SDCV and indian diacritics

[Sebastian Nehrdich <nehrdbsd@googlemail.com>]

Hello everybody,
I am new to this list and didn't do anything with my emacs for some time,
so please be patient if I am not giving the right/enough information. :)

Lately I set up my emacs to work with SDVC via the sdvc.el extension, as
described here: http://www.emacswiki.org/emacs-de/Sdcv

So far it is working alright and it really is a handy tool for me, but one
problem keeps bothering me.
Mostly I am using emacs to set Sanskrit texts wich usually contain a lot of
diacritics.
this might look like this: dṛṣṭvādḍhataṃ rūpamugraṃ tavedaṃ
And the dictionary files contain the words in the ITRANS / Harvard-Kyoto
transliteration sheme, wich is used to set Devanagari-letters in the roman
alphabet without using diacritics at all.
So when I want to run sdcv on my input wich includes diacritis, naturally
it does not work well because the dict-files are in ITRANS/Harvad-Kyoto
transliteration.
Of course, if I replace the diacritics manually and turn the input into
something like this:
dRSTVAdDhataM rUpamugraM tavedaM
wich is in conformity to the ITRANS/Harvad-Kyoto transliteration, sdcv is
working very well. But I do not want to input all my stuff in ITRANS
because I might want to use it later for latex etc.
So my question is: Is it possible to hack the sdcv.el to do this
replacement so I can type with my diacritics, then the sdcv.el converts all
the characters to itrans before the input is sent to sdcv?
Greetings & thanks for any advice on this

Sebastian