Does anyone know if it is possible to use an xml or compiled grammar, like that used by Microsoft SAPI in "command mode", to constrain the words and phrases recognized by the Google speech recognition API? Or is it at this point still just dictation on their terms and try to sort through it later? I'd really like to know... I've found precious little documentation.
Thanks very much, Farley
It's not possible with Google API, but you can use pocketsphinx-android to recognize grammars. Thought the format is JSGF, not XML.