You can outsource the voice recognition. Windows[1] and Mac OS[2] both seem to include speeech recognition APIs these days, and of course, what with it being 2014, there are online APIs[3] and forthcoming browser support[4]. All of them appear to support limited command grammars, which is perfect for this sort of simulation.

[1]: https://en.wikipedia.org/wiki/Microsoft_Speech_API [2]: https://developer.apple.com/library/mac/documentation/Cocoa/... [3]: http://api.ai [4]: http://www.smashingmagazine.com/2014/12/05/enhancing-ux-with...