It is a computerized voice that is made to sound as close to a human voice as possible. There are companies that make and brand these computer voices (everything pre-recorded) and sell them to other companies.

For example, AT&T makes a few computerized text to speech voices named “Samantha” and “Tom.” Let’s say you buy a Mio (GPS) and your’re looking at the languages. You’ll see under the english option “Female”, “Male”, “Samantha” and “Tom”. Male and Female will still read your directions, but not the street name. That’s what Samantha and Tom were created for.

Now, if you’re talking about an actual human voice…

There’s gps companies out there who offer celebrity voices as a downloadable option on the unit. Although it’s a fun idea, these voices will not provide text to speech. Because it’s an actual human voice and not a computer, it would’ve take years to record every street in the United States, not to mention correcting it every time there is a map update.
Make sense?

Hope this helps!