Please see my comment to the question and let me explain some more. I am sorry that my answer is just the opinion and cannot help you to solve your problem, but at least you may get better understanding what you are facing.
I never heard of such feature on the engine supplied by Microsoft for Windows and highly doubt that it exist: I'm afraid you are overestimating the power of this engine. The speech recognition quality, for big grammars, is too low even with perfect pronunciation. When a grammar is small and does not contain words similar in pronunciation, the speech recognition is fairly good, at least good enough to make some voice command interface, but when it comes to dictation (the
System.Speech.Recognition library has a special predefined grammar for that), the recognition becomes nearly hopeless, no matter how well you pronounce.
I am familiar with Indian pronunciation and noticed the following: from the standpoint of pronunciation, English pronounced by people from India (event from different parts of the country speaking different native languages) can be considered simply as a different language. Indeed, if you look at
IPA charts, you will notice that they fairly well describe "Indian" pronunciation (while pronunciation in many other cultures is not adequately represented at all). You will see that people with India simply replace many English sounds (especially like 't', 'd', etc.) with distinct Indian equivalents, and also strongly modify intonation and the rhythm. That said, technically, it wouldn't be a problem to create such a language pack which would adopt "Indian English" pronunciation, as soon as it would be a separate version of the engine, without mixing "English" with "Indian English".
The problem is quite different: I cannot imagine that anyone would take this huge work and do it. Maybe I'm wrong in my assumptions, but I simply don't understand where the funding for such a project would come from. The effect and quality of the final result don't look promising, but the work would be really huge. Who would fund such work if the notion of "Indian English pronunciation" does not have any certain status, there is no an institution which would support this part of culture, standardize it, and so on (am I right?). Who would risk big money for the project supporting something which does not make certain status?
[EDIT]
Thank you for your clarification.
Please see the language packs:
http://windows.microsoft.com/en-us/windows7/products/features/language-packs[
^],
http://windows.microsoft.com/en-us/windows/language-packs#lptabs=win7[
^].
As you can see, there is nothing close to what you want. Moreover, those language packs don't generally come in all features. The only recognition engine I know is in English. Maybe, there are others, I'm not sure. I also heard of some 3-rd party voice recognition packs, so far, only for English and Arabic…
See also:
http://en.wikipedia.org/wiki/Speech_recognition[
^],
http://www.dmoz.org/Computers/Speech_Technology[
^].
—SA
Submit an article or tip