How to isolate/figure out whispers in audio clip?

[grarpamp]

> https://www.c-span.org/video/?323601-1/senate-ceremonial-swearing-vice-president-biden

That entire press session and a few others
were going around the right-pundits due to it being
full of Joe getting creepy with a bunch of girls
and them being cringed out.
Just search creepy Joe Biden / Biden girls / Biden pedo, etc.
Some of them were posted to this list for reference
over in the elections thread.

> Can anyone suggest an audio program (preferably for Linux) that can help
> isolate Biden's voice? Computery techniques I could use

What you want is a parametric equalizer.
It allows you to set a bandwidth around a
chosen center frequency and boost or cut only that.
You'll probably need to use four to six centers...
2 x camera clatter, 2 x background voice and HVAC, 1 x Joe
and play with the widths to get the selective filtration you need.
Use a spectrum analyzer to help pick centers and bands.
There are hardware PEQ's on Ebay, but they usually have
only a couple centers without paying big broadcast money,
can be noisy, don't stack, etc. Similar issues with HW SE's.
These days software is going to be far better and free.
Search for PEQ and SE functions in opensource audio editors
/ nonlinear production, or ask the audio processing / NLP forums.
You'll probably have to hack the source to add in more centers
since most are likely to inherit two as default from the hardware days.

Also consider asking for better raw source from CSPAN, the
press pool, the major media outlets, and FOIA security footage,
all of which were there. What they post online is usually lossy
compressed versions of their raw archival camera feed.