I think you are projecting the limitations of current incarnations of speech into the indefinite future and missing the potential.

Does speech work?
- I use MacSpeech Dictate for all my serious writing, and it works wonderfully. It’s really come of age in the last couple of years.
- I use the Google app on my iPhone to do search. It surprises me at how well it works, given that I didn’t have to train it. Sure beats typing, and even when it’s wrong, I have less work to do to correct it than to type it in.
- I also use the Reqall app to email myself ideas all the time when the inspiration strikes. Without it I would simply not capture the idea and would forget many of them.
- Finally, messages left by people to my Google Voice phone number are transcribed pretty well, considering the low quality of the mic and that callers aren’t even attempting to enunciate for a machine.

So voice works quite well today.

And pulling off what I describe is an easier job than what Dictate/Dragon and Google do. The job of recognizing speech is even easier the smaller the vocabulary it has to recognize. With what I am talking about there are probably fewer than 100 patterns that the device would have to recognize.

As for privacy, and annoying others and looking foolish, that’s up to the user. I personally loathe hearing the half conversation of people speaking to thin air into their Bluetooth earbuds like crazy people, but they do it. Using the device discreetly is always an option.

Talking to a device does not necessitate it talking back. As you point out, that has obnoxious downsides. There is nothing wrong with talking to a device and having it “talk back” visually.

Errors will always occur; the question is what is the overall effectiveness of using the device? I can tell you that an iPhone it’s much more difficult to create a “new appointment with Carol next Wednesday at 2:30 for 90 minutes” than it would be if those words were spoken.

Personally I think this future is an inevitability and more of a revolution than multitouch and gestures that are getting the hoopla today. And I think my claim, that may sounds contentious now, will seem obvious in retrospect. It just takes one really good incarnation to prove the idea.

When you can do your most common tasks faster, and without having to make sure you are in the right mode, and without having to devote your gaze to the machine, I think you will agree.

Philip

So far voice control for things pretty much stinks. Try the new voice-operated car controls, or speak to type software… the same problem always occurs: the computer got it wrong and now it takes 5 steps to retroactively fix the problem that would’ve been easier to type myself anyway.

And what about privacy?
I ride the bus to work, not only do I not want to announce my latest app launch, but I definitely don’t want to hear everyone else speaking their commands. Plus, in crowded situations like this, there’s all sorts of input problems as far as who the device listens to.

For private scenarios (during a phone call for exmaple) voice operations like transcribing to text is fairly easy and reliable, and actually useful, yet even those a laden with errors.

Lastly, using voice tends to require a similar response from the device (talking spawns conversation, so this is natural) but take any of our lovable characters from TV and the movies, I don’t think anyone really wants to have a true conversation with their car. Plus we’re all going to look silly doing it.

Voice just isn’t there yet, and may never be. – my two cents.