo soundbook.  I think I scanned/posted this info once before.

in any case, the point was really about the sonic characteristic of having a mic directly on axis vs. both mics pointing off axis.

You are correct about the mid in MS outputting xy.  I misunderstood or misread what you were saying.  As for on axis, off axis characteristics, I guess that would depend on the mic since some mics have off axis response equal or nearly so to the on axis response as others suffer poorly.         

Here it the quote I was referencing too, cut and pasted the relevant part as it was a long post. 

When you talk about possible "blind spots in the resulting combined pattern," I can only think that you have a somewhat confused mental picture of how M/S recording works; "blind spots" aren't a problem at all, either in theory or in practice. Similarly, what you said about the mono compatibility of X/Y stereo recording is perfectly true--but please note that when you sum L + R from an X/Y (coincident) microphone pair as you discussed, you get the precise equivalent of M in an M/S pair. Or to put it the other way around: L and R are obtained from an M/S pair respectively by adding (L = M + S) and subtracting (R = M - S) the microphone signals. M is the mono sum to begin with, and S is the difference between the L and R stereo channels. It's exactly parallel to the way FM stereo is broadcast. A mono FM receiver gets only M (= L + R), while a stereo receiver gets M but can also detect a subcarrier which is modulated at the transmitter by the L - R signal (= S). In stereo mode, this S signal is matrixed with M to get L and R.

[Nick's Picks]

this is even better:  http://www.wikirecording.org/Mid-Side_Microphone_Technique

[stirinthesauce]

this is even better:  http://www.wikirecording.org/Mid-Side_Microphone_Technique

good stuff, +T 

[Shawn]

this is even better:  http://www.wikirecording.org/Mid-Side_Microphone_Technique

Interesting...

If you make the Mid microphone an omni pattern, then when you decode it the spacing is fixed at a 180º spread - that is, the resulting patterns aim due left and right no matter the volume. But this time, as you adjust Mid vs. Side volume the pattern shifts from cardioid when the microphones are equal, to wide cardioid when there is more Mid, to hypercardioid when there is more Side.

[stirinthesauce]

this is even better:  http://www.wikirecording.org/Mid-Side_Microphone_Technique

Interesting...

If you make the Mid microphone an omni pattern, then when you decode it the spacing is fixed at a 180º spread - that is, the resulting patterns aim due left and right no matter the volume. But this time, as you adjust Mid vs. Side volume the pattern shifts from cardioid when the microphones are equal, to wide cardioid when there is more Mid, to hypercardioid when there is more Side.

Indeed.  I'm going to try this the next time I'm out.

[midside]

While the diagram of an MS matrix with a card looks similar to XY, it will sound different because (as you said) the microphones are not pointing the same way as an XY pair.  Therefore, each capsule does pick up sound differently with different characteristic.  I find that MS tends to add a lot more of the room sound by the side firing figure 8 and can sound much fuller and richer given an acoustically friendly room.  The nice thing about MS is that if the room is too noisy, you can matrix stronger on the mid cap in post to reduce the noise but doing this will narrow the width of the recording.  On the other hand, you will not notice as big of a difference between XY and MS when recording outside.

[BWolf]

i agree that is has something to do with the mics being on axis to the source.  i also think it has a lot to do with the characteristics/frequency reponse of a fig 8 capsule as opposed to the card or omni.  each will color the sound in a slightly different way...

[Nick's Picks]

I recorded last night, on-stage MS w/an omni as the mid.
matrixed on the fly.

I somehow managed to swamp a channel....., the resulting recording is still excellent, but doesn't really image.  wonderful blend/mix of instruments though.

so, to fix it...and tell me if I go wrong here, I need to send it through a ms matrix again, this time its output will be the raw mid and side channels, which I can then swap and re-matrix.  I'll do it 50/50 so that I maintain my original mix.

whoa...
It'll be worth the work though.

[DSatz]

Nick's Picks, if you swap the channels in an M/S recording and matrix at unity gain, you end up with one channel in inverse polarity. If your editing software lets you reinvert the polarity ("phase") of just one channel, that should be simpler than dematrixing and rematrixing the entire recording, and should give you the correct result.

Normal matrixing is (M + S) -> L, (M - S) -> R.
What you apparently did was (S + M) -> L, (S - M) -> -R.

So your L channel should be fine as it is, but your R channel is most likely inverted.

--best regards

[Gutbucket]

pretty sure a card as mid outputs XY.  ...

Nick, what do you mean by 'outputs XY'?

All decoded M/S recordings output some form of X/Y.

I find the technique of M/S fascinating.  Everyone posting here seems to have a good handle on it especially 'as applied'.  Following is a sum up on the more technical aspects of the technique.
--------------------------------------------------
Part 1

There seems to be two questions being addressed in this thread. One is the (easier) technical question:  What 'virtual mic pattern and angle' is produced by various mid mic patterns & matrixing ratios, according to M/S theory?

