Is the term "mid/side" being misused?

[Nonlinear]

Originally Posted by Robo

Yep, I agree and will add Voxengo MSED and Brainworx bx_solo (and presumably other Brainworx plugs) to the list that act like this.

Information in the centre can be isolated and doesn't contain anything in the left and right channels. You can mute/solo the centre or the sides to reassure yourself what is being affected.

I assumed this was Mid Side processing as everyone understands it but now realise it must be a development of the original M/S - and definitely a more useful one by the sounds of it.

Cheers

I'm not sure what is happening in your case but I just tried BX Solo and it behaves exactly like the "sum/difference" approach. If I solo the mid channel and sweep my input full left to full right the mid mix level does not change (it's simply a mono mix of L and R). If I solo the side channel and do the same pan sweep the level goes "max - 0 - max".

So in my test it seems there is nothing magic about BX Solo. It's simply sum/difference like everyone else. I haven't tried Ozone but assume it's the same as well.

[dcollins]

Originally Posted by Nonlinear

So in my test it seems there is nothing magic about BX Solo. It's simply sum/difference like everyone else. I haven't tried Ozone but assume it's the same as well.

I think someone mentioned Adobe Audition as having a real center channel extraction function. But, as JJ mentioned, this function is never going to be perfect. And it has nothing in common with L+R mono sum.


DC

[SweetLossy]

Originally Posted by j_j

Um, the 1/SQRT(2) IS in both the forward (to MS) and reverse (to LR) transforms, and as such, is the same as dividing once by 2.

So I'm not sure what you are asking.

Well, I understand that the 1/sqrt(2) form is symmetric.
But the problem is that we can't use it for processing 24 bit based audio because the result of max L+R is beyond 0dBFS.

So my question is, in what context is it advisable to use this 1/sqrt(2) form?
Thanks!

[FabienTDR]

Originally Posted by Nonlinear

Several plugins these days are advertising "mid/side" capability. But when you read the description it defines "mid" as the sum and "side" as the difference of L & R channels.

I may wrong but that's not the same thing as true "mid/side".

You are right. As stated before, the correct term is SUM/DIFFERENCE.

MID/SIDE is an invention of ad-agencies. It is definitely misleading, miss-used and over-rated. A good example is Sum/Diff processing with non-linear phase EQs who also phase distort both channels differently or Sum/Diff Compression who severely fvcks-up the stereo positioning (the latter has its use in vinyl processing where the whole stereo image will suffer anyway). Sadly, music magazines are full of these strange recommendations.

Originally Posted by Nonlinear

So, is the term "mid/side" processing being misused or do some processors actually process ONLY the middle and ONLY the sides?

Not to my knowledge. All use the following formula:

SUM = L + R
SIDE = L - R

L = ( SUM + SIDE ) / 2
R = ( SUM - SIDE ) / 2

[j_j]

Originally Posted by SweetLossy

Well, I understand that the 1/sqrt(2) form is symmetric.
But the problem is that we can't use it for processing 24 bit based audio because the result of max L+R is beyond 0dBFS.

So my question is, in what context is it advisable to use this 1/sqrt(2) form?
Thanks!

Pick your poison, and why are you working in integer, I have to ask.

[SweetLossy]

Originally Posted by j_j

Pick your poison, and why are you working in integer, I have to ask.

The integer question is for compatibility.
Almost all plugins and of course DA converters work taking into account the integer max limit, so the 0 dBFS limit.

An "1/sqrt/(2) MS Matrix" is not compatible with DA conversion and hardware or software dynamic processors with a 0dB max Threshold. Also non linear processors like tape emulators, saturation, console emulation, analog EQ emulations... etc. It will work OK only in linear devices: pure float EQ or...

... maybe using it for a spectral window?

[DubStudio]

I mainly master to vinyl so I like to think of it as Lat/Vert but yes mid/side is a bit of a mis-nomenclature we are probably stuck with, just like "phase" being used in place of "polarity", though arguably not quite as misleading.

I use a simple thought experiment to get my head around the way it works.

M contains information from both L & R, and so does S. We know M = L+R, but what about S? Well I like to think of there being 2 versions of S... "SL" and "SR":

SL = M/2 - L
SR = M/2 - R

The difference between SL and SR is just a question of polarity. If we give L and R some arbitrary values, say L = 9 and R = 7, we can then give values to M and S (16 and 2).

So M/2 = 8 and SL = -1 and SR = 1. The two numbers will always be the same, but one will be negative and the other positive.

If we then change the value of S before decoding it back to L & R, say by adding 2 to give S = 4. Then effectively we have added 1 to L and taken away 1 from R, thus making the difference between the two greater, even though the sum is still the same:

M/2 + S/2 = 10 = L
M/2 - S/2 = 6 = R

That's why boosting S widens the signal, and reducing S narrows it. Of course being a wave, S will always have a range of positive and negative values associated with it, but the effect is still the same. Its either adding to L and taking away from R, or vice versa depending on whether its value is positive or negative.

