Topic: Rear-facing audience mics: Wow. New links added.

[Gutbucket]

Interesting.   So would one expect the resulting sound to be similar to what it would sound like if you recorded with omnis?

All right now, been kicking around your omni comment for a couple of days:  How does a forward-facing cardioid pair in DIN and a rearward facing supercardioid pair in AB60 = omni?  I'm obviously not good with the maths...so help me understand.

Actually to my way of thinking about all this, it is similar in some important ways to using omnis.  The overall directional sensitivity of the entire microphone array (all the microphones together, considered in combination) is more or less omnidirectional, as the setup is sensitive to sound arriving at the microphone position from most anywhere in the entire horizontal plane.  That includes not just the direct sound from the band in front and the audience members all around, but also a much more detailed sampling of the complex reflections and reverberance in the room.  I think that is probably the biggest contributor to the improvement in depth and spaciousness over setups that are not as omnidirectional overall.  It is similar to both spaced omnis and Blumlien crossed figure-8s in that respect, since those setups also have an overall directional sensitivity which is omnidirectional in the horizontal plane.

However, one important aspect which is quite different than using omnis or Blumlien is the directional channel separation and the control that provides over the overall horizontal sensitivity.  I can adjust the level of the rearward facing microphones as necessary. Lets say I lower the level, and the overall directionality of the array becomes more forward biased.  To be able to change the overall directionality of the array like that, I need sufficient channel separation.  To get sufficient channel separation I need to use directional rear facing microphones, and arrange them in such a way as to minimize pickup of the direct sound in front to a significant degree.   I'll arrange them so the null of the microphones point the at the band or the PA speakers, and/or do other things to reduce the pickup of the main sound arriving from in front as much as possible while focusing on the audience and room sound.  I've posted photos of on-stage setups where I've clamped audience facing cards or supercards so they were hanging down in front of the lip of the stage so the stage itself sort of baffles them from the main sound, increasing the usable separation more than if they were facing the same direction, but up on the stage.  Same goes for miniature omnis taped to the face of the stage facing out into the room, they become directional and that directionality is intentionally used to exclude the main sound.  Omnis further back in the room or mounted on the back wall of the venue aren’t as useful because they contain too much of the main sound along with the audience and room ambience.

That ability to dial in just the right amount of the rear facing information is part of what makes this so powerful.  The optimal level varies, but is pretty easy to decide upon by ear.  It depends of course on the 'room' (indoor or out) sounding good or not and if the audience and ambience contribution is positive or not.  It's not suprising that more level is often good on quieter songs, during audience reaction between songs, for the portions where there is no music playing, and if the forward facing pair is drier and less reverberant in isolation.  Less is good when things get louder and the sound bouncing around the room gets denser, if the audience in back is bad, or if the forward facing channels are more reverberant.  If the recording is worth the effort I’ll automate the level of the rear facing mic(s) with a volume envelope to best optimize it's contribution.

I will say this however- In my experience, I find that some amount of rear facing audience/ambience channels is always better than none.  Every time. Even if it’s only a little bit. Even if the audience back there is bad and I want to maximally reduce it, adding just a touch at makes it sound better more dimensional and more natural.  If I can’t reduce it any farther, I can at least make it less bad.

There are other practical advantages over Blumlien or strait spaced omnis too- more control over imaging and pickup angle (apply the stereo zoom stuff, choose a config for the forward and rearward facing arrays) which means increased freedom of mic placement in the room over Blumlein, and also the ability to use mics with different responses or polar patterns for what ever reasons, but its the mixing control it provides which is the most valuable aspect in my opinion.  It's the practical control and adaptibility via all those aspects which makes it a much more useful way of achieving the beautiful sense of depth, dimension and naturalness (in my opinion a mark of an excellent recording) which omnis or blumlein can also provide but only in more limited situations.

[Gutbucket]

Here's a link to a clip in which the rearward-facing pair was mixed in at -4 dB.  EQ'd the singer's voice a bit; I found his timbre + the mic choice (I guess an SM58) too hashy.  Biggest environmental noise in this clip is the giggling baby.

https://soundcloud.com/backwoodsman-1/ts-clip/s-Ow9cW

Very nice sound I think.  Band performace sounds good too.  Nice one.

