Performance Analysis of Text-Independent Speaker Identification System

In this paper, the text independent speaker identification system is implemented and analyzed using the speaker specific features extracted by applying image processing techniques. The speech is converted into spectrogram, an efficient representation of the speech signal in the form of pattern. Radon Transform (RT) and Discrete Cosine Transform (DCT) are applied on the spectrogram to extract the features. Radon Transform is applied for eight orientations on the acoustic characteristics of the spectrogram. After applying RT, the two dimensional DCT has been applied on Radon projections yields low dimensional feature vectors. The performance of the algorithm has been evaluated with our own created BCS database using sound forge version 5.0. The effect of number of Radon projections and DCT coefficients are analyzed. The demonstration program is coded by MATLAB V 7.0.4. The recognition rate of this algorithm is 96% on BCS database. (c) 2012 Published by Elsevier Ltd. Selection and/or peer-review under responsibility of Noorul Islam Centre for Higher Education