首页 | 本学科首页   官方微博 | 高级检索  
     检索      

语音特征的降维变换与特征鲁棒性
引用本文:王昆仑.语音特征的降维变换与特征鲁棒性[J].新疆师范大学学报(自然科学版),2000,19(3):15-19.
作者姓名:王昆仑
作者单位:新疆师范大学计算机科学系,乌鲁木齐
摘    要:本分析了语音识别中的特征降维变换方法和特征鲁棒性问题,经实践证明:基于CDN分布的降维方法适合非特定人语音识别,具有突出的优越性,通过调整模型参数并使用含噪特征训练模型,解决特征鲁棒性问题。

关 键 词:语音特征  语音识别  降维变换  鲁棒性  时域特征

THE LOWER DIMENSION TRANSFORMATION AND ROBUST QUALITY OF SPEECH CHARACTER
Wang Kunlun.THE LOWER DIMENSION TRANSFORMATION AND ROBUST QUALITY OF SPEECH CHARACTER[J].Journal of Xinjiang Normal University(Natural Sciences Edition),2000,19(3):15-19.
Authors:Wang Kunlun
Abstract:This paper has analysed the question with transformation of characteristic lower dimension and characteristic Robust quality. It is certified by practice that lower dimension means based on CDN distribution suits the speech recognition of independent speaker. It has outstanding advantages. The problem of characteristic Robust quality can be solved by adjusting model parameter and using learning model with voice confusion.
Keywords:Speech character Transformation and Robust quality Speech  recognition of independent speaker
本文献已被 CNKI 维普 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号