Ever hear a computerized voice mail system? You can't mistake it for a real person. Maybe just for a second, if you're not paying attention.
The text-to-speech feature of the Kindle reader sounds like that.
When the Kindle 2 came out about a year ago, some publishers were sure Amazon was trying to do an end run around them by allowing their digital books to be read with that robot voice. The publishers thought that readers would stop buying professionally recorded audiobooks if they could listen to books with computerized speech instead.
I've had my Kindle for almost a while now and I use the text-to-speech feature occasionally, when I'm in the middle of a good book and have to do something that precludes reading, like ironing. Ten or twenty minutes is the most I can listen to the synthetic voice. Or iron, for that matter.
The voice itself isn't terribly distracting, although it does read everything in the same monotone, regardless of what's happening in the story. The distractions for me are the robot's pronunciations. For instance, I am reading a gossipy book about the 2008 presidential campaign, Game Change. I am pleased to find that since last summer when I read Renegade: The Making of a President, the robot has learned that “Obama” doesn't rhyme with “Alabama.”
Robot Gone Bad: "Open the iPod bay doors, Hal."
The robot doesn't know the difference between the past and present tense of “read,” and sometimes gets it wrong. It pronounces “wound” the same way, whether it's a war wound or tightly wound. And the stress on “record” is on the second syllable whether it's to record a message or about setting an Olympic record.
Text-to-speech technology is getting better. Nuance, the company that provides Kindle's voices (there's a male and a female voice ) also has voices in various languages with a variety of accents. Here's a few samples.
They sound pretty good compared to this AT&T demo, in which you can enter whatever text you like and it reads it back to you.