首页> 外国专利> SYSTEM AND METHOD FOR SPEAKER ADAPTATION USING BILINEAR MODEL

SYSTEM AND METHOD FOR SPEAKER ADAPTATION USING BILINEAR MODEL

机译:基于双线性模型的扬声器自适应系统和方法

摘要

PURPOSE: A system and a method for adapting a speaker which used bi-linear model are provided to improve performance of speaker adaptation by controlling the number of estimation parameter according to the number of adaption data. CONSTITUTION: An observation matrix configuring unit(22) comprises observation matrix by using SD model which is comprised by considering mean vector. A bi-linear model configuring unit(23) calculates asymmetric bilinear model parameter by applying SVD about the comprised matrix. The bi-linear model configuring unit comprises a bi-linear model by performing linear transformation as a space which reflects the style of a speaker. If an adaptation data of a new speaker comes in, a speaker suitability part approximates only the style factor to constitute a speaker adaptation model.
机译:目的:提供一种使用双线性模型的说话人自适应系统和方法,以通过根据自适应数据的数量控制估计参数的数量来提高说话者自适应的性能。组成:观察矩阵配置单元(22)通过使用SD模型来观察矩阵,该SD模型考虑了均值向量。双线性模型配置单元(23)通过对所包含的矩阵应用SVD来计算不对称双线性模型参数。双线性模型配置单元包括通过执行线性变换作为反映扬声器风格的空间的双线性模型。如果输入了新讲话者的适应数据,则讲话者适合度部分仅近似样式因子以构成讲话者适应模型。

著录项

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号