Sure!  Here's what I use for docx, and I think it could be adapted to
pdf with pdftotext or whatever you're already using there.  You need a
small shell script that reads from STDIN, writes to a file, and calls
pandoc or pdftotext or whatever, like ~/bin/antiwordx:

    #!/bin/sh

    tmpfile=$(mktemp /tmp/antiwordx.XXXXXX.docx)
    trap 'rm -f -- "$tmpfile"' INT TERM HUP EXIT
    cat > "$tmpfile"
    pandoc --normalize -r docx -w markdown "$tmpfile"

You need a small handler function to call it from Elisp---see attached
file `inline-docx.el`, which assumed you have both the old `antiword`
for old-style .doc files and pandoc for new-style `docx`.

I apologize for the roughness of the code; it should probably use
customizable paths for pandoc and such.

-Brian