Practical issues in the application of speech technology to network and customer service applications

This paper proposes a simple model to characterise the different stages of short telephone transactions. It also discusses the impact of the context o f the caller when entering an automated service. Three different styles of service were then identified, namely, large vocabulary information gatherin g, spoken language command and natural language task identification for hel pdesks. By considering human dialogue equivalents. the requirements for eac h style are considered. Consequently, it is shown that each style pushes di fferent technological limits. Three case studies, selected from current pro ject from BT laboratories, are presented to highlight the practical design issues in these different styles. The styles and case studies presented are : Information gathering - UK name and address recognition. Spoken language command - network service configuration. Natural language helpdesks - BT operator services. It is shown that large vocabulary information gathering systems require hig h accuracy, careful data modelling and well-designed strategies to boost co nfidence and accuracy. Spoken language command requires dialogue and gramma r design and test complexity to be managed. Natural language task identific ation requires large volumes of training data, good learning algorithms and good data generalisation techniques. These styles can be mixed into a sing le interaction meaning that design frameworks of the future will have to ad dress all of the aspects of the different interaction styles. (C) 2000 Else vier Science B.V. All rights reserved.