How to Fix Latency with Dual Boom Mics

A small correction. I fell into the trap also by implying that the phase invert shifts the phase. It doesn't in inverts the polarity, it doesn't shift the phase.
 
For meetings, live broadcasts, or multi person interviews, I find the Dugan Automix on the Sound Devices to be extremely beneficial. It will not cut out "mistakes" in the recording like clothing rubs or coughs or clapping or whatever, but it gives you a perfectly useable mixdown. I haven't tried the the Zoom automix, but the Sound Devices "automix" doesn't sound very natural compared to the Dugan. I use Dugan automix even when I'm mixing lavs or people talking into cardioid mics. I'm surprised that anybody can get away with a straight mix down. Even if you aren't getting phasing issues. Just the additional reverb and roomtone from having multiple open mics in a mix is enough to cause me to feel like the audio needs to be fixed.
 
Is “but the Sound Devices "automix" doesn't sound very natural compared to the Dugan” a typo or are you using Dugan hardware into Sound Devices hardware, and not using the SD builtin Auto Mix?
 
From SD "The Scorpio, 888, 833, 633, and 688 offer two automatic mixing algorithms: Sound Devices’ own MixAssist and Dugan Automix, a collaboration between Sound Devices and Dan Dugan Sound Design, Inc."
 
  • Like
Reactions: jcs
Ah cool thx Scott. I haven’t tried Auto Mix on the F6 though online reports seem to indicate it works ok.

If there’s a way to save an auto mixed output along with the originals, that would address your baked-in concern.
 
I would love to see a photo of Joshua's setup and then from Noiz2 (or any other contributor) on how to better position the mics

cheers
 
Occam's razor: import audio to timeline as discrete mono tracks, cut and crossfade when necessary.

We used to call it "checkerboaring" on a magazine show I worked on. Yes, it's manual... Yes, it's menial... But, you have total control over how and where it's done (ie when they are stepping on each other, coughs, uh-hu's, lip and tongue snapping, etc.). A half hour show (about 17 min) would take maybe 20 minutes give or take, including levels and EQ. You get the hang of it after awhile.
 
Noiz2 said:
Auto ducking or auto mixing if you are lucky and the algorithm is good is about the only "easy" fix. The catch is that generally they are looking for the hottest signal and with two boomed mics that might get problematic, depending on the setup.
Now I don't know if the easiest fix has been mentioned but that would be to use the two tracks as a stereo pair. As long as you don't mix them to mono you won't get that comb filter effect and instead the slight delay will give you stereo placement.
Click to expand...

That's one of the reasons I like the scripting approach used by Vegas as it is not dependent at all on levels to trigger the auto duck, that's the beauty of it. This script is not a standard inclusion with Vegas it is a third party written script. The scrip just informs Vegas that if it sees audio on the chosen "Narrative" track duck the audio on the other track you have nominated for ducking. The script will then set the attack/release and duck levels as per whatever attack/release/dB levels the script is set for.

Chris Young
 
Seanik said:
Occam's razor: import audio to timeline as discrete mono tracks, cut and crossfade when necessary.

We used to call it "checkerboaring" on a magazine show I worked on. Yes, it's manual... Yes, it's menial... But, you have total control over how and where it's done (ie when they are stepping on each other, coughs, uh-hu's, lip and tongue snapping, etc.). A half hour show (about 17 min) would take maybe 20 minutes give or take, including levels and EQ. You get the hang of it after awhile.
Click to expand...
That is what it is called in film also. Your time might be a little optimistic, but it does get faster as you go. It also depends on how much cleanup you do, and that is very project dependant. As a rule I would be removing extra "ums" and such, and on an audio only project taking out all those pauses to think cleaning up false starts and rethink moments.

But even fairly heavy editing shouldn't take more that 2X the run time. Now if there is serious noise reduction or word fixes that can expand rapidly.
 
doubledark said:
I would love to see a photo of Joshua's setup and then from Noiz2 (or any other contributor) on how to better position the mics

cheers
Click to expand...

There is a "rule" that supposedly eliminates or at least vastly reduces the phaseyness. If I remember correctly it's three to one, I'm sure it can be googled. The idea is that the "off" mic is three times the distance to the subject as the "on" mic. I can't say that I personally have ever tried it. Mostly because A) I am going to be editing or B) it's a group so there is no practical way to make the math work. With a two person/ two mic setup it would be pretty easy to do, so might be worth a test.

