What to do with non-talking track in interview?

I’m editing an interview, and I don’t know what to do with the track that’s not talking at any given time — silence it entirely? Use noise gate or something else?

I’ve found lots of posts about cleaning/compressing one singular track, but what to do with the non-talking track in an interview?

Thanks so much!

Most people listening to an interview consider ping-pong sound, one person on the left and the other on the right to be distracting and annoying. Most people want to face the person talking, so ping pong could be very entertaining in a car.

That means most interviews get mixed down to either mono, or stereo with the all performers in the middle. You should try that. Mix down to mono and see if the host’s track makes the composite show too noisy.

This is also a terrific time to find out if the composite show gets so loud it overloads. This could happen if the host likes to interrupt the guest. Always a bad idea. We’re not all Charlie Rose. You need to fix that in Audacity before the show goes out. Audacity can overload in processing and you can fix it with no permanent damage.

After you export the sound file for publication, any crunchy, popping overload damage is permanent.


Do you have the voices on two tracks and they alternate?

Dead silence is most of the time a poor choice because the transition will be audible at certain listening levels.
There’s somewhere a “Punch-in” plug-in where the selected region can be replaced by a noise profile you’ve grabbed before.
This is useful if there’s some unwanted sound in the background - paper shuffling, chair creaking and similar.
The noise gate is certainly a possibility.
You can also apply the Auto Duck effect to silence the other track when the person in one track is speaking and vice-versa.


You can also apply the Auto Duck effect to silence the other track when the person in one track is speaking and vice-versa.

Again assuming they don’t interrupt or talk over each other. Many of these advanced tools require surgical inspection of the whole show with good headphones to make sure the tools and corrections didn’t do anything messy.

I would save a protection copy of the show, smash the two voices together in mono mixdown and listen to the composite. I’m guessing it will sound fine unless, as above, the host likes to constantly shuffle papers or clink his coffee cup. Apply gentle Normalize to the mix and you’re done.


Thanks for the suggestions!

I recorded the interview on two tracks - mine is stereo, and my guest’s is mono. Mine has ambient sound when I’m not talking (not papers shuffling; I think just my computer fan or the fact that I was using a sensitive mic), and my guest’s is pretty silent when he’s not talking.

Since my track has ambient sound, I imagine I want to leave some ambient sound throughout, so that the transitions aren’t super obvious / there’s never dead silence?

What if I create silence when I’m talking, create a third track of low ambient sound, and run it throughout? So it would be 3 tracks: 1) me in stereo, with silence when I’m not talking, 2) my guest in mono, with silence when he’s not talking, and 3) a low-level ambient noise mono track running throughout?

Did you try just smashing them together? That should work if you just leave your mic up during the interview. This is what happens when you record in a noisy room. A simple production job turns into a rescue effort. If the background noise is constant, you may be able to use Noise Reduction to take most of it out of the show.

Record a test clip like this and post it. We may be able to suggest settings.


For the next show, find out why your microphone is noisy.


Thanks Koz!

Indeed, for my next show, I’ll aim to quiet my fan :slight_smile:

I’m not sure what you mean by smashing the tracks together, but what do you think of my proposed solution?

I imagine I don’t want complete silence when there’s no one talking, so I was thinking I’d silence my track between when I talk, and then create a track of low ambient sound to run throughout the interview. So there would be 3 tracks: 1) me with silence when I’m not talking, 2) my guest with silence when he’s not talking, and 3) a low-level ambient noise track running throughout. What do you think?

what do you think of my proposed solution?

Too much work and too high a possibiliy for damage.

I had your configuration right up to the last post. So you have a stereo track of you (two blue waves) and a mono track of the guest right above each other on the timeline like the attached (click the graphic)?

What happens if neither MUTE or SOLO are clicked and you play the show? If you object to your fan noises, I’m betting we can suppress them with Noise Reduction (not Noise Removal). Then, when you export the sound file, Audacity will produce one clean stereo show with both voices in it.

Noise Gates and other tricks like that leave fan noises behind your voice, so your side of the conversation would have pumping fans in it. Noisy rooms kill a lot of home recordings.

I can get a lot more accurate if I could hear what you sound like. Please produce a ten second mono sound clip according to this recipe and post it.


Record it exactly the same way you’re going to produce your show. Don’t turn anything on or off and don’t apply filters or effects.


Screen Shot 2015-08-22 at 11.25.32.png

Wow, thanks for all your help Koz! Really appreciate it.

Attached is a test clip.

Here are my settings:

  1. I’m recording into GarageBand and then importing into Audacity, using this setup: http://www.wikihow.com/Record-Skype-Into-Garageband-With-SoundFlower-and-LineIn
  2. In Audacity, my guest’s track is in mono and my track is in stereo.

I’m realizing it complicates things that I record into GarageBand (I do this because it can pull directly from Skype), and then edit in Audacity?

Either way, my test clip is recorded in Audacity (not in GarageBand).

What do you think? Generally it sounds pretty good, just wondering what the standard is for dealing with one track that has ambient sound (mine), when the other track doesn’t (my guest).

Thank you again and again,

Thank you for the test clip. We can go a long way toward helping when we know what the original sounds like.

I applied Steve’s Custom rumble filter and adjusted the volume with Normalize and I got it to pass ACX AudioBook Testing (attached - last sentence). No compression, no noise removal, no processing. Some of that may be needed in the final show, but it’s a good sign that you don’t have to “rescue” all your raw clips as step 1.

That was the good news. The bad news is even though you have a very nice announcing voice, your microphone system produces an overly crisp and harsh sound. "SSSSo this is me and the “C” sound in “Audacity” can cut paper. We suspect microphone makers think that sounds “professional.” Many listeners think it sounds like icepicks in the ear.

That’s called “essing” and there is a custom de-esser tool about which I know next to nothing.

So we’ll be learning how to do that together.

As we go.

Screen Shot 2015-08-25 at 11.35.26.png

I got rid of a bunch of it with a custom equalizer filter that I can publish (clip attached).

So it’s a race as to which tool can do a better job.


Wow, thank you again Koz.

Really grateful for all your time and thoughtfulness here!

For some reason I can’t seem to download your attachment (?).

Sounds like you’re suggesting I use Steve’s Custom rumble filter and adjust the volume with Normalize, yes? I can’t seem to find something called “Steve’s Custom rumble filter” — where might I find that? And you adjusted the volume in which direction?

I hear you on the mic issue — do you recommend using my built-in mic (on my computer) instead? That’s what my guests are all using, and seems like their audio sounds better.

Thank you again,

I lost your thread because the title didn’t follow the narration wander.


That’s Steve’s rumble filter in zip form. Unzip it to LF_rolloff_for_speech.xml and install it in the equalizer tool.

Adding Audacity Equalization Curves
– Select something on the timeline.
– Effect > Equalization > Save/Manage Curves > Import
– Select LF_rolloff_for_speech.xml > OK. (it won’t open the ZIP. You have to decompress it)
– LF rolloff for speech now appears in the equalization preset curve list.

You install the "DeEsser file (attached) the same way. It’s a custom equalization curve that takes the edge off the SS sounds. After you get the custom curves installed, run them like the following:

LF Rolloff (rumble filter)
– Select the whole clip or show by clicking just above MUTE.
– Effect > Equalization: LF Rolloff for speech, 8191 Length > OK

Run the DeEsser by just substituting SLeppDeEsser.xml. Everything else stays the same.

Yell if you get lost.

SLeppDeEsser.xml.zip (371 Bytes)