[kindms]

kindms and I have been using his 414 XLS's in a "Bi-directional" or Blumlein config for certain situations for some time now. Usually outdoors and with crowds that are hopefully more polite than some. We set them up in a coincident crossed physical config, and set them to figure 8. We have always noticed the openness and ambient quality you are mentioning with your rearward facing MK 41s. The rear lobes of the figure 8s bring out some extra dimensionality to the recording but of course present objectionable audience noise into the mix if you are only using 2 channels.
Recently, at Phish SPAC 2014-07-05 we ran MS with his 414s from the lawn with the stand about 12 feet up. Overall it is a pretty good recording, the overall sound quality and tonality beats the crap off of our "safety" recording made with AKG C460 w/ck 8 short shots. As kindms said about the 414s MS, "the audience is DEFINITELY a part of the recording".
One question: were your rearward facing mics FOH or on stage? a lot of large stage A-list performers' sound crews place on stage shotguns facing toward the audience to get the same effect you hear with your mk 41 channels.

http://www.blueberrydreams.com/Audio/phish2014-07-05akg414.flac16.zip

http://www.blueberrydreams.com/Audio/phish2014-07-05akg414.flac24.zip

I was debating putting these on etree so if this is a big Phish no no (direct link) let me know and I will pull them down.

The links above are the 24 and 16bit M/S we did from the lawn at SPAC 07/05/2014. I think it sounds pretty good for what it is and gives you a good idea of the width of the fig 8 in a M/S setup. for being as far back as we were I was surprised by the recording.

[Gutbucket]

Here's something to keep in mind which you guys may want to try sometime with those 414s-

When considering Blumlein crossed coincident 8's (rigged either as X/Y or M/S) in a situation where it would be advantageous to have a bit more forward-bias to the overall combined sensitivity pattern for whatever reason, the 414's option of a supercardioid pattern allows you to use a supercardioid Mid instead of a figure-8 Mid, which decodes to X/Y hypercardioids rather than 8's.  That still provides much of the same Blumlein-like qualities, but with more emphasis on sound from the front and a bit less on sound from the rear.  The overall combined pattern becomes forward facing subcardioid-like in shape rather than fully omnidirectional in the horizontal plane.  The resulting virtual X/Y hypercardioids are 'one-notch' closer to X/Y crossed-8s than setting up as X/Y supercardioids would be.  X/Y supercardioids may also be a good choice, producing a overall combinded pattern which resembles a forward facing cardioid.

Edit- Excluding the slightly different imaging and SRA aspects, choosing between these options is somewhat analogous to adjusting the level of seperate rear-facing microphones by reducing their level in the mix, by way of the choice of pattern beforehand rather than by fine tuning it afterwards with a fader.

[thunderbolt]

Thanks, Lee, for an excellent and detailed explanation that I can wrap my head around.  Makes sense.  And from my one-time experience, I can understand that it can improve recordings every time.  I now understand why you've miniaturized your setup.  Lugging that stuff is a bit of a PIA!  I might even try playing the file through my home theater amp!

[acidjack]

Interesting.   So would one expect the resulting sound to be similar to what it would sound like if you recorded with omnis?

I don't know about that crowd Corbin, I know of one person that goes there every so often and he's a real dick. 

Hey, you better not come 'round 'cause it's MY temple of sound now!

All right now, been kicking around your omni comment for a couple of days:  How does a forward-facing cardioid pair in DIN and a rearward facing supercardioid pair in AB60 = omni?  I'm obviously not good with the maths...so help me understand.

Well, if you have a perfect 360 degree field of sound (well, theoretically perfect; they never really are) you'd have omni. This forward-backward thing gives you a 360 degree field, albeit with differing (and overlapping) stereo images and with the ability to adjust the sound "behind" in a way you could not with an omnidirectional microphone.

