A speech input recognition unit, an event processing unit, which processes an event including a speech recognition result or a command, an expert unit including a plurality of expert modules each of which processes the event in cooperation with the event processing unit, and an execution history management unit, which manages execution history of the expert modules are provided. The expert unit is, for recording a user speech as message text, provided with a speech processing expert module for producing standard format text and a speech processing expert module for producing free format text and a transmission expert module for sending the message text.