TwoSeven: I have cortana working but I don't use it.

The main thing that stops me is having to say "hey cortana", rather than the name I have given my machine - which would help in a multi-machine environment.

The other thing is the lack of detailed voice training so that it can recognise my English accent. It would be good if I could also add my own voice commands to menus (shortcuts, contacts etc) for the same reason.

Same issues occur with Siri as well which is why I don't use that either.

Good feedback. We're currently working on optimising the multi machine multi room scenario

The voice training and the voice naming (tagging) of machines are the two important ones. Windows used to have a good voice training system in it, but it seems to have vanished.

I must admit, I have a couple of real sense cameras and a Kinect for Windows (I have an interest in NUI's) and I can see the need to be able to issue commands to machines without wanting to sit in front of them - getting cortana to understand msbuild and web deploy for example. Being able to refine individual voice prints/tags I think would also help in a noisy environment with multiple people speaking.

I don't mind learning a bit of sign language for when I need the sound turned down or if I am on a call.

It would also be good if metservice had cortana integration regarding weather.