Mid / side processing...

Don't know if it belongs in the brewery...
Also don't know if this has been discussed before...

Anyway, here goes.
M/S processing is not 100% accurate.
For example, let's take a stereo file with low audio on the left side (below 100Hz) and high (above 10.000Hz) on the right.

The M/S encoding process goes as follows:

L-R   - for the side material. Leaves us with a mono signal containing audio below 100Hz and (polarity inverted) audio above 10.000Hz
L+R  - for the mid material. Leaves us with a mono signal containing audio below 100Hz and audio above 10.000Hz

We'll process the side material with a high cut at 5000Hz, cutting away ALL the right side audio.

Now let's decode the signal.

L-R  +  L+R  = 2L
Now what does this 2L signal contain exactly?
2 X the <100Hz audio
PLUS
1 X the >10.000Hz audio since ALL the right side (polarity inverted) audio was taken away during the processing of the side material.

Try it yourself.....

Is there a way to really separate the side from the mid signal leaving us with three signals, mono, left side, right side?
I can't get my head around this, I've done the math but I think it's not possible...

What else would you expect to happen? You started with two signal channels and only EQ processed one of the channels.

It's one of the reason I never started with M/S. I also thought that I was the only one who was hearing phase shifting issues without seeing it on the gronometer.

Speedskater said:

What else would you expect to happen? You started with two signal channels and only EQ processed one of the channels.

Click to expand...

Well, before I thought about it I believed that M/S processing REALLY separated the mid and the side signals.
I did not expect this weirdness with hard panned signals.

detonator said:

It's one of the reason I never started with M/S. I also thought that I was the only one who was hearing phase shifting issues without seeing it on the gronometer.

Click to expand...

How was the M/S  en/decoding being achieved?

radiance said:

detonator said:

It's one of the reason I never started with M/S. I also thought that I was the only one who was hearing phase shifting issues without seeing it on the gronometer.

Click to expand...

How was the M/S   en/decoding being achieved?

Click to expand...

I have done it by trining different ways. With products like Brainworx, or the encoder/decoder plugin by Voxengo. For me it's a no go.

radiance said:

Is there a way to really separate the side from the mid signal leaving us with three signals, mono, left side, right side?
I can't get my head around this, I've done the math but I think it's not possible...

Click to expand...

Yes, it is possible. In fact I wrote my university dissertation on some algorithms I and several others developed and adapted for this. It's an extremely heavy brute force FFT or Wavelet based algorithm, and it is not possible to achieve this without some loss of quality, as far as we found out back then. We basically take FFT or wavelet scoop of a short slice of audio (STFT), compare the frequency domains of each slice, and decide what is common between left and right. That's how we get the true center of stereo audio, ie. the stuff you hear dead center between speakers. That's a cool feature of our brain, I wonder what algorithm is used there. In hard panned material there is nothing in the middle and we get silence.

The separation is basically as good as you can make your wavelet or FFT windowing and can result to some not so good sounding artifacts. This was maybe 2003, and back then it was not possible to have this algorithm running realtime. Today I think there are several freeware plugins doing it, I forget which ones. And I once even bumped into a low end version of this for winamp, hidden as a "karaoke vocal removal" feature! Usually the karaoke stuff is just cheap mid-side tricks, but this one was processing heavy FFT with even source code available.

I'm pretty sure some TC Electronics high end mastering processor has this feature masquerading inside some surround "width" feature, again I forget which. It's been a while.

Maybe some math geek has already done the ideal version on paper, but nobody discovered it could also be used for audio yet. Similar to how autotune algorithm was originally used for cleaning up seismographic data for oil rigs..

Kingston said:

.....an extremely heavy brute force FFT or Wavelet based algorithm,

Click to expand...

I see, so no analog version for now 

Nah, plenty of people using it in analog domain, lots of mastering houses.  Makes me think you are recording string quartets or something.  I find far more problems from poor spaced stereo mic placement than I do with MS processing.  I have occasionally mixed into a MS bus compressor array, with great results.  Is it accurate?  Not a question that has concerned me, since it's achieved what I wanted, and with no obvious sonic problems.  Your mileage clearly varies. 

emrr said:

Not a question that has concerned me, since it's achieved what I wanted, and with no obvious sonic problems. 

Click to expand...

Same here...however, understanding the math behind it made me think.
Before I always thought that mid and side material was really separated by this (l-r  l+r) process but this is clearly NOT the case.
Does not mean the M/S tool is unusable all at once....far from it. I kinda like what it does....

radiance said:

emrr said:

Not a question that has concerned me, since it's achieved what I wanted, and with no obvious sonic problems. 

Click to expand...

Same here...however, understanding the math behind it made me think.
Before I always thought that mid and side material was really separated by this (l-r  l+r) process but this is clearly NOT the case.
Does not mean the M/S tool is unusable all at once....far from it. I kinda like what it does....

Click to expand...

I think you are making heavy work of a simple concept. Read the "The Inventor of stereo' the life and works of Alan Dower Blumlein".
Focal Press, ISBN 978-0-240-51628-8

Frank

in what ways does m/s micing differ?

12volts said:

radiance said:

emrr said:

Not a question that has concerned me, since it's achieved what I wanted, and with no obvious sonic problems. 

Click to expand...

Same here...however, understanding the math behind it made me think.
Before I always thought that mid and side material was really separated by this (l-r  l+r) process but this is clearly NOT the case.
Does not mean the M/S tool is unusable all at once....far from it. I kinda like what it does....

Click to expand...

I think you are making heavy work of a simple concept. Read the "The Inventor of stereo' the life and works of Alan Dower Blumlein".
Focal Press, ISBN 978-0-240-51628-8

Frank

Click to expand...

A genius, one of a kind.

Best,
Mattia.

MS mic technique is the "real" MS  and can be simulated in the
hardware world, haven't messed with plug in versions yet

M  mid mic  cardioid facing the center.

S  side mid  figure eight facing side to side.

one mic oriented above the other (generally) both capsules very close.

split the side mic into two channels (either with Y cable and go to two console pre amps and
pan those two channels left and right, or buss to two tracks, pan L & R)
flip one of the two out of phase.
 >note if you go to two mic pre amps, with a Y cable (my usual method) you only need phantom from
 one of the two channels.

mid mic (cardioid) to one channel, panned to middle.

bring up the mid mic into the mix.  your main sound.
now bring up the two out of phase channels.   wow!

the more you bring up the side channels the wider the signal
to balance the two side channels, pan em to the middle for a second,
mute the mid channel, adjust level of side mic faders till signal goes away (duh)
pan to the sides again and add in the mid mic again.

what blows me away is some one can snap their fingers o  the left side of the mics,
and it's in the left speaker.  snap that finger on the right side of the mics, and it's
in the right speaker. (ok, depending on which side channel you flipped outta phase).
stuff in the middle is in the middle.

great room mic setup.  great for ensembles in a semi-circle.
just freakin' great.

and you know what's gonna happen if it is  played back in mono.
you only get the mid mic.  (that may limit how wide you wanna go)

three channels for mic MS.  five? or seven? for setting it up from stereo
audio tracks.

MS and ORTF are my two favorite stereo mic setups.

TRY 'EM BOTH

thanks Mr. Bag.

Great post mr bag.

I guess my thinking got more towards the mid /side processing in for example a mastering set up instead of the recording technique.

HI,


  I should ad that the other great advantage of ms recordings is that you can just use the mid channel for mono! XY has to be summed for this. In the old days of tape, often I would be requested to record say percussion in stereo, only to need a track back later. you can just erase the side, and you got da mono . . .


   Kindest regards,


   ANdyP