Topic: Mid-Side recording levels

[thunderbolt]

When recording mid-side, should I match the level of the side (fig. eight) to match that of the mid?  When decoding in a DAW, after copying, inverting, panning, rendering the side tracks, the levels are understandably hot.   I usually have to reduce the side level before mixing to avoid clipping.

So, what do you do?  Should I record with the same amount of gain (to the mid and side mic, with side mic level always being lower) or adjust the levels to make them equal?

[boyacrobat]

i rec them equall and play with the combinations in a daw

g

[Tom McCreadie]

It's pretty normal for the recorded S signal to display significantly weaker than that of the M because the S mic is picking up less of the strong, direct sound from the podium. Just record at nominal 1:1 gain for M:S. (Bear in mind that your mic capsule sensitivities and preamp channel gains may not be very close-matching, anyway.) That will get you in the ballpark, then tweak to taste in post.       

[thunderbolt]

thx guys.

[DSatz]

For the best signal-to-noise ratio, record both channels as you normally would with L/R stereo, i.e. let the peak levels approach, but not quite reach, full scale (0 dB).

In most cases this will require increasing the gain of the "S" channel by a few dB over that of the "M" channel, since considerably less direct sound reaches the "S"-channel microphone, and also because figure-8s are often a few dB less sensitive than comparable microphones of other patterns such as the cardioids that are commonly used for the "M" channel.

--best regards

[noahbickart]

At the great risk of differing with DSatz, whose advice should almost always be followed, I have a different method:

I decode to L/R *on the fly* at 50% Mid / 50% Side. I then try to get these L/R levels to peak at about --6db to -3db, running 24bit. I do this for for 2 reasons:

1. I can listen to the show on the way home on headphones and in the car.

2. Post-Processing takes less time, because the final mx isn't ever very far from 50/50 anyway. There is no difference mathematically in adjusting the M/S ration "inline." (p.s. you can do this for non M/S mastered shows as well)

Running M/S, and trying *both* M and S to peak at -6db to -3db equally might ultimately improve the S/N ratio significantly, which is likely why DSatz likely recommends this. Yet this is almost never a real problem at loud rock shows.

Basically, I find it harder and more time consuming to dial in the right mix in post when you begin with WAY too much side info. So I do it on the fly.

[DSatz]

noahbickart, when someone openly disagrees with me, they're taking what I say seriously, and I appreciate that. Even when I'm making wisecracks, on some deeper level I'm saying something (probably very boring) that I happen to think is true.

You're describing a basically different approach to M/S, is the thing, and as I'm sure you realize, and ultimately our difference of opinion is about that difference in approach. You're setting up an M and an S microphone, matrixing their signals to L/R stereo after the preamp, and recording the L/R stereo result. I'm setting up (as I believe the original poster is also doing) an M and an S microphone, leaving their signals as M and S, preamplifying and recording that--and then going home and processing the M and S signals to L/R stereo later on, while monitoring through loudspeakers. So I end up making two recordings, and if the second one doesn't pan out (no pun unintended), I can always go back to the first one (the M/S original) and do the "post-processing" work over.

My real problem is that some 40 years into this game, I've still never learned the knack of balancing stereo image width vs. reverberation balance while listening to headphones. I'm not sure that I could ever learn to do it. After getting it wrong a few times early on, I gave up trying. By recording the M and S signals in "raw" form I can set the ratio of M to S at the matrix inputs as well as the other pre-matrix processing I typically do, such as bass-boosting the M-channel signal [ETA: whoops--as Tom McCreadie points out below, I meant to write "bass-boosting the S-channel signal"], all while monitoring the result over loudspeakers. I don't have to commit to any particular set of settings until the result feels right and sounds right.

But it certainly is nice to go home and feel that your recording is basically finished, and all that's left to do is some trimming and putting in the fades and duplicating the CDs and/or mp3 files for the clients. If there's a pill that I could take, I'd consider taking it.

And one thing I've found--I think we probably agree implicitly about this--is that there usually is only one M:S ratio setting (or at most a very narrow range of such settings) that sounds good for any given loudspeaker setup--though it differs markedly if you move the loudspeakers farther apart or put them closer together.

--best regards

[noahbickart]

But DSatz, what, aside for some minor s/n improvement do you get my doing it your way?

Given how easy it is to convert from M/S to L/R and back in a DAW, why not just decode on the fly, and convert back to M/S in Post?

I too, following your advice, tend to raise the bass response of the S channel in post. I also tend to give a HF bump to compensate for a windscreen when I use one. But most 2 channel EQ plugins (not to mention compressors, etc) these days can operate in MS mode so you don't need to convert back to do that.

I agree with you that there is usually a small range of acceptable M/S ratio for loudspeakers. I also find that I want a little more M in a headphone mix (I usually create a dedicated headphone mix with more M, less dynamic range, greater loudness, etc for listening on my phone in the subway.)

But both of these mixes are relatively close to the 50/50 starting point.

[Tom McCreadie]

Couple of nitpicking comments to DSatz:

If the L/R didn't turn out appropriate first time qua width, there's no need to "go back to the original M/S" One can do the width readjustment directly on the L/R. Many daws have that convenient functionality. (Under the hood, of course, that's actually doing an L/R > M/S > M'/S'> L'/R' transform.)

