How to add custom SR (Speech Recognition) to Micro

问题:

I have a Speech Recognition application for a Language X.

This language is not supported by Microsoft SAPI. How can I add this SR for Language X to Microsoft SAPI so that I can use it via Microsoft SAPI ?

So What I wonder is

• Can I add a custom SR to Microsoft SAPI?
• If so How?
• If It is a complex process, where to find details?

回答1:

SAPI is just the application programming interface, it only provides the connector for the speech recognition engine. There could be multiple implementations of the same interface.

If default implementation by Microsoft doesn't include language X you have multiple choices. I suppose a choice to call Microsoft and ask them to add language for $1M is not for you.

You can create your own SAPI implementation with the software toolkit which allows you to add any language into a speech recognition system. Such engine is CMUSphinx. You can easily create a speech recognition system in any language using CMUSphinx. For more details see the tutorial.

Once you have the core functionality of speech recognition you need to implement engine functions. For that please read the SAPI engine developer documentation from Microsoft. Essentially you need to create DLL with certain interfaces implemented and install it in your system. After that the recognition will be available through SAPI. There is a sample engine in SAPI distribution which you can use as a base.