Abstract: A one-bit differential detector for a GMSK signal operates independently of an offset frequency. In particular, a decision signal used by a decision circuit to distinguish between logic "0" and logic "1" is always equal to sin[.DELTA..phi.(T)]. The inventive detector has an improved bit error rate performance in comparison to the prior art.

Abstract: A pre-encoder look-up table (LUT) provides output symbols for a statistical coder in response to signals representing a prediction of a prediction signal in a differential pulse code modulation (DPCM) system. The normal prediction signal is used as one address to the LUT and the prediction of that normal signal is used as a second address. The average probability distribution of the signals is determined to construct the LUTs wherein the output symbols are assigned values corresponding to different average probabilities of occurrence of the signal to be processed to thereby reduce the entropy further without adding overhead control symbols to the transmission system.

Abstract: In an NTSC digital television receiver, the composite video signal is digitized with a sampling frequency of 32/13 times the color subcarrier frequency, f.sub.c. The chrominance signal component of this video signal is demodulated using a pair of finite impulse responsive filters having programmable weighting factors. The weighting factors are changed in a 32 step sequence as 32 consecutive chrominance samples are applied to the filters. The circuitry which generates the weighting factors is responsive to a phase alignment signal to determine when the 32 step sequence begins.

Abstract: In field/frame recursive filters for providing noise reduction to composite video signal, inversion of the chrominance component of the composite video signal is obviated by selecting nearest neighbor in-phase field/frame delayed signals for recursion. The system arbitrates between two or more nearest neighbor in-phase signals or combinations thereof and feeds back the signal having an amplitude most closely matching the incoming composite video signal.

Abstract: In systems in which the current digitized level value of a synthesized-signal sample (e.g., a television video signal sample) is determined by summing a transmitted error-signal-sample level value and a delayed previous synthesized-signal-sample level value, the dynamic range of the transmitted error-signal samples are reduced by as much as one-half. This is accomplished by passing the error-signal samples through a first non-linear processor (NLP) before transmission thereof and then passing the summed output through a second NLP. Both NLPs exhibit the same specified transfer function taught by the present invention.