Topic: Mid/Side question

[Duffalo_Soldier]

Hi all,

Thanks for your responses to my 6 stereo pair matrix. I’m still working on that but have been working on a matrix utilising two stereo sources and I ended up going down a rabbit hole with mid/side and I was hoping someone could clarify the following:

I watched a video and someone used Audacity to convert a L/R recording to M/S

Their process was basically as follows:

L+R = Mid (aka Sum) which was then panned to the CENTRE
L-R (aka inverted R) = Side (aka Difference) panned 100% LEFT
R-L (aka inverted L) = Side 2 (aka Difference) panned 100% RIGHT

Mixing the two stereo pairs I have gotten what I think is a nice mix with a good stereo image but I am curious about two things:

1) Are Side and Side 2 the same thing?
2) Do I have to convert the final mix (with two mid channels and either two or four side channels (depending on the answer to question 1 above)) back to standard L/R?

Thanks for your assistance!

[Gutbucket]

There are various ways to transform a Mid/Side signal (which is the same "Sum/Difference") to standard Left/Right stereo.. and back the other way.  The process works in both directions, you can convert Mid/Side (terms which refer to stereo microphone outputs) to Left/Right stereo, and you can convert Left/Right stereo to Sum/Difference (more generic but basically equivalent terms which don't necessarily involve microphones).   

I watched a video and someone used Audacity to convert a L/R recording to M/S
L+R = Mid (aka Sum) which was then panned to the CENTRE
L-R (aka inverted R) = Side (aka Difference) panned 100% LEFT
R-L (aka inverted L) = Side 2 (aka Difference) panned 100% RIGHT

^Except for a couple incorrect statements (L-R does not = inverted R, R-L does not = inverted R) this describes the conversion going the opposite way, converting an existing Left/Right stereo signal into a Sum/Difference stereo signal.  You instead want to produce L/R stereo from a Mid/Side microphone output.

There are easier ways to accomplish it, and I'll describe those in a later post. Even though you are doing this digitally in Audacity, you are essentially doing it the "old school way" as would have been done using a traditional analog 2-bus mixer (2-bus meaning the output of the mixer is 2 channels of Left/Right stereo). Audacity's graphic mixer interface is a digital emulation of a traditional analog 2-bus mixer. Even though there are easier ways of acomplishing the M/S to L/R conversion, doing it this way helps to get a good conceptual grasp on what is going on, and also provides an easy way to tweak things when adjusting the mix.

For clarity, let's ignore the other pairs being including in the mix for now and just focus on conversion of the Mid/Side stereo mic-pair output to standard Left/Right stereo..

[Gutbucket]

Here is the "old school in the mixer" way of converting a Mid/Side output to Left/Right stereo.

1 ) The Mid channel gets assigned to a mono input channel in the mixer. Let's call that channel 1.
2 ) The Side channel gets mult'd to two additional mono mixer input channels, lets call those channels 2 and 3. Mult'd means that the analog signal it gets routed to both channels.  Using a DAW in place of an analog mixer, the equivalent is to load the Side channel file into channel 2, and a copy of the same Side channel file into channel 3.
3 ) Pan channel 1 to center (its signal gets sent equally to both Left and Right mixer output buses).
5 ) Pan channel 2 hard Left, pan channel 3 hard Right. (channel 2 only gets sent to the Left output bus, channel 3 only gets sent to the Right output bus).
6 ) Invert polarity on channel 3
^ Note: When setup correctly, the figure-8 'Side channel' microphone should face Left.  A positive voltage is produced for a wavefront arriving from Left of center, a negative voltage is produced for a wavefront arriving  from right of center.
7 ) Bring up the fader for channel 1.  Both Left/Right output buses will contain monophonic Center channel in equal amounts.
8 ) Bring up the fader for channels 2 and 3.  The output in the Left output bus will be the sum of channels 1 and 2 (L = M+S).  The output in the Right output bus will be the sum of channels 1 and 3 (R = M-S).
9 ) Adjust the relative balance of channel 1 verses channels 2 & 3 (use 2 & 3 in equal amounts) to change the ratio of Mid to Side.  Without using any of channels 2 & 3 you get 100% Mid (mono center).  Without using any channel 1 you get 100% Side signal (L output = S, R output = -S).
^
10) You don't have to use equal amounts channels 2 and 3.  You can adjust them separately to help balance the resulting image.  Note that doing this is not the same as adjusting the stereo balance control of the resulting L/R stereo output.  If the image needs Left/Right balancing, try both ways and listen closely to what they do differently.  You may well end up using a combination of different levels of channels 2 and 3 AND a stereo balance adjustment of the Left/Right output.

Left  = Mid + Side (Sum + Difference)
Right = Mid - Side (Sum - Difference)

To answer your specific questions:
1) Are Side and Side 2 the same thing?
"Side 1 and Side 2" as you describe them are polarity inverted but otherwise identical to each other.  The polarity inversion is the critical thing that makes them different from each other.  But that difference really only manifests as something significant upon being mixed with another channel.
2) Do I have to convert the final mix (with two mid channels and either two or four side channels (depending on the answer to question 1 above)) back to standard L/R?2)
By mixing channels 1, 2, and 3, in this manner you are making the conversion from M/S to L/R in the output bus.  The output of the stereo mixer = Left/Right stereo.

