Subjective evaluation of PCM noise-feedback coder for television

Abstract

An optimization of a PCM coding system has been evaluated. The system consists of a linear signal predistortion filter, a linear network in a noise-feedback loop around the quantizer, and a linear reconstruction network. Synthesis procedures for these filters to minimize the runs weighted noise have been reported [1]. Results of a program of subjective testing show that the subjectively optimum system is nearly the same as that predicted theoretically. With a sampling frequency of twice the signal bandwidth and three-bit quantization, the optimum system gives nearly a 12-dB improvement in subjective signal-to-weighted-noise ratio over a conventional quantization system with no filtering.