Removing breath noises ... is there a plug in?

I’m still trying to get my first audio book audition accepted through ACX. As I listen to my stuff before submission, I find myself noticing breath noises between phrases and sentences that sound pretty bad to me. I then go back and try to remove or reduce them individually – about -10db application works wonders – but this is tedious and time consuming. Is there a plug in or some other technique where this can be done en masse?

Thanks for your help.

Did you get rejected by ACX? Did they say why? Typically, it’s not because of performance issues unless you’re really terrible.

Do you have Wet Mouth Noises, too?

People who record for a while in a quiet room discover all sorts of problems. We live in an insanely noisy word that your head does a good job filtering out. The microphone doesn’t filter anything.

“Is that my refrigerator? I never noticed that before…”


I’ve done about 15 auditions and am batting .000. Nobody from ACX has made that claim, but a couple folks who I asked to review my work (they are not involved with ACX) have mentioned that my breaths are more prominent than they should be.

I also have some problems with mouth noises, but am finding that staying hydrated, using Biotene mouth rinse and the recent de-clicker plug-in can minimize those issues.

Part of the problem is that I, like most talent, am extremely self-critical. Another issue is that I had a long career in radio that ended back in the 1990s and I am trying to establish myself in a medium that isn’t really like radio at all, except for the part about the human voice and the microphone. I’m also used to having some sort of music playing behind me, which would mask some of the noise issues.

Part of it is frustration and a little envy. An old radio pal suggested that I give ACX a try. He put up an audition and was hired for his first gig in less than 48 hours. He’s now got a second gig under his belt. He’s doing pretty well; I am floundering.

They can be automatically attenuated using a “noise gate” plugin , e.g. …[b]Noise_Gate[/b]

You have to be careful adjusting the settings otherwise bits of words can be cut off.

Great! I won’t have a chance to do anything with this for a couple days.

Next question: Could someone suggest introductory settings for the plug-in? I never much concerned myself with just how all those electronic boxes worked; I figured I’d let the engineers and program directors worry about that. Now that I am a one-man band, I need some schooling.

if you can let us hear an example we can suggest settings.
You can attach audio files to posts here if they are less than 1Mb,
or post a link to the audio hosted elsewhere.

.wav or .mp3?

For this purpose either will do. With the 1Mb limit for attachments here, you will be limited to about 10 seconds, mono, 16-bit, 44100Hz WAV.

Let’s try this, then:

suggested noise gate settings.png
There was a version of this noise gate which allowed different values for attack and decay, which would produce better results, but I can’t find it at the moment. Rather than having attack and decay both ~200ms , the attack should be shorter around 100ms or less.

Many thanks. I’ll be out of town tomorrow, but I will try this Thursday. BTW, am I correct in assuming that all the work should be done on the .wav file before export to .mp3?

Hope I can return favors to you good folks one day.

MP3 is a delivery format. It’s an end-of-the-line product, not a step in the middle.

WAV makes a dandy archival format for stuff you may want to edit again someday. If you need to continue an edit or complex production later, then you can Save an Audacity Project. That will preserve all your layers and tracks. Projects do not, however, save UNDO.

Export a protection WAV any time you will curl up and die if the computer crashed taking your current show with it. Save backups to a second media such as an external drive or maybe a USB thumb drive. Projects are brittle and it’s possible to open up a Project and have it not open.

What were the reasons given for your rejections? Companies many times give you something to fix or work on. ACX recently had a users meeting where they explained how they did acceptance testing and what they look for.

We’re generating tools and testing processes because a lot of performers use Audacity and they all ask the same questions.


in a medium that isn’t really like radio at all

But it can be. To do an audiobook you have to be terrific at telling stories around the campfire, but to do a podcast, you have to know how to do sound production, critical timing and radio structure. That sounds more like your speed.

What was your old show? Did you interview people? Did you get fan mail?

I saved “The Morning Stream” as an example of a very popular, upbeat podcast.

But you don’t have to get that crazy. There are a lot of people doing successful podcasts with a good deal less technical polish. You do, however, have to be your own producer and that can be daunting. If you couldn’t figure out what your producer did before, this will be an eye-opener.


At this point, I am only interested in working with ACX. If I can figure out a way to make money with a podcast, perhaps then.

I finally got an audition accepted at ACX. :smiley: And I will send off the first fifteen minutes tomorrow, but for one problem.

Using the settings for noise reduction recommended on the previous page, I find that sometimes the last little bit of a word gets chopped off. What is causing that, and what adjustment should I make to make it go away? I have saved all the raw files, so I can undo any damage.

RE: Mu long and obscure career in radio. I did just about anything you can do on the air at one time or another. News, sports anchor, hockey color commentator, commercial production, disc jockey in multiple formats, airborne and studio traffic reporting. Fan mail? Not even from my mom. :cry:

The plugin on the previous page is “noise gate” , (“noise reduction” is something different).

Lower the gate threshold* of the noise-gate, or increase its decay, or a combination of both, to avoid chopping off the ends of words.

[ * “lower the threshold” = use a negative number with a larger absolute-value , e.g. -40dB instead of -35dB ]

Both Noise Removal and Noise Gate have problems. Noise Gate has to be very carefully adjusted so it doesn’t do what you had, cut off pieces of words. It’s just a slicer. Sound below this setting gets managed downward. Unfortunately, some of your vocal expression may be down there. There is the temptation to smash the background sound to dead zero to make it sound like you’re in a studio. You should avoid that.

Noise Reduction has to be juggled so your voice doesn’t sound martian and honky.

That’s why we recommend very strongly to record in a good quiet environment at the beginning.