To provide a voice mixing method and a system that requires less computing amount with high efficiency in an inter-multiparty speech service.

The voice mixing method compares sampled packets of voice data of a plurality of participants generated at the same time and extracts a sampled packet whose amplitude absolute value is a maximum among compared samples. The extracted sample is extracted as an output and other samples are aborted, and only voice data whose amplitude absolute value is a maximum among voice data generated at the same time are transmitted. Thus, computation complexity in the inter-multiparty speech service can considerably be reduced.