Quote:
Originally Posted by martifingers
Yes it seems fairly impossible just now.
But (and this is a stab in the dark!) would this help?
1. Use Reafir or whatever to build a profile of one instance of the offending sound
2. "Subtract" this from the whole waveform
3. If you are right this should produce periods of near silence that may be easily visible.
4. Mark these as regions, UNDO the correction and go back to delete the unwanted sections.
This might well be a really dumb idea - if so I apologise in advance!
|
Worth trying at least! My result? Not good. I tried it on a small snippet, and the waveforms simply don't match well enough, proving the sounds are very different, even though they look the same. It will work with one laugh, but not the next because of minute shifts.
I have a feeling this is impossible! Need a human brain to identify laughter. Great thought experiment though, thanks!