A digital voice transmission system for transmitting voice signals from a transmitter that processes the voice signal to provide segments having a predetermined duration, each segment providing a block of bits containing a term (ENERG) relating to the energy of the signal within a high-frequency band, characteristic coefficients (COEF) of the voice signal to be transmitted, and a term (SIGNAL) relating to the contents of a low-frequency band of said voice signal to be transmitted, said SIGNAL term being processed within sub-bands of the low-frequency band, and means of transmitting said block to a receiver, said receiver reconstructing the voice signal from the ENERG, COEF and SIGNAL terms contained in each received block of bits, said receiver being characterized in that it includes:
- input means that process the received signal for reconstructing each block of bits;
- buffer connected to said input means and storing at least one block of bits;
- validity checking means connected to said buffer means for checking the validity of blocks received and identifying lost blocks;
- means connected to said validity checking means and said buffer means for determining, for each sub-band, a term T.sub.(k) relative to the period of the signal contained in the sub-band considered;
- means for reconstructing lost blocks from the blocks in the vicinity thereof, said reconstructing means connected to said means of determining T.sub.(k) for reconstructing a sub-band signal of period T.sub.(k) said reconstructing means processing the ENERG, COEF and SIGNAL terms separately;
- and,
- synthesizing means connected to said means for reconstructing lost blocks and to said buffer means for providing the received voice signal.