An approach to high resolution D/A converter utilizing a linear predictive coding

This paper proposes a linear predictive coding algorithm whereby the quality of the decoded signal is improved to realize a high resolution Digital-to-Analog (D/A) converter and describes the performance of the algorithm using computer simulation. The first discussion is the present problems of the digital audio system, one of which is degradation of decoded signal by quantization noise inherent to D/A conversions. Next, some reports published to date for improving the problems, are evaluated to clarify the limitations of their technical applicability. As a result, by regarding the digital signal decoding system as a single function system, the arithmetic means of the input and output groups of signal components are regarded to be equivalent when those signal groups are seen from the viewpoint of their time averages, and proposed here is a digital signal processing algorithm using a linear predictive coding system. Conceptual circuit diagrams to implement the method are shown. The efficacy of this method has been demonstrated by computer simulation, that is, the method increases in effect resolution of the D/A conversion system by at least 1 to 2 bits.