Ive just tested your patch and seems fine to me, I'll come back to this in a moment.
here, Ive re-arranged your patch a bit, as I think it might help you in the future
(actually, I'll say I tested the re-arranged one, but its basically the same, but I need to first be able to clearly see what was going on)
fltenvtest.axp (9.4 KB)
changes:
a) first , Ive cut n pasted your voice patch, into an embedded patcher object, they are easier generally to manage,
so we only have 1 file to deal with 
b) Ive moved the effects out of the voice, and use an audio outlet
this is important, since effects are quite expensive, so unless you are doing per voice modulation, really you do not want 7 copies running.
(i.e. you'll get more polyphony with my patch)
c) execution order.
its really important you use the correct execution order, layout object left to right, top to bottom.
if not you can get weird issues,
note how in my layout all inlets are fed by an object that is processed first (L->R , T -> B)
d) I pushed some params to the parent so i could test it easier
now for testing...
ok, so you can modify the parameters on the top level patch.
seeing an min attack, and a half sustain level, I can clearly hear the filter envelope working
i turned down the filter response to velocity and your lfo, so as to be sure by velocity did not alter the cutoff.
(as your effectively summing them with a mixer)
if you want to be even more sure, if you edit the sub-patch, change the adsr for the amp to -64,-64,64,-64,
then you will clearly here the filter envelope
hope the above helps
enjoy
Mark
