Voice recognition device as a computer interface for motor and speech impaired people.

A voice recognition device has been identified as an interface for personal computer control by patients presenting with high-level spinal cord injuries and mild dysarthria. The device enables a person to bypass the traditional keyboard and activate a computer through voice control. The user simply creates templates of a spoken vocabulary in computer memory. The computer then matches real-time spoken words to stored templates for activation. The system has been clinically tested with one 10-year-old boy who has C1-C2 quadriplegia and a 19-year-old man who sustained a C6 spinal cord injury and dysarthria secondary to head injury. Each patient created a vocabulary for computer storage to run educational software packages and games. Accuracy rates for computer speech recognition were measured in multiple practice sessions. Intelligibility of single words produced by the dysarthric speaker was measured also. Both patients activated computer programs through voice control. Recognition rates ranged from 45 to 60% for the first patient, and 79 to 96% for the second patient. A mean success rate for voice recognition across trials with both patients was approximately 80%. The device increased the dysarthric speaker's articulatory precision. Results indicate that the system might be appropriate for rehabilitation programs though further technologic refinement of the device would increase its effectiveness.