Speaker identification system using Wavelet Transform and neural network

Abstract

The speech enhancement that is concerned with the processing of corrupted or noisy speech signal in order to improve the quality of speaker recognition system is presented. This idea of noise cancellation for the speech signal is processed to increase the speaker recognition system robustness. The presented system is divided into two blocks: 1. Discrete Wavelet Transform DWT and Adaptive Linear Neuron (Adaline) Enhancement Method (DWADE). 2. Wavelet Gender Discrimination (WGD) and Speaker Recognition using Discrete Wavelet Transform (DWT) Power Spectrum Density (PSD). The tested signal is enhanced up to 15 dB by Wavelet Transform and Adaline Enhancement Method that increases the speaker recognition rate. The accomplished speaker recognition rate is about 95%. Back Propagation Feed Forward Neural Network (BPFFNN) perceptron classification methods are used.