Method and system for frequency domain active matrix decoding without feedback

摘要

A perceptually motivated, frequency domain active matrix decoder and decoding method which decodes N audio input signals to generate M audio output signals, where M is greater than N, including by generating M streams of output frequency components which determine the audio output signals, in response to N streams of input frequency components indicative of the audio input signals, determining power ratios from the input frequency components without use of feedback, including at least one power ratio for each critical frequency band in a set of critical frequency bands, and determining gain control values for each of the critical frequency bands from the power ratios including by shaping the power ratios in nonlinear fashion without use of feedback. An active matrix element is steered using the gain control values.

主权项

1. A matrix decoding method for decoding N audio input signals to determine M audio output signals, where M and N are integers and M is greater than N, and N=2, said method including the steps of:
transforming (10, 11) the N audio input signals from the time domain into the frequency domain to generate N streams of input frequency components; determining power ratios (17, 30, 31, 32, 33) from the streams of input frequency components, said power ratios including at least one power ratio for each critical frequency band in a set of critical frequency bands, wherein the set of critical frequency bands is determined in accordance with perceptually motivated considerations; determining gain control values (17, 38) for each of the critical frequency bands from the power ratios including by shaping the power ratios in a nonlinear fashion, wherein the shaping of the power ratios in nonlinear fashion includes a step of exponentiating at least one value determined from at least one of the power ratios with an exponent at least substantially equal to three; operating an active matrix subsystem (16) to generate M streams of output frequency components in response to the streams of input frequency components; wherein the active matrix subsystem (16) is steered using the gain control values; wherein the active matrix subsystem (16) applies multiple sets of matrix coefficients to the streams of input frequency components, each set of matrix coefficients for a different one of the critical frequency bands; and transforming (20) the streams of output frequency components from the frequency domain into the time domain, thereby generating the M audio output signals.