Consonant and vowel sequences as decision units for automatic speech recognition
Abstract
A method is discussed for automatic speech recognition based on the recognition of syllables as an essential intermediate step. The vowelled syllable core is thereby separated from its consonantal beginning and end. These parts are then used as decision units in automatic speech recognition. Since the vowelled core can be easily recognized, tests were preferably made on beginning and ends of syllables. A very general recognition method was used, based on a definition of a similarity measure between speech signals and the classification of an unknown signal by comparison with a sufficiently larger set of known reference signals.
- Publication:
-
In its Speech Signal Processing p 83-105 (SEE N77-13291 04-32
- Pub Date:
- June 1976
- Bibcode:
- 1976ssp..nasa...83S
- Keywords:
-
- Consonants (Speech);
- Decision Making;
- Speech Recognition;
- Vowels;
- Classifications;
- Signal Processing;
- Syllables;
- Voice Communication;
- Communications and Radar