I had a thought while inspired by this little video of a hilarious attempt to use vista’s voice control and I got to thinking about just what is wrong with most attempts at voice control software, and what can be done right.
Essentially, unless you’re physically impaired (and maybe even then) you do not need voice control to work on every aspect of the OS, similarly you do not need it trying to parse input and commands all at once. Taking a lesson from Vim, you should have an input mode, a “command mode”, and perhaps a few more modes. So lets go over an example of what the program might do, comparing it to how I do things in deskbar.
If I want to open up my text editor, I hit alt+F3 and type in text, and it brings me up a list of actions: “Lanch gedit, launch scribes, etc.” all text editors, deskbar also gives me a bunch of other options, such as searching online for text, scrounging through my emails, etc.
With a Voice-deskbar like app, I’d say something along the lines of: “Computer: Open Scribes.”, I’d actually recommend something different from “Computer” such as “Linda” or “Serenity” or something else. Basically it’s a reference to tell the computer you want its attention.
Next you’d have a number of modes: Open, Input, Close, Suspend, Turn Off, Hibernate, Search, Find, Locate, Save, Dictate, Delete, etc. And maybe a visual cue to what “Mode” you’re in. This way, if you told the computer to open Scribes (or Tom Boy), You’d then say “Computer: Input.” and from there
it would save what you had said as voice input into Scribes or Tomboy. Telling it “Computer: Dictate” would instead start the record function, and save your voice as a .ogg file.
As a further example, “Computer: Search Yubnub, gis Firefly” Would search yubnub in firefox and parse: gis Firefly in it (which would in turn tell yubnub to search google image search for firefly. Hence Search is tied to the search engine field in firefox. Find would instead run the Find program in Nautilus, locate would instead use locate or tracker or beagle, respectively.
This is a far more useful application of voice control technology, than attempting to navigate menu items and such which would be better done with either the mouse or the keyboard. And I believe this is actually the right path for voice to take in terms of Human/Computer Interface. E.g., stop trying to do that which is better done by either the mouse or the keyboard.
Edit: Also, instead of using say “Computer” or any other voice command to get the computer’s attention, it could be activated via a modifier key. Such as an unused or seldom used media key, etc.
