Speech Controlled Animation

Andrew Coates

July 17, 2002

 

 

 

Submitted for the degree of Bsc in Computer Science at the University of York

 

 

 

Number of words = 26,355 as counted by the Unix word count command on the latex source file
This includes all of the body of the report but excludes all appendices.

Abstract:

Speech recognition, natural language processing and animation are rarely combined into a single application. The problems each field faces are reviewed and a description of the challenges each may place on an application is provided. A prototype system blending all three technologies is then designed, integrating IBM ViaVoice, Mercury and OpenGL, and implemented resulting in a functioning application capable of allowing control of a virtual character using spoken natural language. The design of the system is given, followed by the details of the implementation. The system is theoretically and empirically evaluated and the results are discussed. The characteristics of the final system are overviewed, along with its applicability and feasibility. The paper is concluded with a discussion of potential extensions to the prototype.