首页> 外国专利> Audio data set creation device for perception test, optimization device, and audio synthesizer unit of computer program and sub cost function for speech synthesis

Audio data set creation device for perception test, optimization device, and audio synthesizer unit of computer program and sub cost function for speech synthesis

机译:用于感知测试的音频数据集创建设备,优化设备,计算机程序的音频合成器单元以及用于语音合成的子成本函数

摘要

PPROBLEM TO BE SOLVED: To obtain a voice data set for perceptual examination so as to specify correspondence relation between optional physical quantity and perceptual evaluation regarding utterance. PSOLUTION: This creation device of voice data set for perceptual examination includes a unit speech element replacement part 40 which replaces an optional unit waveform speech element with other unit waveform speech element and creates a voice corpus 42 after replacement about each piece of utterance voice data included in a voice corpus 20, a featured value/featured value statistic calculation part 44 which calculates a plurality of kinds of featured values about each piece of utterance voice data after the replacement and sub-cost function derivation parts 48A to 48N which create the voice data set for perceptual examination by extracting a set of utterance voice data with large fluctuation of a specific featured value and with small fluctuation of other featured values from the corpus 42 based on the calculated featured value. PCOPYRIGHT: (C)2005,JPO&NCIPI
机译:

要解决的问题:获得用于感知检查的语音数据集,以便指定可选物理量与关于发声的感知评估之间的对应关系。

解决方案:该用于知觉检查的语音数据集的创建设备包括单元语音元素替换部分40,该单元语音元素替换部分40将可选的单元波形语音元素替换为其他单元波形语音元素,并且在关于每个发声的替换之后创建语音语料库42。特征值/特征值统计量计算部44包括在语音语料库20中的语音数据,该特征量/特征值统计量计算部44针对在创建了替换和子成本函数推导部48A至48N之后的每个发声语音数据计算多种特征值。通过基于计算出的特征值从语料库42中提取具有特定特征值的波动较大且具有其他特征值的波动较小的一组话语语音数据,来进行感知检查的语音数据集。

版权:(C)2005,JPO&NCIPI

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