基于模糊C均值聚类的一种说话人识别方法的改进

A Ameliorated Method Of Speaker Recognition With Fuzzy C-means Clustering

模式匹配在整个说话人识别系统中具有重要的作用,其采取的方法将直接影响系统的识别率.本文介绍了一种模糊矢量量化(FVQ)方法,通过对模糊C均值(FCM)聚类算法的分析,提出了基于减法聚类和改进的模糊C均值聚类算法相结合的说话人识别方法,实验表明该方法提高了识别率,是一种行之有效的说话人识别方法.

Pattern matching plays a very important role in the speaker recognition system,whose method can affect the system recognition rate directly.This article presents a method about fuzzy vector quantization(FVQ) and a method of the speaker recognition with subtractive clustering and fuzzy c-means clustering arithmetic by analyzing the arithmetic to fuzzy c-means.The experiment indicated that this method enhanced the recognition rate and is a effective speaker recognition method.