Custom normalization in Windows 7 Speech Recognition

I am using Windows 7 built in Speech Recognition to fill 开发者_运维问答some forms. It works quite well but I have a problem. The speech recognizer does some kind of automatic normalization. For example, if I say "twelve dollars and fifty cents" it writes $12.5. That's good.

However, if I say "15 by 3 metres" it writes that phrase exactly and not what I would like it to write: "15x3 m".

How can I customize the speech normalization process to do what I need to?

There is definitely no way to do this using the SAPI APIs. (The lexicon APIs are close, but aren't parameterizable.)

The only way I know to customize the normalization entries would be to use the Dictation Resource Kit, but that's overkill for this purpose - you would have to rebuild the entire dictation grammar from scratch (since the DRK doesn't include the sources for the default dictation grammar).

you can disable it and implement your own speech-handler, look at this thread here