Taperssection.com

Gear / Technical Help => Battery Boxes, Preamps, Mixers, ADCs, and Processors => Topic started by: yug du nord on November 12, 2007, 12:04:32 PM

Title: M/S ???
Post by: yug du nord on November 12, 2007, 12:04:32 PM
Is there any advantage/disadvantage in recording M/S on the fly with a preamp capable of a M/S config...  or recording M/S straight to a recorder and mix the M/S in post?  Thanks!!!
Title: Re: M/S ???
Post by: Tim on November 12, 2007, 12:12:02 PM
On the fly is a bit of a crap shoot because it can be so difficult to accurately monitor when you're at a show.
Title: Re: M/S ???
Post by: yug du nord on November 12, 2007, 12:24:37 PM
Any difference in the sonic quality between "on the fly" M/S and "post" M/S?  I know with "post" M/S, you can vary the width of the stereo image, is that possible with an "on the fly" M/S rig?
Title: Re: M/S ???
Post by: Patrick on November 12, 2007, 12:29:38 PM
The pros/cons are pretty obvious; decoding on the fly saves lots of post production time but also does not allow you to adjust the ratio of mid to side to your liking.  Some people that are familiar with their m/s setup know what sounds good in certain situations and pull great tapes decoding on the fly.  I am kind of new to running m/s so I prefer to keep the file raw until I can get home and see what sounds best to my ears.

Any difference in the sonic quality between "on the fly" M/S and "post" M/S?  I know with "post" M/S, you can vary the width of the stereo image, is that possible with an "on the fly" M/S rig?

No, they're not different in any way.  On the fly decoding does the same thing as the plugins, but simply does it "in the box."

Also, I was wondering how people decoded their m/s channels.  I actually prefer to do it manually; inverting a copy of the "side" channel, panning it hard L/R and blending it with the "mid."  I guess this is because I don't have any plugins on my computer yet, but the manual way works fine for me.

Title: Re: M/S ???
Post by: KLowe on November 12, 2007, 12:36:33 PM
SF8 has the best and easiest to understand M/S encoder ever!!!..... IMO
Title: Re: M/S ???
Post by: grawk on November 12, 2007, 01:18:18 PM
On the fly can be split back out and then remixed without any problems.  There's not a drawback to mixing on the fly. 
Title: Re: M/S ???
Post by: Roving Sign on November 12, 2007, 01:19:57 PM
On the fly can be split back out and then remixed without any problems.  There's not a drawback to mixing on the fly. 

How?
Title: Re: M/S ???
Post by: Roving Sign on November 12, 2007, 01:24:18 PM
I could see an analog MS decoder that worked on line level signals...then you could record raw MS digitally and then pass the analog signals back through the decoder. That would keep the MS-ing in the analog domain...drawback - ads an A>D>A>D step...
Title: Re: M/S ???
Post by: anhisr on November 12, 2007, 01:41:26 PM
I don't understand why people want to do it on the fly.  Even if you know the venue, there are many variables that makes a difference every night.  Doing it on the fly takes width option out of your hands.  The only advantage is that you have the recording that night.  I use to use SF8 but find Wavelab 6 much easier to use. 
Title: Re: M/S ???
Post by: Brian Skalinder on November 12, 2007, 07:06:13 PM
On the fly can be split back out and then remixed without any problems.  There's not a drawback to mixing on the fly.

Unless one considers double the processing a drawback.  If you don't nail the mix on-the-fly, you have to encode to M-S and then re-decode to stereo.

How?

Most M-S processors (I usually use Voxengo's MSED plugin) provide decoding from M-S to stereo and encoding stereo to M-S.  It's just a reversal of the sum/difference processing matrix.  Blurb from an article Shawn posted a while back, referenced in this TS post (http://taperssection.com/index.php/topic,68498.msg917482.html#msg917482) (the quote link below requires registration).  Description of converting L/R stereo to M/S:

Quote from: http://emusician.com/mag/emusic_front_center/index.html
What you've done is add the two left/right channels together to make a mono signal that creates the mid channel. The side channel was also created by adding the left/right channels together. When the polarity of one of the channels is reversed, however, any sound that was common to both gets canceled.
Title: Re: M/S ???
Post by: grawk on November 26, 2007, 11:29:24 PM
Yes, it's just math. 

