Noise suppression
View Patent ↗A method and apparatus for noise suppression is described herein. The channel gain is controlled based on a degree of variability of the background noise. The noise variability estimate is used in conjunction with a variable attenuation concept to produce a family of gain curves that are adaptively suited for a variety of combinations of long-term peak SNR and noise variability. More specifically, a measure of the variability of the background noise is used to provide an optimized threshold that reduces the occurrence of non-stationary background noise entering into the transition region of the gain curve.
1. An apparatus comprising:
a noise variability estimator determining an amount of variability of background noise (σ n ) in a speech-plus-noise signal and for determining an SNR gain threshold (σ th ); and
a channel gain generator adjusting a gain (γ DB ) applied to the speech-plus-noise signal based on the SNR gain threshold such that
γ DB ( i )=μ g (σ q ″( i )−σ th (σ n ))+γ n
wherein
μ g is a gain slope,
σ q ″ (i) is a channel SNR for channel i,
γ n is a minimum overall gain.
2. The apparatus of claim 1 wherein the speech-plus-noise signal is attenuated a first amount when the variability in background noise is high, otherwise the speech-plus-noise signal is attenuated a second amount.
3. The apparatus of claim 1 wherein the channel gain generator adjusts the gain of the speech-plus-noise signal by increasing the signal-to-noise (SNR) gain threshold when there exists a high variability in the background noise, wherein the speech-plus-noise signal will have a maximum attenuation when the SNR value of the speech-plus-noise signal is below the SNR gain threshold.
4. The apparatus of claim 3 wherein the maximum attenuation is based on the variability in background noise.
5. The apparatus of claim 1 farther comprising:
a channel signal-to-noise (SNR) estimator; and
wherein the noise variability estimator comprises an SNR variability estimator.
6. An apparatus for noise suppression, the apparatus comprising:
a channel signal energy estimator for estimating a total energy of a speech-plus-noise signal;
a noise energy estimator for estimating a noise energy of the speech-plus-noise signal;
a channel signal-to-noise (SNR) estimator having the noise energy estimate and the total energy estimate as an input and outputting an SNR estimate of the speech-plus-noise signal;
a SNR variability estimator for estimating the SNR variability (σ n ) of the noise energy estimate and for determining an SNR gain threshold (σ th ); and
a channel gain generator for adjusting a gain (γ DB ) of the speech-plus-noise signal based on the SNR variability of the noise energy estimate such that
γ DB ( i )=μ g (σ q ″( i )−σ th (σ n ))+γ n
wherein
μ g is a gain slope,
σ q ″ (i) is a channel SNR for channel i,
γ n is a minimum overall gain.
7. The apparatus of claim 6 wherein the speech-plus-noise signal is attenuated a first amount when the SNR variability is high, otherwise the speech-plus-noise signal is attenuated a second amount.
8. The apparatus of claim 6 wherein the channel gain generator increases an SNR threshold when there exists a high SNR variability, wherein the speech-plus-noise signal will have a maximum attenuation when the SNR estimate of the speech-plus-noise signal is below the SNR threshold.
9. The apparatus of claim 8 wherein the maximum attenuation is based on the SNR variability.
10. A method for noise suppression, the method comprising the steps of:
estimating an amount of variability in background noise in a speech-plus-noise signal;
determining an SNR gain threshold that is a function of the amount of variability in the background noise in a speech-plus-noise signal; and
adjusting a channel gain applied to the speech-plus noise signal based on the SNR gain threshold, wherein the SNR gain threshold is the value of a channel SNR where the channel gain begins to increase from a minimum value.
11. The method of claim 10 farther comprising the step of:
attenuating the speech-plus-noise signal a first amount when the SNR variability is high, otherwise attenuating the speech-plus-noise signal a second amount.
12. The method of claim 10 wherein the step of adjusting the gain comprises the step of attenuating the speech-plus-noise signal a first amount when the variability in background noise is high, otherwise attenuating the speech-plus-noise signal a second amount.
13. The method of claim 10 wherein the step of adjusting the gain comprises the step of adjusting a gain in the speech-plus-noise signal by increasing a signal-to-noise (SNR) threshold when there exists a high variability in the background noise, wherein the speech-plus-noise signal will have a maximum attenuation when the SNR value of the speech-plus-noise signal is below the SNR threshold.
14. The method of claim 10 wherein the channel gain (γ DB (i)) is adjusted according to the relations:
γ DB ( i )=μ g (σ q ″( i )−σ th (σ n ))+γ n
and
γ n ≦γ DB ( )≦ 0
where
μ g is a gain slope,
σ q ″ is the channel SNR for channel i,
σ th (σ n is the SNR gain threshold as a function of the background noise variability σ n ,
γ n is a minimum overall gain.