Podcast editing conundrum - any help appreciated!

[rm5]

A small nonprofit has recently got me on board as an A/V producer for their podcast, which is great, but I had no idea what I'd be working with until just a few days ago. Basically, for an upcoming episode, they recorded an interview in a small restaurant (so there is obviously a good amount of background noise). From what I can tell, they've got two mics—one for the host, and one for the guest. With that said, as soon as I got the raw audio, I thought putting this together would be an easy endeavor, but I'm having some strange issues I just can't seem to figure out no matter what I try and who I ask. Right now, I'm using Logic to produce this. I'm really bad at explaining stuff, so please forgive me, and ask for clarification if you need.

Here are the issues/important context - I know, some of these are obvious, but I'm still pointing them out as it's important information:

1. When the two files play back simultaneously on two separate tracks, it sounds absolutely horrendous, so doing that is NOT a solution to my problems.
2. I can't have just one file playing (i.e. just pick one of the two files and call it a day) because then the other person's audio (on the other mic - therefore on the other track) sounds distant and garbled.

My attempted solution - what I thought would work - was to simply make cuts every time the other person speaks, but obviously that doesn't sound good because the two mics are different and the ambience changes. So, what I THOUGHT would be my answer would be to apply crossfades on every cut. And while that sounds fine, doing that on hundreds of cuts would take HOURS.

So... there has to be an efficient way to go about this and make it still sound good, but I just don't know of any at the moment. Also, I think of those people who produce daily podcasts - they've certainly gotta have an efficient workflow for handling these types of things, don't they?

So, anyone got any ideas?

 

[StoneJack]

Re-record the session in studio.

 

[rm5]

Re-record the session in studio.

If it were my thing, that’s what I’d do. But I can’t do that unfortunately.

 

[StoneJack]

Then fix each recording one by one and make something audible

 

[MajorFubar]

Audacity (free DAW) has functions which may help you to reduce the sound of background noise. How to do so is explained here:

Noise reduction & removal | Audacity Support

Removing background noise is core to most audio cleanup operations. Audacity has several tools to aid in this.

support.audacityteam.org

I would suggest running both audio files through it, then pull them into Logic (or, just do everything in Audacity). The result is never audibly transparent, however. You'd need AI tools for that, such as Izotope RX, and they aren't free. And they still aren't perfect. Even Wingnut's MAL isn't perfect.

You might have to resign yourself to the fact that you can't make a pro-sounding production from bad recordings. It's probably wise you bring this to the attention of those you are producing for, sooner rather than later, to manage their expectations.

 

[Avatar74]

If it were my thing, that’s what I’d do. But I can’t do that unfortunately.

So... there has to be an efficient way to go about this and make it still sound good

A way? Yes. Efficient? No.

If they gave you two files, odds are they recorded each "mic" on separate devices (two phones maybe?)... which means you've got two clock sources that don't and won't sync up. This is going to be an absolute nightmare to fix.

I've done this before for a nonprofit and it was weeks of work and I've never done it again.

Is the nonprofit paying you? If not, I'd just say no thanks and move on. It's better for everyone involved that you do this, because they will not like the result that you can deliver in the time that they need.

If they are paying you and you are contractually required to get the job done, then buckle up... You're going to need to do a LOT of manual mixing, dynamics, noise gating, etc. There are tools in logic for this but you can also try to use iZotope RX or Acon Digital Restoration Suite. These are NOT wham-bam-auto-cleanup-wizards. They are dynamics processors, like any other, that require you to do passes to record the automation.

If none of this is making instant sense to you, back away from this project immediately.

 

[rm5]

I've done this before for a nonprofit and it was weeks of work and I've never done it again.

Yep.

Is the nonprofit paying you? If not, I'd just say no thanks and move on. It's better for everyone involved that you do this, because they will not like the result that you can deliver in the time that they need.

No, they aren't.

If none of this is making instant sense to you, back away from this project immediately.

I'm already on the team for the next year, I can't just back away. I have, however, given them an ultimatum to purchase better equipment, which they are doing.

I figured out that I can just use DaVinci Resolve, which seems to be easier with crossfades. Plus, I get the video part of it in there, too. It's working okay so far.

 

[Ben J.]

One kind of approach to this is to try to create a separate background noise track, if you can, by putting together bits of pure restaurant noise and looping. You might even use an external source for this. Then, do your best to balance the general frequency spectum of the two voice tracks and apply heavy gating to each of them to remove as much background as possible. Then combine the three tracks. Leave them all three in the center of the stereo image, and perhaps apply a little ambience reverb to help them gel a bit, and maybe give it some 3D. Just a thought, without having heard the recording.

 

[Avatar74]

I have, however, given them an ultimatum to purchase better equipment, which they are doing.

I hope it's sufficient.

I figured out that I can just use DaVinci Resolve, which seems to be easier with crossfades. Plus, I get the video part of it in there, too. It's working okay so far.

Good luck.