DP's Spatial Maximizer and applied Mid/Side concepts...

[James Steele]

I'm sorry, I deleted the post with the guy holding his ass because he presumably has diarrhea (I actually had to waste time googling to find how to spell that correctly). I know we enjoy humor here, but maybe keep it above say 3rd grade level?

[FMiguelez]

You better go down stairs an confront Doris with your fine, MLC. Like... now, if I were you

[James Steele]

You better go down stairs an confront Doris with your fine, MLC. Like... now, if I were you

That won't happen. MLC is quite the charmer and brought Doris some chocolates last time he stopped by and now she's quite smitten with him.

[MIDI Life Crisis]

Good thing my wife doesn't read this board.

[Phil O]

Yeah, it's kind of a James Bond - Miss Moneypenney relationship.


Gonna do some more testing today and I'll report back in a bit.

[mikehalloran]

Wow... and to think I took a couple of days off to finish my taxes and get the worst passport photo in the world.

This may shed some light on the subject. How M+S saved a recording.

A long time ago in San Francisco, I recorded a chamber music concert at a church right across the street from the Anchor Brewery near the Bay. It was the very last concert I recorded on reel-reel and I was going to use it to test DP 2, sitting in a box but unopened. I also ran the feed to my Fostex FD-8, also brand new. As I often did, I recorded the whole gig using an M+S array.

Right after the intermission, all hell broke loose. I thought it was an accident at the brewery, perhaps, but I then knew. Long-time Bay Area residents will remember the KFOG Kaboom! where they had this big fireworks show fired from a barge in the bay. That was the night. 22 minutes of "the bombs bursting in air" on a barge about 100 yards away. It was deafening. Meanwhile, the musicians kept playing and I kept recording.

My customer said, you can fix this, right? Goodgawd... Even if RX Denoise was yet available, this was going to be a stone bitch bound to disappoint everyone, no matter how long I spent on it.

I loaded the audio from the FD-8 into my new G3 using my 2408 in stand-alone (first time I used that, too). Got out the manual, installed DP, saw the MS decode tool and, wow! stereo.

Then I got to those 22 minutes. Just to see what happened, I panned both sides to the middle. The noise nearly vanished. ?!?!? Soloing the S without processing yielded the same result.

It took a minute to figure this out but here's why: The S channel played the differential signal only. Common signals were nulled. The players were seated R/L but none were in the center. The explosions came from outside the church—all that was left were some residual reflections which remained as differential signals. Common mode rejection did the rest. The end result is that the second half of the concert is in glorious mono with a tiny bit of residual noise that I was able to reduce a few more dB with a gate.

My buddy thought I had performed a miracle—I never told him how I did it. Mono for the second half was an acceptable result.

The M captures the entire signal. The S captures the entire signal with the common elements nulled by being out of polarity—but they are still there unless recorded with a single element ribbon mic.

The processor reverses the polarity of the S channel. Combining these two mono signals and panning them hard R/L recreates the stereo field.

Summing both sides gives you a mono signal without the normal bass rolloff due to phase cancellation—unless the bass source is far away and equal in both channels.

It would surprise me if the Spatial Maximizer isn't behaving correctly when used with two mono sources not recorded in M+S.

[mikehalloran]

The S captures the entire signal with the common elements nulled by being out of polarity—but they are still there unless recorded with a single element ribbon mic.

It is also possible to use the M+S technique using three cardioid mics recorded to three channels. That's a different discussion.

[MIDI Life Crisis]

Wow! He's quoting himself now...

j/k Mikey!

[bayswater]

In theory, it would depend on whether your two mono tracks have any common frequency content between them.

That's exactly what I thought! But that's not really happening...
I am testing with a flute and an oboe, both mono, each panned hard L and hard R. I was expecting to hear VERY LITTLE common material (perhaps only some frequency overlap between the 2, but I'm basically hearing the same thing when I listen to the Side or the Mid.

Is there still some confusion about what Mid Side is? Mid is the sum of left and right (or more specifically half the sum). It is not what Left and Right have in common. One is the union, the other is the intersect. Stubbs says exactly this later (M=L+R) in this thread. So oboe and flute might not have a lot in common, but that doesn't matter.

M=.5*(L + R)
S=.5*(L – R)

M+S=.5L+.5R+.5L-.5R
-->L=M+S

M-S=.5L+.5R-.5L+.5R
-->R=M-S

[stubbsonic]

Here's a thought...

https://www.youtube.com/watch?v=z5ejvH_n4wk

This video seems to reinforce our mistaken notion about the MID channel being only the mono content.

[mikehalloran]

Here's a thought...

https://www.youtube.com/watch?v=z5ejvH_n4wk

This video seems to reinforce our mistaken notion about the MID channel being only the mono content.

Yes. The M is the same as sticking a single mic in front of the source. It can be omni, cardioid or even a figure-8. It depends on the pattern how much it hears.