In some ways it's a bit of a makeshift version of ambisonic, though an ambisonic can do what it does a bit more scientifically, allowing you to adjust to any pattern (theoretically) within the 360 degree field. But this is kind of the same principle, I think, at least in my layman's mind.

[Gutbucket]

That's basically correct.   However, as a user of an ambisonic microphone, I don't think of this as a makeshift version of an ambisonic recoding system.  Partially because this technique can be superior to an ambisonic mic in many instances.  Both in terms of the qualities of the resuting audio and in terms of flexibity of setup for various recording situations.  Ambisonic recording and manipulation is amazing and a fantastic tool, yet it is always limited to coincident X/Y configurations and doesn't provide have any time-difference information.  You are correct in that ambisonics is more mathematically rigorous and allows adjustment of virtual microphones with precise polar patterns pointed in whatever directions you'd like. That's amazing.  But I find doing it this way can often be more rewarding in terms of the resulting listening experience. It provides other freedoms and advantages.  They're really two very different approaches with different aspects.

To make an analogy to 2-channel stereo recording- this technique is analogous to an ambisonic mic (both are multichannel methods of recording information from all horizontal directions) sort of in the same way that spaced omnis are analogous to X/Y (both are stereo recording configurations), and to continue the analogy, they have about the same degree of differences between them as spaced omnis and X/Y do as a well.

[yltfan]

lot of large stage A-list performers' sound crews place on stage shotguns facing toward the audience to get the same effect you hear with your mk 41 channels.

This is what Stan does for Wilco, and I've always thought the mix sounded phenomenal. Although he doesn't use shotguns, I think they are some Sennheiser cards...

[Gutbucket]

I might even try playing the file through my home theater amp!

Yeah do it!  If only for a quick and dirty test.  I considered suggesting it to you but didn't want to climb onto the soapbox and go OT with my blatherings on surround recording and playback.  I know most people here don't care a whit about that stuff and it usually generates more cricket noises than spirited discussion.

However, this is well suited for direct suround playback using the individual channel RCA outputs on the DR-680 into the DVD player RCA inputs on the home theater amp.  Route the Left and Right mic channels to the R & L speakers, the SBD to the center channel, and the audience facing mics to the surrounds.  The main problem is that you'll have no way to fine tune the critical channel balances unless you can do that on your receiver or put some kind of level controls in between the recorder and receiver.  On mine, the analog DVD inputs bypasses the speaker-setup level and delay settings, but leaves the balance and bass/treble controls functional.  The L/R balance is probably fine as is, it's the balance between the center channel and L/R  which is most critical, then the level of the surrounds compared to the front.  That less critical surround/front balance is like the choice of how much to add to a stereo mix, but you can get away with using more of it for surround playback without problems, and don't need to automate the levels as much for best results. 

I'll sometimes use a couple RCA to stereo mini-plug cables to run through a cheapo headphone variable-attenuator-pot cable for simple level control.  Another trick is to do the same by using a handheld recorder instead of the attenuator cable.  Loop the signal through the handheld recorder, line-in to headphone-out, put the recorder in rec/pause, and adjust input/output gain.  You'll have both variable gain and a short delay line (due to the ADC/DAC latency) which can add a bit of delay to the surround channels (where the exact timing is less critical anyway) or maybe just enough delay to somewhat better align the SBD and the L/R channels, which like the level relationship is a more critical adjustment than the surrounds.

Try it with just the 4 mics and no SBD center channel also, which will probably be less problematic.

If you have a muti-channel soundcard and are playing back the files from your editing software you can adjust everything however you need to of course.

[rocksuitcase]

Here's something to keep in mind which you guys may want to try sometime with those 414s-

When considering Blumlein crossed coincident 8's (rigged either as X/Y or M/S) in a situation where it would be advantageous to have a bit more forward-bias to the overall combined sensitivity pattern for whatever reason, the 414's option of a supercardioid pattern allows you to use a supercardioid Mid instead of a figure-8 Mid, which decodes to X/Y hypercardioids rather than 8's.  That still provides much of the same Blumlein-like qualities, but with more emphasis on sound from the front and a bit less on sound from the rear.  The overall combined pattern becomes forward facing subcardioid-like in shape rather than fully omnidirectional in the horizontal plane.  The resulting virtual X/Y hypercardioids are 'one-notch' closer to X/Y crossed-8s than setting up as X/Y supercardioids would be.  X/Y supercardioids may also be a good choice, producing a overall combinded pattern which resembles a forward facing cardioid.

