Yeah, I had that idea with labeling the chessboard too, and forgot it ^^ Thanks for reminding me.
Concerning misunderstandings of the voice recognition engine, you can try to train it so it fits your pronounciation:
https://support.microsoft.com/en-us/windows/use-voice-recognition-in-windows-83f...
To make black react faster, you can edit nodes.txt and set it to a lower value. It will make the chess engine less powerfull, though.