Topic: What to do with Mono soundboard used as bulk of mix?

[adrianf74]

Was at a small room show which was a single acoustic guitar with one mic.  Naturally, the soundboard output was mono (it's a small room). Also recorded was room sound but it's very wookified and doesn't add much since there's nothing not being fed to the sound system.

Two separate recorders were used so there will be clock drift. My general thought is to fade the crowd in and out between sons or wherever else it's required.

What to do with the board feed?  I'm going to Deplosive or Declip in iZotope where they occurred (3 or 4 times). Plan to boost any banter between songs but should any simulated stereo or other such effects be added to liven it up?

Thanks everyone.

[Gutbucket]

Inclusion of some room sound will be helpful in placing the performer in the same apparent acoustic space as the audience between songs, and in adding some depth dimension and room to the otherwise dry, upfront and in-your-face monophonic soundboard recording.  You may want to ride the AUD room pair levels if there is a lot of distracting chatter during the performance sections.  But this is probably what you've already planned on doing.

It may help to add a touch of artificial stereo room reverb to the soundboard.  That can help with the blend and make riding of the AUD level less obvious.  And if the wookieness is just too much that might end up doing most of the work in getting the SBD to sound like it is more naturally "in and of the room".  The trick will be to find the right room reverb algorithm and settings to get the artificial 'verb sounding similar to that of a performance in that actual room without the audience noise.  Even if you nail that really well, I'd probably not mute the AUD entirely but just reduce it a lot in level during the music, since it will add a subtle human liveness that even a perfectly tuned and great sounding room reverb will not.

Enjoy the artistic/creative side of taping.

[rocksuitcase]

Everyone has varied paths here, but I often do this with a mono SBD/AUD mic mix:
keep the SBD mono and add it in "below" the level of the AUD, almost as a centering filler.
In Audacity, I will ensure the LT/RT tracks are 'mono'd, then import the AUD track, adjusting each pair's levels to be similar.
Playing around with the relationship of SBD/mono vs AUD/stereo to suit your ears. (Generally, it comes out like 25-30% sbd in these cases) ymmv
Than I will export the multi-channel mix to a 2 channel mix and do EQ, Compression, limiting, levelling.

Others should have varying ideas.

[Gutbucket]

I'm 100% onboard with Kyle's methodology regarding the majority of AUD/SBD mixes - That is, using the AUD as the primary source and adding just enough SBD, at a significantly lower level, to sweeten vocals and sense of presence as necessary, generally without otherwise processing the mono SBD.  To my taste, many AUD/SBD mixes tend to lean too heavily on the SBD and don't sound as natural as they could/should with a bit more room in there.  That's probably because a taper can't help but focus on the annoying audience chatter portion of a recording (which easily leads to overcompensation with too much SBD), while the same is easier to ignore by other listeners who tend to be more focused on the music rather than the more technical aspects of the recording.

However, for a recording where the AUD portion contains mostly cocktail party chatter and very little sound of primary interest, you'll need way more SBD relative to AUD.  That's when it might help to add a touch of room 'verb to the SBD portion to keep it from sounding flat, sterile, and overly close.  Will generally be best to add the 'verb to the SBD part prior to mixing that with the AUD.. but you can also add most of it SBD and a touch to the overall mix of the two to help blend the two sources and hide the level riding of the AUD a bit better.  Tread lightly though, and double check with a fresh listen later.  The added reverb shouldn't be obvious except in its absence.  It's easy to go overboard.

Even if the AUD is almost all chatter, I think you should keep some small amount of it in there as mentioned previously to improve continuity.  Think Marvin Gaye, What's Goin' On>What's Happening, Brother

https://youtu.be/Y9KC7uhMY9s
^Not the LP version with all the cool background party chatter I intended to link, but an outstanding extended live version with James Jamerson sitting next to Marvin!

[nulldogmas]

Rent out the venue, fill it with silent people, play the SBD through the PA, and use mics to record that. Problem solved.

[rocksuitcase]

Sorry:

it's very wookified

I missed that. As with all things, ymmv. I'd now go with the added reverb thing or possibly a pseudo stereo mix.

[adrianf74]

Thanks Rocksuitcase and, as always, Gutbucket.

This is a very difficult one to wrap my head around.  I'm usually VERY good at pretty much knowing the exact direction I'm going to take with the given elements but appreciate the feedback from both of you.

