Taperssection.com

Gear / Technical Help => Post-Processing, Computer / Streaming / Internet Devices & Related Activity => Topic started by: admkrk on March 30, 2019, 07:08:53 PM

Title: M/S Question
Post by: admkrk on March 30, 2019, 07:08:53 PM
I tried an experiment with M/S and it came out OK, except the right channel is peeking ~4db over the left. I am using mono channels and not doing this in software, so I would think it would all balance out.

I am using Wavelab and have the mid track sent to l/r and also l/r. The right side is sent to r and r, and the inverse is sent to l and l. Dropping the right channel ~8.5db seems to have evened out the peaks, for the most part.

Is it normal for the peaks of an inverse of a track to be lower, or higher, than the original? I was expecting them to be relatively close at least. Could this just be due to bad aiming on my part?
Title: Re: M/S Question
Post by: EmRR on March 31, 2019, 09:55:45 AM
I’m not following your mix method, but what happens if you turn mid off? That should result in a balanced two channel signal. 
Title: Re: M/S Question
Post by: morst on March 31, 2019, 03:45:40 PM
Could this just be due to bad aiming on my part?
Was the sound source symmetrical at all? Guitar on the left and Bass on the right can make it tough to get a balance in an up-close recording.


Were you running downstage center (stage lip?)
Title: Re: M/S Question
Post by: admkrk on March 31, 2019, 05:43:56 PM
Quote from: morst link=topic=190035.msg2295739#msg2295739 date=1554061540
Were you running downstage center (stage lip?)
[/quote

No, I was a bit back. That was the first lesson I learned, need to be close to use this.

Anyway, the problem seems to be with the mid track, and I really do not really understand why. Creating a stereo track gets the same result, louder right channel. To get it balanced, I needed to lower the gain of the right side track by ~11db. I am starting to think this is a Wavelab problem. If I drop the right gain, on the track, all the way down, it still runs louder. I tried one plugin, but it added static, so no good. I am going to try a different one and see if that helps.

I still do not understand why one channel would be different since both are coming from the same track.
Title: Re: M/S Question
Post by: mepaca on March 31, 2019, 09:42:12 PM
Get the free plugin Voxengo MSED.  https://www.voxengo.com/product/msed/   It is the best mid/side convertor ever and who can argue with free.
Title: Re: M/S Question
Post by: admkrk on March 31, 2019, 10:22:38 PM
if I had just a stereo track a converter would work. I do not need one with mono tracks, just duplicate the side and invert it. Anyway, Blue Cat Gain 3 seems to do the trick leveling out the mid track, even if left and right are backwards. -10 db to the left(right) channel and I am pretty evened out. I need to start over and do some work to the mono tracks, so I might be able to even it out a little better that way. If I  can get it down to -5 or less, I will be happy. As it is, the right is a bit too quiet when soloed. Dropping the gain on the right channel of the mid sounds way better than do it on the right side.
Title: Re: M/S Question
Post by: Gutbucket on April 01, 2019, 12:10:59 PM
What EmRR said.

If you mute the Mid, your two duplicated Side channels (one with non-inverted polarity panned hard left, one with inverted polarity panned hard right) should be equal in level.  If you solo those channels and listen to them in mono, either by using the mono monitoring function if there is one in Wavelab or by panning both to center temporarily, the two should cancel each other out. 

^ You can use that mono sum cancellation technique to achieve absolute balance of the resulting stereo image and equivalent virtual pickup patterns on each side if one of your duplicated Side channels is higher in level that the other for whatever reason (even though they shouldn't be) by adjusting gain of one channel agasint other until you achieve maximum cancellation.  Then switch back to stereo monitoring or revert panning of the two Side channels to hard Left/Right.

At that point, after you bring up the Mid and/or any other channels you are mixing in, the L/R balance may no longer be equal, yet should reflects the actual energy balance in the room when the recording was made.

