Constructing modular speech interface to remote applications

Speech technology is becoming popular within personal and enterprise computing and is being used to develop speech interfaces to various applications. In this thesis a prototype speech interface that accesses remote application was developed. This interface uses speech recognition as input and speech synthesis as output and navigates through the remote application called sihlos. New concept of deploying and accessing speech-accessible hyperlinked objects over the Internet and downloading the grammar associated with them is introduced. Methods to access sihlos and choice of speech technology were investigated.; The interface was constructed using Java, IBM's implementation of JSAPI. Different users used the interface and from different locations over the Internet, findings from this investigation are discussed in this thesis.