What Rocksuitcase said is one of the ways I usually go about this.  Depending on the level of wookiness at a show, the amount of soundboard to audience can vary but I usually try and keep it no more than 50/50.  There are times where I've had to almost take out the audience because of unruliness -- this happened to me once at a show by a really well-known artist who allows me to record his shows with the understanding that they don't get released.  During one track, I had to fade the audience to almost nothing in order to get rid of the distraction.  I ended up using some generic audience noise I had to "fill the hole" but it wasn't ideal.  Great show, great performance, great mix (properly mixed for playback rather than what the venue hears), but the ass-hat killed it.

This show I'm working on is most different than anything I've dealt with before.  The two recorders being used (both Sony but an (1) A and (1) M) will have to be brought closer together as there is drift.  Then there's the other issue at hand which is a guy yelling off his rocker sitting about 10 feet next to the audience mics.  There were other times where him and his group were chatting incessantly.  This show as a very intimate setting with about 50 people in a small room where the stage was maybe 30-40 feet from the doors.  The stage mic picks this guy up, even, albeit relatively quiet compared to the artist. 

So, this is more the case of the cocktail party chatter if it weren't for this guy.  Reverb was added to the vocals, naturally, so maybe adding a little more to the overall mix will help.  The point of what you raise with the Marvin clip is pretty neat.  I'm used to the "chatter version" from the LP but what I'm trying to avoid with the final mix here is something that doesn't sound like I'm removing parts of the audience mix intentionally which will sound the case if I drop it out completely at points.  I know I have my work cut out on this.

And before I forget, Nulldogmas' suggestion was perfect -- if only that were a possibility.  People in my town are, generally, pretty far from silent.

Thanks everyone.

[relefunt]

Is there a DeWooker plugin for iZotope?

[adrianf74]

Is there a DeWooker plugin for izotope?

If only...

[hoserama]

Here's how I would attack it:

1. Run the soundboard through the music rebalancer in Izotope RX, split to vocals + everything else. Mix them as independent stems
2. Align the audience recording to the soundboard
3. Clean up the AUD in Izotope RX...zap particularly intrusive crowd noise
4. Load everything up in a DAW. Mix to flavor.
5. For SBD vocals extraction, likely bit of EQ + limiter + reverb
6. For SBD guitar/other, likely bit of EQ + spatializer for depth
7. Get a good sounding soundboard feel, then start mixing in the aud
8. Likely roll off a bit of low-end from the AUD to remove mud but depends on how the AUD sounds. I would keep the AUD at a basic nominal level so it sits a bit underneath the sbd. Then automate the AUD higher between songs. You could do a sidechain compression and tie it to the vocals so you get a few DB of ducking whenever there's vocal chatter between songs.

[nassau73]

You also might try playing around with the Adobe Audition 3.0 plugin - Binaural Audio Panner

[checht]

Concur with Hoserama. Just gotta resign yourself to spending quite a while in spectral repair.

Appreciate the distinction between our tendency to focus on things in the recording that are distractions vs everyone else who listens past that stuff and enjoys the recording.

I have developed less tolerance for wookiness and that impacts a lot of things. Cap selection, post production workflow, general grumpiness.

Also, current crowds are different. When I listen to a mid 80's recording fob dfc 'section', it's almost devoid of wook.

Especially outdoors, just so sweet. Greeks, Frost, Rocks. Sigh.

[wforwumbo]

Well the good news is, you have multiple stems to play with. Having your stereo mic’d room pair plus the SBD means you’ve got a better chance of getting the end product you want.

I would suggest you first consider and concretely define what it is you want to accomplish. I can’t tell you what sounds best, only you can. And I can’t tell you what the “right” way to do something is, either - how much adjustment to make to levels, plug-ins, etc. I further could not tell you what the preferred workload is. All I, or any of us, can suggest is what works best for us. To wit, I encourage you to experiment with as many techniques as possible til you are happy.

What I CAN suggest, is insight into what works for me w/r/t the process at large.

Generally, the question I ask is “what did I think the music sounded like in the moment, and given all recordings I have on hand what is the best way for me to let my recording get out of the way between me and enjoying the music?”

Do you think the aud generally sounds great, but wish you could add a touch of definition or clarity? That’s where Kyle’s method *shines* - get your room pair generally sounding balanced to taste, then use the SBD at very low levels (I generally set the SBD 20-30 dB below my mains pairs when I do this).

