COST278 and ISCA Tutorial and Research Workshop (ITRW) on Robustness Issues in Conversational Interaction
University of East Anglia, Norwich, UK
Recent improvements in speech technology are expected to change the way we communicate, facilitating access to web services and applications (voice portals, multimodal email clients or games) in various mobile situations. One threshold for broad adoption of speech enabled services is their limited interaction robustness. In this paper, we focus on possibilities to increase interaction robustness by enhancing speech output with non-speech sound. First, we describe the relationship of speech and non-speech output and their contribution to interaction robustness from a user-centered perspective. Second, we report on a user study, in which the addition of non-speech sound to Text-to-Speech (TTS) in an Email reader was evaluated. The results indicate that in most cases the addition of tailored non-speech sounds increases the user’s ability to perceive the structure and meaning of the audio content and decreases the subjective workload, without leading to a higher user annoyance. Finally, we provide an outline of further, currently ongoing research work on nonspeech sound enhancement of speech output.
Bibliographic reference. Fröhlich, Peter (2004): "Increasing interaction robustness of speech-enabled mobile applications by enhancing speech output with non-speech sound", In Robust2004, paper 22.