I have considered it, actually, and researched some open source speech
recognition APIs.
Unfortunately the libraries tend to be kind of large (storage
footprint) and only give so-so results. Since voice recognition is
now built-in to the iPhone 4S and all future iOS-based products are
likely to support it, it may not be worth the effort to try to put it
into Frotz. Haven't ruled it out altogether, though.
Craig