It’s worth identifying the distinction between what I will call the “dry” mono sound board and your stereo AUD mics. Your AUD has something the sound board does not - a room. Reflections massively influence your perception of a space, and those reflections do a lot of things to help you localize a mono sound source. Your mono SBD is being pumped through a set of filters - an amplifier, then a speaker, then the greater room impacts. Most of those can be generally compensated (the amp + speaker), but a room is a different beast that’s crazy hard to model. This was actually the origins of reverb algorithms in the mid-late 70s: David Griesinger was a recording engineer and had stereo mains of a piano/violin/cello trio, as well as close mics of all 3. He loved the mains but didn’t hear enough cello; so he went to add the close mic cello, but it didn’t blend with the other instruments in the mains - it was too immediate and direct. He also happened to be incredibly intelligent and the senior engineer at a company called Lexicon, so he researched and created the first concert hall algorithms in DSP. All this to say, Gutbucket’s suggestion of adding a room reverb to the mono sound board is a great way to “stereoize/spatialize” a dry mono SBD recording, especially one with so few musical elements likely played in dual mono over two PAs in a room. If you configure the reverb algorithm right (“size” and “decay” controls), you should be able to get both pretty close.

If you think the SBD sounds great but could use more “immersion” or “space” - the mythical realism of being in the crowd - you could set the dry mono SBD board on its own, and then slowly and quietly bring up the AUD underneath it until the blend sounds good. This is treating the AUD as your “reverberation algorithm”, and if you can get the blend of wet/dry right you can find some good results there too.

iZotope can be a magical tool, but I find it generally ends up being used to polish a less than stellar recording. I’ve heard a lot of tapers here use it to great effect, and I come from a studio production world where it does wonders for restoration work especially, but the DSP engineer in me cries every time I think about what iZotope does to the phase response of any recording. This to say, it’s not a tool I find comfortable or intuitive to use for my own work, but that’s not a commentary on if it’s a good tool or if it’s the right tool for you.

Avoid a binaural audio planner. All it’ll do is induce a head related transfer function, which artificially adds ITDs/ILDs (directional cues based on interaural time and level differences) that don’t exist in the recording and cannot be undone losslessly. And additionally, it adds those cues on top of your headphones or stereo, which themselves induce even more cues for your brain. I’d also avoid any ambisonic encoders, on a mono source it’s just fancifying mid-side encoding that will permanently distort your phase response.

To the meta-commentary on trying to clean up a recording in general… I’ll comment that this is why the most important microphone in my locker is my hypercardioids. I don’t love how they sound, and while I can enjoy recordings made with them I find myself not loving them the way I do my other microphones. But the hypers are a tool - if I’m in a chatty crowd, or a less than ideal location, I know they’ll always make a listenable recording with minimal possible wookery, and if I deploy them right I can do away with “room boom” in the bass and “haze” that too strong of reflections from the back of a hockey rink will bring me. I think my cardioids make much better recordings in general that I consistently truly love, and my subcards in the right spot sound sublime. But I reach for my hypers first, they’re what I prioritize running if I can only have one pair in a compromised spot, and I am always happy I have *a* clean recording when my other mics turn out too chatty. All this to say, I encourage you to prioritize grabbing a pair of hypers.

Addendum edit: it is crucial you get all your recordings time aligned if you’re going to mix them down. You will be in phase hell if you do not. And if that’s too much work, it might be worthwhile just to post the sbd and aud as individual releases, allowing your listeners to choose which recording they’d rather hear - it prevents any permanent phase destruction in either recording.

[morst]

Addendum edit: it is crucial you get all your recordings time aligned if you’re going to mix them down. You will be in phase hell if you do not. And if that’s too much work, it might be worthwhile just to post the sbd and aud as individual releases, allowing your listeners to choose which recording they’d rather hear - it prevents any permanent phase destruction in either recording.

I agree with the first part of this, but experientially disagree with the second part.
Yes, it's very important to get the clocks synced up from separate recorders, and it's very important to control the delay between stage mic (SBD) sources and distant mics (out at FoH for instance.)
But I feel that you just have to get them close, and don't be afraid of phase hell.
Maybe I'm just too deaf about 10kHz to realize that I'm putting listeners through hell, but I have not been getting complaints.


In my experience, trying to get two sources with separate clocks to line up perfectly is nearly impossible with temperature and pressure changes let alone the elusive clock jitter!?
I line up the start and finish of my files, interpolate, and let the chips fall where they may in between those two points.
I do double check that the middle sections are reasonable close, as something like a data gap can throw things WAY off. 
Even for tracks recorded on the same clock, there are still variations to the distance correction based on atmospheric conditions...

Yes, theoretically, you will get all kinds of weird artifacts like comb filtering if you have mismatched phase, but it's not something I can hear.
I bet my buddy digitalist0ne would disagree emphatically with me, but I can't hear what he is hearing.

