Become a Site Supporter and Never see Ads again!

Author Topic: "Deleting" rear lobe hyper info with M/S Theory  (Read 1395 times)

0 Members and 1 Guest are viewing this topic.

Offline noahbickart

  • Site Supporter
  • Trade Count: (28)
  • Taperssection All-Star
  • *
  • Posts: 1746
  • Gender: Male
  • So now I wander over grounds of light...
"Deleting" rear lobe hyper info with M/S Theory
« on: April 22, 2013, 09:40:59 AM »
I know people has discussed this before with the consensus that this was impossible. But, as I was walking home today with by daughter strapped to my back asleep I got to thinking...

What if you record two Schoeps mk41s in say DINa.

Track 1 = L
Track 2 = R.

If you feed these into a DAW with M/S processing, why not pretend that

Track 1 = M
Track 2 = S

If these are then converted from "M/S" to "L/R" without any "M" (e.g. Track 1) info, the resulting 2 channel file will present the front Lobe of the mk41 representing Track 2 in the L and the rear Lobe of the mk41 representing Track 2 in the R.

If you then Cut and Paste the "L" channel into the R channel of new stereo file, you have the front lobe only information of the Track 2 = R mk41.

If the same process is repeated with the original channels reversed, you can derive the front lobe only of the Track 1= L mk41.

Now, I'm not sure you would wan't to do this. The rear lobe information is likely a good thing most of the time. However, does my theory work?
Recording:
Capsules: Schoeps mk41v (x2), mk4v (x2), mk22 (x2), mk3 (x2), mk21 & mk8
Cables: 2x nbob KCY, 1 pair nbob actives, Darktrain 2 and 4 channel KCY extensions:
Preamps:    Naiant Littlebox, Naiant IPA, Naiant PFA, Sound Devices Mixpre6
Recorders: Sound Devices Mixpre6, Sony PCM m10
Home Playback: Mytek DSD 192> Adcom SLC 505> Marantz Ma500 (x2)> Eminent Tech LFT-16; Musical Fidelity xCan v2> Hifiman HE-400
Office Playback: Grace m903> AKG k701

Offline page

  • Trade Count: (25)
  • Needs to get out more...
  • *****
  • Posts: 8392
  • Gender: Male
  • #TeamRetired
Re: "Deleting" rear lobe hyper info with M/S Theory
« Reply #1 on: April 22, 2013, 09:54:32 AM »
What if you record two Schoeps mk41s in say DINa.
<snip>
the resulting 2 channel file will present the front Lobe of the mk41 representing Track 2 in the L and the rear Lobe of the mk41 representing Track 2 in the R.
<snip>
However, does my theory work?

Nope.

2 general problems: You have time delay differences between the caps unless they are coincident in the signal plane. Second, your left front lobe will contain some of the information from the rear right lobe when you do the decode (but with the time delay issue). Specifically, if you take an MS pair and ditch the M track, you are left with a track that has positive and negative information (courtesy of the bidirectional capsule). Doing so with a hyper doesn't change that, it just means that you're negative information is reduced. So your L/R will contain information from both lobes, not separated as you have hypothesized above.

There is some/minor benefit in doing M/S processing on a non-coincident pair, but it's limited and best used sparingly. I give it a shot on about half of my tapes and find sometimes it's worth taking a hit on the stereo image cohesiveness for a more stable and better sounding image, but it's not often and I do so only lightly.
"This is a common practice we have on the bus; debating facts that we could easily find through printed material. It's like, how far is it today? I think it's four hours, and someone else comes in at 11 hours, and well, then we'll... just... talk about it..." - Jeb Puryear

"Nostalgia ain't what it used to be." - Jim Williams

Offline Gutbucket

  • record > listen > revise technique
  • Trade Count: (13)
  • Needs to get out more...
  • *****
  • Posts: 12131
  • Gender: Male
Re: "Deleting" rear lobe hyper info with M/S Theory
« Reply #2 on: April 22, 2013, 11:07:35 AM »
Snarky answer- Rear lobes are a good thing.

Short answer- You cannot derive a less directional pattern than that of the microphone used for the Mid channel.

Slightly more technical answer- Any pattern derived from Mid/Side processing of two coincident microphones will be a 'first-order' pattern.  That is, a pattern that lies somewhere on the continuum between omni > through cardioid > to figure-8.  There is no way with standard decoding to derive a higher order pattern.  Some higher order patterns begin to look something like supercardioids without a rear lobe.

