Analysis of Embodied Conversational Agents in SecondLife for Speech Recognition

Abstract

A virtual world is a computer simulated environment that's interactive and allows the user to become a participant in the computational space, entering a computer world that they have created. This participation gives the user the sense of presence in the environment. In a virtual world people can select or create an animated alter ego, known as an avatar. The world can be inhabited by a single avatar or simultaneously by many creating a virtual community. These virtual worlds and communities are created and used for many different reasons such as aviation and medical training, entertainment, socializing, commerce and marketing, mentoring, and educational purposes, just to name a few. When users decide to use a particular virtual world, one of their main concerns is the quality of various attributes of the software. Some of these attributes are functionality, usability, efficiency, reliability, maintainability and sometimes portability. The focus of this thesis is on a specific virtual world application, Second Life, one of the top 10 online three-dimensional virtual worlds. This study is an intensive descriptive analysis of Second Life's virtual agents and the capability of incorporating speech recognition into Second Life.