For Nick's particular example of a M/S recording using cardioid mid mic decoded with equal levels of M & S matrixing, the resulting 'theoretical' equivalent pattern is virtual hypercardioids with an approximate 130 degree inclusive recording angle. You can find the answers to that question in reference books, via a web search or the links others have posted.  Some common configs:

Omni mid = X/Y cardioids at 180 degrees
Cardioid mid = X/Y hypercardioids at approx 130 degrees
Hypercardioid mid = X/Y tighter hypercardioid pattern at around 105 degrees
Fig-8 mid = X/Y fig-8's at 90 degrees (AKA Blumlein)

That's the idealized situation according to theory.

Some interesting implications:

• If you use a figure-8 for the mid mic you can vary the mic angle when decoding, but not the pattern. In other words, the resulting virtual mic directional patterns don't change at all as the mix ratio is adjusted, they are always a crossed figure-8 pattern, just the virtual angle between them changes.
  
• If you use an omnidirectional for the mid mic you can vary the directional pattern when decoding, but not the angle. In other words, the resulting virtual mic angle doesn't change at all as the mix ratio is adjusted, it is always 180 degree opposed, just the virtual mic pattern changes.
  

..and by extension:

• You can not produce a virtual X/Y cardioid pattern stereo pair at any angle other than 180 degrees (opposed).
  
• Any virtual pair that features a recording angle of less than 180 degrees will always have a pattern that falls somewhere in the hyper cardioid-ish region to varying degrees, between figure-8 and cardioid.
  

--------------------------------------------------
Part 2

The other question is outside the realm of theory and asks "How do the virtual patterns produced by the M/S technique compare to actual microphones set up to correspond to the same mic pattern and recording angle?"

This question goes beyond theory and deals with the characteristics of real world microphones and the subjective results of M/S recording, so for me it is the more interesting question.  Here are some variables not addressed by the theory that are in play here, some may not be a big deal:

• As others have stated the mid mic points at the center of the source, not off-axis as with a stereo X/Y pair and real world mics have non-ideal off-axis responses. In the same way, the side mic points at the walls or the extreme edge of the sound stage.  We all know most mics are designed around their on-axis response and it is easy to extrapolate the "point it at the sound and the mic sounds best" effect to the M/S capsule orientation.
  
• The quality of the summing at the heart of the M/S technique is dependant on how well the matrixing circuit is implemented and balanced (either in hardware or software). 
  
• The proximity of the mic capsules is likely more critical in M/S than X/Y stereo because the matrix summing that the technique is based on happens in the electrical (or digital) realm, where cancellation & peaks in frequency response due to comb-filtering can be more of a problem. With X/Y any summing between signals occurs only in air making potential comb-filtering slightly less of a problem (ignoring the issue of mixing X/Y stereo to mono, of course).
  
• As mentioned in the first point, the pattern response of real microphones varies off-axis, often radically with frequency, especially for cardioid pattern mics. This effects the results of M/S decoding in other ways that may not be as obvious.  Consider how an imperfect real world off-axis mic pattern response also affects the shape of the 'virtual' M/S mic pattern produced.  Different patterns will be produced at various frequencies (just as the pattern varies with the real mics) but these patterns will not have the same polar alignment as that of the individual mics.  If you are using a cardioid mid mic, that mic probably has better rejection to sounds from the rear at mid frequencies and less rejection at low and high frequencies.  When decoded, this response variation will have the effect of decreasing the mid mic matrix contribution for midrange frequencies more than bass frequencies as the source of the sound moves farther off-stage towards the rear. The resulting virtual stereo frequency pattern will be weighted more towards the sides for midrange sounds originating far off-stage. The pattern for low frequency sounds originating far off-stage will point more toward the center. Applause & ambience sounds different.
  
• The frequency dependant polar response of a figure-8 mic usually varies much less off-axis than a cardioid pattern mic, but many figure-8 mics have a different response between the front and back of the capsule (in the case the side +/- signal).
  
• Besides the off-axis pattern response variation of each individual mic discussed above, any overall average frequency response differences between the mid and side mics used will further affect the shape of the virtual mic pattern produced at different frequencies.
  

There are lots of variables.  I think the textbook illustrations of the various M/S patterns are somewhat misleading, in the same sort of way that a single polar response graph showing a nice heart-shape doesn't really tell you what the response of your cardioid mic is at each frequency range, but even more so.  The idealized response plot for a cardioid mic is usually is accurate 'on-axis'.  Which is to say, all the varying polar plots will usually line up at zero degrees and agree with the 'idealized' plot on-axis if not anywhere else.  This is not the case with the 'idealized' polar plots of various M/S configurations, because in reality the response angle will vary for different frequencies due to all the variables mentioned above.

I had to think all this though and ended up with way more than I meant to write.  Though it says nothing about applying the M/S technique to various recording scenarios, I think this is a good list of possible reasons why a decoded 'virtual' M/S mic pair may not sound exactly like the X/Y stereo mic pattern it is supposed to emulate.  I think I now understand what is going on better, but in the end the stuff above is all academic.
[/end ramble]

[midside]

Great information here....you should have my screen name!

[BWolf]

pretty sure a card as mid outputs XY.  I've got a great set of pictures showing all this ...from the new stereo soundbook.  I think I scanned/posted this info once before.

this one?

btw, gutbucket, thats some fantastic information.  i have really enjoyed the way my midside recordings have come out, and have experimented in different m/s ratios.  the information you've provided is awesome and i will keep these items all in mind when i'm out there taping.  thanks again.