Optimum design of subband coder for a first order autoregressive process

This paper provides a method for analysis and design of a subband coder involving the quantization effect. A joint optimization method for the design of filter banks and related dithered quantizers by using a time-domain formulation is proposed. To show the effectiveness of design method, the optimum filters are calculated for a first order autoregressive process in a paraunitary filter bank and results are compared with a non-modified subband coder. Experiments show that by appropriate choice of synthesis filter bank, it would be possible to replace signal-dependent artifacts with signal-independent white noise and also reduce the mean square error of reconstruction signal.