Wednesday, August 30, 2006

Vista 5536 (Pre-RC1) Voice Recognition Rocks

The voice recognition in Vista just simply rocks. I'm having the best geek time playing with it, since I installed this Vista build yesterday.

Some background first. I have Vista on an old Thinkpad (T30) which gives me a Vista rating of 1.0, so it's pretty close to bottom of the Vista performance barrel. But it DOES work...

Today I started playing with the Voice Recognition and am very impressed. I'm having the best time playing with all of its features. (My wife is having a good time laughing at me... "But honey look at how cool this is... and if I say this... Wow, check THIS out... Hey, where are you going? You haven't seen it do THIS yet... "  LOL )

For example, I just started playing with the Mousegrid feature.

When you say "mousegrid" a 9 square numbered grid is drawn on your screen (think Sudoku grid). You then say the number of the square you are interested in. You now get another 9 square in the area... i.e. you are zooming into the area you are interested in. Once you have it narrowed down you can say "click" or "double click" or "mark" (to select an item so you can then verbally drag it somewhere), etc. (Think Blade Runner and the "picture zooming" scene...)

Meaning that even if an app doesn't directly support voice, or have shortcuts, easily determinable text, etc, you can still use voice recognition to control it. For example, using the mousegrid, you can actually draw an image in MSPaint.

The bitmap below was done completely hands free, from launching the app, to selecting tools and colors, drawing the lines & circles, sizing the text box, entering the text, to the file save and close. Done entirely using Vista's voice recognition. Yeah, it's nothing special to look at, and with a mouse it would have taken 1/100% of the time, but that's not the point. The point is that this was done with voice alone (which is cool in my book)!

Okay, okay... I know I'm getting a little (little?) too excited about this. But it really is "officially cool."

Cool in that my lower end notebook does an okay job handling the recognition. Cool in that I've not even really trained my machine yet (I got about half way through the tutorial/training and just HAD to play with it "in the wild"). Cool that while voice recognition has been around for years now, this technology may be finally be close to truly usable.

Sure it's not perfect and some of its mistakes are pretty funny...

Still I have to say this implementation is pretty darn impressive. Good job guys.

No comments: