Notes from @wsc1 at #32 (comment)
For the LPC stuff, just wanted to note
this mail
this chapter
and zc lpc
So one possible way to approach flac LPC (the fixed LPC shouldn't require this BTW) would be
to extend zc/lpc to use line spectral pairs, expose the parcor coefficients, and work with some of the quantisation methods in the chapter.
If that is of interest to you, I'm happy to collaborate.