Mixing multiple lavs of loud talent, speaking over each other

[Heus]

With a small crew we produce a network reality tv show and do post production by ourselves, including audio (all for the sake of absolute efficiency in ressources).

We have mostly dynamic scenes with 2-4 talents, 2 of whoom constantly raise their voice. Since we cover with multiple cams - one of which is almost always wide, and the fact it's mostly unpredictable who speaks, we boom very rarely. So we go with well placed lavs.

Since the talents often stand in close proximity to each other and speak loud, the lavs of the ones not speaking pick up alot of the sound. I try to keep only one lav open at a time, but it's very time consuming when mixing, plus often nearly impossible, since they talk over each other.

 

My workflow has become to leave all lavs open, phase align all lavs with "auto align post", isolate the voices as well as possible (with plugins) and use Acons deverbaerate plug in on each lav (which helps a fair bit with the comb filtering). This takes quite some treble out, which i try to add back with EQ.

 

What are your opinions about my workflow? Any tips? Are there better ways and techniques to implement to get rich voices in these scenarios? Or perhaps any plugins from the likes of izotope that helps with the situation?

 

Thank you all ahead.

[JustusWirth]

If you want a better result you will need to edit all the lavs seperately and only leave the specific lines (and maybe breaths etc.).

In case of overlaps just cut shortly after the the other person starts speaking, it will mask the cut.

There is a reason why dialog editing is a seperate and very important job. Just do it before you start mixing. 

[Olle Sjostrom]

25 minutes ago, JustusWirth said:

If you want a better result you will need to edit all the lavs seperately and only leave the specific lines (and maybe breaths etc.).

In case of overlaps just cut shortly after the the other person starts speaking, it will mask the cut.

There is a reason why dialog editing is a seperate and very important job. Just do it before you start mixing. 

This.

 

There are tools to make your workflow a bit faster, sure, but doing the dirty work of "mixing" in post - ie - playing the sounds that should be heard at the time you need them will ensure you get the cleanest mix. Otherwise you'll have all the noise from the other channels on at all times. And if you use plugins to mitigate noise or reverb issues, you'll end up with a very CPU heavy project and a lot of artifacts. 

[Philip Perkins]

When faced with this sort of project in post I make the director work closely with me about which of the simultaneously speaking talent they want to spotlight at that moment.  If the talent are all on lavs, and yelling close to each other your options for isolating anyone are pretty limited, as you have discovered.  I had a few highly skilled boom ops, who when equipped with a long-gun mic were often able to fish usable dialog out of such a mess IF they knew who to go for and of course were able to get in close.  If the OP's project is improv-verite style and also being shot wide and tight I'd suggest asking about the possibility of painting out overhead booms in post.  If you are going to rely solely on lavs then you are already doing what can be done.

[drpro]

You might check with Dan Dugan, his products could help.  I have had success using them in similar dialog recordings.

David

[Matthew Steel]

Part of what I do is live sound.  When I do panel discussions I often find "Dugan Style" automix useful.  Dan Dugan's original patent has expired so not only are his hardware boxes available, but many mixing consoles have incorporated algorithms inspired by his system.  For post, a plugin would be ideal, but I've never seen it available that way and I am not sure whether the algorithm lends itself to a plugin architecture.

[Philip Perkins]

+1 Dugan.  Still noticeably better sounding than the competition.  The version that can be added to Sound Devices recorders is quite good, the version that can be added to the higher end Yamaha (etc) consoles is even better.  Dugan mixing saved my ass several times over the years.  The higher end Dugan allows some tweaking of the level parameters re: which "voice" is "preferred", but it takes some time to get it working.  Worth a try.

 

[Izen Ears]

This is not a serious suggestion, but I was very surprised when I placed the mic near the belly button - it picked up everyone mostly evenly! But don't do this, I did it as an experiment.

[Philip Perkins]

That's a serious suggestion if it worked for you in that situation--pretty inventive!  In verite doc post often entire conversations have to be taken off one character's lav, balancing the closer voice against the other one.   In that case having the talkers close together would actually help you!

[Olle Sjostrom]

But this is a post discussion…

 

i rigged lavs on myself and camera’s backs once, because of us not knowing where and who was gonna say what, impossible to boom in the place (party). I didn’t do the post on that shoot, but I heard later they actually had some use of those ”back lavs” for off camera lines. 

[Wandering Ear]

If you’re getting that much bleed, are you sure you need both lavs open?  Choosing a primary lav for those moments might sound better than fighting the phase issues, and then bring that other characters lav back in once the overlap is done. 
 

I personally don’t think you’ll find an automated plugin workflow that will avoid having to do the editing work and provide the same results.  There are ways to save time when editing/mixing, but there is no substitute.  Sometimes you have to put in the time to get the results.