Topic: Mid/side decoding on the fly?

[SmokinJoe]

I'm talking about setting up mid/side mics, and then running through a mid/side to XY converter where you record XY (or Left/Right) to the recorder.  It's seems like a good idea, and a cool feature to have in a pre, and maybe there really are a few people who are good at it, but I think it's EXTREMELY hard to get right, and I don't think I can ever get there.  I accept the fact I'll never be supermodel or play guitar like Warren Haynes either.  Who agrees with me? (not just about the supermodel part)  Who thinks they have it nailed?  I think the majority of us shouldn't even try... if you want to do mid/side, record raw mid/side and mix it at home.

I can see on-the-fly decoding would be necessary for video recording folks if they have a mid/side shotgun and want to record L/R audio without post processing the audio.  How well they dial in the "width" is much less important to them... most people who watch the a TV show pay almost no attention to the audio quality.  Tapers are very fussy about dialing in the width.

I've done dozens of raw mid/side recordings which I decoded at home (LSD2, 414's, various ADKs).  I think it's a right of passage that every taper worth his salt should go through, because you'll force yourself into an education in the process. You will beat your head against the wall on the first couple, but then you'll figure out a workflow and it's easy.  If you give up, you dropped out of school on that forced education part.

One day I bought a Mini-MP which had the decoding feature, so I tried it.  The first time trying to use that feature I messed it all up, somehow got something backwards, and I couldn't repair it in post.  I used it a couple more times with the right connections, but I didn't dial in the width correctly.  In that case I did an XY -> Mid/side conversion, then picked it up like it was a raw recording so I could dial in the width.   I thought "maybe I just need to do this enough times until I get the hang of it", but I don't think it's that easy, and here's why...

How many of us can walk into a room, set up and guess where to set the gain knobs perfectly before the music starts, and then not have to touch them after that?  I'm talking regular L/R stuff. I can do it maybe 50% of the time, the other 50% of the time there is tweaking, which of course I have to level out at home.  To get a good "mid side on the fly" recording I think you need to have that skill nailed perfectly, and THEN nail the skill of guessing the width on top of that.  Because if you adjust 1 knob, you better adjust the other one exactly the right amount too, or you just shifted the width of the recording.  Most of us know how to "fix" a recording that has a gain change, or one channel is a little hotter than the other.  To repair a width shift you have to encode XY > M/S, then find the spot where the width went bad, and adjust the side channel by 1.2db (or whatever it is) then decode back to XY.  I can see that taking a few iterations.

If anyone read my recent post under AT4050ST thread, you might think I'm contradicting myself, but I don't think I am.  That's a mid/side mic with an XY decoder in the mic.  I can set that to "simulated 127 degrees" and record Left/Right to the recorder.  If I don't get the left/right gain just right I can tweak that until the left/right are even just like anyone else does with XY mics.  THEN I can encode it back to M/S and play with the width.  To me this is a big difference... with a preamp it's dialing in "how much of the mid mic to do you want, and how much of the side mic do you want?" and the meters are not helpful.  On the AT4050ST mic, the width is fixed,  I just try to bring both gains up evenly but I have meters to guide me.

I think the only way to do it is:
a) start with preamp with mid/side decoding,
b) obviously get your mid/side left/right front/back stuff straight
c) set the preamp conservatively, and don't touch those gain knobs, so the width at least stays fixed.  At that point you have the stereo mic with fixed width.
d) then you make gain adjustments with the recorder, not the preamp.
e) it would be extra handy to have a recorder with a Gain/balance mode instead of just Left/Right.  I think the SD-7xx family has that.  Any others?

Probably someone should tell me "Dude, you are waaaaaay over thinking this.  Take a chill pill and relax".    But that's the goal of this hobby, isn't it?  To perfect the art of getting good recordings.

[Gutbucket]

Sounds rational to me.

