Daron,

You wrote:

: Is this what is known as pole instability in LPC synthesis/analysis or is
this
: another problem in Csound itself?
: Is there a easy solution?

To find out if it is instability you're dealing with, you might want to try
some work I've included in the analysis/resynthesis of Csound. In order to
do that, you'll need to have at least V3.44 or equivalent. The thing is
that lpanal does normally work by storing coefficient filters and does not
care if the filter is stable or not. By specifying -a (or is it -A ;-) in
the command line, you will force lpanal to write an Alternate storage mode
which stores the _poles_ of the filters. When processed in Csound by
lpread, the storage is automatically recognized and pole correction is
done, removing filters instability. The drawback (you need one don't you)
is the it is much slower than the regular one. My advice would be to use it
at the end when you're happy with the treatment and want the definitve.

Let me know if this corrects you're problem, I didn't got much feedback
yet.

Regards,

Marc.

: The results with pitch tracking, 50 poles etc. are excellent, better than
any
: vocode I've ever heard in the
: past and I've heard alot.

May I dare to request the orc/sco ;-) ?