Abstract: A speech processing system receives multiple streams of speech frames. The system selects among concurrent ones of the frames a subset of those frames that are the most relevant, based on pre-assigned stream priorities and energy content of the frames. The selected frames are then decoded and rendered. The resulting signals are mixed. This architecture provides bandwidth scalability and/or processing power scalability.

Abstract: A known cross-coupled adaptive noise cancelling arrangement uses an adaptive noise filter and an adaptive cross-talk filter in a feedback loop for cancelling correlated noise at a primary signal input and reference input. This known cross-coupled ANC does not operate satisfactorily, particularly not for acoustic noise cancellation. This leads to reverberant-like sound signals, in particular in a typical office room with remote noise sources. A different configuration of a cross-coupled adaptive noise cancelling arrangement gives rise to a better performance. The adaptive cross-talk filter is now split into a pre-filter section and an adaptive filter section, the sections using different input signals.

Abstract: In order to decide on the logic value of the output signal corresponding to a received multivalued digital signal, the receiver includes a comparator circuit (3) for comparing the amplitude of the output signal with a plurality of decision thresholds produced by an integrator (24). Based on such comparison, a regenerated signal is produced having the decided logic value and an amplitude which is proportional to the decision thresholds. An adaptation circuit (17) compares the regenerated signal amplitude with that of the output signal, and based on such comparison produces a control signal (z) for either increasing or decreasing the decision thresholds so as to conform the regenerated signal with the output signal, such adaptation being effected by selectively supplying either a positive or a negative voltage to the input of the integrator (24).