Automatic Speaker Recognition :An Approach using DWT based FeatureExtraction and Vector Quantization

Automatic Speaker Recognition (ASR) is the use of a machine to recognize a person from a spoken phrase. It verifies the persons claimed identity from his voice pattern. It is probably the most natural and economical method of restricting unauthorized access to computer network, communication system and multilevel access control. With the ubiquitous telephone network and microphones attached with computers, the cost of speaker recognition system might only be for software. In this paper speaker recognition system using DWT based feature extraction technique and VQ for pattern matching is proposed. The performance of the proposed recognizer is compared with traditional methods and found better in terms of measure of False Acceptance and False Rejection.