Silencing the quieter of two identical tracks?

My co-host and I record on two separate mics in the same room. Previously, we’ve exported the file as a single combined feed, but we’ve recently started isolating the two mic feeds in hopes of improving our sound quality.

In the combined track and when I manually copy and paste the two isolated feeds into a single Audacity file, there’s a slight echo and distancing effect on my co-hosts voice, while I sound fine. On his isolated mic feed his voice sounds fine.

Obviously as we’re in the same room our mics are picking up each other’s voices as well as our own, and a failure to perfectly blend this is presumably causing the problem.

I can go in and manually silence his voice on my mic feed track every time he speaks, but that’s going to be massively time consuming. Is there an function I’m missing that can identify when two tracks in one file have identical sound waves, and automatically silence the quieter of the two? It doesn’t seem a million miles away from the noise reduction functionality, but I can’t seem to figure out how to do it.

Apologies if I’ve explained this poorly, very much a learner when it comes to audio editing.


This sounds like a job for a Noise Gate. Applied to your microphone track it will mute the track whenever the sound level in the track falls below a specified level, thus muting your co-host’s voice on your track. Is that what you want?

You can get a noise gate plugin here. Be sure to read the instructions provided either in the zip file, or provided within the plugin.

This is a Nyquist plugin. For instructions on installing Nyquist plugins look here.

– Bill

What’s supposed to happen is you two record on opposite sides of a desk with elevated middle so direct sound is not likely to leak between the two. Then use good, directional microphones in “fist spacing” and do the whole thing in a echo free room with no outside noises.

If you don’t do that, you get what you got. Leakage between the two microphones or, much worse, echos between the two microphones. You’re right, fixing that in post is a time-consuming nightmare. I don’t know of any automated way to solve this that doesn’t sound worse than what you started out with.

By all means, mess with the sound gate. Please note that gates will not follow you when you announce with expression or theatrical emphasis and by the time you get the gate to follow your voice, some of the other microphone may leak through. And remember you have to do this twice, once for each microphone.

I think they got this solved now, but some gates used to have “tails” and leave a tiny bit of bad sound at the end of each word.

I make that room sound like a nightmare, too. It’s not. You can go a long way by propping up furniture moving blankets along the walls and the floor, and then pile books or towels on the table between you. Most directional microphones have a “hole” directly behind them and so you have to be facing each other to pull this off.

Being obsessive, I made up wooden frames for my blankets.

This is a voice recording I did for a movie.

This was an animation. The pictures come from somebody’s pen, but the voices are all real and somebody has to shoot them.

And at the end of the day, if you get something to work, write back and say how you did it.


Somebody had a picture of what that desk looks like and I need to find it again. Throw a moving blanket over the desk and pile anything solid between you. Leave just enough so you can see each other over top. Books is good. If you’re handy, cut a piece of plywood (or even heavy cardboard) and throw a moving blanket over it.

Or you could decide to continue to do it the old way. That is an option. Keep tight to the goal. Whatever you do to fix it has to sound better than the old way, and then the next level is it has to not be career move workload.


AutoDuck could do that.

…to one track. My impression is you want to tick back and forth between both tracks. It makes my head hurt what would happen if you tried to launch two instances at once - if you even can.

Even if you got one to work, that would mute the opposing track, not the echo of track 2 in track 1. You’re still stuck with that.

I can get you a good deal on furniture moving pads…


And to save you some typing, no you can’t use math cancellation techniques. That and almost all these tricks fall apart if you like to talk over each other. Talk-over can sound pretty cool if done in moderation, but it just kills you in post-production.


No simultaneity required.
Start with two originals : track 1 & track 2
AutoDuck track 1 with track 2, to create a ducked track 1
AutoDuck track 2 with track 1, to create a ducked track 2
The show is the two ducked tracks.

Conjugate the verb “to duck.”

Leaving you with two tracks and the echo of each. What happens if they both talk at once?


What happens if they both talk at once?

One of the problems with athletic post-production is the need to restrict the performance.

“Oop. We both talked at the same time. We’ll have to record that part over again.”

Previously, we’ve exported the file as a single combined feed

If you have older shows available, where are they? We’ve been cranking away at this while guessing about the quality issues you’re trying to solve.

Maybe we should stop guessing. Post the addresses.


Did you keep the two original performance WAV files? It’s not unusual for us to want samples of those instead of any failure edits. And just to cover it, do you get old style performance quality if you just cram those to together as if you had recorded it the old way?


You can adjust the fade times on ducking so it’s primarily going to act when one is speaking and the other isn’t.

The echo is going to be most noticeable when the other person isn’t speaking.

[ billw58’s suggestion of the noise-gate would be simpler strategy and produce a similar result ]

And since we’re solving a specific problem, it would be good to hear one of the original shows.