Dynamic noise reduction using linear model fitting
View Patent ↗A speech enhancement system improves the speech quality and intelligibility of a speech signal. The system includes a time-to-frequency converter that converts segments of a speech signal into frequency bands. A signal detector measures the signal power of the frequency bands of each speech segment. A background noise estimator measures a background noise detected in the speech signal. A dynamic noise reduction controller dynamically models the background noise in the speech signal. The speech enhancement renders a speech signal perceptually pleasing to a listener by dynamically attenuating a portion of the noise that occurs in a portion of the spectrum of the speech signal.
1. A noise attenuation system, comprising:
a modeler configured to fit a first line to a first portion of a sound signal and a second line to a second portion of the sound signal;
a dynamic noise adjuster configured to calculate a difference in slope or coordinate intercept between the first line and the second line, and calculate a dynamic adjustment factor based on the difference; and
a dynamic noise processor configured to attenuate a portion of a noise detected in the sound signal based on the dynamic adjustment factor.
2. The system of claim 1 , where the first portion of the sound signal comprises a frequency portion below a cutoff frequency threshold and the second portion of the sound signal comprises a frequency portion above the cutoff frequency threshold.
3. The system of claim 2 , where the dynamic noise processor is configured to attenuate the first portion of the sound signal below the cutoff frequency threshold based on a constant attenuation factor and the dynamic adjustment factor.
4. The system of claim 3 , where the dynamic noise processor is configured to attenuate the second portion of the signal above the cutoff frequency threshold based on the constant attenuation factor without the dynamic adjustment factor.
5. The system of claim 1 , where the first line is a first linear regression model and the second line is a second linear regression model.
6. The system of claim 1 , where the dynamic noise adjuster is configured to calculate the difference by calculating a difference between a slope of the first line and a slope of the second line.
7. The system of claim 1 , where the dynamic noise adjuster is configured to calculate the difference by calculating a difference between a coordinate intercept of the first line and a coordinate intercept of the second line.
8. A noise attenuation system, comprising:
a modeler configured to fit a first line to a first portion of a sound signal and a second line to a second portion of the sound signal;
a dynamic noise adjuster configured to calculate a difference between the first line and the second line, and calculate a dynamic adjustment factor based on the difference; and
a dynamic noise processor configured to attenuate a portion of a noise detected in the sound signal based on the dynamic adjustment factor;
where the first line is a first linear regression model and the second line is a second linear regression model; and
where the modeler is configured to fit the first linear regression model to the first portion of a power spectrum of the sound signal, and fit the second linear regression model to the second portion of the power spectrum of the sound signal.
9. A noise attenuation method, comprising:
fitting a first line to a first portion of a sound signal;
fitting a second line to a second portion of the sound signal;
calculating a difference in slope or coordinate intercept between the first line and the second line;
calculating a dynamic adjustment factor based on the difference; and
attenuating a portion of a noise detected in the sound signal based on the dynamic adjustment factor.
10. The method of claim 9 , where the first portion of the sound signal comprises a frequency portion below a cutoff frequency threshold and the second portion of the sound signal comprises a frequency portion above the cutoff frequency threshold.
11. The method of claim 10 , where the step of attenuating comprises attenuating the first portion of the sound signal below the cutoff frequency threshold based on a constant attenuation factor and the dynamic adjustment factor.
12. The method of claim 11 , further comprising attenuating the second portion of the signal above the cutoff frequency threshold based on the constant attenuation factor without the dynamic adjustment factor.
13. The method of claim 9 , where the first line is a first linear regression model and the second line is a second linear regression model.
14. The method of claim 9 , where the step of calculating the difference comprises calculating a difference between a slope of the first line and a slope of the second line.
15. The method of claim 9 , where the step of calculating the difference comprises calculating a difference between a coordinate intercept of the first line and a coordinate intercept of the second line.
16. A noise attenuation method, comprising:
fitting a first line to a first portion of a sound signal;
fitting a second line to a second portion of the sound signal;
calculating a difference between the first line and the second line;
calculating a dynamic adjustment factor based on the difference; and
attenuating a portion of a noise detected in the sound signal based on the dynamic adjustment factor;
where the first line is a first linear regression model and the second line is a second linear regression model, where the step of fitting the first line comprises fitting the first linear regression model to the first portion of a power spectrum of the sound signal, and where the step of fitting the second line comprises fitting the second linear regression model to the second portion of the power spectrum of the sound signal.
17. A non-transitory computer-readable medium with instructions stored thereon, where the instructions are executable by a processor to cause the processor to perform the steps of:
fitting a first line to a first portion of a sound signal;
fitting a second line to a second portion of the sound signal;
calculating a difference in slope or coordinate intercept between the first line and the second line;
calculating a dynamic adjustment factor based on the difference; and
attenuating a portion of a noise detected in the sound signal based on the dynamic adjustment factor.
18. The non-transitory computer-readable medium of claim 17 , where the first portion of the sound signal comprises a frequency portion below a cutoff frequency threshold and the second portion of the sound signal comprises a frequency portion above the cutoff frequency threshold;
where the step of attenuating comprises attenuating the first portion of the sound signal below the cutoff frequency threshold based on a constant attenuation factor and the dynamic adjustment factor; and
where the instructions are further executable by the processor to cause the processor to perform the step of attenuating the second portion of the signal above the cutoff frequency threshold based on the constant attenuation factor without the dynamic adjustment factor.
19. The non-transitory computer-readable medium of claim 17 where the step of calculating the difference comprises calculating a difference between a slope of the first line and a slope of the second line.
20. The non-transitory computer-readable medium of claim 17 where the step of calculating the difference comprises calculating a difference between a coordinate intercept of the first line and a coordinate intercept of the second line.