VICI - A speaker independent word recognition system

This report describes the development, operation, and performance characteristics of an Advanced Development Model of a Voice Input Code Identifier (VICI). The VICI is an isolated word recognition system capable of recognizing the English digits and four control words, CANCEL, ERASE, VERIFY and TERMINATE independently of speaker for a large population of General American males. A visual feedback system has been incorporated to allow the speaker to verify the correctness of each digit entry and if necessary to correct a faulty entry by the use of the control word, ERASE, and then enter a new digit. A complete code group of four digits can be accepted by the use of the control word VERIFY or rejected by the word CANCEL. The speaker can view on an alphanumeric display each recognized digit within .1 to .2 seconds after it is pronounced in order to verify the correctness of each digit entry. Live tests involving a total of 30 speakers showed that a four digit group could be entered into the VICI system with verification in as short an interval as 2.8 seconds. Four to seven seconds were typically required for most speakers for a digit group if no errors were made either by the speaker or the system. Ten to 12 seconds were required by most speakers to detect and correct an error and complete the entry of a proper code. It was necessary to employ correction for an average of 7.5% of the 75 digit groups spoken by the 29 participants in the live tests. In every case the errors were correctable and every code was entered properly.