I thought about doing so with my current video, but decided not to develop it that way for a few reasons.
My concept originally began with how Mac OS X's speech recognition can function you can tell the OS to constantly listen using the microphone for a keyword or phrase. Upon hearing that, the OS would then act accordingly to the command stated after the keyword.
The reason why I didn't do this is because of 1) battery life? 2) privacy concerns. I'm sure if I posted a video of such, many of the comments would discuss how the iPhone is always listening...wherever you go. A computer is stationary, but everyone carries their phone everywhere. And with that, the phone is always listening. 3) Many people might have their default keyword be "iPhone". How many people have iPhone's now? I say the word iPhone on my university's bus, and everyone's phone would then ring that familiar sound.
Of course, I have gotten this kind of feedback a lot (re: much more developed speech recognition that should listen in and act accordingly). In the next few days, look out for another video by me that demonstrates such.