A voice signal from a microphone is processed for compressing and coding in a voice compressor. The compressed voice data is input into a transmission switching section. The transmission switching section switches and outputs the compressed voice data to either a UDP transmitter or a TCP transmitter according to a transmission switching signal that the network situation supervisory section outputs. The transmission switching signal allows-transmission using the UDP transmitter when a network is in a situation with a margin, and transmission using the TCP transmitter when a network is in a confused situation responding to network congestion information that is obtained from a network interface. This allows a protocol transmitting the compressed voice data to be selected corresponding to a congested situation of the network, and thereby a voice call without interruption can be obtained even though the network is in a confused and low-quality situation.