Podcast Audio Editing: Remove Breath Noise
Reduce breath noise when recording and methods to remove it in post-production.April 18th, 2020
What is breath noise
As you might have guessed, breath noise is the sound of a speaker's breathing being recorded. There’s no way to prevent people from breathing, but there's a few approaches to handle it in your podcast:
- Reduction during recording
- Removal in post-production
- Leaving it if stylistically appropriate / not distracting
You might not need to remove breath noise
In my opinion, not all podcasts need to get down to zero breath noise.
Likely, most listeners aren’t gong to mind there being natural breath noise from proper recording - as long as it isn’t distracting - (see the How to reduce breath noise when recording section)
A quick way to have a sanity check on this is have a few people listen to your episode - when they’re done ask them if they noticed the breath noise.
If they immediately know exactly what you’re referring to or have feedback on it being distracting/noticeable at all - it’s probably a good call to change how your recording and/or address this differently in post-production.
How to reduce breath noise when recording your podcast
Solid mic technique will be the best option you have to reduce breath noise. Some of these concepts are covered in: Podcast Audio Breakdown: Clipping, Headroom, and the Noise Floor.
Key Points for mic placement related to breath noise:
- Microphone about 6 - 8 inches from your mouth
- Microphone diaphragm pointed directly at your mouth
- Have the microphone positioned off-axis to your mouth
- Use a pop filter - primarily for plosives - but will also help for very heavy breathing
If the off-axis bullet wasn’t clear follow these steps:
- Position mic 6 inches directly in from of your mouth with the diaphragm pointed to your mouth
- Shift the mic 6 inches in any direction up, down, left, right from that center point - where ever feels comfortable to speak during your recording
- Tilt the mic to point the diaphragm back to your mouth
Important: Test Your Mic Placement
Do a few trial runs with different mic placements. The timbre of your voice will change depending on where you positioned the mic for the off-axis placement.
Adjust to what feels comfortable and sounds best to you - and of course, to make sure you’re reducing the breath noise.
Record a few rounds and note the specific mic positioning before changing setups. When you’re finished you’ll have a list of audio samples to compare.
Ok, so ideally you can reduce breath noise as much as possible during recording.
But, if you’ve decided you do need perfectly clean recordings with zero breath noise - then you'll need to handle that in post-production.
Remove Breath Noise in Post-Production
Theres a few ways to approach this - and any one of them could be the best option depending on your situation.
Option One: Remove breath noise with a noise gate
This is a great option if you’re using a DAW such as Adobe Audition which has extremely powerful post-production effects processing available.
In Audition you can add a noise gate by adding a dynamic processing effect and then select the noise gate preset to get initial settings to start from.
(If you're using a DAW that doesn't have very powerful post-production tools, this can be difficult to dial in your gate settings for breath noise without removing actual words.)
Noise Gate Settings to try:
- Low threshold - dial this back and forth until you hear only the breath being removed - if using Audition use the spectrum analyzer
- Slow down the attack (~100ms) - this will prevent cutting off the very begging ins of words
- Slow down the release (~300ms)
If your gate effect has the ability to let you listen to only the audio being cut out - toggle that on, then roll up the threshold as you listen to your audio until you start hearing the beginnings of words come through - then pull back.
If you're using Audition, be sure to use the preview mode of the Spectrum Analyzer to support what you're hearing as you dial in your threshold.
If you start hearing the beginnings of any words coming through, you know you’ve gone too far.
Ideally you want the output of the gated audio to be virtually silent during the majority of your episode - with the occasional light little breaths coming through.
If using this method - I still recommend taking a look at The Best of Both Breath Noise Removal Techniques below to shorten the empty spaces a bit where needed.
Option Two: Reduce breath noise with dynamic EQ - maybe applicable
Reduction with EQ is focused more on the reduction of specifically noticeable aspects of a breath.
This can be a good option if you’ve done all of your mic placement correctly and breath noise is virtually non-existent BUT your guest has a slight whistle when they breath in.
