PURPOSE: To reproduce the feature of a talker in a synthesized voice with fidelity by making a voice signal into character data at a transmission side, sending the data with the feature data of the talker, and performing voice synthesis by using those data at a reception side.
CONSTITUTION: When a voice is inputted from a microphone 1, a speech recognition device 2 recognizes an input voice. And the making of the voice into the character data is performed and the making of it into the feature data is performed by extracting a feature including the speed of the input voice and the magnitude of a frequency. Those voice information data are sent to a transmission path 4 with a transmitter 3. A receiver 5 receives the voice made into the character data and the feature data of the input voice. A voice synthesizer 6 sets received feature data, and performs the voice synthesis from received character data. In such a manner, it is possible to reproduce the feature of the talker with high fidelity in the synthesized voice.