[kuba e]

I think Wforwumbo meant to align the recordings within a certain range. If I remember correctly, our hearing can tolerate a shift of up to about 20-25 ms. Our brain does not distinguish it and takes it as one sound. When the shift is greater, our brain processes it as two separate sounds. When we are out of this range, the matrix will sound unphased. I don't think there will be significant comb filtering. Sbd and aud are different enough signals. But for our hearing it will be confusing. It will sound like an echo, losing clarity and precision.

In my experience, when sbd is slightly shifted forward in front of the aud, the sbd is more pronounced in the mix and the aud is more in the background. And conversely. It's like changing the levels of sbd and aud, but it has a slightly different sound effect. These are delicate things. When making the matrix, I proceed as follows. First I set the correct alignment of sbd and aud, as nicely described by Morst. Then I set the optimal levels of sbd and aud. And finally, I try a slight shift forward and backward, eg +-5 ms. Alternatively, I can fine-retune the levels in the end.

[wforwumbo]

Aligning two source recordings is also important if you plan on processing the channels independently - equalization being an important one. If your recordings are not aligned, and you try to eq anything (either independent sources or the master) you’ll be constantly fighting an uphill battle. Sure many DAWs have plug in group delay compensation but that’s still fighting a losing battle if your sources aren’t aligned to start.

For most of the other points, phase delays are not always constant across frequency. Sure there is some wiggle room from environmental factors, but they’re not linear across frequency and our brains have ways of processing and filtering the information posthaste. The digital processing side of things offers a layer of control that benefits greatly from phase alignment.

It’s not just about combing, it’s also about a psychoacoustic phenomenon called…

… our hearing can tolerate a shift of up to about 20-25 ms. Our brain does not distinguish it and takes it as one sound. When the shift is greater, our brain processes it as two separate sounds. When we are out of this range, the matrix will sound unphased. I don't think there will be significant comb filtering. Sbd and aud are different enough signals. But for our hearing it will be confusing. It will sound like an echo, losing clarity and precision.

In my experience, when sbd is slightly shifted forward in front of the aud, the sbd is more pronounced in the mix and the aud is more in the background. And conversely. It's like changing the levels of sbd and aud, but it has a slightly different sound effect. These are delicate things

… the precedence effect. Also called the law of first wavefront, the way your brain fuses auditory information from delayed signals can get very confusing. What’s posted in the quote is a hand wave, but a very practical manifestation of the precedence effect.

It honestly is not too hard to get your sources lined up, even if you don’t feel your ears are dialed in. Most DAWs come with a sample delay and a corellation meter (and some even come with gonioneters), this process takes maybe all of 5 minutes once you become comfortable with the tools and your ears learn to adjust and identify the shifts.

[kuba e]

I remember doing one recording sbd + aud (hyper, din, in the middle of the room). I aligned the first version exactly according to the waveform and my listening. In the second version, I delayed sbd by 10 ms (maybe a little less, I don't remember it well). My friend, who has perfect hearing and good monitors, confirmed that the second version is better. The sbd didn't sound good, and by delaying it, it fell into the aud. It was not a significant change but it was positive. And when I tried to just turn down the levels of the sbd, it didn't have the same good effect.

I think the same as Morst, there is a certain possible range for syncing sbd and aud. From what I was listening, most of the matrixes came to me well done. I've heard a bad sync on a few recordings. Perhaps some instruments may be sensitive to just minimal unsync, such as cymbals? I'm not sensitive to it and I hope most people will overlook it as well. In addition, much can be forgiven in a live recording if it does not significantly interfere.

I also think comb filtering is negligible when mixing aud and sbd. Theoreticaly, if we mix onstage mics 1m spacing and sbd, we would get into problems for 350hz (wavelength 1m) and higher. But I have not encountered these problems. I think sbd and aud are enough different signals. Maybe I got it wrong. Why should there be a problem with equalization?

[Gutbucket]

[snip] ..These are delicate things. When making the matrix, I proceed as follows. First I set the correct alignment of sbd and aud, as nicely described by Morst. Then I set the optimal levels of sbd and aud. And finally, I try a slight shift forward and backward, eg +-5 ms. Alternatively, I can fine-retune the levels in the end.

^
In my experience, this is a good practice for making a AUD/SBD mix.  In the end, trust your ears.. just be aware that your brain (to which your ears are directly attached, and the filter to rule all filters) can play tricks on you when listening closely in detail for extended periods.  Come back and revisit it after lunch or whatever with fresh ears and give the first few moments of listening the most weight in deciding if it sounds right or not, before you unconsciously adapt to the recording over the course of a minute or two.