A few thoughts-

As you say, adjusting gain while recording is trickier with M/S than L/R.  But it's no big deal if your preamp or recorder can adjust both gains simultaneously so the ratio between channels doesn't change (ie: the Areco preamp with a level & balance control instead of two gain knobs; small recorders with only one gain control which works on both channels; the DR-680 with 'gangable' gain controls for adjusting multiple channels simultaneously irregardless of their relative settings; and I think the SD recorders as you mention, I'm not as familiar with them).  So if you can either set your gain ratios to what you know is good and not touch them (hard to do, though I've gotten better at this over time, also partly made easier if you can be OK with not messing with levels that are only peaking around -18 or -20) or if you can adjust them simultaneously as mentioned above, then things are simpler and easier to manage.

If you are trying to avoid post work and keep things simple (like for the video guys you mention) and decide to record pre-decoded L/R for that reason, but want to adjust width to satisfy your more discerning taper ear, then you can re-adjust width within limits afterwards with another L/R>M/S>L/R stage or with a stereo width plugin or tool that basically does that for you.  As mentioned, you can only do so within reasonable limits, and it would be better to record the raw M/S if you know you will want to fine tune it later, but the ability to do so might be a way of satisfying the need for both simple immediate use (the video guy) and the more discerning listener (taper).

Over thinking it? naw, we don't get to really talk about this stuff in the 'real world', it's what TS is for!

Heading home now to M/S encode the L/R Blumlein stagelip recording I made Wednesday night so I can try using just the resulting forward facing figure-8 Mid channel as a center along with wide spaced omnis for 3 channel L/C/R playback.  Over thinking it!

[George2]

Yes, I've recorded in M/S, and prefer doing it at the time of recording.
I like the new SD MixPre D, as when set to  MS, one fader becomes gain for both channels, and the other fader controls the width.
No trim pots to fiddle with, so it's very fast and repeatable.
1. Position Mid mic.. leave Side mike off, and get a good mono signal.
2. Turn up Side. Typically use less Side  then Mid.
3. I like to record it just a touch too wide in the headphones. When playing back in DAW you can pan each side in to narrow it a bit, and make a new file... no need to go about "recoding' everything, or using plugins.
4. Again, use more mid than side. 60/40.... never 50/50.
5. Use High Pass Filter on the Side Mike.

[Gutbucket]

3. I like to record it just a touch too wide in the headphones. When playing back in DAW you can pan each side in to narrow it a bit, and make a new file... no need to go about "recoding' everything, or using plugins.

Definitely easy to pan both towards center some if necessary, just recognize that doing that is not the same as changing width by adusting the M/S ratio.

5. Use High Pass Filter on the Side Mike.

Why? To reduce handling or wind noise?  Doing so will also make the bass below the filter's corner frequency pregressively more mono, which is fine if that's what you want.  However, one post-processing M/S technique is to essentially do just the opposite- boost the low frequencies of the Side signal.  A few reasons why someone may want to do that- it can compensate for the reduced low frequency sensitivity common with most figure-8s, it can increases bass 'envelopment and openness' and can helps align the apparent direction of the highs with the lows across the playback image between speakers. 

Whatever sounds best..

[DigiGal]

M+S = L   
M+(-S) = R         

so L+R=(M+S)+(M-S) = 2M (mono sum)

[Gutbucket]

Definitely easy to pan both towards center some if necessary, just recognize that doing that is not the same as changing width by adusting the M/S ratio.

If you pan decoded LR to center on each channel (L and R), then you get a signal that is L+R, which is M.  The main difference is you are limited in *increasing* stereo spread as you can't directly decrease M beyond what it was recorded at (you would have to convert back to MS), but you can definitely make it narrower in the same manner as reducing S.

Good point, I stand corrected.  Possible to narrow only but not widen. 

In digiland you can swap back and forth between MS and LR as often as you like, practically without loss.  In analog that's not true, each required amplifier stage will accumulate noise and distortion.  So I would not record decoded LR unless absolutely necessary for monitoring; indeed, I would probably crank up the S to maximize its SNR.

