attack/decay CV modifier?

[Paul Perry]

Given an envelope (from an envelope follower or generator), I
want to have independant control of the attack and decay. Is it
worth differentiating the CV to determine whether the slope
is 'up' or 'down', then multiplying the differentiator output 
by a factor depending on whether the attack or decay is to be 
increased or decreased, then re-integrating to produce the 
modified envelope?
I know it's a long shot.. but perhaps someone on this list has 
done this already?
Has anyone done anything on slope detection? Is there a trivial
hack with a sample and hold runing at say 50Hz and comparing 
with the previous sample? (my high school calculus is coming
back to haunt me!)
Any and all ideas appreciated! I have a mathematician to help
me, but I need a starting point.

paul perry melbourne australia