The goal is creating the most engaging recording for the majority of listeners, both casual and more attentively serious.

I see making the independent sources available valuable as a related but essentially different mission.  The goals there being preservation for posterity, tapers listening critically to the individual components, and folks who may wish to play around with making their own mix.

[Gutbucket]

Semi-related OT comment on source alignment-  Years ago I developed something of an ear for alignment by way of simultaneous playback of two sources directly from the (separate) digital recorders on which they were recorded rather than from within DAW software, aligning the two audibly by way of quick jabs of the play/pause button on the leading source.  It took a lot of trial and error, familiarity with the equipment, and careful listening to get it right.  It also made me acutely sensitive to slight misalignment either way.  Making it more challenging, most of the time the two sources were two AUD microphone pairs, sounding more similar to each other than AUD and SBD. This unintentionally ended up being something of a party trick when friends were present, as they'd immediately hear when then alignment was correct, yet when it was off they had only a vague idea of by how much, and no idea of which source was leading. 

Interestingly, once aligned in this manor, the recording would remain closely aligned for the entire length of the file, despite rather actually because of the difference between the two clocks.  Playing back using the same machines on which the recordings were originally made eliminated the differential between the two clocks and the need to stretch or shrink one source to match the other when using a common clock reference.  The clock in each recorder only needed to be highly accurate with regards to itself - playing back at the same imperfect rate at which it recorded.  Or more precisely, both clocks needed only to be "true to itself" and have similar error related to temperature differences and whatever other outside influences impacted them between the recording and playback events. 

Doing the same with three sources was a much greater challenge that I attempted only rarely.  Way too much of a PITA!  In that case I only got one chance to get the third source aligned perfectly before having to pause and realign the other two sources again prior to a second attempt.

All of that gave me a good feel for how close was really close enough in terms of perception (in contrast to phase precision in regards to processing) and the perceptual effect of miniscule delay differences.

[adrianf74]

Thanks everyone for your help with this.

I was able to get my way through all of it to a mix that seems pretty decent all-around with a fake stereo soundboard track that doesn't sound too fake. 

Possibly the only other issue I'm dealing with is an overly bright acoustic guitar (it was the nature of the guitar being used in the set).  Curious if anybody has any suggests in the iZotope world to help dull this without affecting vocals or is this simply a matter of it's better to stick with what you've got.

Again, thanks everybody for your feedback and thoughts on this.  It was extremely helpful.

[checht]

Thanks everyone for your help with this.

I was able to get my way through all of it to a mix that seems pretty decent all-around with a fake stereo soundboard track that doesn't sound too fake. 

Possibly the only other issue I'm dealing with is an overly bright acoustic guitar (it was the nature of the guitar being used in the set).  Curious if anybody has any suggests in the iZotope world to help dull this without affecting vocals or is this simply a matter of it's better to stick with what you've got.

Again, thanks everybody for your feedback and thoughts on this.  It was extremely helpful.

This is a job for music rebalance!

Use it to separate stems, then work your eq on just the guitar, then mix back together.

[kuba e]

Thank you Gutbucket for a nice explanation. It's very interesting how to train syncing. I will try it in DAW, I will cover the waveforms. I'll try it on audience recording that I recorded with two pairs of microphones. And then I can check how far I'm from the perfect alignment.

I totally agree with your description of the purpose of making final recordings. I belong to the first group. It would have to be a really big exception for me to try to mix it myself. But Taperssection has a big advantage. Those who are experienced and belong to the second group can write to the taper about the original tracks. And then it can be interesting to compare two different mixes.

Adrianf74, how did you create fake stereo soundboard track? I once successfully used IR (impulse response) reverb on a dry sbd. Lots of IR types are available for free download on the Internet. For example, simulations of rooms of different sizes, halls, etc. You can also play with the reverb nicely and tune everything.

[morst]

...
Interestingly, once aligned in this manner, the recording would remain closely aligned for the entire length of the file, despite rather actually because of the difference between the two clocks.  Playing back using the same machines on which the recordings were originally made eliminated the differential between the two clocks and the need to stretch or shrink one source to match the other when using a common clock reference.  The clock in each recorder only needed to be highly accurate with regards to itself - playing back at the same imperfect rate at which it recorded.  Or more precisely, both clocks needed only to be "true to itself" and have similar error related to temperature differences and whatever other outside influences impacted them between the recording and playback events. 

