首页> 外国专利> REVERBERANT ENVIRONMENT DETERMINATION DEVICE, REVERBERANT ENVIRONMENT DETERMINATION METHOD, AND PROGRAM

REVERBERANT ENVIRONMENT DETERMINATION DEVICE, REVERBERANT ENVIRONMENT DETERMINATION METHOD, AND PROGRAM

机译:无菌环境确定装置,无菌环境确定方法和程序

摘要

PROBLEM TO BE SOLVED: To provide a reverberant environment determination device for enabling even an operator who is not familiar with voice recording to determine a reverberant environment.;SOLUTION: The reverberant environment determination device comprises: an acoustic model recording part 140 for recording acoustic models created from recorded voices in different reverberant environments and model names; a reverberant knowledge recording part 160 for recording the model names and the reverberant environments; a read-out text recording part 150 for recording a read-out text; a voice recognition part 110 for voice-recognizing an input voice by using each acoustic model of the acoustic model recording part, and for generating an acoustic model-categorized recognition result including the model name and a text of a voice recognition resu a recognition rate calculation part 120 for generating an acoustic model-categorized recognition rate including the model name and a recognition rate from the acoustic model-categorized recognition result by using the read-out text of the read-out text recording part; and a reverberation determination part 130 for generating, as a determination result, the reverberant environment of the reverberant knowledge including the model name of the acoustic model which has output the highest recognition rate among the acoustic model-categorized recognition rates by using the reverberant knowledge recording part.;SELECTED DRAWING: Figure 1;COPYRIGHT: (C)2018,JPO&INPIT
机译:解决的问题:提供一种混响环境确定装置,以使即使是不熟悉语音记录的操作员也能够确定混响环境。根据录制的声音在不同的混响环境和型号名称中创建;混响知识记录部分160,用于记录型号名称和混响环境;读出文本记录部分150,用于记录读出的文本;语音识别部分110,用于通过使用声学模型记录部分的每个声学模型来语音识别输入的语音,并用于生成包括模型名称和语音识别结果的文本的声学模型分类的识别结果;识别率计算部分120,用于通过使用读出的文本记录部分的读出的文本,从声学模型分类的识别结果中生成包括型号名称和识别率的声学模型分类的识别率;混响确定部分130,用于产生混响知识的混响环境,该混响知识包括通过使用混响知识记录在声学模型分类的识别率中输出最高识别率的声学模型的型号名称零件图;图1;版权:(C)2018,JPO&INPIT

著录项

  • 公开/公告号JP2017207546A

    专利类型

  • 公开/公告日2017-11-24

    原文格式PDF

  • 申请/专利权人 NIPPON TELEGR & TELEPH CORP NTT;

    申请/专利号JP20160097994

  • 发明设计人 YOSHIDA AKIHIRO;IJIMA YUSUKE;

    申请日2016-05-16

  • 分类号G10L15/10;G10L15/06;

  • 国家 JP

  • 入库时间 2022-08-21 13:11:30

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