Hi,
I have written something that may be useful to some and is somewhat
complementary to what is being discussed here.
[Well 'written' is not quite accurate  -- Nick and Stefan helped me get it
together]
Its a transliteration system that Ive used with the itrans devanagari (ie
Hindi/Sanskrit etc)

I expect that in any language in which
1. there is a somewhat accepted standard roman form
2. that roman form is available as an emacs input method
3. the roman is easier to write but painful (for natives of course!!) to
read

this will be useful

ie make a buffer (needs to be visiting a file) containing
OM
and call rpm-apply-iinput-method
and you should get a buffer containing
ॐ

Regards
Rusi

-- 
http://www.the-magus.in
http://blog.languager.org