I picked up Meldaproduction's MMultiAnalyzer yesterday based on bitflipper's heads up deal and was specifically interested in the collisions part of this plugin. It didn't take too long into this that I found some conceptual errors with getting mid-side to act as I expected, so I  delved back into his post from over a year ago, and still not getting proper results. One part that may be biting me is that this is a sandbox project, so I am not sure about the history of bounces that occurred between stereo/mono over time as I was playing with things (and messing around yesterday may only have exacerbated this).
 
Background is simple. Two tracks (guitar and vocal) recorded in mono which have significant collisions in 100-1K Hz range. The only M/S EQ I am using is from Ozone 5. The goal is to notch the mid from the guitar and moved it to the sides, but playing with various things (bouncing to stereo, interleave back and forth from stereo to mono, and M/S EQ) MMultiAnalyzer actually showed more vocal on the sides.
 
Rather than get into what I am doing wrong, it is probably better to start from scratch to learn the technique in SONAR correctly. Given a situation with two mono tracks with collisions, what is the best method to leave one centrally panned and push the other to the sides using M/S EQ?
 
Edit - I just found a portion of my confusion. Using just the vocal track and analyzing with MMultiAnalyzer, I found that whether the track is mono or stereo is immaterial to what MMultiAnalyzer is seeing. However, the Interleave button is rather odd (below are the components with "content" when using MMultiAnalyzer in SONAR Platinum):
 
Interleave?  Left  Right   Mid    Side    VOX track
Stereo        Y      Y      Y      N
Mono          Y      N      Y      Y
 
I definitely "do not get" why mono interleave has no right-side component.
 
Edit2 - To make matters worse, I just did the same test on the guitar track with different results. Again, being mono versus stereo was irrelevant, but in this situation the interleave was irrelevant too:
 
Interleave?  Left  Right   Mid    Side    Guitar track
Stereo        Y      Y      Y      Y
Mono          Y      Y      Y      Y
 
In both cases, no effects (other than MMultiAnalyzer) in the track, and PC was off  Why are these different? Channel Tools inserted says the guitar track has no right-side component same as the VOX track, yet MMultiAnalyzer is seeing them differently.

Those interleave tables look like a nonsense to me.  For L/R or M/S encoding, two separate signals are required.  A single signal is mono and cannot be L/R or M/S encoded (if you like, a mono signal is half of an L/R or M/s encoding).
 
This would be my table:
 
Interleave   Left  Right   Mid    Side
Stereo        Y      Y      Y      Y
Mono          N/A    N/A    N/A    N/A

 
I wouldn't use M/S processing to address 'collisions', instead I'd notch out some frequencies from one or both of the signals to give both some space in the frequency spectrum.  It might be that the arrangement needs to be changed, or the registers which the colliding signals are using could be moved up or down.
 
I'd use M/S processing to shift bands of spectral energy either out towards the sides, or in towards the middle.  For example, when cutting a vinyl master, heavy low frequency cuts (HP filtering) were often applied to the side channel, to prevent large low-frequency differences in the L and R channels from causing the needle to jump out of the groove.  Another use of M/S processing is to narrow, or widen the stereo image.
 
> what is the best method to leave one centrally panned and push the other to the sides using M/S EQ?
 
Did you mean "leave centrally panned" or "moved towards the centre" for the first bit of that question?  I'm guessing 'moved towards the centre', as 'leave centrally panned' involves simply doing nothing.
 
I'd set up an M/S encode/decode matrix for each part (i.e. separate matrices).  Cut the side channel for the part which needs to be moved centrally, cut the mid channel for the part you want pushed to the sides.  Cut completely to create 'mid-only' or 'side-only' parts.
 
HTH (?)

Thanks Ston. The bigger issue I was "seeing" was that I was messing around with the MMultiAnalyzer just looking at different settings etc. I think that I actually hit on a flaw in the program, since the "collisions" it shows are accurate, but I am not clear as to what the program is "interpreting" as L, R, M, S (the error may indeed lie with me or my interpretation of what these "really are").
 
It occurred to me on the way home today that I have a much more accurate graphical representation of this data available (at least for the M/S visual) - R-Mix. I simply added that onto the Master buss so I could visually see things and everything was operating "as expected," so where MMultiAnalyzer is great for collisions requiring notching, it doesn't appear overly accurate on channel info (then again, I have not RTFM yet either).
 
Back to the OP situation, I agree that notching is the #1 method. For this situation, I had been notching the guitar slightly, and giving them more width than the lead vocal. Vocal was mono (centered), and the guitar was converted to stereo and given more width. With R-Mix this looks exactly as I was intending.
 
I honestly have not played with M/S EQ at all, so need to fiddle with this specifically to see its effects on things. I think I still am carrying a conceptual error of how to best employ it, which may be best to learn through trial and error here.

ston's right in that M/S analysis and EQ won't help much for a project consisting of just two mono tracks. Mono tracks don't have Mid and Side components any more than they have Left and Right components. Once you combine those tracks on a bus and fiddle with their panning, only then do you start to get L/R and M/S data to play with. Even then, M/S EQ won't start being much help until you've added some more tracks. M/S EQ is really meant for busses and stereo tracks.
 
