Experimental Comparisons of Data Entry by Automatic Speech Recognition, Keyboard, and Mouse
DEFENCE SCIENCE AND TECHNOLOGY ORGANISATION SALISBURY (AUSTRALIA) ELECTRONICS AND SURVEILLANCE RESEARCH
Pagination or Media Count:
The objective was to determine the conditions under which Automatic Speech Recognition ASR is an efficient choice for data entry. In particular the focus was on data entry tasks that are part of constructing military messages. The ADF Formatted Messaging System utilises a structured formatting system to constrain the semantics of a message but also includes a field for unlimited and unstructured text. Hence the data entry tasks involved range from form-filling to free dictation of short phrases. In the experiments, ASR and manual input modes are compared for three data entry tasks textual phrase entry, selection from a list, and numerical data entry. To effect fair comparisons, the tasks minimised the transaction cycle for each input mode and data type and the main comparisons use only times from correct data entry. The results indicate that for inputting short phrases ASR only competes if the typists speed is below 45 wpm. For selecting an item from a list, ASR offered an advantage only if the list length was greater than 15 items. For entering numerical data, ASR offered no advantage over keypad or mouse. The general conclusion for formatted data entry is that a keyboardmouse interface designed to match the data to be entered will be more time efficient than any equivalent ASR interface.
- Voice Communications