Noise Removal w/o Quality Change (& Removing swallowing?)

Audacity (2.0.3) on Macbook Pro (OS 10.8.4)

A friend showed me how to use the Noise Removal tool to take away “room noise.” I could hear the improvement. But when I used it to remove swallowing or some other sound in one spot, then did select all to search for that everywhere in recording, it seemed to alter the sound of the entire recording unfavorably, almost a muted effect on my voice. (I am recording voice, similar to podcasting.)

Any tips on optimal use of noise removal and when not to use it?


PS. Any tips on dealing with. i.e., removing swallow sounds (something i do alot when I get nervous unfortunately :unamused:

Any tips on optimal use of noise removal and when not to use it?

“No,” and “Never.”
There’s a standing line that by the time you realize you need noise removal, it’s too late.

Noise Removal is a patch job that in certain very limited circumstances can slightly improve the quality of a performance. There is a long list of rules for use and the thing people want most – improving a voice – almost never works right.

The two sentence version of Noise Removal:
Select a portion of noise by itself and use that as the profile. Run the tool a second time and subtract the profile sound from the show.

What could be simpler? The problem comes when you try to remove hiss (for only one example). The tool also tries to remove the hiss sounds from your voice (Sister Suzy Sells…), so provision is made to “go around” voices when applying the tool. This is what “Frequency Smoothing” and other settings do. This gives you hissy snake voices in a velvety-silent show.

The list of problems like that is long. You are urged very strongly to produce a show that does not need noise removal at all. Make sure the volume of the voice is high enough so you don’t have microphone hiss problems. Throw pillows and quilts around so you don’t have room echo and dog barking problems, etc. Separate yourself from the computer fan noise.

As a side issue, we can’t easily change the character of a speaking voice. It’s a Frequently Asked Question. It’s painfully difficult to remove “mouth clicks” and other physical speaking problems (a recent post). If you have a noisy swallow, then learn to turn away from the microphone when you do it. Announcers have been doing this for years. In broadcast facilities, there might be a “cough button” which temporarily disables the microphone.

We wrote a piece on Noise Removal.

And if I wasn’t Debbie Downer enough, even if you do figure out a way to patch yourself into a successful podcast or performance, remember you’ll have to do it the same way each and every time. This puts you in Post Production category where producing the show can take up to three times the length of the show.

One other Post Production comment made by an audio type friend of mine. “You’ll hear a lot more trash in your headphones than will ever appear in the show.” I’m hoping you’ll have a background music or natural theatrical sound in your pieces. Those go a terrifically long way to hiding live recording problems. If your voice is going to appear completely by itself, then you need to record in a studio or a room you can make into a studio.


Thanks again, Koz. What comprehensive answers!

So, I get it about selecting a portion of noise by itself.

Do I need to change the settings below get noise profile? I have not been and just clicking ok. ( Only did this for what my friend called “room noise” between my sentences and it flattens the line in the recording and seems to remove the subtle buzz). I read the page you linked me to and sorry to say the settings and how to use them confused me.

Also, what is

natural theatrical sound?

Thanks again.


The settings below “Get Noise Profile” make no difference to the “noise profile”, they affect the “removal” pass of the effect, not the profile itself.
The default settings should work reasonably well, but to get optimal results you will need to fiddle with those settings. It’s difficult to give “rules” for how to set them - much is down to experimentation and experience.

The biggest difference comes from the “Noise Reduction (dB)” setting.
Increase that setting to remove more noise.
If the result sounds unnatural reduce that setting. It is less bad to have a bit of noise than the bubbly honking sound of too much noise removal.

That’s a “Koz’ism”.

Who will be happy to explain it. May be an accidental split descriptor there.

Natural sound is the dog barking in the background of your show during the recording. Theatrical sound is the jet taking off behind your voice which is inserted later because that’s what was written in the script.

Also see background theatrical music behind such shows as This American Life and in particular, the Radiolab shows. See attached clip.