View Single Post
Old March 14 2012, 09:38 PM   #4
Lindley
Moderator with a Soul
 
Location: Fairfax, VA
Re: Universal Translator a reality?

So basically, voice recognition (including some kind of principal component analysis describing the deviation of the speaker's voice from the model), then translation, then text-to-speech except using the above PCA parameters to try and make the voice synthesis sound like the original speaker?

Seems straightforward enough. That's a lot of potential weak links though.
__________________
Lead Organizer for EVN: Firefly.
"So apparently the really smart zombies have automatic weapons!"
-Torg, Sluggy Freelance
Lindley is offline   Reply With Quote