Granted Patent B2

US 7,110,554 · App. 10/214,057 · Granted Sep 19, 2006

Abstract

Claims (35)

1. An adaptive signal processing system for improving a quality of a signal, comprising:

a first weighted-overlap-add (WOLA) analysis filterbank for receiving a primary information signal in the time domain and for implementing block processing including a weighting window limited in length to transform the primary information signal into a plurality of oversampled sub-band primary complex signals in the frequency domain;

a second WOLA analysis filterbank for receiving a reference signal in the time domain and for implementing block processing including a weighting window limited in length to transform the reference signal into a plurality of oversampled sub-band reference complex signals numerically equal to the number of primary signal sub-bands;

a plurality of sub-band processing circuits for processing the oversampled sub-band complex signals to improve a quality of an output signal, each including a complex adaptive filter for performing adaptive filtering and a feedback circuit, each complex adaptive filter having adjustable parameters, each feedback circuit including an adaptation circuit for adjusting the adjustable parameters based on the result of the processing and an emphasis filter for improving convergence of the adaptation circuit; and

a WOLA synthesis filterbank for combining the outputs of the sub-band processing circuits to generate the output signal in the time domain.

2. The adaptive signal processing system of claim 1 wherein each complex adaptive filter is a digital filter having the adjustable parameters, each feedback circuit adjusting the adjustable parameters of the digital filter based on the result of the processing.

3. The adaptive signal processing system of claim 2 wherein the filter is a FIR filter for filtering the sub-band reference signal, and the feedback circuit includes a summer for summing the output of the FIR and the sub-band primary signal, and wherein the adaptation circuit includes module for adjusting, in response to the output of the summer and the sub-hand reference signal, the adjustable parameters of the FIR filter.

4. The adaptive signal processing system of claim 1 wherein the adaptation circuit includes a LMS circuit, and the emphasis filter is provided for improving convergence of the LMS circuit.

5. The adaptive signal processing system of claim 1 further comprising a Voice Activity Detector (VAD) for accepting frequency domain and time domain inputs, and controlling the sub-band processing circuit.

6. The adaptive signal processing system of claim 2 further comprising a Voice Activity Detector (VAD) for accepting frequency domain and time domain inputs, and controlling the sub-band processing circuit.

7. The adaptive signal processing system of claim 3 further comprising a Voice Activity Detector (VAD) for accepting frequency domain and time domain inputs, and controlling the sub-band processing circuit.

8. The adaptive signal processing system of claim 4 further comprising a Voice Activity Detector WAD) for accepting frequency domain and time domain inputs, and controlling the sub-band processing circuit.

9. An adaptive signal processing system for improving a quality of a signal, comprising:

a first analysis filterbank for receiving a primary information signal in the time domain and transforming the primary information signal into a plurality of oversampled sub-band primary signals in the frequency domain;

a second analysis filterbank for receiving a reference signal in the time domain and transforming the reference signal into a plurality of oversampled sub-band reference signals numerically equal to the number of primary signal sub-bands;

a plurality of sub-band processing circuits for processing these signals to improve a quality of an output signal; and

a synthesis filterbank for combining the outputs of the sub-band processing circuits to generate the output signal,

wherein the sub-band processing circuit minimizes some error metric, which includes The value of the squared error E 2 as described in the following equation:

E 2 =( YW−X )( YW−X )*

wherein the “Y” denotes the sub-band primary signals, the “X” denotes the sub-band reference signals, the “W” denotes an adaptive weighting filter, the “E” denotes a summation of the “X” and “Y” filtered by the “W”, and the “*” operator denotes complex conjugation.

10. The adaptive signal processing system of claim 1 wherein the primary information signal includes an audio signal and a first noise and the reference signal includes a second noise, and the first noise is correlated to the second noise.

11. The adaptive signal processing system of claim 5 , wherein the VAD operates on the primary information signal, the reference signal or a combination thereof, and the oversampled sub-band primary complex signals, the oversampled sub-band reference complex signals or a combination thereof.

12. The adaptive signal processing system of claim 1 , wherein each of the WOLA analysis filterbanks provides an input to each sub-band processing circuit, which is isolated in frequency.

13. The adaptive signal processing system of claim 1 , wherein the sub-band processing circuit performs a least mean square (LMS) algorithm, a recursive least square algorithm, or a combination thereof.

14. The adaptive signal processing system of claim 1 , wherein one of the primary information signal and the reference signal is noise, and the other includes a signal of interest with the noise.

15. The adaptive signal processing system of claim 1 , wherein the sub-band processing circuits include an emphasis filter for whitening the input of the sub-band processing circuits.

16. The adaptive signal processing system of claim 2 , wherein the filter includes a Wiener filter for filtering the oversampled sub-band reference complex signal, and the feedback circuit Includes a summer for summing the output of the Wiener filter and the oversampled sub-band primary complex signal, and wherein the adaptation circuit includes a LMS circuit for controlling the Wiener filter in response to the output of the summer and the oversampled sub-band primary complex signal.

17. A method of improving a quality of a signal, comprising the steps of:

at a first weighted-overlap-add (WOLA) analysis filterbank, implementing block processing including a weighting window limited in length to transform a primary information signal in the time domain into a plurality of oversampled sub-band primary complex signals in the frequency domain;

at a second WOLA analysis filterbank, implementing block processing including a weighting window limited in length to transform a reference signal in the time domain into a plurality of oversampled sub-band reference complex signals numerically equal to the number of primary signal sub-bands;

at each of sub-band processing circuits, adaptively processing the over-sampled sub-band complex signals, including performing adaptive filtering at a complex adaptive filter which has adjustable parameters, the processing step including, at a feedback circuit included in the sub-band processing circuit, implementing adaptive algorithm to adjust the adjustable parameters based on the result of the processing, and improving convergence of the adaptation algorithm; and

at a WOLA synthesis filterbank, combining the outputs of the sub-band processing circuits to generate an output signal having a signal of interest of improved quality and in the time domain.

18. A method of claim 17 , wherein at each of the sub-band processing circuits, the processing step includes the step of adjusting the coefficients of the complex adaptive filter based on the over-sampled sub-band primary complex signal and the over-sampled sub-band reference complex signal.

19. The adaptive signal processing system of claim 1 , wherein the complex adaptive filter Includes a first adaptive filter for the over-sampled sub-band primary and reference complex signals and a second adaptive filter for the over-sampled sub-band reference complex signals, the emphasis filter and the adaptation circuit being allocated only to the first adaptive filter.

20. The adaptive signal processing system of claim 9 , wherein each of the first and second analysis filterbanks is a WOLA analysis filter-bank.

Assignments (8)