Setting the track interleave to stereo for mono clips (or vice versa) compounds the confusion, but what happens is pretty simple. When you set the track interleave to stereo but with mono data, SONAR merely creates two identical copies of that mono data and sends them hard left and right. What you hear sounds exactly the same, because left and right are identical.
 
Sometimes, plugins can get confused and act strangely when the interleave is set to something they don't expect. For example, some plugins (and this might include Ozone, I haven't checked) expect a stereo input and do funny things when you send them a mono signal with a mono interleave.
 
MMultiAnalyzer has no problem dealing with a mixture of stereo and mono tracks. It sees a mono input and treats it as such, and then outputs mono to the next plugin in line. When those analyses are then routed to another instance of MMultiAnalyzer (one that's showing multiple tracks) it's sent in stereo so that the pan positions of any mono tracks is respected in the multi-track spectrum.
 
Reading this back, I fear I've only muddied the waters more. I'd suggest pulling up a full mix with lots of tracks and throwing the analyzer into that richer pool of panoramic frequencies to get a better picture. 

Michael, have you seen Dan Worral's excellent Pro-Q tutorials?  Even though it's showing off Pro-Q, the techniques apply to any M/S EQ. (Warning: it may make you really, really want Pro-Q after you see these!)
 
Here's a link: http://www.fabfilter.com/video/fabfilter-pro-q-advanced-eq-techniques
 
[EDIT] Sorry, that's not the video I had in mind. This is the one I was thinking of.
 
 

Hi Dave, thanks for that info. I have seen Dan Worral's video several times which is what actually brought the idea to mind (same exact concept he was doing with guitars there). I think I am not grasping the M/S itself. Conceptually, M/S is another "view" of L/R, and seems it would be possible to lower the center component (to zero potentially).
 
The setup I am playing with is simple, mono guitar track converted to stereo, stereo interleave (so this is a "duplicate mono" track sent into Ozone). Track FX are Ozone 5 Imager, followed by Ozone 5 EQ. R-Mix on the Master just to look at results. The Imager alone seems to do what I am intending (for the most part) as it is widening the guitar from center. When playing with M/S EQ for "normal values" it doesn't seem to alter M/S a great deal. I then started to play with more extreme values for boosts/cuts and graphically (R-Mix) a boost/cut to either M/S in an extreme setting "pinches" that EQ band back to center, creating a "hourglass" look to the stereo field.
 
This is the effect I am not grasping now. If I extremely cut center, it would seem that the center would create a "hole," but from the "stereo perspective" this make more sense... since this is simply a mono track given width, as the EQ is being used extremely, it is balancing out the L/R components which is basically overriding the Imager and driving the audio back to a balanced L/R situation (i.e. "mono" on dead center). Is that a more accurate assessment of what is happening?
 
Being an engineer, I tend to see things in a "here-to-there" scenario, so the extreme values give me the perception that a 3dB cut in mid, with a corresponding 6dB boost on sides is simply a baby step to the "hourglass effect." Now that R-Mix has come to mind, I was chuckling to myself and thought "if I want a whole in the center, I could simply widen the track and punch a hole in it with R-Mix (literally), but thought that is what an extreme mid-cut would do (hence the conceptual error I am carrying around).
 
 

Ozone's imager is applying short delays to create artificial L/R differences, which you can look at as increasing the Side component, since the S in M/S is by definition the differences between left and right channels.
 
When you use M/S tricks (changing levels, applying EQ, inserting delays, or adding effects) you're exaggerating whatever differences exist between left and right. The greater the starting difference, the more you can do with M/S processing. What M/S processing can't do is manufacture width.
 
For example, you can insert a simple M/S balance plugin such as Voxengo MSED into the master bus and lower the Mid by 3dB while raising the Side by 3dB. IF there are already significant differences between the left and right channels, this small change can have a huge effect on width perception. By the same token, if the mix isn't wide to begin with, this trick will do little or nothing.
 
If you find that M/S techniques aren't yielding much payoff unless you resort to extreme settings, that means your mix isn't wide enough to begin with. That's when you go back into the mix with the goal of making things wider, e.g. double-tracking and hard-panning rhythm guitars, being more aggressive with panning in general (e.g. LCR), using ping-pong delays and reverbs, applying chorus, adding wide synth pads or strings. Once you've got some organic width in your mix (meaning greater left/right contrast) via conventional mixing techniques, then you can really start to have some fun with M/S EQ to exaggerate those differences.

Thanks Dave, that explanation makes much more sense, especially the part of the side component being the differences. It definitely makes sense with what I was doing since this particular track is truly mono in essence. The Imager does create artificial width, but the amount is fairly minimal, which I think is why EQ caused it to "collapse" toward center.
 
This particular project only has mono tracks thus far, so was a bad choice to look at, but I will definitely revisit this on projects which have stereo material when I get time. Thanks again for that explanation.

Actually, a project consisting entirely of mono tracks can benefit just as much from M/S processing as any other, if not more.
 
One of the most common mistakes made nowadays is thinking that stereo tracks result in a more stereo-sounding mix. Rookie producers will load up a project with rich stereo synth tracks and are puzzled when it all ends up sounding like fuzzy mono rather than the nice wide sound they were expecting. In that case, M/S EQ might be indicated as a remedial measure, but it won't save the mix.