Just finished a simple demo using the WAMI speech recognition API. It's pretty cool and very easy to work with. The commands in my example are structured like this:

schedule <activity> with <employee> at <starttime> for <duration> hours.

activity: 'call' or 'meeting'
employee: any name in the list
starttime: one two three etc... (only whole hours for simplicity)
duration: one two or three

To perform a command, just click the "Hold to talk"-button and say something. Here are two example commands you can try.
  • "Schedule meeting with Mike at eleven for three hours."
  • "Schedule call with Arnold at eight for one hour."

Click this link to try it out: