@jb101 would a multiband gate NOT fix That problem? maybe a de esser...
 
you could also (if the breaths are visual) create a volume automation point on the breath and fade it, no need for chopping then

Yeah, in practise, I've found that there's no truly reliable automated way of doing this. These are audiobooks - I'm currently working on one around ten hours long; apparently the next one they're giving me is about 30 hours.
 
So if I was to try to do it automatically, I actually have to listen to the thing in full, possibly more than once, to make sure not starts and ends of words are being chopped off (which they will be) and no breaths are missed (which they will be). And then carefully repair those bits by pulling out plosives and what have you onto a track that doesn't have the gate on. At that point, the hourly rate is worse than flipping burgers for a living.
 
It turns out to actually be far quicker to whizz through the wave visually identifying the breaths, which are extremely recognisable, and delete them by hand. The performance is already approved, so I only need to actually listen to sections where I've done quite a few edits close together, or anywhere that looks dubious. Doing it like that makes the job viable.

THanks John, 
 
I understand now. I thought that maby a gate expander was new for you and the other forum members easliy overlook this as it can been seen as a basic tool....

I had a good old go at that method, and also tried using Izotope Breath Control. Which is very impressive but still leaves you with the issues of having to manually edit places where it doesn't quite work. It'd be ideal for shorter material, just doesn't suit the logistics of what I'm doing.

This reminds me of my feature request here - http://forum.cakewalk.com/Option-to-create-a-crossfade-when-splitting-clips-m3125452.aspx 
 
I think it's a big oversight that Sonar doesn't have any way, when splitting a clip, to automatically create a fade, and what you're asking for is a very similar thing - as soon as you create a new audio clip by cutting an existing one you almost always want it faded in some way.

May have been mentioned and perhaps there are pitfalls I am not seeing but what about a Gain Automation envelope instead of making splits? By cutting out the breaths you are essentially stopping input to the track which I think is the same as reducing the gain to zero (which means it should affect any FX tails or whatever which is why a volume envelope would be undesirable). Slicing things out would be more difficult and time consuming than writing the envelope or doing it live with a controller.
 
And then there is the Mute Tool. Not sure if that applies any kind of quick fading to avoid abrupt stops.
 
Also I though the Split at Zero Crossings setting was supposed to deal with such issues... aaand how about gating (which could be automated at those points as well if need be)?
 
Just brainstorming.
 
Cheers.

And I just thought of a wacky workaround using the Comping feature... which again may have been brought up already.
 
You could have your final take/comp, then record silence for the during of the song (using the same mic/room/etc if you want to use the same incoming sound of "silence" that would be produced by the singer naturally).
 
Then just use the Comp tool to promote the silent track sections anywhere a breat occurs. The comping feature creates automatic crossfades so it does the work for you. You can then adjust the X-Fades as needed but I find it's pretty darned good automagically (and the first time I've ever really found x-fade useful in any way... I used to do ALL my fades manually because X-Fades have always sucked no matter what program/version I've used until X3 comping came along).
 
 

I was wondering when I first saw this thread how long the track was. Unfortunately whenever a human is involved, trying to "automate" anything can get dicey.

The problem with automating a gain envelope is that you will be affecting the level fed to any dynamics processors.This could cause undesirable effects further down the chain. In essence, attack/release/threshold/gain reduction will be totally unpredictable.
 
The downside to using any sort of mute, or cutting/splitting clips up and removing the breaths entirely is that the transition from an inherently noisy signal to total digital silence could and often does, sound quite unnatural and distracting.
 
I'd use a volume envelope and experiment with how much the breathy sections need to be attenuated whilst still retaining some "air" around the signal.

Bristol_Jonesey
The problem with automating a gain envelope is that you will be affecting the level fed to any dynamics processors.This could cause undesirable effects further down the chain. In essence, attack/release/threshold/gain reduction will be totally unpredictable.
 
The downside to using any sort of mute, or cutting/splitting clips up and removing the breaths entirely is that the transition from an inherently noisy signal to total digital silence could and often does, sound quite unnatural and distracting.
 
I'd use a volume envelope and experiment with how much the breathy sections need to be attenuated whilst still retaining some "air" around the signal.

Yeah... I'm thinking more and more a "silent" but not silent take (using an input signal from the mics with nothing going on) would be the best and easiest way and I will probably do this from now on just in case.
 
Failing that just making a copy of a period of silence on the actual vocal track the pasting it into a new lane wherever a breath needs to be yanked out then comping would be a way to workaround recorded a long silence take.