You can make the switch back and forth in digiland without loss (and analog noise and distortion), but if you change ratio drastically while doing so you'll get loss-  take the extreme case: If I adjust the ratio fully one way or the other (all Mid or all Side) there is no way of getting it back in the next stage.

[Gutbucket]

True, in that situation any practical amount of adjustment back and forth would cause no degredation.

I'm all for tweaking things to improve the balance for those who listen closely enough to care.

[noahbickart]

I tend to record a decoded M/S signal for two following reasons:

1. Decoding to and from M/S is really easy today with AU and VST plugins.

2. I like to listen to the show on the drive/subway/walk home.

I understand that if you record to maximize both M and S, the subsequent likely reduction in S can improve your s>n ratio, but I always feel like the post processing is so much shorter when you lay down a normal 50/50 mix and only have to tweek it a few clicks one way or the other.

[NOLAfishwater]

the Tascam mixer screen allows you to easily dial in the mid/side. I personally love doing it onstage with a subcard as mid

[grrrayson]

5. Use High Pass Filter on the Side Mike.

Why? To reduce handling or wind noise?  Doing so will also make the bass below the filter's corner frequency pregressively more mono, which is fine if that's what you want.  However, one post-processing M/S technique is to essentially do just the opposite- boost the low frequencies of the Side signal.  A few reasons why someone may want to do that- it can compensate for the reduced low frequency sensitivity common with most figure-8s, it can increases bass 'envelopment and openness' and can helps align the apparent direction of the highs with the lows across the playback image between speakers. 

Whatever sounds best..

Indeed: whatever sounds best is the ultimate rule. "If it sounds good, it is good."

However, realize that bass frequencies are more mono by nature even acoustically, as demonstrated by the fact that many stereo systems don't lose a sense of stereo width by having one mono sub. Therefore, rolling low end off of the side signal can help you clear up muck where you don't need low end and can't even notice the directionality of it, help the low end feel punchier by being more centered, and it can actually sometimes make things sound wider (odd but true).

Consider that the Schoeps mk8 has a dramatically rolled-off low end but still sounds very good and like it matches with other capsules from cardioid to omni (all of which have a bit more low end) as the mid in an M/S array.

M+S = L   
M+(-S) = R         

so L+R=(M+S)+(M-S) = 2M (mono sum)

I think that's the only case in which I've used algebra since 8th grade.

the Tascam mixer screen allows you to easily dial in the mid/side. I personally love doing it onstage with a subcard as mid

Yes! I am in love with the Schoeps mk21/mk8 combo. I had some hard times recently and sold all my Schoeps stuff including the bodies but kept those mk21 and mk8 capsules.


To the original question, if in doubt and they are matching mics and you have some sort of switched/repeatable gain settings, just record the M and S at equal gain, decode on the way in, and it'll sound pretty good. If your gain knobs aren't so precise or you're using mics that don't match and you can't get a good listen at the show, leave it for later.

However, I virtually always give the side channel an extra squee...about half a decibel more. I've gone through this dozens of times in various situations, decoding on the way in, decoding after, adjusting it with my eyes closed, listening on speakers or headphones or whatever, and again I virtually always like the side channel goosed just a little. I record acoustic stuff, which may be different from electric shows, but I doubt it, as I think it has more to do with the physics of how the mics work. I think part of it is that the figure 8 patterns of most mic systems tend to be slightly less sensitive than the other patterns and another main part of it is that I personally just like the ear candy of having the S channel cranked up just a little.

Since I've always come back to this same conclusion, with matching mics I decode on the way in with the side channel bumped up .3 decibels or so and move on. It's still easy to tweak in post if necessary, as in the digital realm you can convert between L/R and M/S an infinite number of times without changing the signal at all.

If you're not sure where to start with gain settings while decoding on the way in, set the M to where it looks/sounds good first since that's where most of the signal will live and then just bring the S up to that. Also, if you bring up only the M channel and you hear a weird out-of-phase mess, you'll know you inadvertently switched a connection somewhere. (I've been there/done that and not caught it until after, which does suck...)

