MIME-Version: 1.0 Server: CERN/3.0 Date: Sunday, 01-Dec-96 19:52:13 GMT Content-Type: text/html Content-Length: 1655 Last-Modified: Friday, 09-Feb-96 04:11:12 GMT Senior Honors Project at UMass

Design and Implementation of Dynamically Generated Synchronized Computer Speech And Facial Animation

Abstract:
By integrating visual and audio information using computers, a new form of human-computer interface has been created. The current research is in the design and implementation of a user interface whereby a computer can produce speech (audio) and a moving image of a human face synchronized to the speech (video). The result is the appearance of a human talking to the user. The program is given text to speak (for proof-of-concept purposes, the numbers "0" - "9"); it then creates the speech, automatically determines the boundaries between phonemes (to synchronize the video), and uses morphing to smoothly transition between the images used for each phoneme.

A PostScript copy of the paper

Sample Output (in Video For Windows .AVI format):
"Zero"
"One"
"Two"
"Three"
"Four"
"Five"
"Six"
"Seven"
"Eight"
"Nine"
"Hello Professor Hill"