Primi Isolated Words Spectrogram Classification by Support Vector Machine Based on Immune Genetic Algorithm

Abstract

We propose a method for Primi isolated words spectrogram classification by support vector machine based on immune genetic algorithm (SVM-IGA). Firstly, time-frequency spectrograph of Primi isolated words is generated by Short Time Fourier Transform (STFT). Secondly, binary feature is extracted by binarization spectrogram. Thirdly, spectrogram classification is realized by IGA-SVM. The experimental results show that the predictive accuracy rate of Primi isolated words spectrogram classification was 88~91%. Compared with the speech signal classification, the spectrogram classification by SVM-IGA is better.