Slightly more technical answer on Page's notes about non-coincident sources (and Jon's addition/subtraction operation statement)- Mid/side processing of non-coincident sources is completely lossless, unproblematic and transparent as long as the input ratio = the output ratio.  But that doesn't do you any good if the whole point is adjusting the ratio. Pragmatically, the range of usable adjustment for non-coincident microphones is smaller than for coincident sources which have no phase difference between channels.   The potential problem in changing the ratio is comb filtering between the two channels due to the phase differences (think mixing spaced omnis to mono).  Increasing the Mid over the Side ratio is the same as panning both channels closer to center.  Increasing the Side to Mid channel ratio is identical to increasing typical 'stereo width enhancement' which is a common alternate function to standard Left/Right pan/balance on the stereo channels of many editors.

On Saturday I mixed down an organ trio session I recorded last week using two different non-coincident microphone arrays, and ended introducing a very small amount of 'stereo width enhancement' to the 2 channel master buss to push the slightly forward sounding drums back ever so slightly in the center of the stereo image and to help further 'detach' the guitar on the left and the organ on the right from sounding quite as locked into the position of the speakers.  I only dialed in the very slightest amount, which was subtle but quite beneficial.  Trying to add much more would have quickly made a mess of things.

volition > vibrations > voltages > numeric values | numeric values > voltages > vibrations> virtual teleportation time-machine experience
made easy- >>Improved PAS table<< | made excellent- >>click here to download the Oddball Microphone Technique illustrated PDF booklet<<

Offline noahbickart

  • Site Supporter
  • Trade Count: (28)
  • Taperssection All-Star
  • *
  • Posts: 1746
  • Gender: Male
  • So now I wander over grounds of light...
Re: "Deleting" rear lobe hyper info with M/S Theory
« Reply #3 on: April 22, 2013, 12:11:48 PM »
I guess what I don't understand is that if one can separate the lobes of a figure-8 evenly such that one half goes left and the other goes right (even if I'm muting the M channel, I can hear a full soundstage).

Why doesn't this work for a hyper? The Front Lobe is + and the rear is -, just like a figure - 8!

Recording:
Capsules: Schoeps mk41v (x2), mk4v (x2), mk22 (x2), mk3 (x2), mk21 & mk8
Cables: 2x nbob KCY, 1 pair nbob actives, Darktrain 2 and 4 channel KCY extensions:
Preamps:    Naiant Littlebox, Naiant IPA, Naiant PFA, Sound Devices Mixpre6
Recorders: Sound Devices Mixpre6, Sony PCM m10
Home Playback: Mytek DSD 192> Adcom SLC 505> Marantz Ma500 (x2)> Eminent Tech LFT-16; Musical Fidelity xCan v2> Hifiman HE-400
Office Playback: Grace m903> AKG k701

Offline Gutbucket

  • record > listen > revise technique
  • Trade Count: (13)
  • Needs to get out more...
  • *****
  • Posts: 12131
  • Gender: Male
Re: "Deleting" rear lobe hyper info with M/S Theory
« Reply #4 on: April 22, 2013, 12:46:06 PM »
Jon beat me to it. posting anyway..

I think that's just an illusion.  You can't "seperate the lobes" of the figure-8.  Used as the Side microphone the two point left and right simultaneously and are always interdependent.  One side cannot be isolated from the other and whatever mixing is done to one is also done to the other.  The only difference between the two is the polarity of the signal.

If you mute the Mid channel you'll hear only the mono side channel information.  It may sound like a full soundstage, but that is probably due to it sounding significantly different than the Mid channel.  Since it sounds more reverberant it may sound like it has an actual stereo soundstage with different information on each side, but it's actually only the mono side information.
volition > vibrations > voltages > numeric values | numeric values > voltages > vibrations> virtual teleportation time-machine experience
made easy- >>Improved PAS table<< | made excellent- >>click here to download the Oddball Microphone Technique illustrated PDF booklet<<

Offline page

  • Trade Count: (25)
  • Needs to get out more...
  • *****
  • Posts: 8392
  • Gender: Male
  • #TeamRetired
Re: "Deleting" rear lobe hyper info with M/S Theory
« Reply #5 on: April 22, 2013, 02:04:42 PM »
your "separation" comes in terms of the cancellation with the coincident second channel that contains the same information but of a different polarity. So think of it in terms of an omni and a fig8; have a channel with +/- and a channel with +/+. Combine them and you get 2+/0, thus a cardioid on the left side. Invert the omni so it's now -/- and you end up with 0/2- which gives you information on the other side. This is all assuming that the levels are as they were captured, etc. Other adjustments will tinker with the outcome, but that's the basic MS theory behind it.

