An address data entry system with a multimodal interface including speech recognition
Osamu Yoshioka, Kazuhiro Arai, Noboru Sugamura, Shigeki Sagayama · Systems and Computers in Japan · 1999
In this paper, improvement of the efficiency of data entry operations by adding a speech input function is reported. We implemented the multimodal interface on the desktop terminal for performance evaluation. An efficient and pleasant-to-use communication between man and machine is expected to result from the combination of speech input and a multimodal interface which can compensate for defects of speech recognition. In this paper, we apply the system to address data entry and evaluate its performances. Users can enter the address data by speech input, by menu selection from a display of ten words in one field, or by one-by-one selection display of the next candidate word, with input by keyboard. In the evaluation experiments, 25 operators performed 22,860 address data entry tasks and were instructed to use the speech input function in half of the tasks. From the evaluation results, the increase of operation time due to an increased number of entry data was reduced, and operation time was shortened by about 12% using the speech input function. The recognition rate of speech containing pauses is found to be significantly lower. © 1999 Scripta Technica, Syst Comp Jpn, 30(9): 64–73, 1999