To provide the multimedia public telephone system where smooth and sure communication is conducted without the need for a translator even when a caller and a called party use different languages.

A voice of a talker is received by speech signal reception means 11-14 via a communication system and in the case of selecting 'translation' by a translation conversion switch 15, a voice recognition device 16 recognizes the received speech signal and a speech recognition generator 17 converts the recognized speech into character data, a speech/character changeover switch 18 changes output mode of speech/character, a translation conversion processing section 19 translates the data and the result of translation is sounded in speech or displayed as characters on a display device. Conversely, the speech signals of the talker or entered characters are recognized as speech or character and it is translated and speech is synthesized and the result is transmitted to the opposite party via the communication system through a means.