METHOD TO SIMPLIFY AN SVM-BASED SPEECH AND MUSIC CLASSIFIER FOR SELECTABLE MODE VOCODER CODEC BASED ON CONTRIBUTIONS OF SUPPORT VECTOR

摘要

The present invention relates to a simplification method of a support vector machine-based speech and music classifier, a simplification method of a support vector machine-based speech and music classifier according to the present invention facilitates implementation of a classifier into an embedded system by reducing the amount of calculation while maintaining the performance of classification of a support vector machine (SVM)-based speech/music classifier for a selectable mode vocoder (SMV). According to the present invention is able to reduce the amount of operation which is followed by an SVM-based classifier by simplifying a support vector of the SVM-based speech/music classifier for an SMV codec, and in particular, the implementation of the classifier into an embedded system is easy. [Reference numerals] (AA) Calculating the average contribution of the support vector;(BB) Calculating the relativce contribution of the support vector;(CC) Removing the support vector having the smaller contribution compared to the threshold value