I have wgpluck mostly working (there were several bugs in the
distributed code).

The reason I say "sort of", is because while I fixed the tuning to be
acceptable for most cases, the way I fixed it isn't optimum. Also there
is at least one dangerous design problem in the code same quantity being
calculated in two different modules (not too bad, except in one case it
is being used to determine memory allocation!).

Maybe if I get a chance I'll clean everything up and sumbmit it, but in
the mean time if someone wants to try this opcode, I could individually 
send the two source files. Again, it works, it's just not ideal and not
cleaned up. 

wgpluck2: This opcode seems to serve the same purpose as wgpluck, but
rather than using an allpass filter to fix up the tuning, uses
oversampling. I would have used this, except that I thought the
performance might not be good enough for realtime on my Pentium 100
machine. It looked like one could easilly be looking at 20 to 40 X
oversampling. Probably very accurate, but not fast.

Larry Troxler