VOICE SOUND RECOGNITION.
Abstract
The goal is to determine the feasibility of recognizing a large vocabulary (250 words) by using the Single Equivalent Formant parameters. Five speakers (three male and two female) uttered thirteen of these 250 words. The thirteen words chosen were difficult words for recognition since they contained a maximum of phonetic similarity such as from, some, he, she, be, me, etc. The over-all recognition rate was 95.4%, error rate 1.5% and non-recognition rate 3.1%. Word recognition was based on phoneme category code sequence of the words. The results show that the Single Equivalent Formant and the recognition of phoneme categories is a practical parameter upon which to base a 250 word vocabulary recognizer.
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 01, 1966
- Accession Number
- AD0802997
Entities
People
- Casimir F. Piotrowski
- Louis R. Focht