NOISE ROBUST SPEECH RECOGNITION METHOD INSPIRED FROM SPEECH PROCESSING OF BRAIN
摘要
Presented is a new method of extracting the feature information of speech. Kernels which represent the main frequency behavior of a base layer responding to sound is defined. Inputted speech is represented by at least one kernel among the kernels. Information about the represented kernel is used as the feature information of the inputted speech.
申请公布号
KR101408902(B1)
申请公布日期
2014.06.19
申请号
KR20130033841
申请日期
2013.03.28
申请人
KOREA ADVANCED INSTITUTE OF SCIENCE AND TECHNOLOGY