Edit- Excluding the slightly different imaging and SRA aspects, choosing between these options is somewhat analogous to adjusting the level of seperate rear-facing microphones by reducing their level in the mix, by way of the choice of pattern beforehand rather than by fine tuning it afterwards with a fader.

Lee, I would like to add that we ran the MS with two different patterns between set 1 & 2. Set 1 we had the top mic was facing forward and set to cardiod. the "side" mic was on the bottom facing sideways and set to figure 8. for set 2 the top mic was changed to supercardiod. (we had a 3 taper poll, and airbladder cast the deciding vote to "experiment" vs, make a consistent recording- knowing there were plenty of other tapers onsite)
Two pics:(the bottom mics are AKG C460B with CK8 short shots)

[Gutbucket]

Ahh okay.  I mistakenly misread the posts above and thought you all had setup as M/S with using a figure-8 Mid, which is Blumlein.

[Edited to remove what is better developed and less potentially misleading in my following post below]

If the only difference between sets was the pickup pattern of the Mid 414, that makes for a good comp of those M/S Mid configurations in that particular situation.  Subjectively, what are your impressions of the difference in sound between the two?  I'll try and give it a listen tonight.

[rocksuitcase]

Ahh okay.  I mistakenly misread the posts above and thought you all had setup as M/S with using a figure-8 Mid, which is Blumlein.

With the cardioid Mid, the M/S decode = virtual X/Y supercardioids (ish).
With the supercardioid Mid, the M/S decode = virtual X/Y hypercardioids (ish).
With a fig-8 Mid the M/S decode = virtual X/Y 8's (Blumlein).

Forgive me since you all are probably well aware of that, I just want to clarify for the thread.

If the only difference between sets was the pickup pattern of the Mid 414, that makes for a good comp of those M/S Mid configurations in that particular situation.  Subjectively, what are your impressions of the difference in sound between the two?  I'll try and give it a listen tonight.

IMO, the second set with the sub-card Mid sounds "better" then the first set card Mid. in one sentence, it "feels" smoother with less HIgh Freq annoyances.
Hopfully kindms can add his thoughts here.
BTW, Lee, we were doing all of this with a "What would gutbucket think?" mentality. We were talking stereo zoom and all sorts of things which you have influenced us to consider while doing our recording!
 

[Gutbucket]

Cool, hope that stuff was fun and not too much of a drag.   

Okay, so it was subcard Mid 2nd set, rather than supercard.  Lots of options with 5 switchable polar-pattern choices on the 414's!  I downloaded the files and the txt file confirms subcard 2nd set.

Here's a better general take on the M/S virtual microphone patterns using different Mid patterns than the simplified version I posted above.  A 50/50 M/S decode ratio of a recording made using the following Mid patterns with will produce the following approximate virtual X/Y pairs and – the following approximate overall combined sensitivity patterns: 

Bare with me here, I know most members here understand the virtual pattern and angle relationship with Mid/Side, we’ve had extensive threads on it before.  It’s consideration of the overall combined sensitivity patterns of the pair that I’m really trying to get at here.  Here's the clincher- notice that a subcardioid Mid and a supercardioid Mid will produce aproximately the same overall combined pattern sensitivity!  Even though they obviously produce very different virtual left and right pickup pattern and angles.

> With an omni Mid, the M/S decode = a virtual X/Y cardioid pair with a 180 degree angle between them - combined sensitivity of the pair is omnidirectional.

> With a subcardioid Mid, the M/S decode = a virtual X/Y supercardioid-ish pair at a ~150 degree angle - combined sensitivity of the pair is forward biased and sort of wide-cardioid like in shape.

> With a cardioid Mid, the M/S decode = a virtual X/Y pair of mics with patterns between super and hypercardioid at a ~130 degree angle - combined sensitivity of the pair is forward facing cardioid like.