Waves will let you change the width on the fly even on an already decoded track.  But even if you have to run it through 2 cycles, it's still not that big a deal.
Title: Re: M/S ???
Post by: DSatz on November 27, 2007, 12:41:40 AM
Any X/Y (coincident) stereo recording that has been stored digitally can be re-encoded as M/S without a conversion to analog. Similarly any M/S stereo recording that is stored digitally can be rematrixed to left/right stereo, with different matrix parameters if desired--again without any conversion to analog being necessary.

--best regards
Title: Re: M/S ???
Post by: Gutbucket on November 27, 2007, 01:39:55 PM
FYI, FM stereo and MP3 are also M/S encoded formats, the L/R signals are derived when they are decodeded by the tuner or player.

Any X/Y (coincident) stereo recording that has been stored digitally can be re-encoded as M/S without a conversion to analog. Similarly any M/S stereo recording that is stored digitally can be rematrixed to left/right stereo, with different matrix parameters if desired--again without any conversion to analog being necessary.

--best regards

This gets to something I've wondered-

If one is planning on adjusting the M/S ratio in post, one would traditionally record the M & S feeds directly without decoding in the field, then adjust the L/R output later post.

Let's assume that post work is going to be done with the original digital files in a DAW. There is no additional D>A>D conversion required to do the matrixing and all the summing is done via a plugin or DAW routings.  In that case, is there in any reason to record the raw M/S signals and NOT just record the L/R matrix output with the recorder if the preamp or recorder has that capability?  You could then at least listen to the resulting recording (without a decoding matrix) before preforming the DAW work even if the matrix ratio could use further adjustment.  That's a plus.

The question is.. Does the additional matrixing (done by an analog circuit I assume in the preamp or recorder to produce the L/R signals and done digitally in the DAW from L/R to M/S) impact the quality of the audio? 


Trying to think through potential issues other than the one above..   Do the 'on-the-fly' M/S decoders in preamps and recorders allow for independent ratio adjustment of the decoded output or is that function typically just handled by the input gain of each channel?

And.. do any of you fellows adjust your non M/S coincident recordings in post using these techniques?
Title: Re: M/S ???
Post by: Jhurlbs81 on November 27, 2007, 03:32:50 PM
Quote
In that case, is there in any reason to record the raw M/S signals and NOT just record the L/R matrix output with the recorder if the preamp or recorder has that capability? You could then at least listen to the resulting recording (without a decoding matrix) before preforming the DAW work even if the matrix ratio could use further adjustment.  That's a plus.

If you are recording the mid and side channels seperately, and then play it back without a decoding matrix you'd have all mid in one channel and all side in the other.  Not sure how listening to this would help you with you matrix ratio. Not to mention that when you encode it you will be creating a third channel.  M, S-, S+

Quote
The question is.. Does the additional matrixing (done by an analog circuit I assume in the preamp or recorder to produce the L/R signals and done digitally in the DAW from L/R to M/S) impact the quality of the audio? 
I would say no, but I'm sure someone would argue that point.  You are processing the audio, but you kinda have to if MS is your goal.
Title: Re: M/S ???
Post by: Gutbucket on November 27, 2007, 04:58:19 PM
Quote
In that case, is there in any reason to record the raw M/S signals and NOT just record the L/R matrix output with the recorder if the preamp or recorder has that capability? You could then at least listen to the resulting recording (without a decoding matrix) before preforming the DAW work even if the matrix ratio could use further adjustment.  That's a plus.

If you are recording the mid and side channels separately, and then play it back without a decoding matrix you'd have all mid in one channel and all side in the other.  Not sure how listening to this would help you with you matrix ratio...

