Topic: Oddball microphone technique (OMT) - part 3

[Gutbucket]

Summarizing-

Pseudo-random phase relationships between channels can be beneficial or not.  With regards to "beneficial or not" some listeners may deem what others hear as being beneficial as being an "phasey mess", while others may deem a more strict phase-relationship as "flat, boring, and lacking a sense of space and dimentionality".  There is room between extremes to find whatever works best and sounds right to you, and there are a number of ways of managing the phase-interactions between channels with this in mind, both by way of the microphone setup and by way of how the resulting channels are treated and mixed.  The goal for me is an optimal combination of both coherent-phase direct-sound imaging and pseudo-random phase ambient-sound immersion.

We work within a large set of constraints.  I've worked to develop my OMT setup in such a was as to try and optimize the phase-interactions between channels via the physical arrangement of all the microphones in the array, which in turn probably gives me a bit more freedom in mixing and not needing to do things like applying low-pass to the omnis in comparison to a setup unable to achieve the same amount of omni spacing that may produce a better end result by low-passing them.  Do whatever sounds most-right given the constraints in which you are working, directed by your own preferences.  I'm not here to say "do it like I do", but rather, "consider these things" in an effort help everyone achieve the results they want.

[rocksuitcase]

https://taperssection.com/index.php?topic=195541.msg2342957#msg2342957

Here is an OMT6 we did last year in an acoustically great room- The Egg in AlbanyNY
https://archive.org/details/jh50f72019-09-24.akgomt
Recording Info:
Source MDAUD:
ch1|2     AKGck22 omni spread 6 feet apart > Grace V2 >Tascam DR680|SD
ch3|4     AKG ck8 shotgun fwd 0'|AKGck61 cardioid rear 180' >Tascam DR680|SD
ch5|6     AKG c426 (Blumlein:fig8, 90) >V3 >HDP2 (24/48) >Tascam DR680|SD
Location: Row O center 70 feet from stacks about 5 feet up.
Transfer: SD >Audacity >CD WAV Editor >TLH >FLAC
Photos of the rig:

How much of the rear mic did you use in the final mix?  Looks like you're near the back wall so I'm curious what it was picking up.

I didn't take the exhaustive notes I usually do during mixdown, but iirc, I kept in the rear channel, but about -18dB down from the rest of the channels.
And in keeping with Lee's commenting on all of our mixing techniques, I can say, by the time I get to choose rear mic or not, or at what level, I've already processed outside channels as a stereo pair, levelling L/R. Then the inside (center/rear channels) get "mixed" in in by mono'ing each center channel (select track, select mono with Audacity) and then levelling it in relationship to the outside channels. In this case, I had three forward facing channels, the ck8 and the 426 fig 8' pair. So I had to "tread lightly" when combining them. On this recording I did not use a LPF, but did use EQ reducing 400Hz on down about 4-5 dB on the omnis.

[kuba e]

Rocksuitcase, thank you for your recording from Egg. It sounds great. Thank you also for sharing you mixing procedure.

Gutbucket, I really like what you wrote about lpf/hpf crossovers. My opinion is the same. I'm glad, now I'm sure I didn't listen in the wrong way, but that it depends on personal preferences and that one way may suit someone and the other way someone else.

I am using Raper and it is easy to setup crossover there. It is possible to link hpf and lpf and changing their crossover frequency by one slider. (If anyone is interested, I can write instructions on how to do it.) So the implementation is not limitation for me. But I just prefer full range mix for the reasons of sound, which Gutbucket explained.

There is something that is between full range and lpf/hpf crossover. It is "tilt eq". You can enhance the bass and suppress the treble on one track (omni) and do the opposite on the other (xy). You can also link two "tilt eq" in Reaper, so it is possible to sweep crossover frequency to see what is happening.
I love this. I don't use it in the final mix, but it is great that it shows how omni and xy contribute to the resulting sound. We just enhance /suppress a low/high of omni/xy. But everything still sounds together. It says a lot more than just turning omni/xy on/off.
It is useful for me because it shows the weaknesses of omni and xy in the final mix. And when I know the weak points, I can fix it by individual standard eq. (If you have no possibility to link a crossover frequency, the crossover can be set before and then only turn the two tilt eq on and off.)

