Quality prediction models for telephone-based spoken dialogue systems
Sebastian Möller · 2004
The spoken interaction between a human user and an application based on speech technology (e.g. for railway information or telephone banking) gains ground in modern telephone networks. The underlying applications contain speech recognition and understanding components, a dialogue manager, a response generation component, and a component for output of pre-recorded or synthesised speech. The interaction between these components is rather complex, and it is difficult to decide how well individual components have to perform to guarantee an acceptable overall quality for the user.