Remove Mouth Noises in Voice Over or Podcast Audio Pt.2
We learned how to recognize different types of mouth noises in our dialogue recordings in the previous post.
We talked about using both our ears and eyes to spot them in Part 1.
That's because first being able to recognize the pesky, sticky sounding clicks is key in allowing us to get rid of them!
Let's continue on to learn the different techniques to remove mouth noises using the mouth de-click module in iZotope RX when re-recording isn't an option.
Remove Mouth Noises From Entire Recording
Firstly, how do we remove mouth clicks and smacks from an entire voice recording without damaging the recording with over processing?
It's really easy to go overboard with many of the corrective modules in RX especially when you're dealing with your entire recording.
Typically, you really want to err on the conservative side when processing to preserve the overall character of your voice recording.
In my experience the "default" preset for the mouth de-click module is just too heavy-handed for an entire recording when you listen to the before and after.
It will take away more than just the mouth noises.
(Longer section of my voice recording)
However, if you're in a real time crunch, you might trade quality for time and treat the recording more heavily.
Just be aware that there is a quality trade off when using heavy processing.
Oftentimes, I've found that heavy processing with the mouth de-click module will reduce quite a bit of the higher frequencies, making the sound more dull.
Consequently, to remove mouth noises from a long section of audio or an entire recording, try doing a gentle pass with the mouth de-click module.
This will catch some of the worst clicks, save you time and preserve the overall character of your dialogue.
How To Set Up The Module
OK, but what settings do we use for this gentle pass in the mouth de-click module?
1. First go to the "transparent removal" preset. It should look like this:
2. Next, adjust the "frequency skew" from: -2.0 (more low frequency detection) to 0.0 (even between low and high frequency detection):
- in my dialogue audio, I tend to have more of the high frequency mouth clicks than the lower mouth smacks, so I use 0.0 to detect more highs
- however, listen to your recording before this process to see what stands out to you as the biggest mouth click problem: higher click sounds or lower sounds?
- which is the most prominent and distracting in your audio?
3. Finally, set the "click widening [ms]" from: 0.0 to 1.0 (this repairs the signal around the click by widening the repair area according to RX).
4. Here's what the final settings should look like to remove mouth noises for an entire recording:
5. Now you're ready to select the entire recording and press "Compare" on the mouth de-click module:
- this will render a "before" and "after" version of your processing for you to listen to without applying the effect.
This is your chance to listen and see your recording with the processing in place and without it - very powerful!
As you listen to the processed or "after" version, ask yourself, does it sound better?
Did it remove some of the mouth noises and leave the recording sounding mostly intact?
This is the goal with most of the processing in RX - transparency.
You want to remove the problem areas in your voice recording without it being noticeable (as much as possible).
If the steps above didn't achieve what you wanted, try experimenting with the settings.
Focus especially on tweaking the "Sensitivity" and "Frequency Skew" settings to get things sounding how you want them to.
Mouth De-click Best Practices
Let’s say you’ve used my gentle pass settings or experimented to find your own settings to remove mouth noises from your entire recording. Great work!
You’re going to find that it does catch quite a few of the nasty mouth clicks, but it won’t get them all.
Some areas will be especially bad and the gentle pass settings just won't be strong enough.
These areas will have to be treated individually and with a heavier setting (we'll cover this in another post).
This is where you could try running another gentle pass over the entire recording to see if it picks up more mouth noises in the second pass.
Quite a few of the RX modules (corrective processing effects) do benefit from more than one pass, so it's worth a shot.
Just be sure to double-check your overall sound using the "Compare" function to see if the second pass is helping or removing more useful high frequency content.
You have to make a judgment call informed by careful listening and any enhancements that could help the voice recording's sound further on in processing.
For example, you might be able to regain some lost high end from removing mouth noises using an EQ move later on in your processing steps.
See my series on EQing your voice recording from start to finish for more on how to enhance with EQ.
But the safest and probably more natural sounding method would be to process with RX gently to keep the good sound that's already there rather than trying to recreate it.
Avoid over processing. And this means that you may not want to do that second gentle pass with the mouth de-click module!
No one, but YOUR EARS can tell you for sure what will be best for your situation.
Summary
Remember, to remove mouth noises from a long section of audio or an entire recording, use a gentle pass with the mouth de-click module.
Next, for the gentle pass, set up the mouth de-clicker in iZotope RX to preserve your overall sound while getting rid of mouth noises (see my tested settings above).
Finally, aim for transparency in all of your RX processing including removing mouth noises.
Try to remove the problem areas while leaving the rest of the voice recording "untouched".
Try out this technique today and let me know how it's working for you in a comment below!
Cheers, Bryan
Your Dialogue Editing Mentor
Transform Your Dialogue Recordings to Sound Professional
Grab My 20-page Guide Packed With Tips To Easily Improve Your Audio Today!
When you sign up, we'll send emails once every 2 weeks with valuable free content. We promise to keep your info safe.