发明名称 VOICE RECOGNITION PROCESSING SYSTEM FOR LARGE-VOCABULARY
摘要 PURPOSE:To shorten the time of nor response after voice input by a user and to reduce a metal load by providing a switching means which switches and sends out the output of a preliminary selecting means and the output of a main matching means. CONSTITUTION:When a voice is inputted, the features of the voice are obtained by a feature extracting means 11. The preliminary selecting means 12 matches the features with features registered in a word dictionary 13 roughly to select candidate words. The normal matching means 14 matches the features obtained by the feature extracting means 11 with features registered in the word dictionary 13 in detail to give order to the candidate words. While this main matching is carried out, the candidate words obtained by the preliminary selecting means 12 are sent to a word selecting means 15 through a switching means 17 to display one or plural words by a display means 16 to the user in sequence; and the user selects a desired word by using the word selecting means 15 and rejects different words.
申请公布号 JPH03266898(A) 申请公布日期 1991.11.27
申请号 JP19900067641 申请日期 1990.03.16
申请人 FUJITSU LTD 发明人 SANADA TORU
分类号 G10L13/00;G10L15/28 主分类号 G10L13/00
代理机构 代理人
主权项
地址