Methods and devices for updating filter coefficients during echo cancellation
View Patent ↗Methods and devices for updating filter coefficients during echo cancellation are disclosed. A communication device includes a processor configured to operate a background filter configured as an adaptive filter, operate a foreground filter configured as a fixed filter; and update the foreground filter with coefficients from the background filter responsive to monitoring a history of both a near-end signal energy and an estimated far-end signal energy for a pre-determined time window in which conditions of each of the near-end signal energy and the estimated far-end signal energy are satisfied. A related method includes monitoring a near-end signal energy, monitoring an estimated far-end signal energy, performing an echo cancellation, and transferring coefficients from an adaptive filter to a fixed filter responsive to a history of both the near-end signal energy and the estimated far-end signal energy satisfying conditions for a pre-determined time window.
1. A communication device configured for echo cancellation, comprising:
a microphone configured to capture a near-end signal;
a speaker configured to produce sound from a far-end signal; and
a processor operably coupled with the microphone and the speaker, the processor configured to:
operate a background filter configured as an adaptive filter having background filter coefficients;
operate a foreground filter configured as a fixed filter having foreground filter coefficients;
update the foreground filter coefficients with coefficients from the background filter responsive to monitoring a recent history of both a near-end signal energy and an estimated far-end signal energy for a predetermined time window in which conditions of each of the near-end signal energy and the estimated far-end signal energy are determined to be satisfied over the predetermined time window having a fixed length (L) that is a last L number of samples taken at any given time for the near-end signal energy and the estimated far-end signal energy; and
generating an echo cancelled signal using the foreground filter.
2. The communication device of claim 1 , wherein the processor is configured to:
count a first number of times that the near-end signal energy is above a first pre-determined energy threshold during the predetermined time window and comparing the first number against a first count threshold to determine whether a first condition is satisfied; and
count a second number of times that the estimated far-end signal is below a second pre-determined energy threshold during the predetermined time window and comparing the second number against a second count threshold to determine whether a second condition is satisfied.
3. The communication device of claim 2 , wherein the first count threshold is defined as a first fraction of the predetermined time window, and the second count threshold is defined as a second fraction of the predetermined time window.
4. The communication device of claim 3 , wherein the first fraction (F x ) is between an interval of 0.7≦F x ≦1 when a first count value and the first count threshold are based on a number of samples being counted above the first pre-determined energy threshold.
5. The communication device of claim 3 , wherein the second fraction (F a ) is between an interval of 0≦F a ≦0.05 when a second count value and the second count threshold are based on a number of samples being counted below the second pre-determined energy threshold.
6. The communication device of claim 1 , wherein the background filter is a recursive least squared (RLS) adaptive filter.
7. The communication device of claim 6 , wherein the RLS adaptive filter implements an algorithm selected from a group consisting of a covariance-matrix-based RLS algorithm and its square root versions that use a tapped-delay line parameterization and the matrix inversion lemma, an order-recursive RLS algorithm that uses a lattice structure for the filter parameterization and error computation, and a fast fixed-order RLS algorithm.
8. The communication device of claim 1 , wherein the processor is configured to operate a state machine to control the RLS adaptive filter and update the foreground filter with coefficients from the background filter.
9. The communication device of claim 8 , wherein the state machine controls an amplitude control block of the near-end signal after transferring the coefficients has been initiated.
10. The communication device of claim 1 , wherein the communication device is a caption-enabled device specifically configured for use by a hearing-impaired user to receive and display text captions from a communication service during a communication session with a far-end communication device.
11. A method for cancelling echo in a communication device during a communication session with a far-end device, the method comprising:
monitoring a near-end signal energy of a near-end signal;
monitoring an estimated far-end signal energy of a far-end signal;
performing an echo cancellation on an input signal that is input to an echo canceller;
determining if the near-end signal energy satisfies a first condition over recent history having a predetermined time window having a fixed length (L) that is a last L number of samples taken at any given time for the near-end signal energy and the estimated far-end signal energy;
determining if the far-end signal energy satisfies a second condition over the predetermined time window;
transferring coefficients from an adaptive background filter to a fixed foreground filter responsive to satisfying the first and second conditions for the pre-determined time window; and generating an echo cancelled signal using the foreground filter.
12. The method of claim 11 , wherein monitoring the near-end signal energy includes generating a first count value for a number of samples of the near-end signal energy on one side of a first energy threshold during the pre-determined time window.
13. The method of claim 12 , wherein monitoring the estimated far-end signal energy includes generating a second count value for a number of samples of the estimated far-end signal energy on one side of a second energy threshold during the pre-determined time window.
14. The method of claim 13 , further comprising determining whether the first and second conditions for the near-end signal energy and the estimated far-end signal energy are satisfied by:
comparing the first count value to a first count threshold value to determine if the first condition is satisfied; and
comparing the second count value to a second count threshold value to determine if the second condition is satisfied.
15. The method of claim 14 , wherein the first count threshold value is a first fraction of the number of samples in the pre-determined time window, and the second count threshold value is a second fraction of the number of samples in the pre-determined time window.
16. The method of claim 14 , wherein the first count value is for the number of samples of the near-end energy that is above the first energy threshold during the pre-determined time window, and the first condition is met if the first count value is greater than the first count threshold.
17. The method of claim 14 , wherein the first count value is for the number of samples of the near-end energy that is below the first energy threshold during the pre-determined time window, and the first condition is met if the first count value is less than the first count threshold.
18. The method of claim 14 , wherein the second count value is for the number of samples of the estimated far-end energy that is below the second energy threshold during the pre-determined time window, and the second condition is met if the second count value is less than the second count threshold.
19. The method of claim 14 , wherein the second count value is for the number of samples of the estimated far-end energy that is above the second energy threshold during the pre-determined time window, and the second condition is met if the second count value is greater than the second count threshold.
20. The method of claim 11 , wherein monitoring the near-end signal energy of the near-end signal and the far-end signal energy of the far-end signal each comprise storing sample values in a circular buffer configured to replace an oldest sampled value with a newest sampled value such that a number of sampled values stored in the circular buffer at a time corresponds to the predetermined time window.
21. A non-transitory computer-readable medium that, when executed by a processor, is configured to operate an echo cancellation module, comprising:
an adaptive background filter configured to generate an adaptive echo canceled output from a line-in signal and a line-out signal; and
a fixed foreground filter configured to generate an echo canceled signal from the line-in signal and the line-out signal, wherein coefficients for the fixed foreground filter are updated with coefficients from the adaptive background filter responsive to satisfying conditions indicating that a near-end signal is generally active and a far-end signal is generally inactive over a recent history of both a near-end signal energy and an estimated far-end signal energy during a pre-determined time window having a fixed length (L) that is a last L number of samples taken at any given time for the near-end signal energy and the estimated far-end signal energy.