基于改进的模糊ART的语音信号端点检测与切分

Speech signal endpoint detection and separation based on improved fuzzy ART

分析并讨论了模糊ART在处理样本数较大的分类问题时出现的连接权向量的"饱和"问题及某些特定情况下分类不准确的问题,在此基础上提出了改进的IFART算法,并将其运用于语音信号的端点检测与切分,实验结果表明该算法在不同的噪声环境下都能获得较好的分类效果。

This paper analyzes and discusses the saturation problem of the connection weight vector of the fuzzy ART when the processing sample's number is bigger and the inaccurate problem of classification with some special case, and proposes the improved IFART algorithm on this foundation, it is applied to the speech signal endpoint detection and separation, the experimental results show that this algorithm can have better classification effect under the noise environments of difference.