It's mostly great which makes the minor faults that much more frustrating. For example, for me it can't recognize the word "add" 19 times out of 20. It'll hear either "and" or "had" or "Ed". So I'm trying things like using a finger to close one nostril while I say it and that somehow helps a little. Repeating the same sentence multiple times is frustrating. Do we have any controls to fiddle with other than allow personalized recognition? Any other hints on how to speak to help improve recognition?