> With a supercardioid Mid, the M/S decode = a virtual X/Y hypercardioid-ish pair with a ~110 degree angle between them - combined sensitivity of the pair is again forward biased somewhat and sort of wide-cardioid like in shape.

> With a fig-8 Mid the M/S decode = a virtual X/Y pair of figure 8s with a 90 degree angle between them (Blumlein) - combined sensitivity of the pair is back to omnidirectional again.

Your examples are highlighted in bold.  Now this all assumes equal gain settings for both the Mid and Side channels when the recording is made, and a 50/50 matrix ratio when decoding.  If you adjust the recording gains for each channel separately (perhaps recording the Side channel with additional gain so that both M and S channels peak similarly) that will change the ratio just like adjusting the matrix ratio when decoding.  In that case the angles between the virtual X/Y microphones vary and the shapes of the virtual polar patterns do too.  Standard M/S stuff.  Since we usually end up adjusting the M/S ratio while listening, we may end up with somewhat different patterns and angles making the shapes and angles I list above somewhat academic. Still it helps me in trying to relate the differences in virtual patterns with the differences in sound of the samples.

Unlike M/S, Ambisonic decoding and Schoeps double M/S decoding (which is basically horizontal-only first order ambisonics) unlinks the independency of virtual pattern and virtual angle, allowing either pattern or angle to be changed without affecting the other.   When I play around with different ambisonic decodes and can adjust microphone pattern and angle independently, I often find myself homing in on two prefered general settings: Either a crossed supercardioid-like pattern with about a 110 inclusive angle, or a subcardioid pair with an approximate 150 degree inclusive angle.  The imaging between those is very different and the tonality is different, but a certain shared similarlity of a different kind between those two otherwise very different configurations has struck me more than once.  In those situations I could be happy with of those two settings, and not with others. 

I think that relates to a critical importance in considering the overall combined sensitivity of pair.  It's partly why I similary consider not just the level balance between the direct/reverberant sound but the tonality balance and the corellation relationship between the direct and reverberant components the most important aspect of a recording.  I now realize it's also a basic foundational element behind working up the PAS table and revisiting it again earlier this year.

[rocksuitcase]

Cool, hope that stuff was fun and not too much of a drag.   

Are you kidding? It was incredible fun; I don't let these things like recording techniques get in my way of enjoying the music, but I think about them often. 

I get what you are saying about the approximate combined polar patterns of the M-S the way we did it at Phish. And to remind/inform you that it is Phish from the lawn and the recording is quite a true representation of the lawn experience.

What we were discussing early in the first set was to try to recall one-three people who were yelling and screaming and their locations to the array so we might get a sense of the overall "accuracy" of the recording. So when I was listening, there was this one woman who let out the hoopinest, hollerinist scream at the transition of Wingsuit into Piper who was about -10 degrees off axis and in front. I hear her in front on this recording.

we were unable to use equal levels of gain on each channel due to the volume of the crowd noise. for set 1, the S channel was about 1 notch lower (3db?-Tascam HDP-2) for set 2 we were about 2 notches lower on the S channel. kindms had even remembered that I had told him a while ago to have the gain even in both channels, but the crowd was overwhelming the S mic so we had to do it.
Kindms can tell us what ratio he mixed them in as he did the Post production.

As an experiment, it was great, we learned a lot. I"m not sure how much we will use the M-S for audience taping in rock n roll festival type environments.

Tell us what you think.
 
 

[Gutbucket]

Will do. 

I tend to like spaced configs in those kind of environments. Maybe M/S as part of a multi-mic setup included in that, unsually not so much on its own.  But there are always exceptions to the rule, and to self-determinded preferences too.

Heh, I've been known on occasion to walk around the setup between acts at a fest with a dog clicker announcing clock locations both close up and far away, so I could listen back later and assess things.  I think I actually dig the odd looks I get from tapers who are probably rolling their eyes as well as others who have no idea what this odd fellow is up to.  All part of the carnival.  Eh, it's geeky fun.