Since that "as recorded reality" may not be ideal, you can intentionally use more level in one Side channel than the other to rebalance things.  That works sort of like panning, but varies polar pattern along with virtual pan angle for each side like Mid/Side ratio adjustment always will unless using a figure-8 Mid.  You end up with a slightly more supercardish pattern panned a bit closer to center on one side and a slightly more hypercardish pattern panned farther away from center on the other.. or vice versa.

The other way to adjust stereo panning is be to render the 3 Mid/Side channels to a single L/R stereo channel then adjust panning in the typical way.  That will retain the identical virtual pattern on each side which was "set" when converting from M/S to L/R, and simply varies the level of L against R.
Title: Re: M/S Question
Post by: Ronmac on April 01, 2019, 02:05:11 PM
I’m not following your mix method, but what happens if you turn mid off? That should result in a balanced two channel signal.


Me either....

Was this a M/S recording done with a mono mid mic and a Fig. 8 for sides, or are you decoding a stereo file to M/S for mixing?
Title: Re: M/S Question
Post by: Gutbucket on April 01, 2019, 03:31:21 PM
Is it normal for the peaks of an inverse of a track to be lower, or higher, than the original
No. A channel and an identical copy of that channel with polarity inverted should be absolutely identical to each other in every way except for polarity.  That's why they should cancel out when summed to mono or both are panned to center.

But after converting M/S to L/R is would not be uncommon to have some content in one channel higher in level than the other.

Quote
I am using Wavelab and have the mid track sent to l/r and also l/r. The right side is sent to r and r, and the inverse is sent to l and l.
  This is the part I don't follow and which others may not be understanding.

Given the way you are going about doing the Mid/Side to Left/Right conversion manually:
~Mid channel should routed to both the Left and Right output buss or panned to center.
~Side channel should be duplicated so as to populate two channel strips.  Assuming that when making the recording you used the standard M/S setup with the front of the fig-8 (positive polarity lobe) facing Left, the original copy should be routed or hard panned fully Left.  The polarity inverted copy channel should be routed or hard-panned fully Right.
~Mute the Mid channel. Adjust the channel gains of two Side channel copies so that they are identical.  Mono or center-pan the two Side channels to confirm that they cancel each other.  If they don't cancel, perhaps something other than a simple polarity inversion occurred when you copied them.  Adjust the gain relationship between the two until they do cancel.  Then lock the two Side channel faders together so that their relative gain relationship remains the same and both move together as a group.
~Unmute the Mid channel and adjust stereo width by the relative balance of the Mid channel against the pair-locked Side channels.

Quote
Could this just be due to bad aiming on my part?
If you are seeing that imbalance between Left and Right channels after doing the conversion, then it could be how the mic array was aimed, or it could reflect the Left/Right energy balance in the room from the recording position.
Title: Re: M/S Question
Post by: admkrk on April 01, 2019, 05:33:42 PM
Let me start with this part.

Quote
I am using Wavelab and have the mid track sent to l/r and also l/r. The right side is sent to r and r, and the inverse is sent to l and l.
  This is the part I don't follow and which others may not be understanding.

I have a dialog with a group that says "Audio is sent to" and check boxes for Left front and Right front, there are others but they are greyed out unless I change modes. Another group says "Audio is also sent to" with the same check boxes. I am not sure if that is any clearer, or not. Basically, if I understand right, I am doubling up on the side channels by sending both outputs to the same channel, and sending one each left and right for the mid. All three are mono tracks.

I switched the two side channels to mono and they do not completely cancel each other out. The lows are pretty much gone, but the highs are still there. When I switched back to stereo, it went back to being ~6 db hotter in the right channel, so there is definitely something funky going on with Wavelab. A restart put then back even again. I think I need to play around and figure out what is causing this before I go any farther.

Title: Re: M/S Question
Post by: heathen on April 01, 2019, 05:39:46 PM
Can you share a clip of your source files so we can try our own mid-side processing and see if we run into the same issues?
Title: Re: M/S Question
Post by: Gordon on April 01, 2019, 06:48:49 PM
This thread reminds me of why I've never done M/S!   :o :scared:
Title: Re: M/S Question
Post by: admkrk on April 01, 2019, 09:35:28 PM
Can you share a clip of your source files so we can try our own mid-side processing and see if we run into the same issues?

