To cut or to shrink, that is the question

I’ve got five hours of audio to edit and the two main problems are breath sounds and plosives. I think the usual way to remove breath sounds is to cut them and then insert room tone to fill up the time that was lost by cutting. This works but involves a huge number of mouse clicks and menu choices. I have hundreds of breaths to remove and I’m looking to simplify. Can I just highlight the breath sound and then choose Effect/Amplify and enter a negative number? It seems to work when I enter -12 dB. I realize this will create a wider range between the loudest and softest parts of the track. Is this going to bite me when I go to normalize? Any other downside? This is a very efficient method as far as mouse clicks/keyboard work because when you choose “Effect” you will see, right above “amplify” another choice, “repeat amplify” and it will execute the last Amplify command that you used. So you don’t have to re-enter -12 every time. If -12 is not enough could I go -15 or more?
As far as plosives are concerned I have been highlighting them and then using -6 dB reduce amplify and that usually takes care of it. Sometimes -9 is required. Again, this is tedious but it sounds good to me. Pros and cons?

“Breathing” is normal for humans. Unless the breath sounds are excessive or “peculiar” in some way, you may not need to remove them.

Plosives are not usually a problem if you use a pop shield, position the mic carefully and (optionally) roll off frequencies below about 80 Hz.

A noise-gate or an expander can be used to attenuate, (or remove), all the breath sounds automatically, rather than individually. [ Like Steve says, removing the breaths completely sounds abnormal ].

If you must correct each breath manually, Audacity’s recent addition “spectral selection” is one way to do that . Select the breath on the spectrogram then apply “spectral edit shelves” at say -10dB to that area of the spectrogram.

The “spectral selection” tool can deal with individual plosives too : just select the loud bassy part of the spectrum which is the plosive thud.

[ Time saving Tip: to repeat the last effect press the keys “CTRL” +“R” ]

All those and you can drag-select a portion of your show and silence it with Control-L. That will produce a dead silent portion where the offending sound used to be and the exact same size.

If you have a natural amount of background noises, silence can sound a little weird— bullet holes in the show. That’s why a more natural thing is jam a little Room Tone in there.

I swear there was a technique where you could select some damage and the system would figure out how much room tone to put in there. Nobody is amazed you need to do this. This patching is fairly common.

Looking.

Koz

Thanks to Trebor for the tip about spectral selection. I’m now ready to tackle all the plosives in my audiobook so I read the manual pages on Spectrogram view. To say the least, I’m daunted by all the parameters that must be set correctly. I also have trouble interpreting the views.
-----"the loud bassy part of the spectrum’----
I’m not sure what that would look like on the spectrum view.Can anyone suggest the settings I need for this? A screen shot would be gold. I assume that once I get all the parameters set I just roll through the project, select the plosives and repeat the effect on each one.

“Loud” = white-hot bits on spectrogram are the loudest. “bassy” = everything below 200Hz …
removing bassy thud on plosive with Audacity's ''spectral edit shelves'' tool.gif
You’ll have to to increase the “window size” of the spectrogram from the default setting of 256 to 2048 to see detail in the bass frequencies …
Audacity Preferences showing spectrogram settings.png
If you have a lot of such thuds to remove you could try a multi-band compressor which can be set to ensure no bassy sounds go above user-defined volume threshold. Such a tool will process all of the audio in one go, i.e. it does not need to be applied individually to each instance. Paul-L’s De-Esser plugin can also be used as a De-Thumper if it is set to operate in the bass frequency range : 20Hz-200Hz , ( but Paul-L plugin is comparatively slow, even on two frequency bands ).

Thanks Trebor. It looks like the spectral selection method would be perfect but time consuming. I like the idea of using a gate or compressor for that reason. I’ll be conducting experiments over the next week or so and let you know which method works best for me.