kyleb2112 wrote:

 

So in the simplest terms, if you've got 10-second waveform with peaks all at 5 db, and you amplify a two second chunk to 10 db, there's no way to then just invert the whole thing so the loud part is the quiet part?  I can't be the first guy to want to do that .  

You might be the first guy that actually wanted to keep the results of it! Generally, doing that makes the result sound so unnatural that nobody wants to listen to it anyway.

What you might be better off doing is to use the silence finder to mark your file appropriately. What you'd have to do is to reset the thresholds so that it marked everything above whatever level you think is appropriate. There are several possibilities then, I think - but exactly which one will be best I can't say until I've tried it, and that will have to wait until tomorrow. The basic idea though is that you should be able to reduce everything between the marks down to a much lower level, or even delete it altogether and replace it with silence. What effectively you are doing is gating the signal, but you aren't using a gate the way it's normally used, which is to cut out low level crud.

Maybe I'm old fashioned but it often seems to me to be quicker in the long run just to go through and manually fade the levels on the bits you don't want to hear.  I tend to describe them as "make a fresh coffee and just do it" jobs.  You'll be amazed at how quick you can work through even a relatively long file...and you're guaranteed to have exactly the right bits faded, something nothing just relying on relative levels can ever do.

Bob Howes wrote:

 

Maybe I'm old fashioned but it often seems to me to be quicker in the long run just to go through and manually fade the levels on the bits you don't want to hear.  I tend to describe them as "make a fresh coffee and just do it" jobs.  You'll be amazed at how quick you can work through even a relatively long file...and you're guaranteed to have exactly the right bits faded, something nothing just relying on relative levels can ever do.

That's really the bit I should have developed more last night. What was in my mind was to create a favourite that did exactly what you want to the shouting, and have it as a keystroke. All you do then is to highlight the bit you want to apply it to, and hit the key. If you can identify those sections by level, then fine - but if you can't, then do it manually, as you suggest.

The main problem you would have with attempting to get rid of the yelling is that you are simply bringing down everything in the waveform at that point rather than just deleting or quieting the yelling. 

For me I find that simply using a limiter to cut off the peaks of the unwanted noise in the mix and bring those levels back down to the same level as the rest of the waveform works quite well in certain situations. 

Also, if this unwanted sporadic noise is at a certain frequency range then the multiband compressor can be used as a single band compressor to drop the level of this noise.  Setting frequency range, compression ratios, thresholds and attack/release times to taste can have good results depending on the content of the waveform.