摘要:
A system that reconstructs independent signals (voice 1, voice 2) from degenerate mixtures estimates independent Auto Regressive (AR) processes from their sum. The system includes an identification system (20) and an estimator (22). A mixture (X) of two signals is inputted into the system and through the identifying and filtering processes, two estimates of the original signals are outputted. The identification system includes an ARMA identifier, a computation of autocovariance coefficients, an initializer and a gradient descent system. The estimator includes filtering.
摘要:
A method and apparatus that performs blind source separation usign convolutive signal decorrelation. More specifically, the method accumulates a length of input signal (mixed signal) that comprises a plurality of independent signals from independent signal sources. The invention then divides the length of input signal into a plurality of T-length periods (windows) and performs a discrete Fourier transform (DFT) on the signal whithin each T-length period. Thereafter, estimated cross-correlation values are computed using a plurality of the average DFT values. A total number of K cross-correlation values are computed, where each of the K values is averaged over N of the T-length periods. Using the cross-correlation values, a gradient descent process computes the coefficients of a FIR filter that will effectively separate the source signals within the input signal. To achieve an accurate solution, the gradient descent process is constrained in that the time-domain values of the filter coefficents can attain only certain values, i.e., the time-domain filter coefficient values W(τ) are constrained within each T-length period to be zero for any time t > Q. In this manner, a unique solution for the FIR filter coefficients is computed and a filter produced using these coefficents will effectively separate the source signals.
摘要:
A signal processing system includes a plurality of detectors for detecting signals, each detector receiving signals from plural sources ; and a processor for receiving the detected signals and producing a reconstruction filter for filtering the detected sigals to produce reconstructed source signals in which each source signal is reconstructed without interference from the other source signals. The reconstruction filter is produced such that cross-polyspectra of the reconstructed signals are forced to approach zero.
摘要:
A system that reconstructs independent signals (voice 1, voice 2) from degenerate mixtures estimates independent Auto Regressive (AR) processes from their sum. The system includes an identification system (20) and an estimator (22). A mixture (X) of two signals is inputted into the system and through the identifying and filtering processes, two estimates of the original signals are outputted. The identification system includes an ARMA identifier, a computation of autocovariance coefficients, an initializer and a gradient descent system. The estimator includes filtering.
摘要:
In a multi-subscriber radio system (1) in which the signals are given a subscriber-specific form, after a transmission via one and the same channel (4), the mixture of signals received at a subscriber's station (5) is separated in two steps a manner tailored specifically for the subscriber concerned. In the first step channels are equalised (7). In the subsequent second step the signals are filtered (8), whereby the filter coefficients are determined from the signal form specific to the subscriber concerned. The process of the invention can be used for the down-link of a mobile telephone system operating on CDMA.
摘要:
A method of recovering source signals (y1, y2) from an equal number of mixed signals (x1, x2) including components of the source signals. Channel one (x1) and two (x2) mixed signals are multiplied by channel one (w1) and two (w2) adaptive weights to produce channel one and two product signals, respectively. A channel one filtered feedback signal is generated from a channel two approximation signal (u2) and a channel two filtered feedback signal is generated from a channel one approximation signal (u1). The channel one filtered feedback signal is added to the channel one product signal to produce the channel one approximation signal (u1) and the channel two filtered feedback signal is added to the channel two product signal to produce the channel two approximation signal (u2). A nonlinear function (35, 36) is applied to the approximation signals (u1, u2) to produce output signals (y1, y2). The adaptive weights and the filtered feedback signals are adjusted (37) to maximize entropy of the output signals (y1, y2).
摘要:
Le système (10) de séparation de sources traite des signaux d'entrée formés par des mélanges de signaux primaires issus des sources et estime les signaux primaires de manière à ce que les estimations ne divergent pas même en présence de signaux d'entrée non stationnaires. Le système comprend un premier sous-ensemble (12) de séparation qui délivre des premières estimations des signaux primaires, un second sous-ensemble (14) qui détermine adaptativement les coefficients de séparation, un troisième sous-ensemble (15) qui normalise les premières estimations et délivre des premières estimations normalisées servant au calcul des coefficients de séparation. Un module de sortie (17) permet de délivrer des estimations ayant entre elles un même rapport de proportionnalité que celui existant entre les signaux primaires. Un module de sélection (19) évite à une estimation d'être dupliquée sur plusieurs sorties dans le cas où certains signaux primaires sont absents ou très faibles. Applications: Système de séparation de signaux de mélange, antenne, autoradio, chaîne Hi-Fi, télévision, système de téléconférence.
摘要:
A method of recovering source signals (y1, y2) from an equal number of mixed signals (x1, x2) including components of the source signals. Channel one (x1) and two (x2) mixed signals are multiplied by channel one (w1) and two (w2) adaptive weights to produce channel one and two product signals, respectively. A channel one filtered feedback signal is generated from a channel two approximation signal (u2) and a channel two filtered feedback signal is generated from a channel one approximation signal (u1). The channel one filtered feedback signal is added to the channel one product signal to produce the channel one approximation signal (u1) and the channel two filtered feedback signal is added to the channel two product signal to produce the channel two approximation signal (u2). A nonlinear function (35, 36) is applied to the approximation signals (u1, u2) to produce output signals (y1, y2). The adaptive weights and the filtered feedback signals are adjusted (37) to maximize entropy of the output signals (y1, y2).