Controllable, Expressive Speech Synthesis



Devised by Vincent Wan

Written by Vincent Wan, Javier Latorre and

BalaKrishna Kolluru

Story by Kayoko Yanagisawa

Cartoons by Patrick McCarthy


Outline:  Most of today’s speech synthesizers read speech in a neutral (boring) way. That’s why you wouldn’t want a computer to read you a story. But what if a computer could read as well as a real person.  This exhibit explores this possibility. It allows you to create a story for a short comic strip and have a synthetic voice read it back to you in characters and emotions of your choosing.

 Research:  Choosing the best emotion for a piece of text is tricky even for a human. That is why we have professional narrators to read audio books. However there are so many books and texts out there with new content being created daily in the form of newspapers and magazines that it is practically impossible to have all content read aloud by a human voice. But what if a computer could read as well an average person? How would you use such technology?