Method for improving audio classification accuracy through mixed component clustering Fisher scoring algorithm

Abstract

The invention discloses a method for improving the audio classification accuracy through a mixed component clustering Fisher scoring algorithm. The method includes the steps that all class of GMMs are united, and Gaussian components are combined into one Gaussian; a CGMM is formed; Fisher transform is performed on the CGMM; the Fisher score is solved to obtain the equal length characteristic. According to the method for improving the audio classification accuracy through the mixed component clustering Fisher scoring algorithm, each class of GMMs are united; the Gaussian components are combined into one Gaussian; the CGMM is formed; Fisher transform is performed on the CGMM; and the Fisher score is solved to obtain the equal length characteristic. The method combines the advantages of a generative mode and the advantages of a discriminant model, the differentiating characteristics among classes can be described, details can be well differentiated, and particularly when the fragment length of an extracted characteristic is small, high classification accuracy can still be achieved. Through the method, the classification accuracy of six voices can reach 77 percent.

Description

technical field [0001] The invention belongs to the application field of Fisher score algorithm, and in particular relates to a method for improving audio classification accuracy by using mixed component clustering Fisher score algorithm. Background technique [0002] Currently, Fisher scores, which are generated based on generative information, try to extract more information from individual generative models than just their output probabilities. The purpose of the Fisher score transformation is to analyze how the score depends on the model, which parts of the model are important in determining this score, and thus gain information about the internal representation of the data. From this point of view, it seems natural to compare the similarity of two data points by stretching the direction of the model parameters, that is, treat the score function of the two points as a function of the parameters, and compare the two gradients. If the two gradients are similar, it means t...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide a method for improving the accuracy of audio classification by using the mixed component clustering Fisher score algorithm. It is intended that the existing methods are easy to ignore the distinguishing details of some categories. When the length of the audio segment is short, The effect obtained will be worse, and the problem of low processing efficiency

Images