[Gutbucket]

Continuing on from there with the mix.. add the other stereo microphone pairs by loading them into additional input channels, panning each mono input channel hard Left or hard Right.  You will then have 3 mono input channels assigned to the Mid/Side microphone pair and two mono input channels assigned to each additional microphone pair.  Adjust the stereo balance of each pair individually in isolation.

Besides checking the stereo balance of each pair in isolation, also check that the left channel is actually Left and the right channel is actually Right in each of the individual stereo pairs.  Correct that if needed.  To swap Left/Right on the regular stereo mic pairs, just hard pan each channel of the pair the opposite way.  To swap Left/Right on the Mid/Side pair, invert polarity of channel 2 instead of channel 3.  Be sure to only invert polarity on one of those channels, not both.

Once each individual pair has been balanced and confirmed as being correctly assigned Left/Right you can start to balance them against each other in the mix.  After arriving at a decent mix, go back and play around with readjusting the Mid/Side ratio. With the other stereo channels included in the mix, you may end up wanting more or less stereo width from the Mid/Side pair than you did when listening to it by itself in isolation.  Try it with no side at all, and with differing amounts of side as you work the balance between the different pairs.  One way I like checking mix balance is to mute the Side channel entirely (by muting both channels 2 and 3 in this particular case) and dial in a good balance of all remaining channels, then unmute the Side channel and adjust its level to best effect.  When doing it that way you may end up using a different amount of Side than when dialing it in initially without the other channels in the mix.  The more microphone channels you end up juggling in the mix, the more possible balance variations there end up being, so try a bunch of different things in your pursuit of the best mix.  Having so many possible balance options may seem daunting at first, but by switching back and forth you can home in on what works best.. and that often ends up being rather different than the initial mix, even when it started off sounding pretty good to begin with.

[Duffalo_Soldier]

Thank you for the extremely detailed response, Gutbucket!

I think I have slightly confused matters, I have a bunch of left/right stereo pairs to begin with. I don’t have any that are mid/side so what I am saying is that I have converted to Mid/Side in Adobe Audition (Audacity was simply what the person on the video used to demonstrate the process I outlined in my original post). I then, added the right signal to the left signal to create the Mid channel and then created two further side channels by adding the inverted right channel to the left signal and panned that left and then another side channel by adding the inverted left channel to the original right signal and panned that left.

I understand the answer to my first question that the two sides are the same but opposite polarities (because panning them both the same side resulted in silence so clearly “out of phase” with one another so on that front I am good.

Basically I had two (or three if you include “side 2”) channels for what was the first original stereo pair and panned the middle to the centre, and the other side channel(s) hard right (and hard left for “side 2”)…and then maybe the second original stereo pairs Mid to the centre or slightly off centre and the side channels 50% left and 50% right and quite like how it sounds. Possibly adding some compression to the mids and/or different EQing to the Mid and Side Channels and I like how it sounds overall. So my question for 2) is basically - can I just mix those 4 (or 6 if you include the “Side 2”’s) down to a new stereo pair or do I need to convert all the Mid/Sides back to being L/R?

[Gutbucket]

Ah, ok you are essentially going the the opposite way, making a Sum/Difference stereo-width manipulation of an existing Left/Right stereo signal.

Just mix it down, the output from the master bus that you are hearing is the L/R stereo output that has been "enhanced" by way of the Sum/Difference processing you've set up.


The easy way to do the same is to use a stereo width plugin.  Such a plugin can make the stereo width narrower, which is equivalent to simply panning both Left and Right channels closer to center.  Not much special about that, but it can also make the stereo width "wider than normal" which in essence "pans farther than hard Left/Right" by introducing some Left channel with inverted polarity to the Right output and some Right channel with inverted polarity to the Left output.  To do that simply without using a plugin or a built in stereo widener control (not sure about Audacity, but a number of DAWs have such a function built in, often as part of the stereo balance control on any stereo input channel, or maybe just the master bus), you can copy the left channel to a open mixer channel, invert polarity, and pan it fully right.  Do the opposite with the right channel.  Then bring each of those two new channels up a little bit in level and the stereo output of that mixed with the original channels will go "overwide" or what is sometimes referred to as "super-stereo".  You only be able to add so much before it gets "overly phasy sounding" and starts getting less wide again in a weird way.  If your goal is just a bit of widening, you don't need to also make an additional mono Sum channel (Mid).  Just mix those two inverted polarity / opposite panned channels with the original L/R channels. Essentially, this method just increases the amount of Side (Difference) signal in the mix.

If your goal is more extensive Sum/Difference processing, then it makes sense to convert Left/Right to Sum/Difference, do what ever you wish to do in that mode (change stereo width, or otherwise process the center imaging content differently than the content that images out toward the left and right sides - it's a very powerful way of manipulating stereo) and then convert back to Left/Right again.  Seems to me the process you describe is more akin to this.

^ Many stereo plugins (stereo width adjusters, EQ's. compressors, reverbs, whatever) can be switched run in Sum/Difference mode rather than Left/Right mode.  Some of them will refer to that as Mid/Side mode rather than Sum/Difference mode, which does somewhat better describe which part of the resulting stereo stereo image it is targeting.  Using such a mode in a plugin makes all this very easy since both the L/R to Sum/Difference conversation that precedes the primary plugin effect and the Sum/Difference conversion back to L/R again afterward is done automatically inside the plugin.