I really love M/S for many reasons.

[Gutbucket]

However, realize that bass frequencies are more mono by nature even acoustically, as demonstrated by the fact that many stereo systems don't lose a sense of stereo width by having one mono sub. Therefore, rolling low end off of the side signal can help you clear up muck where you don't need low end and can't even notice the directionality of it, help the low end feel punchier by being more centered, and it can actually sometimes make things sound wider (odd but true).

I think you're over simplifying things.

It's true that our directional perception diminishes at low frequencies and below about 40Hz a single sub can work equivalently on a perceptual basis (though on a practical basis there may be other advantages to using more than one mono sub in that range).  Our directional perception of low bass above ~40Hz is not very good either, yet the feeling of 'movement and space’ can be distinguished and the difference between mono and stereo subs up to 80 or 100Hz can be quite noticeable.  But most domestic systems with single subs crossed over to the mains at higher frequencies are mostly about practical and cost compromises of accommodating smaller speakers.  Take a look at how most mastering engineers set up their systems.  They know what’s going on, their livelihood depends on it, and most who use subs above 40Hz run them in stereo. 

Now the corner frequency where a figure 8 microphone’s response typically starts to roll off and the range typical of a low end Side boost which we’re talking about here concerning of the is generally higher still.  Those effect should be audible even with a single mono sub crossed over at an unquestionably compromised high frequency of say 120Hz (compromised in terms of the limits of human hearing perception.. in engineering terms such a cross over frequency may be an appropriate compromise for that particular system).  I agree that mono-ing bass below about 40Hz is of no consequence and often advantageous, it’s the region above that where I begin to differ.

How much does all this matter? Well that's a personal decision, but the acoustics of it is demonstrable. Besides the being familiar with the literature on all this, I’ve heard the difference between stereo subs and mono subs crossed at 80Hz in the same high quality system.  I have full range speakers with response down to about 35Hz and no subs and I hear these low frequency effects when manipulating Mid/Side EQ differences in my recordings.

What I’m trying to point out is that the prevalence of stereo system with mono subs is not a particularly good argument against a low boost in the Side signal.

Consider that the Schoeps mk8 has a dramatically rolled-off low end but still sounds very good and like it matches with other capsules from cardioid to omni (all of which have a bit more low end) as the mid in an M/S array.

Sure it sounds good.  I'm certainly not arguing otherwise, only that there is a chance it might sound even better.  No way to know until you try it.. and the appropriateness of doing so changes with each situation.  The few times I’ve played around with EQing the Side and Mid signals differently I usually prefered narrowing the highs and widening the bass somewhat- everything seemed to line-up and solidify nicely.  With acoustic material in a great sounding room, doing something like bumping up the low end of the Side channel is more likely to be advantageous than for PA amplified stuff that often has more than enough low bass to begin with, often of sloppy quality, or with acoustic recordings with lots of low HVAC, road rumble or other gunk down there.  But on the good ones it can make them even better.

Keep in mind that when EQing the Mid and Side signals differently, any adjustment will also affect the overall timbre balance of the recording unless you introduce a corresponding inverse adjustment in the other.  So to be fair in the comparison, cut the Mid by as much as you boost the Side, or vice versa, if you only want to affect the apparent width at that frequency range and not the overall tone.

Agreed with you on all the rest,
Cheers!

[H₂O]

Just get a recorder with a built in mixer (ie hs-p82, 788, possibly 680)

If it supports ms decoding you should be able to record 2 tracks raw and the mix tracks decoded.

This way you get both versions

[Charlie Miller]

M+S = L   
M+(-S) = R         

so L+R=(M+S)+(M-S) = 2M (mono sum)

Is there gonna be a test on this?

[barrettphisher]

Use your v3!  Crack it open and use the On board decoder. 

[DigiGal]

M+S = L   
M+(-S) = R         

so L+R=(M+S)+(M-S) = 2M (mono sum)

Is there gonna be a test on this?

Be sure to bring your #2 pencils