Statistical properties of high quality stereo signals in the time domain

Bit-rate reduction algorithms are examined for compact disks, containing high-quality stereo signals sampled at 44.1 kHz and 16 bits/sample. Reducing the bit rate for the stereo signals requires knowledge of the statistics. Probability densities and joint probability densities provide parameters for individual samples. Short-time energies and correlation coefficients express similarities and linear relationships of blocks of samples. It is shown that by replacing the original signals in the left and the right channel by half of the sum and half of the difference, a small reduction of the amplitudes takes place. Linear relationships in the channels could not be found. The similarities of the short time energies together with the retaining of the energy after a transform indicate possible correlations in the frequency domain