Low-complexity comfort noise generator
View Patent ↗A comfort noise generator ( 104 ) suitable for use in a communication system includes a finite impulse response (FIR) filter ( 136 ), a random number generator ( 140 ), and a coefficient updater ( 138 ). The coefficient updater ( 138 ) determines an updated set of filter coefficients ( 142 ) based on the signal frame of the input signal ( 102 ). The updated set of filter coefficients ( 142 ) is output to the FIR filter ( 136 ). The FIR filter ( 136 ) shapes a white noise signal ( 146 ) supplied by the random number generator ( 140 ) to provide a simulated background noise signal, or comfort noise signal ( 122 ). The comfort noise signal ( 122 ) is selectively output from an echo suppression system or corresponding method to overwrite or suppress reflected residual echoes.
1. A comfort noise generator for an echo suppression system comprising:
a finite impulse response (FIR) filter;
a coefficient updater for receiving a digital signal frame and for generating and outputting an updated set of filter coefficients to the FIR filter, the updated set of filter coefficients depending on a best fit subframe of the digital signal frame and a current set of filter coefficients; and
a random number generator for generating a white noise signal and for outputting the white noise signal to the FIR filter,
the FIR filter for shaping the white noise signal to correspond to the digital signal frame based on the updated set of filter coefficients received from the coefficient updater provide a comfort noise signal from said FIR filter to the echo suppression system.
2. The comfort noise generator of claim 1 , wherein the digital signal frame is a digital background noise signal frame.
3. The comfort noise generator of claim 1 , wherein the coefficient updater further comprises:
a buffer for queuing and outputting a current set of filter coefficients;
a correlator connected to the buffer for receiving the current set of filter coefficients from the buffer, for correlating the digital signal frame with the current set of filter coefficients to determine the best fit subframe of the digital signal frame, and for outputting the best fit subframe of the digital signal frame; and
an integrator for combining the best fit subframe of the digital signal frame with the current set of filter coefficients to produce and output the updated set of filter coefficients to the FIR filter.
4. The comfort noise generator of claim 3 , wherein the coefficient updater further comprises a speech detector coupled to the correlator, the speech detector for detecting the levels of the digital signal frame.
5. The comfort noise generator of claim 3 , wherein the integrator is for combining the best fit subframe of the digital signal frame with a linear combination of previous sets of filter coefficients to produce the updated set of filter coefficients.
6. The comfort noise generator of claim 3 , wherein the digital signal frame comprises a set of 80 data samples and the current set of filter coefficients comprises a set of 50 data samples.
7. The comfort noise generator of claim 3 , wherein the coefficient updater further comprises:
a window for receiving and conditioning the best fit subframe of the digital signal frame from the correlator and for outputting the best fit subframe of the digital signal frame to the integrator.
8. The comfort noise generator of claim 7 , wherein the window is a spectral estimate enhancement window.
9. The comfort noise generator of claim 3 , wherein the integrator further comprises:
a first attenuator for receiving the best fit subframe of the digital signal frame from the correlator and for attenuating the best fit subframe of the digital signal frame by a first attenuation factor;
a second attenuator for receiving the current set of filter coefficients from the buffer and for attenuating the current set of filter coefficients by a second attenuation factor; and
an adder for combining the best fit subframe of the digital signal frame that has been attenuated by the first attenuation factor with the current set of filter coefficients that has been attenuated by the second attenuation factor to produce the updated set of filter coefficients, and for outputting the updated set of filter coefficients to the buffer to replace the current set of filter coefficients.
10. The comfort noise generator of claim 4 , wherein the correlator correlates the digital signal frame with the current set of filter coefficients to determine a best fit subframe of the digital signal frame when the speech detector detects that the digital background noise signal frame is below a predetermined threshold.
11. The comfort noise generator of claim 10 , wherein the predetermined threshold corresponds to a noise energy of a voice signal.
12. The comfort noise generator of claim 11 , wherein the predetermined threshold is within 0.5 dB of a continuously measured noise floor.