Not sure if that helps anyone to think of it that way, but it helped me! Incidentally, it is possible to set up a type of matrix whereby you can control SL and SR independently. Its actually a lot more simple than the standard M/S matrix, but I guess its not that popular as it requires an extra channel.

[Tim Farrant]

Originally Posted by evosilica

Ähhhhm... NO

A mono signal can never be in the Side channel only, you can pan it whereever you want. There's gotta be some 180° phase thingy ionvolved to make it side only

No, this is the mis-understanding. If the signal is hard left or right, it cannot be in the middle, can it ??? The "middle" contains signals common to left and right. It has nothing to do with phase at all, except you use polarity reversal to extract the components in the MS encode/decode process.

[Alexey Lukin]

Sorry, but if the signal is only present in "L" or "R", it will be both in "M" and "S". It's easy to see from the formulas. That's the issue with "non-ideal separation"...

[Greg Reierson]

Originally Posted by Tim Farrant

It has nothing to do with phase at all, except you use polarity reversal to extract the components in the MS encode/decode process.

If by "nothing" you mean "everything", them I completely agree.

GR

[Tim Farrant]

Sorry, I'm thinking of a stereo spreader which is of course a different thing altogether !!

[FabienTDR]

I investigated further and it seems that my assumption that the term "M/S" is just another inaccurate marketing buzz word was wrong. At least for certain contexts.

In fact, all serious articles and literature available to me mention the term "Mid/Side" only in the context of microphone techniques. In this special case, the "Mid" is really the mid and the term absolutely accurate. The technique also only uses M/S decoding, an LR stereo signal is actually never encoded to "Mid/Side" in this context.

On the other hand, the proper term for encoding is always Sum/Difference, the generated signals only represent the mono sum and the stereophonic difference (not the mid and side).

The whole confusion is most probably related to the fact that both Mid/Side and Sum/Difference techniques use the same decoding algorithm.

[dcollins]

Yes, looking at it as "sum" (Left plus Right) and "difference" (Left subtracted from Right) is the best way.

The combination of the two channels is the sum, what is different between the two channels is the difference.

Agreed is has become needlessly complicated..................


DC

[SweetLossy]

Originally Posted by dcollins

What is common to the two channels is the sum, what is different between the two channels is the difference.
DC

A hard Left panned instrument it's not in the Right, so it's not common to the two channels but it's present in the sum.

It's better to try to understand it from a polarity point of view:

"+" = in-phase
"-" = out-of-phase (180º flip).

Sum: (+L, +R)
Difference: (+L, -R)

Sum = all the mix elements. (Mono audio stream).
Difference = Left elements in-phase, Right elements out-of-phase. (Mono audio stream).

So, the consequence of Difference it's that "in-phase & equal level" elements of both Left & Right disappear.
In Side (Difference) you have all the mix elements except the "phantom center" elements.

[Joelistics]

As opposed to what was mentioned earlier, Brainworx plugins work on an L+R=M basis. Hard panned signals are represented equally in both M and S. Goes for most M/S plugins i have worked with

Waves Center is a bit of an interesting one that seems to work using a different formula. Hard panned signals not represented in M and the unit seems to use some dynamic processing to further seperate the channels.. Interesting in mixing although not usable in mastering...

[narcoman]

Originally Posted by Nonlinear

Yes - absolutely!

My point of this post is that the notion that "mid/side" isolates the mids and sides is not how it actually works. It's just a different "X-Y" plane, if you will, that is not as obvious as sometimes described.

That's what I have learned from this discussion and I appreciate all your inputs. Thank you!

It's not a different X Y and at all. It's a mono plane ( !!) with a decoding phase plane. It has more in commodity a mathematical ansform than it does a Cartesian grid. The S signal ( whether you call it Side, Stereo, Vert etc) is a special phase signal that repositions the information stored in M. any information NOT in M but present in S will be decoded as absolute anti phase..... The stuff that disappears in mono.

[j_j]

Originally Posted by narcoman

It's not a different X Y and at all. It's a mono plane ( !!) with a decoding phase plane. It has more in commodity a mathematical ansform than it does a Cartesian grid. The S signal ( whether you call it Side, Stereo, Vert etc) is a special phase signal that repositions the information stored in M. any information NOT in M but present in S will be decoded as absolute anti phase..... The stuff that disappears in mono.

Um, an axis transformation and a mathematical transformation are precisely the same thing.

Yes, it's hard to visualize an 8192 point FFT as an axis transform, but that's what it is, a transform that constitutes rotating the signal in 8192 dimensions.

This is why a transform's inverse is its complex conjugate, in fact, because it rotates the data back to the original axis set.

