[misterhouse-users] Voice Command mode

[Matt B. <ma...@re...>]

I've lurked for a while and looked at the code, but held off
on actually implementing misterhouse until after completing
a move. Now that's done, and our local misterhouse
installation is in its infancy; we're very happy with it so
far.

One thing I haven't implemented yet, but one of the features
my wife is most looking forward to, is voice command. I've
read through the documentation on it, and skimmed some of
the code, and there's just one thing bugging me about the
way voice commands are implemented. Creating a new command
is all very simple and well done, but the problem is that
there doesn't seem to be anything to differentiate a voice
command from ordinary speech. 

This may not be a problem for most systems - I suspect the
VR commands are mainly intended for use while sitting at a
computer or microphone - but one of the more useful aspects
of the system is the possibility of adding microphones or
intercoms in various rooms to allow voice control from
anywhere in the house. In particular, my wife wants a "memo
pad" feature where she can walk through the kitchen telling
the computer items for the shopping list, which the computer
will then store in a list format. Conceptually, this is not
difficult, just something like (pseudocode only)

VoiceCommand("Start Memo")
speak("Ready")
until voice("Stop Memo") {
  $memo=<VOICE>
};

The actual code, of course, would likely be much more
hairy... but anyway, the point is that it might be nice to
include a "voice command mode" that notifies mh that it's
time to listen - otherwise, it ignores speech. Say the name
of the computer by itself, pause for a tone or other
evidence the computer is listening, then give the command.
Close the command mode through an explicit command, or
through a more complex ruleset involving non-recognized
command phrases - i.e., noticing that the human isn't giving
commands anymore. So now, the voice commands to create that
shopping list would be something like

Human: "Computer"
Computer: "Waiting."
Human: "Start memo."
Computer: "Ready"
Human: "Fourscore and seven..."
Human: "Thank you, Computer."
Computer: "You're welcome."

Of course, the command phrases could be easily configured...
anything from "Rosie" for Jetson's fans to "Hey you" to
musical chords. And the replies could be just as
configurable - or even randomly selected from lists of
possible responses. 

I'll probably start on this myself before long, but I didn't
want to reinvent the wheel if this has already been done or
started.