If a voice call has missing data it makes it hard to understand the speaker. Therefore, we commonly allow time for a limited number of retransmissions before playing the sound to the listener. If the network path (in each direction) has total packetization delay of 15ms, total propagation delay of 25ms, and queuing delay varying between 0ms and 10ms, how large (in milliseconds) does the playback buffer need to be if we want to allow for one retransmission?