[heathen]

With regards to "beneficial or not" some listeners may deem what others hear as being beneficial as being an "overly phasey mess", while others may deem a more strict phase-relationship as "flat, boring, and lacking a sense of space and dimentionality".  There is room between extremes to find whatever works best and sounds right to you, and there are a number of ways of managing the phase-interactions between channels with this in mind, both by way of the microphone setup and by way of how the resulting channels are treated and mixed. 

In the interest of defining the extremes, what configuration would represent the most strict phase-coherent relationship?  A coincident stereo setup like XY or mid-side?

[Gutbucket]

^ Yes, sampling from only a single point in space using any coincident array.
That includes X/Y, M/S, Blumlein (which is just a sub category of those), Dual M/S, or ambisonic microphones.  They all ideally produce no phase differences between channels because they are derived from a single point in space.

A spaced pair of omnis crossed over to a coincident center pair produces phase differences only below the cross over point, and within the transition region of the crossover.  Below the crossover its the interaction between 2 points in space, in the transition region its the interaction between 3 points in space- the two omnis and the single center position shared by the coincident pair.

[Gutbucket]

There is something that is between full range and lpf/hpf crossover. It is "tilt eq". You can enhance the bass and suppress the treble on one track (omni) and do the opposite on the other (xy). You can also link two "tilt eq" in Reaper, so it is possible to sweep crossover frequency to see what is happening.
I love this. I don't use it in the final mix, but it is great that it shows how omni and xy contribute to the resulting sound. We just enhance /suppress a low/high of omni/xy. But everything still sounds together. It says a lot more than just turning omni/xy on/off.
It is useful for me because it shows the weaknesses of omni and xy in the final mix. And when I know the weak points, I can fix it by individual standard eq. (If you have no possibility to link a crossover frequency, the crossover can be set before and then only turn the two tilt eq on and off.)

I love tilt EQ!  If one thinks of EQ filters behaving more "musically" at low Q-factors, a tilt EQ might be considered a zero-Q filter.  Not sure of the mathematics of that, just extrapolating from the use of peak filters of higher and lower Q.  Tilt EQ is in a sense, sort of an ideal tone control.

The way you describe using complementary tilt filters on the omnis and the center X/Y pair is essentially the creation of a crossover with a transition region that spans the entire audible frequency range.  It would produce the most phase interaction between the two pairs at the center hinge-point of the tilt and gradually reduced interaction above and below that point.  I've not tried this but it appeals to me and sounds like a great application of the tool.

[Gutbucket]

Is it easy to to do this kind of inverse-function linking for other types of filters in Reaper?

I've posted in the past (not sure if in the OMT threads or elsewhere) about my wish for this type of real time inverse-function filtering as a basic functionality for mixing.  Reduce from this channel (or pair or group) whatever I am adding to this other channel (or pair or group) in an energy preserving way with respect to the output.

Applied to EQ- If both channels have equivalent energy across the same frequency ranges, but otherwise a different quality of content, this would allow for a single knob turn in deciding where the perfect balance lies rather than adding some here and cutting some there in an attempt at manually compensating so as to keep the overall output otherwise equal.  Would be especially useful for exploring what more complex EQ compensation curves in the trade off between channels.  I say basic functionality because I would use this for many things, including making level balancing between channel pairs easier.  Take for example deciding on the level of the rear-facing channel or pair in the mix since we were just talking about that.  I'd love to use a single control to essentially cross-fade between all channels without any rear channels and all channels including the rear channels, without the overall output level being changed.  Could be applied to any filtering applied across multiple channels,

[kuba e]

There are two ways to link plugin parameters in Reaper. I use only the simple one. The simple way is that the linked plugins are all in one track. One track can have multiple stereo independent channels. It's the same as multiple tracks, but this way linked plugins are only under one track. I do this by sending stereo tracks omni, xy, rear pair to the main track with 3 independent stereo channels. Then I put required plugins on each stereo channel and link all their parameters. Output from the main track has no limit. You can freely send 3 independent stereo channels where you need. Or you can mix them straight there.
It is also possible to link plugins that are in different tracks (not channels), but it is complicated. To link plugins in one track with multiple channels is easy.