The minute you start to de-correlate information (e.g. spacing mics, turning the focus, etc), that combination effect doesn't emphasize and deminish to the same effect as it does with a coincident pair.
"This is a common practice we have on the bus; debating facts that we could easily find through printed material. It's like, how far is it today? I think it's four hours, and someone else comes in at 11 hours, and well, then we'll... just... talk about it..." - Jeb Puryear

"Nostalgia ain't what it used to be." - Jim Williams

Offline noahbickart

  • Site Supporter
  • Trade Count: (28)
  • Taperssection All-Star
  • *
  • Posts: 1746
  • Gender: Male
  • So now I wander over grounds of light...
Re: "Deleting" rear lobe hyper info with M/S Theory
« Reply #6 on: April 22, 2013, 02:35:12 PM »
Got it now.

Thanks everyone. This thread is further proof that ts.com is the best little corner of the internet there is. I'd +t all of you if I could...
Recording:
Capsules: Schoeps mk41v (x2), mk4v (x2), mk22 (x2), mk3 (x2), mk21 & mk8
Cables: 2x nbob KCY, 1 pair nbob actives, Darktrain 2 and 4 channel KCY extensions:
Preamps:    Naiant Littlebox, Naiant IPA, Naiant PFA, Sound Devices Mixpre6
Recorders: Sound Devices Mixpre6, Sony PCM m10
Home Playback: Mytek DSD 192> Adcom SLC 505> Marantz Ma500 (x2)> Eminent Tech LFT-16; Musical Fidelity xCan v2> Hifiman HE-400
Office Playback: Grace m903> AKG k701

Offline F.O.Bean

  • Is a 4 channel slut and
  • Trade Count: (125)
  • Needs to get out more...
  • *****
  • Posts: 40657
  • Gender: Male
  • Taperus Maximus
Re: "Deleting" rear lobe hyper info with M/S Theory
« Reply #7 on: April 22, 2013, 02:51:02 PM »
Damn this is one confusing thread
Recording Rigs:
Schoeps MK4's & MK41's ->
Schoeps & NBob 250/05 KCY's ->
Naiant +60v/Low Noise PFA's ->
DarkTrain Right Angle Stubby XLR's ->
Sound Devices MixPre-6 & MixPre-3

Portable Playback Rigs:
Sony NW-A35 Walkman ->
Linum G2 Bax & Westone Epic MMCX Cables ->
Westone UM Pro 30 (V1 & V2) IEM's ->
Comply Isolation+ Memory Foam Tips
OR
Sony Walkman -> FiiO E11K HP Amp ->
Optoma NuForce Primo 8 & EDC IEM's ->
Comply Isolation+ Tips

http://www.archive.org/bookmarks/diskobean | http://www.archive.org/bookmarks/Bean420 | http://bt.etree.org/mytorrents.php

Offline page

  • Trade Count: (25)
  • Needs to get out more...
  • *****
  • Posts: 8392
  • Gender: Male
  • #TeamRetired
Re: "Deleting" rear lobe hyper info with M/S Theory
« Reply #8 on: April 22, 2013, 02:59:43 PM »
Now, what you *really* want is 4 channels, 2 cardioids coincident to each other, 180 degrees apart, and record stuff in what amounts to a B Format. Invert the rear facing ones on each pair and you can adjust your pattern (but not angle) by adjusting the volume on that back "lobe." Be a baller and go for 8 channels and run two sps200s (adjust angle and pattern, but not spacing).

