An electret condenser microphone based system is employed to pick up the laryngeal vibrations corresponding to each word
The corresponding MFCC feature vectors of the laryngeal vibrations for each word is found out and stored in a database.
The feature matching is done to find the word subject intended to speak. During testing phase feature extraction of the testing word is done and it is compared with the features of the already existing words.
The correct word identified using the feature matching process is then passed into the Microsoft SAPI which produces the sound.
Team Lead
Market Lead
Hardware Designer
Developer Associate