When linking parameters, scale [-100% + 100%] and offset [-100% + 100%] can be set. If you want an inverse link [scale -100%], the parameter should have symmetric range. It is possible to use the offset, but doesn't work for all cases. For example, I was not able to set inverse link for the parameters that have range [-120db + 6db]. For this case, I didn't find solution. I have not yet understood all in detail.

The next nice thing that is possible in Reaper is to link basic control buttons of tracks - solo, mute, volume, pan, width. You can make these links between any tracks, there are no restrictions. You can also make an inverse links e.g. mute/unmute.

So far, I haven't found an easy way to compensate the overall volume when mute/unmute track, e.g. "rear" track. I tried to set the rear track with a compensation track to compensate overall volume that had the opposite polarity than the main track. But it's complicated. For big changes, I had to re-tune the ratio of the rear track and the compensation track. And that was pretty annoying. Sends between tracks were also not clear.
The second solution is to make a compensation track with the same polarity as the main track. Initially, manually set the rear and compensation tracks to the same volume. Then link their volume control and set mute control as invert. When the rear track is switched on, the compensation track is switched off and vice versa. That could work. All that remains is to manage clearly and correctly the sends from the main track to the compensation track and back.

EDIT: I tried both methods. Both methods work well.

I don't know the terminology, I hope it's not confusing. If you are interested, I can find nice tutorial videos about this. Or I can create some templates.

[fireonshakedwnstreet]

Fascinating discussion. I keep coming back to this thread again and again!

Anyway, I implemented the EQ advice that was graciously shared and am linking to an MP3 file of the entire recording: https://drive.google.com/file/d/1sHykMK-WD3l6FKt5TWQyYKeC86ugskUt/view?usp=sharing

EQ was as follows:
Split Omnis - +9b under 250 Hz/+9 db over 8000 Khz
XY Supers - Left channel panned to 30% left/ right channel panned to 70% right
Front/Rear Cards - Front panned slightly right of center

Mix and rendered in Audacity: omnis (-2); supers (-3); Front card (-3); Rear card (-8). Track amplified to -0.1.

Am I on the right track here? Thanks Team OMT

[kuba e]

Thanks for recording. What is good to try is to compare the two version you have done. Load both recordings to Audacity and switch between them when playing. To compare it right it is necessarily to set the levels that both recordings sound equally loud. This is very important because different levels can influence us a lot. What sounds a little louder comes to us as being better.

Audacity is good. But if you like OMT, maybe it is good to try some another DAW. In most DAW, you can set up the operations you did in Audacity, all at once, not step by step. And you can change all parameters during playback, so you immediately hear how it affects the whole mix (or just the individual track, there are a lot of possibilities). And when you're happy with how it all sounds, you can render it into e.g. flac or mp3. This way you can also make different versions and compare them.

[fireonshakedwnstreet]

Thanks kuba. I have played around in Reaper, but it honestly gives me a headache! I have the manual and am going to go through it during what is going to be a long break from taping. Audacity is sloooow but really easy to use.

[Gutbucket]

Kuba e- Good advice above. And thanks for the info on inverse linking parameters in Reaper. I may look into Reaper again just for that. I'm sure much has changed since I looked at an early version.

Fireon..- Will try and listen tonight with better headphones or speakers, but listening on the crappy Samsung in ears I have here at work right now (annoying because they need to be hand-held in place to work decently) and streaming the Morning Dew's through two browser tabs to compare, I hear the image well centered now, but I find myself missing the sense of width the previous version had.  This version seems overly center heavy.  I also preferred the tonality of the original in an overall sense, but difficult to really to judge tonality correctly on these 'phones, especially bass.  Yet in comparison something seems off, sort of like there is more mid-range predominance or a sort of can resonance sound.  Could be a side effect of less width..

XY Supers - Left channel panned to 30% left/ right channel panned to 70% right

^
I'd try keeping these as widely panned as possible. 

I'd try it a couple different ways:
1) Try it without any forward-facing center mic, balancing the image by panning the Left X/Y channel somewhat towards center, but leave the Right channel panned hard Right. 
2) Keep the X/Y pair fully hard-panned Left/Right and use panning of the forward-facing center mic to balance the image. 