I probably should have cut off a smaller bit of the start, but here (https://send.firefox.com/download/e0c3b04aa715c07b/#oUuBsMG8hUp4w8juOsYkKQ) is the first song, if this works.

<edit> Obviously the side track is called Right, and pointed right. </edit>
Title: Re: M/S Question
Post by: morst on April 01, 2019, 10:01:06 PM
This thread reminds me of why I've never done M/S!   :o :scared:
this thread reminds me why I never tried wavlab!
Title: Re: M/S Question
Post by: Gutbucket on April 02, 2019, 08:49:12 AM
M/S isn't difficult or overly complicated.   The hardest part is describing the issue in text on a forum!  The biggest constraint to doing it is having a fig-8 microphone to use. 

I don't use Wavelab so I'm unsure what shortcuts it provides, but it's very easy in Samplitude with mono input Mid and Side files- you don't need to duplicate, route/pan the mono Side channel, just specify to Samplitude that it is a Side channel and adjust ratio to taste using the relative position of the Mid and Side faders.

Blue Cat Gain 3 seems to do the trick leveling out the mid track, even if left and right are backwards.

Standard Mid/Side mic setup orients the Side mic so that it is pointing to the Left side rather than the Right.  To flip it back you can invert polarity on the Side copy panned/routed Left instead of the one panned/routed Right.  Or you could just flip or re-pan the resulting L/R output.
Title: Re: M/S Question
Post by: noahbickart on April 02, 2019, 10:03:57 AM
You people are all crazy for trying to explain m/s decode as if we still used analog consoles.

Run M as L and S as R. Then Use a vst plugin like the free MSED referenced above.
Title: Re: M/S Question
Post by: Ronmac on April 02, 2019, 12:06:42 PM
I downloaded your file and dropped it into Reaper to see what was up.

Placed "mid" on Track 1 and "right" on Track 2

Panned Track 1 to "C"

Panned Track 2 to 100% "L"

Duplicated Track 2 to Track 3, panned 100% "R", flipped polarity

Everything sounded as I would expect it should. Note that Track 2 and Track 3 were equal amplitude, as you would expect.

Definitely a problem routing the tracks on your end...
Title: Re: M/S Question
Post by: admkrk on April 02, 2019, 04:55:39 PM
That is what I was figuring would be the case, unfortunately this is what I tend to get.
(https://i.imgur.com/oqABRO1l.png)
if I solo the mid track it looks the same, and why I was confused. A restart puts it back where it should be.
(https://i.imgur.com/FjUlmsXl.png)
unfortunately, at some point, it always goes back to the first image.

I do have Reaper, so maybe I will just give that another shot, instead of fighting with Wavelab.
Title: Re: M/S Question
Post by: admkrk on April 07, 2019, 03:20:30 PM
I think I am finally starting to like Reaper. If you are interested in the final final mix; https://archive.org/details/donna2019-03-23 (https://archive.org/details/donna2019-03-23).
Title: Re: M/S Question
Post by: Gutbucket on April 08, 2019, 09:33:08 AM
Sounds like stereo!

..and the SOTSMP amphitheater.
Title: Re: M/S Question
Post by: Ronmac on April 08, 2019, 09:44:54 AM
I had a quick listen and the image sounds to be stable. Success!
Title: Re: M/S Question
Post by: admkrk on April 08, 2019, 06:28:08 PM
Thanks, I have not listened outside of the PC yet, it will get loaded the next time I redo the thumb drive for the truck. I do not know what the problem with Wavelab was, but Reaper made a huge difference. It is probably going to take me a while to get it setup how I want, but otherwise so far so good. My biggest complaint so far is the waveform does not get redrawn when you change the volume envelope, and I do not think there is a fix for that.

I probably could have avoided a bit of chatter if I had setup more centered in the cage, instead of on the side. It kind of gives you something to listen to during the long breaks when they are waiting for sit-ins though.