N.B. This is probably a touch more higher math than you wanted, but you should be clear on what an axis rotation does.

Btw, M/S is in fact a 2 point FFT, in fact it's a 2-point almost-anything transform.

[narcoman]

Originally Posted by j_j

Um, an axis transformation and a mathematical transformation are precisely the same thing.

Yes, it's hard to visualize an 8192 point FFT as an axis transform, but that's what it is, a transform that constitutes rotating the signal in 8192 dimensions.

This is why a transform's inverse is its complex conjugate, in fact, because it rotates the data back to the original axis set.

N.B. This is probably a touch more higher math than you wanted, but you should be clear on what an axis rotation does.

Btw, M/S is in fact a 2 point FFT, in fact it's a 2-point almost-anything transform.

I was dumbing it down, the point being it's a phase system against a planar system and therefore not the same thing..... both may describe the same series but they are not the same thing. Changing "X" doesn't have a direct equivalent in M/S without decoding back to XY.

[SweetLossy]

Originally Posted by narcoman

I was dumbing it down, the point being it's a phase system against a planar system ...

Based on psychoacoustics and not only in math formulas.
What it's OK for radio or telecommunications is not OK for sound because our perception.

The important thing for the users of a M/S matrix it's to understand that the Side channel is a mono sum of the Left channel and the phase inverted Right channel.

It's very easy.
Why to complicate it more?

[bicarbone]

Originally Posted by SweetLossy

A hard Left panned instrument it's not in the Right, so it's not common to the two channels but it's present in the sum.

It's better to try to understand it from a polarity point of view:

"+" = in-phase
"-" = out-of-phase (180º flip).

Sum: (+L, +R)
Difference: (+L, -R)

Sum = all the mix elements. (Mono audio stream).
Difference = Left elements in-phase, Right elements out-of-phase. (Mono audio stream).

So, the consequence of Difference it's that "in-phase & equal level" elements of both Left & Right disappear.
In Side (Difference) you have all the mix elements except the "phantom center" elements.

Hi, thank you. Eventually someone is explaining things close to the way I understand it.

So, mid is the mono audio stream obtained by summing L+R, whereas side is the mono audio stream that got canceled out in the aforementioned summing (due to being 180° out of phase with that first sream).

In a few words:

mid=what you get when you hit the mono button on your console,

side=all the stuff that's completely out of phase and therefore vanishes when you hit that mono button.

Am I right? (sorry for my english).

[evosilica]

Originally Posted by bicarbone

In a few words:

mid=what you get when you hit the mono button on your console,

side=all the stuff that's completely out of phase and therefore vanishes when you hit that mono button.

Am I right? (sorry for my english).

100% correct

[Joelistics]

Originally Posted by bicarbone

Hi, thank you. Eventually someone is explaining things close to the way I understand it.

So, mid is the mono audio stream obtained by summing L+R, whereas side is the mono audio stream that got canceled out in the aforementioned summing (due to being 180° out of phase with that first sream).

In a few words:

mid=what you get when you hit the mono button on your console,

side=all the stuff that's completely out of phase and therefore vanishes when you hit that mono button.

Am I right? (sorry for my english).

Not as I understand it, no.

Try panning a 1kHz tone hard left and then isolate the center channel followed by the side channel. They should both play back mono and at equal volume now.
This is because you add what is in the left channel (everything) to what is in the right channel (nothing) to get M. To get S you take what is in the left channel (everything) and subtract it from what is in the right channel (nothing). Ergo hard panned signals are represented equally in M and S channels and isolating gives you the same result as hitting the mono botton for BOTH M and S!

[evosilica]

Originally Posted by Joelistics

Not as I understand it, no.

Try panning a 1kHz tone hard left and then isolate the center channel followed by the side channel. They should both play back mono and at equal volume now.
This is because you add what is in the left channel (everything) to what is in the right channel (nothing) to get M. To get S you take what is in the left channel (everything) and subtract it from what is in the right channel (nothing). Ergo hard panned signals are represented equally in M and S channels and isolating gives you the same result as hitting the mono botton for BOTH M and S!

That doesn't change the fact that:

M = everything in phase (0°)
S = everything out of phase (180°)

So, he is right!!

[DrDeltaM]

Mid/Side (Sum/Difference) is not as intuitively to understand as the sonically strict discrete distinction between Left and Right. That's because Side is a mathematical signal, not a sonically truly relevant signal (as in 'something you'd listen to by itself').

Basically, you can look at it like this:

The more equal something is in Left and Right speakers, the louder it will be in Mid (Sum), of course in relation to its own volume.

The more different something is in Left and Right speakers, the louder it will be in Side (Difference), also in relation to its own volume.

While the Mid (Sum) signal can stand on it's own as a mono-ed stereo signal, Side (Difference) only makes real sense in relation to it's matching Mid (Sum) signal.

