 |
ESCA - NATO/RSG 10 Workshop on
|
 |
|
ESCA - NATO/RSG 10 Workshop on
|
|
Speech Recognition technologies have been developed largely in terms of criteria relevant to the evaluation of the technology itself, such as vocabulary size, speaking style, and speaker characteristics. The application of this technology to the automation of existing tasks such as command-and-control, or office-dictation, have encountered a range of usability problems not previously considered important. Such factors include the type of feedback, the nature of user and system training, adaptation, and the methodology required for correction and repair of incomplete or inaccurate communication. A case study derived from extensive testing of the different types of interfaces employed for office dictation using a large-vocabulary speech recognition system is given. A new classification of speech Interfaces is proposed concentrating on the granularity of speech input, the type of feedback given, and the adaptive nature of the interface. This is presented as a step towards the establishment of some principles of dialogue design for speech interfaces in general..
Bibliographic reference. Sharman, R. A. (1993): "Speech interfaces for computer systems: experience with an automatic dictation system", In AST-1993, 113-116.