专利快速检索-快速检索全球专利，免费商用专利数据库-IPRDB

1. 发明公开

KR20100003672A SPEECH RECOGNITION APPARATUS AND METHOD USING VISUAL INFORMATION 审中-公开
标题翻译：语音识别装置和使用视觉信息的方法
公开(公告)号：KR20100003672A
公开(公告)日：2010-01-11
申请号：KR20080063669
申请日：2008-07-01
申请人： DUNET , KOREA IND TECH INST
发明人： KO WOO HYUN , HAN JAE SUNG , LEE SANG MOO , NAM KYUNG TAE , SHIM JUN HONG , CHA NAM JU , KIM SOON KI , SHIN SOO JEONG , JI SANG HOON
IPC分类号： G10L15/25 , G10L15/00
摘要： PURPOSE: A voice recognition device using video information and a method thereof are provided to set a voice recognition variable by identifying a speaker or an age and sex of the speaker from the video information. CONSTITUTION: A variable setting unit(150) sets a voice recognition variable based on information about a video where a speaker is photographed. A voice recognition unit(140) recognizes voice information inputted from the speaker by using the set voice recognition variable. A video identifying unit(130) identifies the speaker corresponding to features extracted from the video information. The variable setting unit sets up the voice recognition variable corresponding to the identified speaker. If the speaker is unidentified from the video information, the video identifying unit identifies an age or sex of the speaker based on the features extracted from the video information.
摘要翻译：目的：提供一种使用视频信息的语音识别装置及其方法，用于通过从视频信息中识别扬声器或说话者的年龄和性别来设置语音识别变量。构成：可变设置单元（150）基于关于照相机的视频的信息来设置语音识别变量。语音识别单元（140）通过使用所设置的语音识别变量识别从扬声器输入的语音信息。视频识别单元（130）识别与从视频信息中提取的特征相对应的扬声器。变量设置单元设置与所识别的扬声器对应的语音识别变量。如果扬声器从视频信息不识别，则视频识别单元基于从视频信息中提取的特征来识别扬声器的年龄或性别。

IPRDB

热门服务

关于我们

友情链接

联系方式