Systems and methods for selectively replaying voice data. Voice data received in a communication session is both buffered by the device and played by the device. When input is received from a recipient, the voice data currently being received is no longer played and buffered voice data is replayed to the recipient. While the buffered voice data is replayed, the device continues to buffer any voice data received by the device to ensure that the recipient can listen to all of the voice data in the communication session. By only storing voice data in the buffer, replaying the buffered voice data can catch the real-time voice data. At this point, the device reverts to playing voice data being received while continuing to buffer the voice data.