Ketchup

[lazyfatbum]

Nintendogs has pretty advanced voice recognition and it's available in multiple languages.

Alot of companies are using voice recognition for the tech help services. It picks up words based on key sound structure. The sound it begins with, the sound it ends with, the constanents and syllables used and the tonal qualities.

If you threw out a sentence like "Hand me the shoe" a voice regnition system could pick up any of those words based on how it's structured. The game would understand that in that instance you are going to have a set finite amount of options:

Hand me the:

Shoe
Fork
Ball
Lawnmower
Pencil
Dildo

that this particular part of the game will allow you to ask for, it could then look for a set number of commands, again finite.

Hand
Throw
Drop
Hide
Break
Use
Combine
Give

And then have a finite set of secondary actions.

Me
You
Them
They
Their
Him
Her
Kid
Child
Man
Woman
Guy
Girl
Person
I

Now that it knows those particulars, the rest of the sentence doesn't matter.

"Use (the) dildo (on the) girl"

"Throw (the) fork (at) him"

"break their lawnmower"

I would agree that having a conversation is way off but the illusion of one isn't. Since it's a video game you could have a realistic depiction of a conversation because it would be extremely limited and finite. When you're driving along and you get pulled over, you cant talk to the cop about your favorite coffee for example, there's a structured set of of words you can use only here. Later in the game you'll be running through an apartment and again use a set finite set of words asking people to hide or open their doors, hold the elevator, etc. That the game would be looking for.

It wouldn't be full on conversation but the illusion of it (multiple voice commands in strings) would offer alot to the look and feel of a conversation and more importantly it can be done today using existing technology

Am I wrong?.