Right, that's the drawback of recording the M/S signals instead of L/R signals  -ie. you can't listen to it until it's dematrixed. But I'm talking about micing in M/S and recording the L/R signal that is being dematrixed in the field before the recording device.  You can listen to that recording without decoding.

Quote
Not to mention that when you encode it you will be creating a third channel.  M, S-, S+

Incorrect. There is no third channel. You are confusing the S signal with how the matrix summing is done.

Quote
Quote
The question is.. Does the additional matrixing (done by an analog circuit I assume in the preamp or recorder to produce the L/R signals and done digitally in the DAW from L/R to M/S) impact the quality of the audio? 
I would say no, but I'm sure someone would argue that point.  You are processing the audio, but you kinda have to if MS is your goal.

That's really my point.  You have to dematrix a M/S recording at some point.  Is it audibly inferior to dematrix it right away before recording realizing that you'll need to re-matrix again in the DAW to make adjustments?  That way you can listen to the recording as is, or choose to adjust it in the later with the L/R>M/S>L/R technique.

If there really isn't a detectable difference, we can eliminate the 'adjust-ability after the fact' argument for M/S vs. other coincident recording techniques (like DSatz mentions above), leaving available mics, off-axis response and derived polar patterns the sole determining factors to consider when deciding to mic (that's mic, not record) in M/S.  Any coincident technique is adjustable after the fact.. but is there a sonic penalty in some cases?

That's why I ask if many are doing this kind of post adjustment with their non-M/S recordings.  They are already doing this.

Good discussion, all.
Title: Re: M/S ???
Post by: Gutbucket on November 27, 2007, 05:32:09 PM
Just had a thought..

Let's turn this whole thing on it's head.
Could I mic in X/Y and use the M/S function in my preamp or recorder to record the M/S signals? 

Assumming for a moment that the extra matrix steps DO deteriorate the sound appreciably, I could then adjust in post without the need for extra matrixing (if that matters at all ;), it's probably splitting hairs),

More importantly, this could open up some interesting mic techniques.. like choosing polar patterns and mic angles based on other issues than the resulting stereo sound stage.

For instance, if using mics with less than stellar off-axis respose in a 120deg. X/Y setup, I could reduce the angle between mics for more tonal accuracy towards the center, then dial back in more stereo separation by increasing the S signal ratio later to compensate.
 
Unfortunatley I don't think the matrix built-in to preamps and recorders will work 'in-reverse'.
Title: Re: M/S ???
Post by: Will_S on November 27, 2007, 06:21:04 PM
Just had a thought..

Let's turn this whole thing on it's head.
Could I mic in X/Y and use the M/S function in my preamp or recorder to record the M/S signals? 

Mathematically, I don't see why this wouldn't work.  If the decoder is outputting the sum of the inputs as the left channel, and the difference between the inputs as the right channel, this should work equally well at creating L and R from M+S and M-S or 2*M and 2*S as L+R and L-R.

Quote
Assumming for a moment that the extra matrix steps DO deteriorate the sound appreciably, I could then adjust in post without the need for extra matrixing (if that matters at all ;), it's probably splitting hairs),

Actually, here's where I can see problems might crop up.

Say you want to use XY cards as your microphones, and you want to use a fairly tight included angle to minimize off-axis frequency response problems.  You will end up with a mid that is much higher in level than the side channel, and you have no means of independently adjusting the gain of mid vs. side since your preamp gain controls are only working on left and right signals.  If you are using a 16 bit recorder, that means less than 16 bits used to capture the master side signal.  Potentially that could lead to problems with noise as you boost the side channel in post (as you probably would want to, given the narrow mic angle you started with).

