Abstract

Digit recognition is important in many applications such as automatic data entry, PIN entry, voice dialing telephone, automated banking system, etc. The speaker-independent speech recognition system for Malayalam digits is presented in this paper. One of the most important uses of speech recognition technology is creating digit recognition software for the blind. By enabling them to use spoken commands or inputs to engage with digital devices and retrieve information, such a system might empower people who are visually handicapped. The system uses machine learning techniques for recognition and Enhanced Mel Frequency Cepstral Coefficient (EMFCC) as a characteristic for signal processing. On a test set of continuous digit recognition tasks, the system achieved 96.45%-word identification accuracy using 45 male and female voices ranging in age from 15 to 50.

Description