I am scoping out a custom dictation application to be built using MS SAPI 5. I would like to be able to change the grammar (topic) of dictation dynamically based on what is being recognized. For example, if my dictation application deals with car repair, then, if I detect the speaker talking about engine, I want to bring in a dictation topic optimized for recognizing engine part names, as opposed to cabin upholstery.
Anyone know if this is possible?
Thanks.
-Raj
I believe your biggest hurdle will be in developing a "fool proof" method of identifying what topic is being discussed. To reference your own statement, " talking about engine", if you simply listen for engine and key off of that word you would not be able to for instance use the word engine to represent both a car engine and a software gaming engine. I have used a couple of speech synthesizers. The ones i have used wait for specific commands to begin listening. Perhaps you could have a combination of start listening commands.
USER "Computer, start listening."
COMPUTER "Ready to Listen."
USER "Car engines."
COMPUTER "Loading Car Engine Library."
Something like this might be a reasonable approach to your problem while still allowing yourself the flexibility of adding libraries. You could also utilize this approach to implement a default library. If the second command given isn't a recognized library then the program could use the default library.