Unnatural voices
by Bob on November 15, 2007
There are spooky and ghostly sounds we hear sometimes, and then there is the new world of un-natural, artificially synthesized voices when we call companies and answering machines. These odd, seemingly gender-less voices, also make anouncements on the train and in the elevator.
These voices are eerie and also annoying due to their flat tone and odd sentence formations. I'd rather have a live human being still doing the talking.
Speech Synthesis is an easier problem than Speech Recognition in Engineering and Computer Science, specifically in Artificial Intelligence. But it's still hard to make it sound really human, unless very specific tricks and techniques are used. Then there's Emotional Speech Recognition being worked on too. To be able to ascertain the speaker's emotional state and sentiments from the speech patterns as stored in digital format. It's a subclass of AI problems known and involved in Affective Computing.
A classic computer voice which sounds all too computer-like is HAL from Stanley Kubrick's 1968 film "2001: A Space Odyssey".
It gets a little funny when an elevator gets caught opening and closing on a floor with no real people in it, and the computer voice just keeps saying for an eternity, and with some impunity, "Second Floor, Going Up", on and on and on and on, in the same monotone pseudo-voice. It has no sense, and awareness of its repeating message, of course.
Corporations keep real people away from their real people (the elusive customer service representative) by throwing a massive set of computerised and baffling button choices for the caller. And un-natural voices.
I think we should employ more real people and keep it humanly engaging. I wish the corporations agreed.