I don't know of any way to adjust the spacing later using just room mics; I've played with doing specific delay timings later in post on coincident pairs and it just doesn't work like I want. Mathematically it makes sense why it doesn't (tldr; I want to delay some sounds but not others), but that's the last limiting factor. You can do it with spot micing easily enough though (dupe the channel, adjust panning to "place" it, reduce the volume in the "other" channel, then add fractional milliseconds of delay to get it to pop out).
"This is a common practice we have on the bus; debating facts that we could easily find through printed material. It's like, how far is it today? I think it's four hours, and someone else comes in at 11 hours, and well, then we'll... just... talk about it..." - Jeb Puryear

"Nostalgia ain't what it used to be." - Jim Williams

Offline Gutbucket

  • record > listen > revise technique
  • Trade Count: (13)
  • Needs to get out more...
  • *****
  • Posts: 12131
  • Gender: Male
Re: "Deleting" rear lobe hyper info with M/S Theory
« Reply #9 on: April 22, 2013, 05:04:43 PM »
Damn this is one confusing thread

Things gets really confusing by introducing spacing into the equation.

The minute you start to de-correlate information (e.g. spacing mics, turning the focus, etc), that combination effect doesn't emphasize and deminish to the same effect as it does with a coincident pair.

To explain that in a bit more detail-

If the microphones are truly coincident, the relationship between signal level and polarity is the same across all frequencies and the derived polar patterns remain consistent regardless of frequency.  That's standard Mid/Side stuff and confusing enough.

If there is space between the microphone capsules, the phase relationship between the two mic signals varies with frequency and that straight-forward relationship becomes very complex.  The phase relationship between signals in that case is also determined by the relationship between the wavelength of the frequency in question and the microphone spacing.

Now down the rabbit hole, Alice-

Now, what you *really* want is 4 channels, 2 cardioids coincident to each other, 180 degrees apart, and record stuff in what amounts to a B Format. Invert the rear facing ones on each pair and you can adjust your pattern (but not angle) by adjusting the volume on that back "lobe." Be a baller and go for 8 channels and run two sps200s (adjust angle and pattern, but not spacing).

I don't know of any way to adjust the spacing later using just room mics; I've played with doing specific delay timings later in post on coincident pairs and it just doesn't work like I want. Mathematically it makes sense why it doesn't (tldr; I want to delay some sounds but not others), but that's the last limiting factor. You can do it with spot micing easily enough though (dupe the channel, adjust panning to "place" it, reduce the volume in the "other" channel, then add fractional milliseconds of delay to get it to pop out).

All traditional B-format manipulation is standard Mid/Side matrix math with more channels than two, and like Mid/Side is constrained to first-order patterns (no magic lobe elimination possible), but gains the freedom to derive patterns and the directions they point independently of each other.

Using traditional B-format matrix techniques there is no way to simulate spacing.  The virtual patterns are always coincident.  The only way around that is actually putting microphones in two separate locations as Page suggests with two Soundfield mics.  That only gets you that one particular spacing however, it’s not adjustable later like the polar patterns and angles.

The possible paradigm changer is using advanced analysis and manipulation techniques other than standard matrix mathematics.  There is a B-Format decoder now on offer called Harpex, which uses advanced techniques I don’t pretend to understand well at all, that simulates the spacing between virtual mic capsules and higher-order patterns. It simulates near-spaced configurations, spaced omnis, greater than 1st order supercardioid directionality and horizontal only HTRF simulations.  In a way it becomes trickery instead of pure Mid/Side matrix mathematics, but if it’s convincing and gives good results maybe that doesn’t matter so much.  As I understand it the technique slices up the audio in to thousands of frequency dependant bits and determines a direction vector for each, then does something like Page mentions.. probably applying source direction specific delays and M/S matrixing based on each frequency bin before putting Humpty Dumpty back together again.

I’ve not tried it yet myself but many in the ambisonic recording community speak highly of the results on the sursound mail list and at Gearslutz.  As you might imagine, very few using it really understand the basis of  how it works, but the interface is easy and visually informative.  The AES article and patent describing how it works is linked at the Harpex website, and is written in typically obscure patent language.  If anyone does B-format work and is interested they have a trial download and a page that lets you upload and decode small samples to get an idea of how it sounds.  It’s not inexpensive, and so far I’ve resisted the urge to spend enough time to check it out.
« Last Edit: April 22, 2013, 05:08:00 PM by Gutbucket »
volition > vibrations > voltages > numeric values | numeric values > voltages > vibrations> virtual teleportation time-machine experience
made easy- >>Improved PAS table<< | made excellent- >>click here to download the Oddball Microphone Technique illustrated PDF booklet<<

 

RSS | Mobile
Page created in 0.189 seconds with 37 queries.
© 2002-2018 Taperssection.com
Powered by SMF