Vista Speech Recognition (under Parallels on the Mac) designed to foil verbal viruses?

I noticed a couple referrer links regarding the speech recognition for Vista running under Parallels post, linked to my blog - I figured I could elaborate a little.

I had it running in my office and a couple friends came by and we were all telling it things to do, “open paint”, etc. I figured out you can navigate around a webpage by saying the proper titles of the various headings etc. Then it occurred to me, why can’t the Mac control Vista via sound. I mean, I can have a text file of commands and have the Mac read it in voice mode and then see if Vista would perform the tasks. I played with this a little bit and couldn’t get it nailed due to a couple things I noticed.

It appears to me that there is intentional adjustment of the commands given, i.e. if I verbally spoke the same series of commands, the response order would change, vague commands would suggest choices (item1 vs item2) that would change in order from time to time, even though I used the exact same verbal command series each time. I’m guessing that in order to avoid voice exploits in Vista (playing an mp3 as a trojan horse) the OS mixes up the responses to foil any attempts down that path. But those are just my ideas based on my observations.

Anybody on the design side want to comment? I’d love the feedback.

Leave a Reply