I think the way it must work is that the off sound (dropping at the square of the distance) dropps low enough relative to the on sound that the phasing drops into the background noise.
Of course this "rule" dates to magnetic tape and may be counting on a noise floor to drop into...
 
Noiz2 said:
There is a "rule" that supposedly eliminates or at least vastly reduces the phaseyness. If I remember correctly it's three to one, I'm sure it can be googled. The idea is that the "off" mic is three times the distance to the subject as the "on" mic.
Click to expand...

That’s the correct ratio. It’s also used for spacing multiple mics for things like choirs in live sound.
 
Seanik said:
Occam's razor: import audio to timeline as discrete mono tracks, cut and crossfade when necessary.

We used to call it "checkerboaring" on a magazine show I worked on. Yes, it's manual... Yes, it's menial... But, you have total control over how and where it's done (ie when they are stepping on each other, coughs, uh-hu's, lip and tongue snapping, etc.). A half hour show (about 17 min) would take maybe 20 minutes give or take, including levels and EQ. You get the hang of it after awhile.
Click to expand...
This!

I certainly love the idea of auto ducking, auto mixing, etc., but usually find myself just cutting. Totally reliable, great sound. Yes, it’s a brute-force method, very granular, but reliable great sound is worth the investment of time.

It’s helpful if you’re recording in a quiet environment. If there’s ambient sound things take a little more time. Dialog editing is a disciplined skill you *can* get really fast at, especially in a tool like the multitrack interface of Audition. Do select all and choose “lock in time” for preservation of sync when you’re cutting fast. With lock turned on, you can drag clips to different tracks during playback, trim them to dialog, while keeping that playhead moving.

And, yes, we call it checkerboarding too - even if you’re recording a single boom to a single clip, checkerboarding across tracks gives you track-level EFX for each voice
 
SethB said:
And, yes, we call it checkerboarding too - even if you’re recording a single boom to a single clip, checkerboarding across tracks gives you track-level EFX for each voice
Click to expand...

Checker boarding originally was so mixers (with no automation) had enough of a gap between clips to get in and out. Mixing analog till the 90's or late 80's was a one reel at a time no pauses no punch in/ no automation performance. So you had to have enough space (and mixers, the people) to read ahead on the Cue sheet and set levels and make fades while watching picture and the time counters so you didn't miss a cue coming up. If you goofed or something turned out not great you had to decide if it was bad enough to redo the entire reel and hope you did better.

In the age of DAWs and mixing in the box, it's more about insert flexibility and ease of finding stuff. You have thirty (or often a whole lot more) tracks and a director breathing down your neck is is a LOT faster if your tracks are organized so you can find that door close quick.
 
I love that the idea of checkerboarding prompted some sentimentality here, lol! It makes me appreciate my former colleagues/mentors all the more. But seriously, I really do find it to be the best practical approach to dialogue editing, even in a rush job. It's really not as arduous as you think.

Joshua Milligan said:
As of now my solution has been to go through and key frame the two audio tracks so that when one person isn’t speaking, their mic is keyed out. This works, but it takes forever and considering the amount of videos I’m doing, there has to be another way.
Click to expand...

I would suggest to stay away from keyframing... go heavy handed first, then do the finessing.



So here's one method of "checkerboarding" (ala Adobe Premiere) that is fairly quick, easy, and effective (IMHO):

Lay down each channel as a discrete mono track (e.g. A1=host, A2=guest, etc.)

First pass:
- Normalize your two tracks individually, including levels, noise reduction, basic EQ, etc. (but don't get too heavy on the effects, especially if they're a matched set of mics)
- While simply eyeballing the waveforms (no need to scrub or playback... yet), use the Razor tool to cut at moments where it's obvious that one talent stops speaking and the other starts (don't worry about accuracy, just ballpark it)

Second pass:
- Select all of the "quiet" segments that you just created, and delete them in one shot
- Copy/paste both tracks onto new tracks; mute and lock the new tracks (this is just for safety)
- Now drag your A1 down to your A2 (or vice-versa, either way) to combine both tracks into one track

Third pass:
- Add a crossfade at each seam (8-15 frames, usually... depending on the noise floor difference in mics, etc.)
- Now, scrub/playback and use Rolling Edit tool (N) to adjust each crossfade.
- Add any global effects as needed (EQ, compressor/limiter, etc.)

Once you get the hang of this, assuming there aren't any major problems to fix, you should be able to do this in roughly the project runtime, give or take. No algorithms, no artifacts, no futzing with parameters that you don't understand. You have full visibility on the process and are in full control the entire time. 51% of video is audio (at least)... there's no shame in working for it.
 
Hey fellas! I hope you all are doing well.

I'm getting ready to film another round of these doctor interview videos with dual boom mics (Sennheiser MKH 8040 cardioid mics) just like in the original post I made when I started this thread. I've done a fair amount of work for this doctor since my original post, but all of the shoots have been with just a single interviewee. But with this upcoming shoot I'm doing at the beginning of June, I'll be needing to film interviews with two subjects again.

I never really got a clear answer from this thread on exactly how to tackle this problem. There were some very insightful and helpful posts here, but there were many opinions on methods to go and no real straight answer on how to solve my issue. Some methods called for post production procedures which I tried, but they didn't quite work, and others just simply went over my head (since I'm not a sound guy).

As stated in the original post, I cannot run lavs on these subjects because they specifically request to wear their doctor clothing which is just awful with lav mics, so I have no choice but to boom with stationary overhead mics. And I've never been a fan of splitting two subjects with one stationary boom because I want to get my mics as close to their mouths as possible which you can't really do with one single mic (unless you have a sound guy to move the mic during the interview which I won't).

This means I'll be right back where I was last time, trying to tackle the audio latency issue that I'm getting that is forcing me to spend countless hours in post fixing. And while spending time ducking audio in post is fine for most things, it's not fine when I'm shooting 30+ interview driven videos in a day.

The difference this time is that I haven't actually shot it yet unlike last time where I was asking for help after the fact. As such, I wanted to bring up some posts on here that discussed using something like a Zoom F6 for auto-mixing the audio which was mentioned in this thread several times. I wanted to start a discussion on this to see if that's a route I should go, or if there's another recorder/mixer or anything similar that I could purchase to help me run two booms for interviews, but while controlling the latency I'm getting that is making these great mics sound so bad in a two mic setup. Thanks guys!
 
Last edited:
Surely the snag here is really mic positioning. You are only thinking maximum subject, and forgetting minimum other capture. It means knowing your mics well enough to know where their nulls are, and putting the other subject in the nulls. Some have quite sedative spots off-axis and it sounds like you have your other speaker in these areas by bad luck. I tend to set these if I am single crewed before the talent appears with me rigging them from the direction where the other person will coming from, so I’m aiming using my voice in the headphones, so while adjusting them I talk, pointing at where the mouth will be but tuning fir minimum me, if you get what I mean. The 3:1 rule is a good start, but you need polar pattern interaction to minimise spill. It’s a variation of what live sound people do with floor monitor placement. That mic does not want to hear that excruciatingly loud wedge, so if you swap a mic you move the wedge out of the rear lobe that makes it feed back. SM58 will be happy with one wedge straight in front but an SM86 might need one wedge each side, or it takes off. Same thing we are talking about in - put the speaker, human or a wedge in the null, wherever that it!
 
Paul is right about careful mic placement and use of nulls, though, of course, we know nothing about the location (which may be very reflective). A combination of good placement, any deadening of reflections (if necessary and possible), and automix would seem what you need. Sound Devices automix (MixAssist) has permeated down to the affordable MixPre series recorders, and there's some detail on it here: https://www.sounddevices.com/automatic-mixing-101/ . Testing in advance - possible this time - seems the way forward.

Cheers,

Roland
 
I'm thinking that a combination of the mic placement as well as MixAssist or AutoMix will be what I need to fix this problem. I found this video by Curtis Judd that is basically tackling my very issue using MixAssist from a Sound Devices MixPre recorder and from a Zoom F6. Both offer an auto mixing feature that looks to do exactly what I need in order to help me solve my problem and save countless hours in post.

My decision now is which to get, but I'm leaning more toward the Zoom F6 because I have always loved Zoom products and because I like that it comes with 6 XLR inputs whereas you have to spend quite a bit more to get that many XLR inputs on the MixPre recorder.

Link: https://www.youtube.com/watch?v=oTLfgQ3qSek
 
I own the mixpre and part of the reason I got it was for the automix feature. Besides that mixpre has a lot of other useful features. All that being said you could fix this problem by using lavs. I know you said you have clothing noise issues but this is usually due to improper lav mic placement technique. You could invest some time and some accessories to improve this area of your skills. If it were me I'd get a mixpre with auto mix, use lavs and have one shotgun just as a safety.
 
You must log in or register to reply here.
Back
Top