Ask-A-Linguist Message Details
|Subject:||George Zimmerman and voice recognition|
First, thanks for this site. It is an awesome resource for us amateur linguists.
So they are talking in the news about using voice recognition to see if it was George Zimmerman screaming on the 911 tape. Here are my questions:
What parameters are used in this kind of analysis? I assume pitch is a feature but what other parameters (right word? I mean factors such as verbal quirks or goodness knows what else) are there?
Can someone disguise/train their voice to skew the data?
How accurate would you say the voice recognition technology is at this point?