Speech Recognition in Public Speaking

By:
Professor Jill Hewitt,
Caroline Lyon,
Carol Britton,
Dorota Jagielska,
Barry Mellor
To add a paper, Login.

This paper explores the usability issues involved in the use of
speech-to-text displays at public meetings and in the classroom. A system, "SpeakView", has been developed which provides real-time captions for live performance using speech recognition software. It is based on a previous project which developed a real-time television captioning system, now in use by two UK television companies. More recently this mode of communicating has been applied in a number of lecturing and public announcement situations, particularly where there are members of the audience who are not native English speakers. The paper will describe the product and explore the changes which have been made following preliminary usability trials. The results of a series of experiments are presented. These have been designed to assess the acceptability of the captions, to evaluate the user interface and to study the enhancement of understanding which the captions can provide. The potential for using the system in the community will be explored


Keywords: Speech, Speech Recognition, Human Computer Interaction, Accessibility
Stream: Human Technologies and Useability
Presentation Type: Paper Presentation in English
Paper: A paper has not yet been submitted.


Professor Jill Hewitt

Head of School, School of Computer Science, University of Hertfordshire
UK


Caroline Lyon

Senior Lecturer, School of Computer Science, University of Hertfordshire


Carol Britton

Associate Head, School of Computer Science, University of Hertfordshire


Dorota Jagielska

Researcher, School of Computer Science, University of Hertfordshire


Barry Mellor

Chief Executive, Speechsoft


Ref: T06P0097