PURPOSE: To make the equipment small, to attain high quality and to simplify the processing by referring only to a band onto which signal power is concentrated and detecting voiced/unvoiced voice signal when a decoded voiced/ unvoiced voice signal is detected.
CONSTITUTION: Let an SB-ADPCM code inputted from each subscriber be A-D, they are separated into ah, al-dh, dl for each band by SB-ADPCM decoders 12a-12d respectively. The signals are inputted to band mixing sections 101, 102 of each band in time division and processed by (N-1) adder circuits 81, 82. Then the result of (N-1) addition is outputted in time division and the result is applied to a control input of level adjustment means 111, 112. Then a level calculator 15 makes processing when the result of output of a voice sound detector 14 indicates a voiced sound and keeps a preceding output in the case of non-voice signal. Thus, the signal obtained in this way is fed to the means 112, 111, in which level control is implemented. Then each signal outputted from the mixing sections 101, 102 is sent via coders 13a-13d.
EBISAWA HIDEAKI
NAITO HISASHI
Next Patent: AUTOMATIC RESTORATION SYSTEM FOR TRUNK LINE TO TRUNK LINE CONNECTION