A method for voice chat in a virtual environment includes receiving a plurality of audio streams corresponding to a plurality of avatars in the virtual environment, determining relative distances between the avatars in the virtual environment, calculating change in volumes of the audio streams, depending on the determined relative distances between the avatars, normalizing volumes of the audio streams depending on the calculated change in volumes, mixing the normalized audio streams to create a plurality of mixed audio streams, wherein one mixed audio stream corresponds to one avatar, and relaying the mixed audio streams to corresponding avatars.