Can I improve Google speech API recognition by give him a words list (in my case the request of user is very predictable) to make recognition more accurate?
Correct answer is: no, you can't. =(