0

I have an application where I am using the voice recognition in google to get the user voice input as text. This works, no problem - I get the text.

Is it possible to send my TEXT STRING to be parsed by the same logic/engine as google's voice actions and get the AndroidOS to execute correct action?

Alternatively start an intent that will prompt the user for a google voice action - the spech will be executed as such (e.g listen to...) but if none match where given by google voice action it will broadcast back to my app with the result text to handle as overflow.

ziggestardust
  • 211
  • 2
  • 13

1 Answers1

1

Parsing the returned text and determining what the user really meant (rather than the text of what they said) is a complex problem. On the simplest side, you can maintain a table of keywords and search the returned text for keyword matches and have predetermined mappings for the meanings of certain keywords. More advanced systems use statistical classification techniques to identify the correct actions from the transcribed speech from a user. If you want to explore advanced classifiers, you might want to look at Mallet.

Sorry, I don't think Google gives you access to their voice action classifier, just the recognizer.

Michael Levy
  • 13,097
  • 15
  • 66
  • 100