Boosting the bass of the M signal. Typo? Did you not mean bass boost of S?. Sure, a cardioid M might need a tad more bass, but the Fig-8 S is even wimpier. [And let's leave "shuffling" dogs lie for now  :-)]

[DSatz]

> Given how easy it is to convert from M/S to L/R and back in a DAW, why not just decode on the fly, and convert back to M/S in Post?

You're offering perfectly valid arguments in favor of an approach that I neither use nor want to use, nor does the original poster in this thread appear to want to use. But I don't want to try convincing you that my way is better; I truly don't know that it would be any better for you.

Some of the main differences between our situations, that may help account for our different preferred approaches, include:

• You seem to feel comfortable using headphones for deciding about aspects of a recording that I don't trust headphones (or myself using headphones) to decide. Try to imagine starting from that premise if you can; it will explain a lot.
• When I record M/S--which actually isn't all that often--it's because I specifically want to take the recording home and set the stereo image width vs. reverberation balance there. Usually that's because it's my first time in a new recording situation (a hall I've never recorded in before, or a type of ensemble that I haven't worked with in that setting), and for whatever reason I can't record a rehearsal beforehand. So I have no way of knowing in advance which M-to-S ratio is going to sound best over speakers.
• I have no immediate use for a recording of a rough L/R mix. I have a portable headphone amp with a matrix built in, which I use for monitoring during the recording, just for the assurance that I'm getting something that makes some kind of sense in L/R stereo. (Again please keep in mind that that's all I would ever expect to get when using headphones.)
• Often the M/S tracks are the main microphone pair in a recording where there are also spot or support mikes, so a post-production mix is a given from the start.
• I don't have a "DAW"--I have a homebrew PC that's showing its age, plus conventional audio editing software (Sound Forge, Adobe Audition). Converting from M/S to L/R or the reverse is relatively time-consuming. So the most efficient approach by far is to start directly from M and S signals rather than to derive them from already matrixed L/R tracks.

--best regards

P.S. Tom: You're right, I meant to say bass boost for S. Damn--I've made exactly that same mistake before.

[mepaca]

Hello DSatz-
   For me, your knowledge and insight is the most valuable asset of this forum (taperssection that is).
What headphone amp with mid/side decoding  do you use? For those of us who decode in post, my favorite
plugin in is the free Voxengo MSED. It can be used to decode or can be used inline to encode and decode a
stereo channel.

[DigiGal]

One time-honored technique is to only monitor the [mid] mic on location during acquisition by using a special headphone adapter cable or headphone mix that puts only the [mid] channel to both ears.

[noahbickart]

Hello DSatz-
   For me, your knowledge and insight is the most valuable asset of this forum (taperssection that is).

Without a doubt. +T.

[Tom McCreadie]

DSatz, you prefer "setting the image width vs. reverberation balance" in post, but we should keep in mind that those two are not _independent_ variables. We can fully control the image width but then have to live with a concomitant impact on the direct/reverb balance (assuming no artificial reverb added.). For full control of imaging _plus_  a pleasing reverb balance, a critical variable is the correct main mic placement, e.g. mic stand nearer to or further from the podium. And if that mic placement is poor, no amount of subsequent width twiddling can redress the situation. So recording direct to MS is OK (my usual workflow too), but a quick 'n dirty monitoring L/R live, even through headphones, does make it easier to nail that main array mic placement. Failing that, one should at least try to monitor the M feed in mono.

 My "DAW" is also a pc with Audition. Its inbuilt stereo width control function, together with Voxengo's MSED plugin. allows hassle-free working
Edit: just noticed Digigal had mentioned mono monitoring       

[DSatz]

People, thank you very much for the kind words, and the same right back at you. This is a combined reply to several postings.

- mepaca, the little AEA "MS38 Lite" portable headphone amp (http://www.ribbonmics.com/aea/ds_MS38_Lite.html), with a switchable, variable M/S matrix built in, was unfortunately discontinued about ten years ago. It runs on a pair of 9V batteries, is sturdy and quiet, and has enough power. I've never seen one come up for sale used, and Wes apparently doesn't have any way to get more of them made--I've asked.

- Tom McC., I agree with you completely. In conventional M/S recording, the reverberation balance and the stereo image width are determined in tandem by the M-to-S ratio going into the matrix--there's only one "knob" (whether it's in hardware or software) to turn for both at once. As a result, you don't have the freedom to set each parameter to your liking--you always have to pick a compromise. And the two parameters work in opposite directions, narrowing your range of usable settings to the point where usually, there's hardly any choice, as I mentioned a few messages earlier. For example you can widen the stereo image in playback by increasing the amount of "S" that you're using--but that also decreases the proportion of direct sound in the result, just when you'd probably want to increase it, and vice versa.

The "Double M/S" approach, a three-mike system which was discussed in several threads on this forum a few years back, offers a way out of that bind, as does the four-capsule Ambisonic system. Both are also useful ways of making surround recordings. But they're still coincident miking techniques, and whenever I have a choice I usually prefer stereo miking with some space between the capsules; the localization may not be as precise but the listening experience seems more pleasurable to most people, myself included.

--best regards