摘要 |
<p>In a speech recognition conversation system having a plurality of recognition conversation servers, there is no way to select and decide one of the recognition conversation servers. A client terminal (10) transmits capability information stored in a terminal information storage unit (140) to a recognition conversation selection server (20). The capability of the client terminal (10) includes CODEC capability (CODEC type, CODEC compression mode, etc.), speech data format (compressed speech data, feature vector, etc.), a recording voice input/output function, a synthesis speech input/output function (no synthesis engine, an intermediate expression input engine exists, a character string input engine exists, etc.) and service contents. The recognition conversation selection server (20) receives capability information transmitted from the client terminal (10) and uniquely decides an optimal recognition server from the capability information of a plurality of recognition conversation servers stored in a recognition conversation server information storage unit (230) and the service content requested.</p> |