BUMP!
Was looking to see if I had commented on mono SBD + mono AUD mixes anywhere on this site and got a nice throwback to a KILLER Gutbucket post.
I love this.

So true and still has potentially handy implications...

Sync issues between two separate clocks?

TRY THE ANALOG HOLE!

Just play 'em back on the original clocks, and redigitize them with a common clock!
Shift one left < - > right to account for randomized playback start times, and Phase is Phun!

OK
The reason I was looking for what I already said, was to see if I ever mentioned my mono SBD + mono AUD mix that I made of Dude Of Life, and which I plan to use on a lot more cassettes with this source config...
Problem: cassette does not have really tight speed control, so sync of cassette source even to a digital one is very difficult. Cassette to cassette far more so, no doubt.
But I have all these Lch SBD + Rch AUD master cassettes to mix down....

Check what I did with the Dude.

Mono SBD is center channel.
I made a SIDE channel out of the mono AUD by copying the channel then inverting the one I panned RIGHT, while panning the original AUD channel to the LEFT.

Now it's mid/side SBD/AUD!

Shift left < - > right to match, mix evenly, and when you play it in mono, it's ONLY the SBD!!!

https://archive.org/details/DudeOfLife1995-04-03


tag: #Reversible

[Gutbucket]

[snip..]I have all these Lch SBD + Rch AUD master cassettes to mix down....

Check what I did with the Dude.

Mono SBD is center channel.
I made a SIDE channel out of the mono AUD by copying the channel then inverting the one I panned RIGHT, while panning the original AUD channel to the LEFT.

Now it's mid/side SBD/AUD!

Shift left < - > right to match, mix evenly, and when you play it in mono, it's ONLY the SBD!!!

https://archive.org/details/DudeOfLife1995-04-03


tag: #Reversible

Right on. Looking forward to giving that a listen. 

I do this alot with the center and back channels of my four channel LRCB (left/right/center/back) recordings.  Works really well.  On a few of them where the L/R pair is missing or unusable, it's exactly the same as you are doing except I'm assigning my Center forward-facing mono mic channel > Mid, and my Back rear-facing mono mic channel > Side.

When I have the full 4 channel LRCB file set available I basically do the same. In that case the L/R channels get hard-pan assigned L/R as usual. The C/B channels get run through the Mid/Side decoder with the Center channel routed to Mid and the Back channel routed to Side.  That automatically assigns the Center channel equally to Left and Right, and adjustment of the M/S ratio provides control over how much "spatialized" rear-facing room/audience gets included. 100% Mid = L/C/R stereo with no Rear facing content in the mix.  Changing the Mid/Side ratio to include more Side channel brings the rear-facing room and audience ambience in.  Folding the L/R stereo mix down to mono cancels out the rear-facing stereoized ambience - upon mono playback the mix is automatically drier consisting of just L/C/R without any B.

That's all easily done on the computer of course, but the cool thing is that it also makes for quick and easy non-computer playback of my raw LRCB recordings.  First step is renaming the two stereo files originally recorded on the DR2d's SD card to the name structure recognized by the R44 when backing it up.  I then pop the card into the R44 and and use it's rudimentary built in mixer and Mid/Side controls, first getting the L/R pair balanced on its own with the Center/Back channels muted, then engaging the Mid/Side decoder on the C/B pair with it set initially to 100% Mid and bringing up the level of the Center channel to find the best balance of it with L/R, getting a nice even L/C/R mix. I then play with the M/S ratio control to add just the right amount of stereoized rear-facing content from the Back channel as special sauce.

Works really well.

[Gutbucket]

^ Getting deeper in the weeds of it (read on if interested, please skip if too arcane!)-

A few years back in a couple other threads I speculated about using about a different method of deriving a Side channel signal from the mono Back mic channel.  Here's the deal-  I'm using four baffled omnis so all four channels have extended low frequency content in them.  The L/C/R channels are all mixed together with the same polarity, so no problem there.  But after going through a standard Mid/Side matrix, the Back channel gets added to the L mix bus in positive polarity reinforcing the shared LF content in L, while being added to the R mix bus in inverted polarity attenuating shared LF content in R.  The idea is to avoid asymmetrical reinforcement of low frequency content in the Left channel and attenuation of low frequency content in the Right channel as Back channel content is added to the mix.

