Stereophonic ADPCM voice coding method

A new stereophonic voice coding method is presented, where a conventional ADPCM method is applied after reducing correlated signals between right and left microphone output signals by an adaptive filter called a cross-channel correlation canceller. To evaluate the method, computer simulations were carrier out by using actual stereophonic voice sound. Simulation results showed that about 6-10-dB improvement of segmental S/N was attainable by reducing cross-channel correlation using the cross-channel correlation canceller. The method was actually implemented as a stereophonic ADPCM codec by using two Toshibas DSP chips. By means of the codec, 3.4-kHz stereophonic voice sound is transferred at 64 kbps. Subjective tests of the codec were carried out on 33 listeners. The test results showed that the proposed method provides more listener satisfaction than double-channel conventional 32 k-ADPCM (CCITT G.723)