Regarding the naming Mid Side versus Sum Difference, the latter one probably is the most clear one. Still, I choose the Mid Side nomenclature myself, as this seems the most general used, and the most familiar one to many people.
in the end, they're both exactly the same.

[bicarbone]

Originally Posted by Joelistics

Not as I understand it, no.

Try panning a 1kHz tone hard left and then isolate the center channel followed by the side channel. They should both play back mono and at equal volume now.
This is because you add what is in the left channel (everything) to what is in the right channel (nothing) to get M. To get S you take what is in the left channel (everything) and subtract it from what is in the right channel (nothing). Ergo hard panned signals are represented equally in M and S channels and isolating gives you the same result as hitting the mono botton for BOTH M and S!

Alright, true, you got me! But who would send a track consisting of a 1khz tone panned hard left to a mixing or mastering engineer?

That would be a nice story: "hey, I had an attended session at Metropolis £££££ and the guys there just did magic to my new radio hit "Test Tone" like they fixed the stereo image and stuff and now it sounds open and in yo' face like on both speakers and shit"

[evosilica]

Originally Posted by DrDeltaM

The more equal something is in Left and Right speakers, the louder it will be in Mid (Sum), of course in relation to its own volume.

The more different something is in Left and Right speakers, the louder it will be in Side (Difference), also in relation to its own volume.
.

Not really. The more different something is in L and R (correlation = 0), the more it will be in equal amounts in M and S.

It has to be equal in L and R but inverted in one channel (180°) to be only in S (correlation = -1)

[DrDeltaM]

Originally Posted by evosilica

Not really. The more different something is in L and R (correlation = 0), the more it will be in equal amounts in M and S.

It has to be equal in L and R but inverted in one channel (180°) to be only in S (correlation = -1)

I mean correlation -1 as the most different, but I understand that I wasn't clear enough on what I defined as 'different'. I should have said opposite.
Basically we're saying the same.

Let me state it more complete and hopefully more clear now then:

The more sounds are equal in L and R (the more they go to correlation = 1), the louder they are in M compared to their appearance in S.
The more sounds are opposite in L and R (the more they go to correlation = -1), the louder they are in S compared to their appearance in M.
Only those sounds that are completely equal in L and R (correlation = 1) are in M only and not in S.
Only those sounds that are completely opposite in L and R (correleation = -1) are in S only and not in M.
All other sounds appear both in M and S. All sounds that are only in L or only in R (correlation = 0) appear in M and S at the same level.

[Joelistics]

Originally Posted by DrDeltaM

Let me state it more complete and hopefully more clear now then:

The more sounds are equal in L and R (the more they go to correlation = 1), the louder they are in M compared to their appearance in S.
The more sounds are opposite in L and R (the more they go to correlation = -1), the louder they are in S compared to their appearance in M.
Only those sounds that are completely equal in L and R (correlation = 1) are in M only and not in S.
Only those sounds that are completely opposite in L and R (correleation = -1) are in S only and not in M.
All other sounds appear both in M and S. All sounds that are only in L or only in R (correlation = 0) appear in M and S at the same level.

Bravo! Beautifully put.

[Babaluma]

Originally Posted by DrDeltaM

I mean correlation -1 as the most different, but I understand that I wasn't clear enough on what I defined as 'different'. I should have said opposite.
Basically we're saying the same.

Let me state it more complete and hopefully more clear now then:

The more sounds are equal in L and R (the more they go to correlation = 1), the louder they are in M compared to their appearance in S.
The more sounds are opposite in L and R (the more they go to correlation = -1), the louder they are in S compared to their appearance in M.
Only those sounds that are completely equal in L and R (correlation = 1) are in M only and not in S.
Only those sounds that are completely opposite in L and R (correleation = -1) are in S only and not in M.
All other sounds appear both in M and S. All sounds that are only in L or only in R (correlation = 0) appear in M and S at the same level.

Seems like we all finally nailed it in easy to understand terms. Bravo!

[bicarbone]

Originally Posted by DrDeltaM

I mean correlation -1 as the most different, but I understand that I wasn't clear enough on what I defined as 'different'. I should have said opposite.
Basically we're saying the same.

Let me state it more complete and hopefully more clear now then:

The more sounds are equal in L and R (the more they go to correlation = 1), the louder they are in M compared to their appearance in S.
The more sounds are opposite in L and R (the more they go to correlation = -1), the louder they are in S compared to their appearance in M.
Only those sounds that are completely equal in L and R (correlation = 1) are in M only and not in S.
Only those sounds that are completely opposite in L and R (correleation = -1) are in S only and not in M.
All other sounds appear both in M and S. All sounds that are only in L or only in R (correlation = 0) appear in M and S at the same level.

Brilliant . I should print this.