Overall, the whole situation is kind of a mess. I think you'll mainly have to decide what you want the commands to be - a thin layer on top of whatever is available or a tightly controlled standard interface. My vote would be for the former, but if you implement the latter, we can always continue to use the native methods of course.
Thanks for the feedback. I will go ahead with the former in that case, so that the voice-related commands are available to voiced users and higher.
