首页> 外国专利> MASK CALCULATING DEVICE, CLUSTER-WEIGHT LEARNING DEVICE, MASK-CALCULATING NEURAL-NETWORK LEARNING DEVICE, MASK CALCULATING METHOD, CLUSTER-WEIGHT LEARNING METHOD, AND MASK-CALCULATING NEURAL-NETWORK LEARNING METHOD

MASK CALCULATING DEVICE, CLUSTER-WEIGHT LEARNING DEVICE, MASK-CALCULATING NEURAL-NETWORK LEARNING DEVICE, MASK CALCULATING METHOD, CLUSTER-WEIGHT LEARNING METHOD, AND MASK-CALCULATING NEURAL-NETWORK LEARNING METHOD

机译:掩码计算设备,簇权学习设备,掩码计算神经网络学习设备,掩码计算方法,簇权学习方法和掩码计算神经网络学习方法

摘要

A cluster-weight calculating unit (322) calculates weights for a mask calculating NN in which at least one layer is decomposed into a plurality of clusters, wherein the weights individually correspond to the plurality of clusters, on the basis of signals of the voice of a specific speaker by using a cluster-weight calculating NN. A mask calculating unit (302) calculates a mask for extracting feature quantities of the voice of a specific speaker from feature quantities of observation signals of the voices of one or more speakers on the basis of the feature quantities of the observation signals of the voices of the one or more speakers by using the mask calculating NN weighted with the weights calculated by the cluster-weight calculating unit (322).
机译:簇权重计算单元(322)基于语音的信号来计算用于掩模计算NN的权重,其中,至少一层被分解为多个簇,其中,权重分别对应于多个簇。使用聚类权重计算NN的特定说话者。掩码计算单元(302)基于一个或多个说话者的语音的观察信号的特征量,计算用于从一个或多个说话者的语音的观察信号的特征量中提取特定说话者的语音的特征量的掩码。通过使用由簇权重计算单元(322)计算出的权重加权的掩码计算NN对一个或多个说话者进行加权。

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