Morst, I suspect this will not be a problem at all for what you are doing, since your mono SBD and AUD channels are likely not phase correlated sufficiently for that kind of asymmetrical reinforcement/attenuation to occur.  I also don't think it will be a problem with my LCRB recordings at mid and higher frequencies due to the mics being spaced apart somewhat, baffle-mounted and facing different directions.  Sure enough when doing these quick mix playbacks on the R44 I don't really hear it as a problem in the low frequencies where it theoretically would be most noticeable.  However most of the content I'm doing this with is classical, jazz and other primarily acoustic material. No significant low subwoofer content.

So in my speculations about what might be the best strategy for optimally mixing this material down to L/R stereo, I've thought and played around a little bit with using a phase-rotation tool in place of simple polarity inversion as a way to create the Side signal.  The idea being to rotate phase equally across all frequencies by some positive phase angle for the Back channel content that gets mixed into the Left mix bus and by some negative phase angle for the Back channel content mixed into the Right mix bus.  A 90-degree phase rotation each way creates a 180-degree total difference like a simple polarity inversion but the phase shift is now symmetrical with respect to L/R, eliminating any tendency of the low bass to pull toward the Left as increasing amounts of Back channel is added to the mix.  Taking it further, different degrees of phase rotation might be used, with some interesting implications.  A symmetrical phase rotation of less than +/-90 degrees will not only alter the perceived width (in a similar but different way to adjusting the Side ratio), but also the degree to which that rear-facing content cancels out when L and R are summed to mono - the ambient rear content is no longer totally canceled out completely. Applying no phase rotation at all simply mixes the Back channel into L and R equally as a mono center-panned source.  Which isn't bad, but not quite as engaging as stereoizing it in an appropriate way.  Can also play around with the time alignment in addition to the phase rotation..

More arcanity-
This polarity/phase-rotation stuff has implications when playing back using matrix surround techniques, from simple Hafler sum/difference setups, to the various Dolby ProLogic incarnations, DTX matrix modes, Logic 7, etc..  With your SBD/AUD Mid/Side mixes, the SBD channel will be reproduced as Center content and the AUD as Surround content.  Depending on the matrix and speakers the playback output will probably wrap all the way around the room.  With a simple Hafler setup the front channels will reproduce the full stereo mix and the rear channel or channels just the AUD.  If present, a physical center speaker will reproduce just the SBD.  Altering the total degree of symmetrical phase rotation will change the surround chanel vector effecting how much your AUD content, or in my case Back channel, is mixed into the side and front playback channels.

[morst]

BRAVO!

The idea being to rotate phase equally across all frequencies by some positive phase angle for the Back channel content that gets mixed into the Left mix bus and by some negative phase angle for the Back channel content mixed into the Right mix bus.  A 90-degree phase rotation each way creates a 180-degree total difference like a simple polarity inversion but the phase shift is now symmetrical with respect to L/R, eliminating any tendency of the low bass to pull toward the Left as increasing amounts of Back channel is added to the mix.

[Gutbucket]

Found the thread where we were talking about this last time a few years back.

Here's the post where I link a few examples of using phase-rotation trick instead of polarity flip to create the Side channel.. in that case as a way of modifying the image distribution of a recording where the drums were all the way over to one side and the guitar/bass over on the other.  It's just an few quick and dirty examples made using headphones.  Never listened over speakers.
Re: Have you ever taped in front of a delay stack before?

If you back up from the linked post a bit it reads as pretty much a repeat of my posts about it here in this thread.  You mention that Dude Of Life recording too. 

I never did get around to making a dedicated thread for the discussion of this stuff.

[morst]

Good find, going back two years or so there.
I forgot about that moe. tape of stage + stack!


Anyone who wants to mess with this technique could remix it from my posted source!
https://archive.org/details/moe1997-03-15-flac16
 

[Gutbucket]

[snip..]I have all these Lch SBD + Rch AUD master cassettes to mix down....

Check what I did with the Dude.

Mono SBD is center channel.
I made a SIDE channel out of the mono AUD by copying the channel then inverting the one I panned RIGHT, while panning the original AUD channel to the LEFT.

Now it's mid/side SBD/AUD!

Shift left < - > right to match, mix evenly, and when you play it in mono, it's ONLY the SBD!!!

https://archive.org/details/DudeOfLife1995-04-03


tag: #Reversible

Right on. Looking forward to giving that a listen.

Got a chance to listen to that last night.  Sounds good to me.  Also gave a listen to the Moe recording and I think the same process would benefit that one similarly, primarily by bringing the vocals to the center.

