PURPOSE: To transmit sound data as character line data.

CONSTITUTION: A sound inputted from a sound input device 23 is sampled by an FFT device 21 at 22kHZ to be A/D converted to the PCM data of 16 bits. Then, spectrum analysis is performed at every time tn (n=1, 2,...) according to a parameter such as a humming window, etc., and the power of the sound data for respective frequencies are calculated. The calculated power for respective times and respective frequencies are stored and preserved successively in a character line table storage area of a RAM 14 as the character line sound data. Then, the character line sound data are transmitted to other sound transmission devices through a communication controller 15 and a communication network 2 successively. On the other hand, the sound transmission device receiving the character line sound data reproduces sound waveforms at every time tn from respective frequencies and the power by an inverse FFT device 22 to output the contents of the receiving sound from a sound output device 24.