Methods, apparatus and communication systems for allocating and managing voice channels are provided. After receiving first voice data, a voice server can search for a voice channel corresponding to a first user of the first voice data. When successfully finding the first voice channel corresponding to the first user, the voice server can write the first voice data to the first voice channel. When successfully allocating a second voice channel that is currently already-timed-out to the first user, the voice server can write the first voice data to the second voice channel, and mark the second voice channel as assigned to the first user. Utilizing efficiency of the voice channels is therefore improved and deploying of hardware resources can be reduced.