To enable a device which has small memory and processing power to perform accurate general voice recognition.

The system and method for voice communication which are used for a communication network system, such as the Internet, have plural acoustic recognizers, embedded in a mobile electronic communication device, which recognize voice information and generate a 1st group of relative language information and plural language recognizers, embedded in a data processing device such as a server, which recognizes the relative language information of the 1st group and generate a 2nd group of relative language information to recognize the voice information more accurately, in the style of decentralized voice recognition processing.