REDUCED-DELAY SUBBAND SIGNAL PROCESSING SYSTEM AND METHOD

• US 20130287226A1
• Filed: 03/15/2013
• Published: 10/31/2013
• Est. Priority Date: 04/30/2012
• Status: Active Grant

• Current Assignee: Synaptics Incorporated
• Original Assignee: Conexant Systems Incorporated (Synaptics Incorporated)
• Inventors: Kerner, Yair
• Granted Patent: US 9,319,791 B2
• Time in Patent Office: Days
• US Class Current: 381/98
• CPC Class Codes: G10L 2021/02082 the noise being echo, reverberation of the speech
• CPC Class Codes: G10L 21/0232 Processing in the frequency domain
• CPC Class Codes: G10L 21/038 using band spreading techniques
• CPC Class Codes: H03H 17/0213 Frequency domain filters using Fourier transforms
• CPC Class Codes: H04B 3/20 Reducing echo effects or singing; Opening or closing transmitting path; Conditioning for transmission in one direction or the other
• CPC Class Codes: H04M 9/082 using echo cancellers echo cancellers per se H04B3/23
• CPC Class Codes: H04R 2430/03 Synergistic effects of band splitting and sub-band processing
• CPC Class Codes: H04R 3/04 for correcting frequency response

Abstract

A method for signal processing, receiving a time domain signal having a sample-rate Fs and generating N time domain signal bands, each having a bandwidth equal to Fs/N. Receiving the N signal bands and transforming a first time domain signal band to a frequency domain at a first resolution and a second time domain signal band to the frequency domain at a second resolution, where the first resolution may be different from the second resolution. Determining one or more first filter coefficients using the frequency domain components from the first signal band and one or more second filter coefficients using the frequency domain components from the second signal band. Transforming the first and second filter coefficients from the frequency domain to a time domain. Applying the first and second time domain filter coefficients to the first and second time domain signals, respectively.

Claims

1. A method for signal processing, comprising:

receiving a time domain signal having a sample-rate Fs and generating N sub-band signal bands;

transforming a first sub-band of the N sub-band signal bands from a time domain to a frequency domain at a first resolution and a second sub-band signal band of the N sub-band signal bands from the time domain to the frequency domain at a second resolution, where the first resolution is different from the second resolution;

determining one or more first filter coefficients using the frequency domain components from the first sub-band and one or more second filter coefficients using the frequency domain components from the second sub-band;

transforming the one or more first filter coefficients from the frequency domain to a time domain;

transforming the one or more second filter coefficients from the frequency domain to a time domain;

applying the first time domain filter coefficients to the first sub-band of the N sub-band signal bands; and

applying the second time domain filter coefficients to the second sub-band of the N sub-band signal bands.

2. The method of claim 1 wherein determining the one or more first filter coefficients using the frequency domain components from the first signal band comprises determining one or more finite impulse response filter coefficients.

3. The method of claim 1 wherein determining the one or more second filter coefficients using the frequency domain components from the second signal band comprises determining one or more finite impulse response filter coefficients.

4. The method of claim 1 wherein the first resolution is lower than the second resolution if the first time domain signal band is a higher frequency that the second time domain signal band.

5. The method of claim 1 further comprising performing filtering on the first time domain signal band.

6. The method of claim 1 further comprising performing acoustic echo cancellation processing on the first time domain signal prior to transforming the first time domain signal band to the frequency domain.

7. The method of claim 1 further comprising performing finite impulse response (FIR) filtering with least mean squares (LMS) adaptation processing on the first time domain signal prior to transforming the first time domain signal band to the frequency domain.

8. The method of claim 1 further comprising smoothing the one or more first filter coefficients and the one or more second filter coefficients.

9. The method of claim 1 further comprising changing an order of the first time domain filter coefficients.

10. A system for signal processing comprising:

a sub-band analysis system configured to receive an input signal and to generate a plurality of sub-band output signals;

a coarse filter configured to receive the sub-band output signals and to generate coarse filtered sub-band outputs;

a plurality of fine spectral decomposition systems, each configured to receive one of the coarse filtered sub-band outputs and to generate a frequency domain sub-band output;

a plurality of filter design systems, each configured to receive one of the frequency domain sub-band outputs and to generate a frequency domain sub-band filter; and

a plurality of frequency to time domain converters, each configured to receive one of the frequency domain sub-band filters and to generate a time domain sub-band filter.

11. The system of claim 10 wherein the sub-band analysis system comprises a uniform discrete Fourier transform filter bank.

12. The system of claim 10 wherein the plurality of coarse filters each comprises an echo cancellation system.

13. The system of claim 10 wherein each of the plurality of fine spectral decomposition systems comprises a windowed fast Fourier transform system.

14. The system of claim 10 wherein each of the plurality of filter design systems comprises a beam forming system.

15. The system of claim 10 further comprising a plurality of buffers, each configured to receive one of the coarse filtered sub-band outputs and to store the coarse filtered sub-band output.

16. The system of claim 15 wherein each of the time domain sub-band filters is applied to one of the stored coarse filtered sub-band outputs to generate an output.

17. The system of claim 15 wherein the outputs are combined.

18. A system for signal processing comprising:

a sub-band analysis system configured to receive an input signal and to generate a plurality of sub-band output signals;

a filter configured to receive the sub-band output signals and to generate coarse filtered sub-band outputs;

a plurality of buffers, each configured to receive one of the coarse filtered sub-band outputs and to store the coarse filtered sub-band output;

a plurality of fine spectral decomposition systems, each configured to receive one of the coarse filtered sub-band outputs and to generate a frequency domain sub-band output;

a plurality of filter design systems, each configured to receive one of the frequency domain sub-band outputs and to generate a frequency domain sub-band filter;

a plurality of frequency to time domain converters, each configured to receive one of the frequency domain sub-band filters and to generate a time domain sub-band filter;

wherein the sub-band analysis system comprises a uniform discrete Fourier transform filter bank;

wherein the plurality of coarse filters each comprise an echo cancellation system;

wherein each of the plurality of fine spectral decomposition systems comprises a windowed fast Fourier transform system;

wherein each of the plurality of filter design systems comprises a beam forming system;

wherein each of the time domain sub-band filters is applied to one of the stored coarse filtered sub-band outputs to generate an output, and the outputs are combined.