发明名称 CONTINUOUS VOICE RECOGNITION DEVICE
摘要 PURPOSE:To decrease the frequency of processing for prediction by collating an input signal with a standard signal and outputting recognition element piece candidates, connecting the recognition element piece candidates and generating a recognition element piece series, and predicting following element pieces and finding and recording recognition element piece candidates and following element pieces as index contents. CONSTITUTION:A collation part 2 collates an input signal with element pieces and outputs the recognition element piece candidates, and a connection part 3 connects the recognition element piece candidates to generate the recognition element piece candidate series. A prediction part 4 predicts recognition element pieces which can follow the recognition element piece candidates and a recording part 5 records the recognition element piece candidates and following element pieces as the index contents respectively. When the following element pieces following the recognition element piece candidate series are predicted. The recognition element pieces as the contents of a recording part are outputted to the connection part if the recognition element piece candidate series is present in the indexes of the recording part, but when the series is not present, on the other hand, the prediction part predicts the following element pieces. Therefore, the contents of the recording part 5 are only referenced to predict the following element pieces.
申请公布号 JPH0527793(A) 申请公布日期 1993.02.05
申请号 JP19910184472 申请日期 1991.07.24
申请人 MATSUSHITA ELECTRIC IND CO LTD 发明人 HORIOKA ATSUSHI
分类号 G10L15/12;G10L15/18 主分类号 G10L15/12
代理机构 代理人
主权项
地址