发明名称 |
METHOD, DEVICE AND PROGRAM FOR SPEECH RECOGNITION AND RECORDING MEDIUM |
摘要 |
PROBLEM TO BE SOLVED: To provide a speech recognition method for neglecting speech other than a recognition candidate, and to provide its device. SOLUTION: The speech recognition method collates input speech and an acoustic model from among recognition object words or a word row contained in speech recognition syntax, and outputs the words or the word row with highest likelihood as a recognition result. An arbitrary speech receiving recognition category receiving any speech is described in speech recognition syntax, and a speech recognition method is proposed for recognizing the words or the input speech except for the word row recognizable in the speech recognition syntax as arbitrary speech by the arbitrary speech receiving recognition category. COPYRIGHT: (C)2006,JPO&NCIPI
|
申请公布号 |
JP2005326497(A) |
申请公布日期 |
2005.11.24 |
申请号 |
JP20040142640 |
申请日期 |
2004.05.12 |
申请人 |
NIPPON TELEGR & TELEPH CORP <NTT> |
发明人 |
YAMAGUCHI YOSHIKAZU;IMAMURA AKIHIRO |
分类号 |
G10L15/20;G10L15/10;G10L15/14;G10L15/18;(IPC1-7):G10L15/20 |
主分类号 |
G10L15/20 |
代理机构 |
|
代理人 |
|
主权项 |
|
地址 |
|