So while you don’t have noticeable “breath” noise, you do have a noticeable whistle every time they breath.
You can approach this the same way you do sibilance - which I went through in this article on dynamic compression / EQ.
Basically, remove that whistle by honing in on those frequencies and smashing them into oblivion when they go above a certain threshold using a dynamic EQ plugin.
Option Three: Reduce breath noise manually - buckle up for a few hours
Ok here we go. Hardcore mode.
This is the most time consuming option - but if you’re going for perfectly clean audio - this is the way to go.
First, here are two common manual approaches that I don’t believe go far enough to get optimum results.
Highlight and delete
This is when you highlight the breath noise in your editor, then hit delete. Satisfying… But not the cleanest method.
- You will have pops if your cut splits a waveform
- It’s going to sound unnatural - We are used to hearing pauses for breath in sentences - there’s a natural flow and rhythm to speaking and breathing. A thoughtful sentence with breaths can turn into soundling like a hurried race to the finish when you completely remove the natural spaces.
Highlight and automate loudness to silence
This is slightly better than the highlight and delete method because you’ll remove the potential for audio pops by technically having an audio fade so you don’t have any split waveforms.
- The same way the delete option sounds unnatural because of a lack of pauses, this method sounds unnatural because of extended pauses. While we’re used to the moments of pause in conversation - we’re used to knowing why they occur because we hear that slight breath (or see in person). But by completely muting the breath you now have an overly extended silent section.
- As a listener you don’t know why it’s silent - just that it keeps becoming silent at random times for too long.
- Noting that this is mitigated if you have voice training - to take quick breaths or to time your breaths between interactions.
The best of both breath noise removal techniques
The technique I argue for is a combination of the previous two:
- Highlight the breath waveform and automate down to zero (if possible - use a noise gate to replace manual automation)
- Delete a portion of that silent section
By automating the level to zero, you won’t have the audio pops that can happen with the deleting.
The important part of this step is to delete a portion of that breath noise that is now silent. This transforms that over-extended silence into a natural pause.
A good rule of thumb is to cut about 30% - 50% of the muted section. But this will depend on the pace of your speakers. If a speaker takes looooong exteeeeended breaths you’ll probably cut more of that than if a speaker takes relatively short breaths.
Finding that sweet spot is all about listening to the flow. Make your edits then listen back to the section - with your eyes closed - and listen to what your ears are telling you.
Does this feel natural? Did you notice where your cuts were? Does it feel rushed or too drawn out or just right?
If you don’t notice anything in particular listening back after your edits - you’re doing it right. These should be transparent edits to listeners.
Breath-noise removal plugins
This is also a totally valid option - and optimal if you have very clean recordings to begin with. I’ve listed this all-in-one plugin option last as both a glimmer of hope - (to not have to spend a few hours cutting / automating levels) and because I’ve seen mixed results.
The most popular I’ve seen is - DeBreath Vocal Plugin | Waves
The biggest determining factor on if this will work for you is if you have clean separation between your breath noise and spoken content.
An issue I’ve seen using breath noise reduction plugins is often the breath noise waveform and next word or previous word would be connected.
This would lead to unnatural cuts of the breath noise, missed breath noise, or false positives - removing actual waveforms of content.
Again, having a solid mic technique will greatly increase the effectiveness of using this type of plugin because you’ll be able to have a greater difference between the breath noise loudness level and the content loudness level.
Should you always completely remove breath noise
As mentioned I don’t think all podcasts need to fully remove breath noise. I think there’s a natural, approachable element of breath noise - but only when you don’t consciously notice it. So in some cases it can be fine to leave in depending on the style of your podcast.
But all podcasters can benefit from the mic techniques listed and if you’re in a pinch realizing your recorded episode is littered with breath noise - you can clean it up and get it production ready.
If this has been helpful consider signing up to my email list where I send out helpful tips to improve your podcast and let you know when I’ve written new articles.
Thanks for reading and happy podcasting :)