Mid Side won't null?

[Storyville]

I've set up some Mid-Side fader groups.

Group 1 - L&R centered
Group 2 - L-R panned left
Group 3 - R-L panned right

All elements in each group are pulled down exactly 6db.

Group 1 is sent to an aux channel mono as the "mid." And left as is.
Group 2 & 3 are sent to a stereo channel and pulled down -2.5db.

The Main original stereo track is phase inverted and pulled down 2.5db. In theory, shouldn't this null out completely? Or am I missing something?

[Cellotron]

Mid is L+R as you have it.
But Side is only L-R (the two channels mono'ed together but with the Right channel polarity inverted) - not two groups as you have it.

Best regards,
Steve Berson

[Storyville]

But that would be a mono signal. How would I get the sides back in stereo again?

[Greg Reierson]

Steve is correct.
L+R=M
L-R=S

and to get back it's the same trick:

M+S=L
M-S=R

The stereo information is encoded in the S channel with reference to the M channel. You can't get the stereo info back without both.


GR

[Cellotron]

For Mid/Side encoding from L/R source:
L + R = Mid (aka "Sum")
L - R = Side (aka "Difference")

For decoding of Mid/Side to L/R:
Mid kept up the middle.
Side multed to 2 channels:
- first instance panned hard Left
- second instance panned hard Right and polarity inverted

So seems to me you're not decoding correctly.

Best regards,
Steve Berson

[Greg Reierson]

I think Steve and I posted responses at the same time. Two different methods to get the same result.

If you have an MS matrix, the formula I listed is how it works. If you're doing it in a mixer, Steve's method is the way to go.


GR

[Loki]

Except to clarify Greg's,

M+S = 2L
M-S = 2R

So the amplitude will be double on decode (take that into account and half the amplitude)

For Steve's, I'm a little worried about the math depending on what system you are using. Assuming you are using PT HD with the stereo mixer (dithered or normal) sending M to a stereo out (LR) channel will drop L and R by 2.5db; using the dithered surround mixer plugin as your mixer you'll get a drop by -3db. You need a drop by -6db for both M & S out to make this work right. So pull down the output faders to compensate for what you need depending on which mixer you're using. (PTLE is strictly -2.5db panning law)

If you go mono to 2 mono outs by multing, the signal will stay a constant level so drop both L & R mono faders by -6db on each mono aux out. (Same for hard panning duplicated M tracks each going to L or R.) You have to take the same consideration to get the S right (by multing the side to 2 monos, it'd be dropping both by -6db and remembering to flip polarity on the channel Side out)

The algebraic math looks like this:
encoding to:

L+R = M
L-R = S

decoding by (1) substituting above formulas in second step (2) distributive property applied in the third step and (3) then solving:

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

Since the amplitude is doubled, drop by -6db each side. And as above stated, be careful of the panning laws applied by your DAW if you send a mono to a stereo bus and leave it panned center.

[korbes]

I just did some M/S experimenting yesterday using Voxengos free MSED plugin. Take a look, you might find it usefull.
Another free M/S plugin I like is Brainworx bx_solo.

[JALFK]

hey guys, for anyone who didn't know, you can do pretty much anything you want in M/S just using the inbuilt "gainer" plugin in Logic (or any equivalent plug that has channel polarity and a mono button). In the case of mid side there is no "better sounding plugin" or "more powerful", as it's what you do with the signal after splitting it into M/S channels. After you've split, you can use anything you want on it, compression, expansion, EQ, reverb, delay, anything!

It's cool to get your head around the routing of it, then you can do it when and with whatever you want! So simple, very powerful.

[narcoman]

Originally Posted by Storyville

But that would be a mono signal. How would I get the sides back in stereo again?

the "sides" arent stereo. The SIDE is the information used to add and subtract to a mono M signal to decode stereo. It is not in itself - stereo.

[deltasigma]

One of the beauties with M+S is that the stereo field can be widened or narrowed by adjusting the amplitude relationships between the sides and the mid.

I tend to use my ears to find the optimal position rather than the maths. Also if using close mics it can help with other phase issues (or even more dangerously, using more than one stereo mic setup!).

It has to be one of my favourite stereo mic techs (good old Blumlien!)

However, that said, it is important to be aware the maths. Great bit of algebra there Loki.

D

[Storyville]

Originally Posted by JALFK

It's cool to get your head around the routing of it, then you can do it when and with whatever you want! So simple, very powerful.

Yes. I have no problem doing it with M/S encoder & decoder. Because it does it for me and I don't have to think. I want to have a clear idea of what is going on at the core of it.


L+R = mid. First question. If you have a hi hat that's panned hard left in your mix, and you combine L+R to get the mid info, wouldn't you end up a hi hat in the mid signal? So anything you do to the mid would then raise up the level of a hi-hat in the mid signal wouldn't it?
L-R = side, any information shared by L and R (the mid info) would null out and only the side would remain. So you still end up with the hi hat in this signal as well.

So, to continue on to decoding. So, if you keep the side and mid signal both in the center, you get L+R+L-R, or 2L. Without worrying about pan laws, you could shoot that sum to the Left and subtract 6db to get the original Left signal on the Left speaker. Now we have the regular level of the hi hat on the left.

For the right, invert the side and you get R-L. Then add that to the mid signal. L+R+R-L. You get 2R. Pan that hard Right, subtract 6db, you get the original Right signal. But here's the rub. Let's say you gained up the mid signal. In that mid signal, you now have an increased level of hi hat. So when you decode back to the right, wouldn't you get a little bit of hi hat on the right channel?