I find that an omni gives too much of the room and audience noise. In a studio or on a sound stage it sounds pretty good.

[magicd]

Hey all. I've been watching this thread all week. My excuse for not posting is I was on the road. Actually it was also because I was not 100% sure of the answer.

One of the things I love about audio engineering is the constant learning. Hats off to Stubbsonic for the clarification. I think he's nailed it and along the way I have learned a fundamental concept about M/S.

Hopefully I can add one little tidbit to the convo. The Side band on the Spatial Maximizer has an HP filter control. What this does is redirect low frequency from the side band back to the mid band. This is useful for vinyl mastering as well as pre-prep for MP3 or other types of data compression encoding. So if the Side band HP knob is turned up, yes some of the side band low frequency will now be part of the mid band signal.

Cheers!
Dave

[Phil O]

May I expand on this?

Ok, an important caveat first. I'm no expert on this topic. I'm simply trying to answer FM's question. If I make an error here please don't get in my face. Just point out the error and move on, please.
Also, I wanted to post some graphics with this, but I'm very tight on time so I have to keep it simple.


A quick review of mid-side recording:

A bi-direction (figure 8 ) mic is placed sideways so that the front of the capsule points left and the rear points right. A cardioid mic is placed facing forward (an omni can be used here as well).

Let's deal with two sound sources A on the left and B on the right.

The Cardioid (or omni) mic will pick up both and it's output will be (A+B).

The bi will pick up both as well, but the rear side inverts the signal so it's output will be (A-B).

Now, to convert this to L-R Stereo, we use three channels. I'll call them Mid, S1 and S2.

We place the Cardioid signal on Mid. (A+B)

The bi signal on S1. (A-B)

The bi signal on S2 and invert it. -(A-B) = (B-A)

Now we sum:

Mid + S1 = (A+B) + (A-B) = 2A = Left Stereo Channel

Mid + S2 = (A+B) + (B-A) = 2B = Right Stereo Channel

Notice that the signal is X2. 20Log2 = 6.02dB


But what if we start with a L-R stereo signal and want mid-side?

From above:

Mid = (A+B)

so: 1/2 (Left + Right) = 1/2 (2A + 2B) = (A+B) = Mid

or: Mid = 1/2 (Left + Right)

Again from above:

Left = Mid + S1

so: S1 = Left - Mid

or: S1 = Left - 1/2 (Left + Right)

remember, S1 was the signal we got from our bi-directional mic (side info)

and Mid was the signal from the Cardioid (mid info)

Notice that we have two signals 2A (hard left) and 2B (hard right) but our Mid signal is (A+B), so it is NOT nonexistent.

[stubbsonic]

The Side band on the Spatial Maximizer has an HP filter control. What this does is redirect low frequency from the side band back to the mid band.
Dave

"back to the mid band" implies that it either was there or should be there (i.e., "back where it belongs"). I suppose a case could be made for the latter.

Ok, so that does explain the inter-channel bleed. When I turned the high pass all the way down to 20Hz, it reduced the amount of bleed, but not completely. The remaining bleed is easily attributed to the slope of the HP going up into the audio range.

Too bad, though. It's still a deal breaker for using the plugin, as I don't want to lose width in this particular way.

[stubbsonic]

Regarding the crosstalk issue, the highpass filter on the side is the culprit, but they really should have an OFF button for it.

I wanted to take a slightly different angle to the MS discussion. We've been relating it to the microphone technique because it kind of works by way of illustration... in a way.

Before we had matrix and de-matrix plugs, if we wanted to do Mid-Side processing -- the mastering processing technique, and not the mic technique, we had a set of tasks to do.

First, take stereo mix and just pan both channels center to make it mono. (Trim plug works well for this.) There's the "mid" channel.

Second, take a copy or mult of the stereo mix. Put it into two mono channels (rather than a stereo channel). Invert the phase of the Right channel. Pan center to hear this SIDE channel as a mono track. Note: this can be done with the TRIM plug in so you don't really NEED two mono channels.

With these two steps you have effectively created an unmatrixed Mid track and Side track. Both are mono.

To rematrix, leave the mono signal as is, but just repan the side mixes channels hard left and hard right. Now when you push up the side channels you can adjust the width.

I'm beginning to think that referring to the MID channel as such "the middle"-- is the cause for our confusion. It is not "the mono components of the mix" it is "the mono mix". I think, in part, it is due to one slight difference between M/S mic'ing and M/S processing. With an M/S mic set up, IF you use a cardioid mic for your mid, you will get a more direct signal from the source with LESS stereo ambience from the sides than the figure-8 mic would get. Theoretically, Mid-Side processing is more akin to a M/S mic set-up with an omni mic in the center-- as (theoretically) it hears the side/stereo non-preferencially.

We should probably refer to the MID channel instead as the MONO channel, and understood that the SIDES channel will through phase cancellation remove hard-panned signals from MONO channel as it is re-matrixed.