发明名称 VOICE PROCESSOR AND COMPUTER PROGRAM FOR VOICE PROCESSING
摘要 PROBLEM TO BE SOLVED: To discriminate as non-voice signal a sound, such as machine sound which has commonalities of pitch and pitch duration with human speech sounds. SOLUTION: A pitch is extracted from an input sound signal and pitch information, to which the extracted pitch is made to correspond is, output by each elapsed time of the input sound signal (pitch extraction section 203a); variations in the pitch are extracted from the output pitch information and output as a pitch variation quantity by calculating the variations in the pitch (variation calculation section 203b); than the kind of input sound signal is determined such that the input sound signal is a voice signal, when the output pitch variation quantity is larger than a predefined value and as being a non-voice signal, when this is smaller (variation determining section 203c). COPYRIGHT: (C)2007,JPO&INPIT
申请公布号 JP2007156361(A) 申请公布日期 2007.06.21
申请号 JP20050355403 申请日期 2005.12.08
申请人 TOSHIBA TEC CORP 发明人 YOSHIZAKI KEISUKE;IKUMI TOMONORI;KAKINO TOMONARI;SEKINE NAOKI
分类号 G10L15/02;G10L11/00;G10L11/04 主分类号 G10L15/02
代理机构 代理人
主权项
地址