This also strikes me as a potential problem with micing Mid-Side but recording as L-R, then dematrixing and remixing in post.  If you recorded left and right at 16 bit, the side channel you derive will have less than 16 bit resolution (even if your DAW does its internal processing at 32 bit, it still can't get that level of precision taking the difference between two 16 bit sources) which again might lead to problems when you boost the side channel in post.
Title: Re: M/S ???
Post by: Gutbucket on November 27, 2007, 07:07:31 PM
Quote
...Assuming for a moment that the extra matrix steps DO deteriorate the sound appreciably, I could then adjust in post without the need for extra matrixing (if that matters at all ;), it's probably splitting hairs),

Actually, here's where I can see problems might crop up.

Say you want to use XY cards as your microphones, and you want to use a fairly tight included angle to minimize off-axis frequency response problems.  You will end up with a mid that is much higher in level than the side channel, and you have no means of independently adjusting the gain of mid vs. side since your preamp gain controls are only working on left and right signals.  If you are using a 16 bit recorder, that means less than 16 bits used to capture the master side signal.  Potentially that could lead to problems with noise as you boost the side channel in post (as you probably would want to, given the narrow mic angle you started with).

That's a concern I'm thinking about too.  Yes the preamp gain would reflect the needs of the X/Y arrayed mics capturing L/R.  However, if I'm using a separate preamp and a recorder with independent gain control for each recording channel, I could then optimally set the recorder's levels for the M & S signals independently, even if that meant the 'correct' M/S ratio was off.  This would mean a better use of recording bits than recording the L/R signal for anything other than a perfectly even 50% matrix ratio. 

Now that could be a problem if there is only one gain stage between mics and all-in-one recorder.  (is this the case for the SD7xx?, assuming it could do the 'reverse' M/S thing?)   

Quote
This also strikes me as a potential problem with micing Mid-Side but recording as L-R, then dematrixing and remixing in post.  If you recorded left and right at 16 bit, the side channel you derive will have less than 16 bit resolution (even if your DAW does its internal processing at 32 bit, it still can't get that level of precision taking the difference between two 16 bit sources) which again might lead to problems when you boost the side channel in post.

If that's a problem it would be an argument that the post procedure does in fact deteriorate the sound, and would apply to doing the matrix>adjust>dematrix thing in post with any source.

Good thoughts.
Title: Re: M/S ???
Post by: Will_S on November 27, 2007, 08:00:26 PM
Interesting stuff for sure.

Another thing to think about is that the polar patterns and effective angles you get back will of course be constrained by the mics and angles you choose for your left and right channels (or mid and side).

Personally, I *like* the sound of conincident hypers at ~120° better than I like cards at 90° most of the time.  So I like the fact that that's the baseline pattern I get back from typical card-fig8 mid-side micing with a 1:1 mix of mid and side.  I wouldn't so much like the baseline I'd get back from a pair of cards.  And if I used cards at a reasonable narrow angle to keep the center sound tonally balanced, then for any effctive mic angle I wanted to get back from my M/S decoding I'd end up with fatter pickups on my virtual mics than I would with conventional M-S micing.  [But then maybe I'd really like playing around with a recording originally made with XY hypers...]

or instance, if using mics with less than stellar off-axis respose in a 120deg. X/Y setup, I could reduce the angle between mics for more tonal accuracy towards the center, then dial back in more stereo separation by increasing the S signal ratio later to compensate.

Except that if what you really wanted was the sound of cards at 120° but you recorded with more narrowly angled cards and boosted the side in post, you'd get back something more like fat hypers at 120°.  But this could be a cool way to go if you wanted to run hypers but only had cards at your disposal.
Title: Re: M/S ???
Post by: Gutbucket on November 27, 2007, 08:17:09 PM
^^^
Like minds think alike.

This is why this stuff never seems to grow old to me.
Title: Re: M/S ???
Post by: Ozpeter on November 28, 2007, 04:41:50 PM
I've routinely taken recordings with hardly any stereo separation due to someone having recorded with too narrow and angle with an X/Y pair, and fed the result through Voxengo MSED in the PC - and the improvement is usually dramatic.  In fact, I find that the Rode NT4 usually benefits from this treatment - the normal two dimensional sound can suddenly acquire the extra third dimension of spaciousness and reverberant warmth.