13. The comfort noise generator of claim 1 , wherein the random number generator comprises a 16-bit linear feedback shift register.
14. An echo suppression system for a communications device comprising:
an acoustic echo canceller for receiving a digital audio signal, for removing estimated echoes from the digital audio signal, and for outputting a modified digital audio signal;
a comfort noise generator including a finite impulse response (FIR) filter for receiving the digital audio signal and for approximating the digital audio signal to produce a simulated signal, the comfort noise generator further comprising a coefficient updater for receiving the digital audio signal and for generating and outputting an updated set of filter coefficients to the FIR filter, the undated set of filter coefficients depending on a best fit portion of the digital audio signal and a current set of filter coefficients;
a soft switch for switching between the modified digital audio signal and the simulated signal to output a transmit output signal; and
an echo suppression controller for receiving the digital audio signal and a far-end receive signal and for controlling switching of the soft switch between the modified digital audio signal and the simulated signal based on values of the digital audio signal and the far-end receive signal.
15. The echo suppression system of claim 14 , wherein the echo suppression controller is further for determining an attenuation factor based on levels of the digital audio signal and the far-end receive signal and for outputting a signal representative of the attenuation factor and a signal representative of an inverse attenuation factor.
16. The echo suppression system of claim 15 , wherein the soft switch further comprises:
a first variable gain attenuator for receiving the signal representative of the attenuation factor from the echo suppression controller, for receiving the modified digital audio signal from the acoustic echo canceller, and for attenuating the modified digital audio signal based on the attenuation factor to produce a first attenuated signal;
a second variable gain attenuator for receiving the signal representative of the inverse attenuation factor from the echo suppression controller, for receiving the simulated signal from the comfort noise generator, and for attenuating the simulated signal based on the inverse attenuation factor to produce a second attenuated signal; and
an output adder in communication with the first and second variable gain attenuators for combining the first attenuated signal and the second attenuated signal to produce the transmit output signal.
17. A method for suppressing echoes in a communications device, the method comprising:
receiving a digital audio signal;
removing estimated echoes from the digital audio signal to provide a modified digital audio signal;
generating a simulated signal using a comfort noise generator comprising a random number generator and a finite impulse response (FIR) filter; and
selectively switching between the modified digital audio signal and the simulated signal to produce a transmit output signal,
wherein the generating a simulated signal further comprises:
queuing a current set of filter coefficients;
correlating the digital audio signal with the current set of filter coefficients to determine a best fit subframe of the digital audio signal: and
combining the best fit subframe of the digital audio signal and a current set of filter coefficients to produce an undated set of filter coefficients.
18. The method of claim 17 , wherein the selectively switching between the modified digital audio signal and the simulated signal further comprises:
attenuating the modified digital audio signal by an attenuation factor;
attenuating the simulated signal by an inverse attenuation factor; and
combining the attenuated modified digital audio signal and the attenuated simulated signal to produce the transmit output signal.
19. The method of claim 17 , wherein the generating a simulated signal using a comfort noise generator comprising a random number generator and a FIR filter further comprises attenuating the best fit subframe of the digital audio signal and the current set of filter coefficients before combining the best fit subframe of the digital audio signal and the current set of filter coefficients.
20. The method of claim 19 , wherein the generating a simulated signal using a comfort noise generator comprising a random number generator and a FIR filter further comprises conditioning the best fit subframe of the digital audio signal before attenuating the best fit subframe of the digital audio signal.
21. The method of claim 17 , wherein the generating a simulated signal using a comfort noise generator comprising a random number generator and a FIR filter further comprises replacing the current set of filter coefficients with the updated set of filter coefficients when the digital audio signal is within a predetermined threshold.
22. The method of claim 21 , wherein the generating a simulated signal using a comfort noise generator comprising a random number generator and a FIR filter further comprises replacing the current set of filter coefficients with the updated set of filter coefficients when the digital audio signal is within 0.5 dB of a continuously measured noise floor.