Granted Patent B2

US 8,705,759 · App. 12/749,136 · Granted Apr 22, 2014

Abstract

Claims (45)

1. A computer implemented method for removing a reverberation component from a noise component of an input signal, comprising:

estimating, by a processor, the noise component of a power density spectrum of the input signal by filtering the input signal with a blocking matrix, wherein output from the blocking matrix includes the noise component and the reverberation component;

estimating, by a processor, the reverberation component in the noise component;

removing, by a processor, the reverberation component from the noise component to obtain a modified estimate of the noise component,

wherein estimating the reverberation component comprises using an adaptive filter to simulate a power impulse response of the reverberation component in the power density spectrum of the input signal.

2. The computer implemented method according to claim 1 , further comprising adapting the adaptive filter to minimize a difference between the reverberation component and the noise component.

3. The computer implemented method of claim 1 wherein the reverberation component includes (i) an early reverberation component following sound of an event causing reverberation and (ii) a late reverberation component occurring later in time than the early reverberation component.

4. The computer implemented method of claim 3 wherein the adaptive filter is used to estimate the early reverberation component, which is removed from the noise component.

5. The computer implemented method of claim 3 further comprising using a post-filter to suppress the late reverberation component.

6. The computer implemented method of claim 3 further comprising estimating the early reverberation component based on, a correlation in time with a wanted signal component of the input signal.

7. The computer implemented method of claim 3 further comprising choosing a length for the adaptive filter to simulate a portion of the power impulse response.

8. The computer implemented method of claim 7 wherein the portion of the power impulse response corresponds to the early reverberation component.

9. A computer implemented method for removing a reverberation component from a noise component of an input signal, comprising:

estimating, by a processor, the noise component in the input signal by filtering the input signal with a blocking matrix, wherein output from the blocking matrix includes the noise component and the reverberation component;

estimating, by a processor, the reverberation component in the noise component;

removing, by a processor, the estimated reverberation component from the estimated noise component to obtain a modified estimate of the noise component,

wherein estimating the reverberation component comprises:

determining an estimate of a zero average noise component with a temporal average of zero based on the estimated noise component.

10. The computer implemented method according to claim 9 , further comprising, before determining the estimate of the zero-average noise component:

detecting whether a wanted component is present in the input signal, and

determining a smoothed noise component based on the estimated noise component if no wanted component is detected;

and wherein determining the zero-average noise component is also based on the smoothed noise component.

11. The computer implemented method according to claim 9 , wherein estimating the reverberation component further comprises a computer process for:

determining an estimate of a zero-average input signal with a temporal average of zero based on the input signal; and

performing the step of filtering the input signal using the estimate of the zero-average input signal.

12. A computer program product, tangibly embodied a non-transitory computer readable storage medium, for removing a reverberation component from a noise component of an input signal, the computer program product including instructions operable to cause a data processing apparatus to:

estimate the noise component of a power density spectrum of the input signal by filtering the input signal with a blocking matrix, wherein output from the blocking matrix includes the noise component and the reverberation component;

estimate the reverberation component in the noise component using an adaptive filter to simulate a power impulse response of the reverberation component in the power density spectrum of the input signal;

remove the reverberation component from the noise component to obtain a modified estimate of the noise component.

13. The computer program product according to claim 12 , further including instructions operable to cause the data processing apparatus to adapt the adaptive filter to minimize a difference between the reverberation component and the noise component.

14. The computer program product according to claim 12 further including instructions operable to cause the data processing apparatus to:

determine an estimate of a zero-average noise component with a temporal average of zero based on the estimated noise component to estimate the reverberation component in the noise component.

15. The computer program product according to claim 14 further including instructions operable to cause the data processing apparatus to, before determining the estimate of the zero-average noise component:

detect whether a wanted component is present in the input signal, and

determine a smoothed noise component based on the estimated noise component if no wanted component is detected;

and wherein determining the zero-average noise component is also based on the smoothed noise component.

16. The computer program product according to claim 14 further including instructions operable to cause the data processing apparatus to:

determine an estimate of a zero-average input signal with a temporal average of zero based on the input signal; and

perform the step of filtering the input signal using the estimate of the zero-average input signal to estimate the reverberation component in the noise component.

17. The computer program product according to claim 12 the reverberation component includes (i) an early reverberation component following sound of an event causing reverberation and (ii) a late reverberation component occurring later in time than the early reverberation component.

18. The computer program product according to claim 17 wherein the adaptive filter is used to estimate the early reverberation component, which is removed from the noise component.

19. The computer implemented method of claim 17 further including instructions operable to cause the data processing apparatus to suppress the late reverberation component using a post-filter.

20. The computer implemented method of claim 17 further including instructions operable to cause the data processing apparatus to estimate the early reverberation component based on a correlation in time with a wanted signal component of the input signal.

21. The computer implemented method of claim 17 further including instructions operable to cause the data processing apparatus to choose a length for the adaptive filter to simulate a portion of the power impulse response.

22. The computer implemented method of claim 21 wherein the portion of the power impulse response corresponds to the early reverberation component.

Assignments (7)