[Groff] utf8 to groff_char filter

[James K. Lowden]

I'm doing something that's got to be rare if it's been done before: I'm
dictating a groff document.  

The dictation program lets me say things like "mdash", and the output
is UTF-8 (so says file(1)).  I prefer to work in ASCII.  preconv doesn't
do what I think I want. It produces outputs that aren't highly mnemonic
e.g. 

        \[u00E2]\[u0080]\[u0093]

I'm not sure what that's supposed to be, but what I'd like to see is 

        \[em]

I'm therefore look for (if you will) "ffort": something that converts
machine-readable codes in to human readable codes.  

Looking at the groff_char manual, it seems like such a translation is
possible, even if many Unicode points map to the same groff symbol
name.  My questions:

1.  Does such a utility exist? 
2.  Am I the only one who wishes for such a thing? 
3.  Is the problem well specified?  Do you see any ambiguity in the
proposed translation?  

It doesn't seem terribly difficult, but maybe I'm overlooking
something.  

Thank you.  

--jkl