SPEECH RECOGNITION USING MFCC AND DWT

Abstract

Human speech serves a very essential function as a critical biometric feature for authentication and identification. It delivers improved security, easy authentication, and significant cost savings. By contrasting the speaker's speech signal with pre-stored speech signal in the database and retrieving the major elements of the speaker's speech signal using Mel-frequency cepstral coefficients, this paper aims to build a system for speech recognition using dynamic time wrapping (DTW)algorithm, which is among the most essential element in determining high accuracy. Because it includes generating coefficients from the user's speech which are unique to each person, the Mel Frequency Cepstral Coefficients (MFCC) algorithm is widely recommended as a feature extraction technique for doing speech recognition.