摘要
为提高清浊音判决的准确度,本文提出了一种利用计算机仿真产生训练样本,采用多指标综合评价方法,运用支持向量机产生判决准则的新方法。该方法克服了传统判据标准的缺陷,试验表明,该方法的分类性能要优于其他传统方法。
For improving the accuracy of U/V decision,this paper provides a kind of training sample generated from computer simulation and new decision method produced by multiple index comprehensive evaluation way and support vector machine.This method overcomes the shortcomings of traditional decision standard.The test shows the clssification performance of this way is better than other traditional one.
出处
《煤炭技术》
CAS
北大核心
2010年第9期170-172,共3页
Coal Technology
基金
邯郸市科学技术研究与发展计划项目(项目编号:0921103118)
关键词
清浊音
训练样本
支持向量机
unvoiced/voiced
training sample
Support Vector Machine(SVM)