Re: [silpa-discuss] English to telugu

[Santhosh Thottingal]

On Tue, May 25, 2010 at 3:36 PM, Suji Arivazhagan <address@hidden> wrote:
> While i run the silpa project for transliterate i got an error
>
> Action: English to telugu
>
> Word :udhyam(telugu word), normal dictionary words i call transliterate like
> kite man etc...
>
> Error : could not fine the word udhyam in dictionary
>           0.0468590359552
>
> How to retify?

This kind of transliteration is not implemented so far. Currently, the
algorithm uses a pronunciation dictionary based approach for getting
the transliteration of English words. The dictionary used is CMU
pronunciation dictionary and obviously it won't contain words like
this(Indian names, places or transliterated telugu words).

But I understand that an ideal transliteration should handle all
these. We just started and we might do it in future once we figure out
a good algorithm for this.  You can also suggest some approach for
this.

A very valid and important use case: User has a  list of names -
Indian names in English. She want to get the names transliterated to
one of the local language. Anybody think that a rule based approach
will work out for this? thoughts?

Thanks
Santhosh