Development of Voice Recognition Control Systems
by
Todd Whiteman and Michael Ostrego
Advisor: Dr. T. L. Stewart
Speech provides one of the most convenient and rapid channels of communication
between man and machine. However, there are many barriers which inhibit
wider applications of such communications. From a technological viewpoint,
major obstacles to speech recognition come from problems related to speaker
independence and continuous speech. This presentation discusses the proposed
development of a control system using speech as the controlling input.
It also examines various markets involving voice recognition systems, current
problems involving voice recognition, and various approaches toward voice
recognition. Early research indicates that current speech recognition control
systems are costly and not readily available. For this reason, this project
proposes to develop a system with a limited vocabulary and a projected
recognition accuracy of around 90%. These constraints decrease cost and
increase the portability. The system will be able to recognize a series
of one word commands and have a relatively fast recognition time.
|