Google just came out with a voice recognition kit that can operate locally or, with addiitonal capabilities, via their online API. Could be interesting to see it integrated with OpenHab if it could emit text to an interpreter.
There’s no mention of software in the article. I think a lot of people would be interesting in this project if it meant not using Internet based API’s for voice interactions. Keeping everything local would add another level of security. I would be very surprised if Amazon haven’t built something similar already.
I would like to mention here that a local alternative is Snips as Google, Amazon & Co. will not be involved there in any way:
Snips provides Private-By-Design, Decentralized Voice Assistant Technology and Solutions
It is already proven to be working with openHAB: