The Expressive Speech Project is developing advanced speech synthesis technology for use in training simulations, emphasizing both high quality and rich expressive prosody. We utilize limited domain synthesis techniques, applied to expressive speech samples. We augment the standard limited domain synthesis approach by selecting units with desired prosodic characteristics. We plan to augment this with pitch modification techniques to further increase the expressive range of the synthetic voices. Our work is inspired by expression in drama.
This project is sponsored by the USC Institute for Creative Technologies, and is conducted in collaboration with the USC Speech and Image Processing Institute

more...

October 4, 2003 -- We recently collected speech data for a new expressive voice, and are in the process of building a new synthesizer based on that data.

pedagogical agents, speech synthesis

• W. Lewis Johnson
  
• Catherine M. LaBore
  
• Richard Whitney
  
• Jamison Moore
  
• Shrikanth Narayanan, USC School of Engineering
• Murtaza Bulut, USC School of Engineering

• Army Research Laboratory

• Johnson, W.Lewis, Dramatic Expression in Opera, and Its Implications for Conversational Agents. In AAMAS 2003 Workshop on Embodied Conversational Characters
  


• Burut, Murtaza, Shrikanth Narayanan, and Ann Syrdal, Expressive speech synthesis using a concatenative synthesizer. In Proc. of ICSLP, (Denver, CO), 2002
  


• Johnson, W.L., S. Narayanan, R. Whitney, R. Das, M. Bulut, and C. LaBore, Limited Domain Synthesis of Expressive Military Speech for Animated Characters. In Proceedings of the IEEE Text-to-Speech Synthesis Workshop, 2002