To double check myself I also went back and listened to the Charlie Hunter examples I did a while back that I linked above, which have varying degrees of phase rotation applied to the channel assigned to Side, and I'm pleased to find I remain in agreement with my original assessment of them.  I specifically chose that CH recording because the image distribution in the original is so extreme (not wrong as is, just extreme).  The examples I added later which use less than 180 degrees of total phase rotation, particularly the ones in which the phase rotation is applied somewhat asymmetrically on each side (more rotation used on one side than the other) are the ones I like best, because that allowed me to position the drums, bass, guitar, musicians voices and audience reaction in a way I found most pleasing, determined by ear.  They aren't easily "reversible" in the same way though.

Somewhat unlike morst's recordings which were made with one mic pointed at the PA stack and the other at the stage and therefore are not an attempt to convey some kind of "accurate documentation" of what it sounded like from the audience, the original version of the CH recording is actually the more accurate reflection of what would have been heard by someone standing at the recording position. My altered versions are an example of a way to provide something of a more balanced listening experience.. one which seeks to be more pleasing and relatable to listeners that weren't present at that live event, or who maybe were present but were positioned farther back in the audience.  In that way I fully recognize that what I'm doing actually represents a deliberate distortion of the stereo image away from "accurate" toward "pleasing", and I'm totally okay with that!  Other tapers who are motivated more by the attempt of "accurately documenting how it sounded in the room" are free to disagree of course.  Personally I think "pleasing" is a both a more reasonable and a more successfully achievable goal than "accurate", but that's a philosophical discussion. ; )

[Gutbucket]

I remember doing one recording sbd + aud (hyper, din, in the middle of the room). I aligned the first version exactly according to the waveform and my listening. In the second version, I delayed sbd by 10 ms (maybe a little less, I don't remember it well). My friend, who has perfect hearing and good monitors, confirmed that the second version is better. The sbd didn't sound good, and by delaying it, it fell into the aud. It was not a significant change but it was positive. And when I tried to just turn down the levels of the sbd, it didn't have the same good effect. [..snip]

Was going back through this thread and realized that I'd intended, but never actually got around to replying to this post way back when.

Here's something that I think provides insight into the observation made by kuba e above, has informed how I think about all this, and which I believe may be relevant to mixing AUD and SBD..

Check out the paper by Gunter Theil that is attached to the bottom of this post titled:  "Room Related Balancing Technique".  It discusses an interesting technique for mixing orchestral section spot mics in with a main stereo pair microphone feed.  When reading it, substitute a typical AUD recording for the orchestral main stereo pair of microphones and a SBD feed for the spot mics.  After all, a SBD feed is essentially a mix of a bunch of spot mics.

This technique treats the main pair (AUD) as the primary source, reinforced by the spot mics (SBD) as secondary source, used at a lower level.  Although that's the opposite of the original intent of this thread, doing it that way around is my preference whenever possible rather than to treating the dry SBD as primary and using the AUD to "spatialize it" with the addition of some room and audience sound, as that can sound upfront and clear, but almost always a bit flat, artificial, and not natural souding me.  Useful when necessary, but I prefer to preserve the depth, imaging, and naturalness of a good main stereo pair or mic array when possible. More on the fundamental difference in approaches in wforwumbo's post earlier in the thread- https://taperssection.com/index.php?topic=198253.msg2367215#msg2367215

The goal of Room Related Balancing Technique is to preserve imaging and depth cues inherent in a good main stereo AUD pair by having that content arrive first, providing first arrival information, and treat and the spot microphone signals (SBD) so that they act like early reflections reinforcing the main pair signal.  To do so, the dry spot-mics are first fully aligned with the main pair (or calculations made to figure out the delays necessary to bring both into alignment), then the spot mic signals are muti-tap delayed by a few milliseconds with respect to the main pair.  Each delayed copy is panned to a random position so that they collectively emulate the ever-so-slightly delayed arrival of early reflections from random directions.  The spot mic signals are essentially fit in between the initial arrival of the main pair and the reverb tail of the main pair as would be discrete early reflections.  In that way they provide the desired increased clarity and bring up the level of content not well enough represented in the main pair alone, but without screwing up the initial arrival information from main pair that provides good stereo imaging and depth cues.

In application to what we are doing, we'd not mulit-tap delay the SBD feed and randomly pan those delayed copies around in the same way, although we might do that if working with the individual SBD channels.  Instead we'd most likely do exactly what kuba e describes - delay the SBD just enough that it is a few milliseconds behind the AUD.  AUD pings our awareness first, then SBD comes in and reinforces the AUD withing the precedence-effect perceptual window, inhabiting a Goldilocks zone by not arriving too early so as to obscure the main mic cues, not or too late to blur or become echoic.

Cool stuff and a good read-