the toughest part,for me, is duplicating that first grain where it
has to start half way through so that you dont get that volume ramp
from the grain window. if anyone has an example of this please let
me know.

#P window setfont “Sans Serif” 9.;
#P window linecount 1;
#N vpatcher 0 44 1020 768;
#P window setfont “Sans Serif” 9.;
#P message 576 189 60 196617 reverse $1;
#N comlet polarity;
#P inlet 576 160 15 0;
#P newex 528 413 35 196617 % 0.;
#P newex 566 377 58 196617 r totaltime;
#P outlet 520 599 15 0;
#P user number~ 212 389 361 404 9 3 3 2 0. 0. 0 0. 60 0. 0 0 0 221
221 221 222 222 222 0 0 0;
#P newex 177 172 59 196617 select bang;
#P newex 265 197 64 196617 * 0.022676;
#P newex 265 222 35 196617 sig~ 1;
#P newex 265 246 26 196617 +=~;
#N comlet Audio;
#P outlet 378 382 15 0;
#N comlet PitchMult;
#P inlet 549 160 15 0;
#N comlet GrainLegnth;
#P inlet 338 160 15 0;
#N comlet position;
#P inlet 177 151 15 0;
#P newex 492 244 33 196617 / 1.5;
#P newex 406 350 138 196617 grain.pulse~ sample window;
#P newex 407 319 138 196617 grain.pulse~ sample window;
#P newex 338 246 70 196617 train.shift~ 3;
#P newex 408 279 138 196617 grain.pulse~ sample window;
#P connect 5 0 12 0;
#P connect 9 0 13 0;
#P connect 12 1 11 0;
#P connect 11 0 10 0;
#P connect 10 0 9 0;
#P fasten 12 0 9 0 182 241 270 241;
#P connect 6 0 1 0;
#P connect 0 0 8 0;
#P connect 2 0 8 0;
#P connect 3 0 8 0;
#P connect 1 0 3 0;
#P fasten 18 0 3 0 581 339 411 339;
#P connect 1 1 2 0;
#P fasten 18 0 2 0 581 308 412 308;
#P connect 1 2 0 0;
#P fasten 18 0 0 0 581 275 413 275;
#P fasten 9 0 3 1 270 343 453 343;
#P fasten 9 0 2 1 270 301 454 301;
#P fasten 9 0 0 1 270 270 455 270;
#P connect 4 0 3 2;
#P connect 4 0 2 2;
#P connect 6 0 4 0;
#P connect 4 0 0 2;
#P connect 16 0 14 0;
#P connect 13 1 16 0;
#P fasten 7 0 3 3 554 343 537 343;
#P fasten 7 0 2 3 554 315 538 315;
#P fasten 7 0 0 3 554 273 539 273;
#P connect 15 0 16 1;
#P connect 17 0 18 0;
#P pop;
#P newobj 225 270 43 196617 patcher;
#P window clipboard copytext patcher #E;
#P window clipboard copycount 1;

I think you should try to use 4 instances of [2d.wave~] and use them with granular technique.

I’ll post a patch later to clearly explain the idea.(I’m not on my computer now)