Speaker recognition using discrete wavelet transform and artificial neural networks

Abstract EN

In recent years biometrics has emerged as applied scientific discipline with the objective of automatically capturing personal identifying characteristics that distinguish one individual from another and using the measurements for security, surveillance, and forensic application.

Speaker recognition is the process of automatically recognizing who is speaking based on individual information included in the speech waves.

This paper presents the speaker identification method based on Discrete Wavelet Transform (DWT) and Artificial Neural Networks (ANN).

In this study the DWT is used to extract a speaker's discriminative features from the mathematical representation of the speech signal.

These feature vectors are used to train a feedforward neural network which is used to model the speakers and make the decision task.

A database of 20 speakers (10 male and 10 female) has been used with a vocabulary of Kurdish words.

The system led to 100% identification rate for text-dependent and 80% identification rate for text-independent.