Key variations on 2)
a) Try the center mic hard-panned right and slowly raise its level until the image is centered.  You may only need a little bit of it in this way to get the image centered. This is likely to retain maximum image width.
b) If you like more of the forward-facing center mic, increase width of the X/Y pair as described previously using a Mid/Side readjustment or stereo width controller to compensate for using more forward-facing center mic panned less hard to right.

My suspicion is that the variation in bold might work best in this case, and is simple to try. 

Otherwise, adjusting width of the X/Y pair (most likely increasing it significantly) by using a Mid/Side tool or stereo width controller is what I suggest for an X/Y center pair angled so as to Point At Stacks.  Here is the thinking behind that: We are using a coincident pair in the center for all the reason's previously discussed, so that means either an X/Y pair or a M/S pair there.  An X/Y pair in PAS is angled so as to maximize the pickup of direct sound from the PA (a good thing for this pair) by putting both mics directly on-axis with the PA.  However that translates to an overly narrow angle and image width, so we compensate for the narrow microphone angle by increasing the stereo width of the pair during mixing.  Does that make sense?  Using a Mid/Side pair instead of X/Y in the center requires conversion to X/y anyway, so its width adjustment is already a part of the working process.

[fireonshakedwnstreet]

Thanks gut. Will give that a try!

[Gutbucket]

Thinking about that a bit more, I now realize that mixing-wise it might be easier (and perhaps more enlightening) to mix an X/Y pair that needs width readjustment more like a Mid/Side center pair, using 3 mixer channels.

1) Use a Mid/Side tool to convert the PAS X/Y center pair to Mid/Side.  Put Mid on one channel.  Duplicate Side and put it on two channels. 
2) Bring up the omnis alone. Confirm they are relatively well balanced in terms of energy if not in stereo image.
3) Pan the Mid channel to center and bring it up to find a perceptually correct balance between the omnis and center.  Pan the center as necessary to correct any center image imbalance.
4) Pan one of the Side channels hard left.  Pan the other Side channel hard right and invert its polarity. Link the Side channel faders or manually keep them adjusted to the same level.
5) Now bring up the two Side channels together to increase width from the hard center.  Use as much as you need to find an optimal width blend between omnis and center pair. Play around with it.

The cool thing about this is that you balance omnis and mono center first, then add center width as necessary.  It might be easier to find the optimal blend of image width this way.  Its also enlightening to be able to mute the Side channels when you have additional stereo pairs in the mix.  You can also affect balance and centering of the Side information by intentionally not keeping the two Side channels at the same level. Once you find a Side balance that works best, link the Side faders or otherwise keep their relative level relationship the same as you play around with the amount of Side in the mix.

This provides you with two ways of adjusting center image balance- Panning the direct-arriving sound from the Mid channel, and panning the indirect-arriving Side information by way of the relationship between the two Side channel faders. Think of it as panning the center point, adjusting the width out to the left, and adjusting the width out to the right, all separately from each other.


I'm using three stereo pairs plus the center Mid/Side pair.  I typically keep the omnis hard-panned, my forward facing mid-spaced pair hard panned, and often the rear-facing pair hard panned.  Sometimes I pan the rear facing pair more toward center to get a more seamless ambience across the back, but they are at a lower level and that's more of a find tuning thing.  Typically the only thing that's not hard left or right is the center Mid channel. It's very interesting to be able to have the whole mix up and dialed in and mute and unmute the Side channel.  A lot of magic seems to come from the contribution of that figure 8 side channel.  Deriving Side channel from the X/Y pair should do the same, although with a PAS X/Y pair you will probably need increased fader level from it to achieve the same result.  If you need to amplify the Side channel more than the amount of travel your fader gives you that's ok.

Other than potentially using different amounts of ambient Side on the left and right, this is essentially the same as using a stereo width control and balancing the center and omnis with that control set to mono prior to increasing width to whatever works best.

Hope this makes sense.

[kuba e]

I understand you Fireon. I also started with Audacity and then I moved to Reaper. Beginnings are hard because it's a slightly different philosophy than how to work in Audacity. Patience is needed. Maybe do the main things in Audacity and gradually,when you have the time, try some simple things in the Reaper. Small step by small step. I'm sure there are other DAW's, maybe you could find one that would suit you. If you can move from Audacity, it'll make it easier for you to work with OMT in the future.

Gutbucket, if you need anything with the Reaper, I'd be happy to advise on what I know.