首页> 外国专利> Self-development type voice language pattern recognition system, self-organization method for constructing a neural network structure and its construction program used in the system

Self-development type voice language pattern recognition system, self-organization method for constructing a neural network structure and its construction program used in the system

机译:自主开发型语音语言模式识别系统,构建神经网络结构的自组织方法及其构建程序

摘要

PPROBLEM TO BE SOLVED: To provide a self-development type voice language pattern recognition system capable of efficiently performing pattern recognition processing for voice language data at a high recognition rate and self-adaptively developing its system configuration. PSOLUTION: Mutually relative neurons among respective neurons included in a self-organizing neural network structure structured in a pattern recognition section 12 are coupled with each other through a weight coefficient to be related in steps corresponding to the hierarchical recognition level of the voice language data. Specifically, each neuron is included in one of a plurality of network layers corresponding to hierarchical levels (phoneme level, word level, phrase level, and sentence level) of the voice language data, respective neurons included in the respective network layers are mutually coupled to only neurons included in network layers adjacent to each network layer. PCOPYRIGHT: (C)2006,JPO&NCIPI
机译:

要解决的问题:提供一种自我开发型语音语言模式识别系统,该系统能够以高识别率高效地对语音语言数据执行模式识别处理,并自适应地开发其系统配置。

解决方案:模式识别部分12中构造的自组织神经网络结构中包括的各个神经元之间的相互相对的神经元通过权重系数彼此耦合,以与语音的分层识别级别相对应的步骤相关语言数据。具体地,每个神经元被包括在与语音语言数据的等级等级(音素等级,单词等级,短语等级和句子等级)相对应的多个网络层之一中,包括在各个网络层中的各个神经元相互耦合。仅包含在与每个网络层相邻的网络层中的神经元。

版权:(C)2006,JPO&NCIPI

著录项

  • 公开/公告号JP4899177B2

    专利类型

  • 公开/公告日2012-03-21

    原文格式PDF

  • 申请/专利权人 独立行政法人理化学研究所;

    申请/专利号JP20050328805

  • 发明设计人 保 谷 哲 也;

    申请日2005-11-14

  • 分类号G10L15/16;G06N3;G06N3/02;

  • 国家 JP

  • 入库时间 2022-08-21 17:37:06

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