Babble noise suppression

Patent metadata

Field	Value
Publication number	US-10783899-B2
Application number	US-201616073740-A
Country	US
Kind code	B2
Filing date	Nov 18, 2016
Priority date	Feb 5, 2016
Publication date	Sep 22, 2020
Grant date	Sep 22, 2020

Abstract

Systems and methods are introduced to perform noise suppression of an audio signal. The audio signal includes foreground speech components and background noise. The foreground speech components correspond to speech from a user's speaking into an audio receiving device. The background noise includes babble noise that includes speech from one or more interfering speakers. A soft speech detector determines, dynamically, a speech detection result indicating a likelihood of a presence of the foreground speech components in the audio signal. The speech detection result is employed to control, dynamically, an amount of attenuation of the noise suppression to reduce the babble noise in the audio signal. Further processing achieves a more stationary background and reduction of musical tones in the audio signal.

First claim

What is claimed is:

1. A method of performing noise suppression of an audio signal, the audio signal including foreground speech components and background noise, the method comprising: determining, dynamically, a speech detection result indicating a likelihood of a presence of the foreground speech components in the audio signal; and computing, dynamically, spectral weighting coefficients based on the speech detection result determined and applying the spectral weighting coefficients computed to the audio signal to suppress the background noise in a dynamic manner.

2. The method of claim 1 , further comprising: computing, dynamically, a dynamic noise overestimation factor based on the speech detection result determined, wherein the spectral weighting coefficients are computed based on the dynamic noise overestimation factor; determining periods of speech pauses and periods of speech activity in the audio signal as a function of the speech detection result determined; and increasing a value of the dynamic noise overestimation factor for the periods of speech pauses determined relative to the value of the dynamic noise overestimation factor for the periods of speech activity determined, wherein increasing the value of the dynamic noise overestimation factor enables the spectral weighting coefficients computed to increase suppression of the background noise relative to an amount of suppression of the background noise for the periods of speech activity determined.

3. The method of claim 1 , further including estimating a power spectrum of the audio signal based on a transformation of the audio signal from a time domain to a frequency domain, wherein the speech detection result determined is a function of a combination of feature values determined in the time domain, frequency domain, or a combination thereof.

4. The method of claim 3 , wherein the combination of feature values includes kurtosis and at least one other feature value.

5. The method of claim 1 , wherein the background noise includes stationary and non-stationary noise components.

6. The method of claim 5 , wherein changes in a power spectrum of the audio signal over a time interval are less for the stationary noise components than for the non-stationary noise components.

7. The method of claim 5 , further including: computing, dynamically, a dynamic noise floor; and selectively lowering the dynamic noise floor based on frequencies corresponding to the non-stationary noise components, wherein computing the spectral weighting coefficients is further based on the dynamic noise floor computed and selectively lowered.

8. A method of performing noise suppression of an audio signal, the audio signal including foreground speech components and background noise, the method comprising: determining, dynamically, a speech detection result indicating a likelihood of a presence of the foreground speech components in the audio signal; and computing, dynamically, spectral weighting coefficients based on the speech detection result determined and applying the spectral weighting coefficients computed to the audio signal to suppress the background noise in a dynamic manner, the method further including: identifying one or more spectral weighting coefficients from the spectral weighting coefficients computed based on contextual information from neighboring spectral weighting coefficients; and post-processing the spectral weighting coefficients computed by setting first values computed for the one or more spectral weighting coefficients identified to second values, the second values enabling a stronger attenuation of the background noise than the first values, and further wherein the applying includes applying the spectral weighting coefficients computed and post-processed.

9. The method of claim 1 , further comprising: pre-processing the audio signal to pre-emphasize spectral characteristics of the audio signal and wherein: the speech detection result indicates the likelihood of the presence of the foreground speech components in the pre-processed audio signal; the determining and the computing are performed for a given time interval of the pre-processed audio signal; and the applying includes applying the spectral weighting coefficients computed to the pre-processed audio signal in the given time interval.

10. The method of claim 1 , wherein the foreground speech components correspond to speech from a user speaking into an audio receiving device and further wherein the background noise includes babble noise, the babble noise including a composition of multiple background speech components from other speakers.

11. A system configured to perform noise suppression of an audio signal, the audio signal including foreground speech components and background noise, the system comprising: a soft speech detector configured to determine, dynamically, a speech detection result indicating a likelihood of a presence of the foreground speech components in the audio signal; and a noise suppressor communicatively coupled to the soft speech detector to receive the speech detection result determined and configured to compute, dynamically, spectral weighting coefficients based on the speech detection result determined and apply the spectral weighting coefficients computed to the audio signal to suppress the background noise in a dynamic manner.

12. The system of claim 11 , wherein the noise suppressor is further configured to: compute, dynamically, a dynamic noise overestimation factor based on the speech detection result determined, wherein the spectral weighting coefficients are computed based on the dynamic noise overestimation factor; determine periods of speech pauses and periods of speech activity in the audio signal as a function of the speech detection result determined; and increase a value of the dynamic noise overestimation factor for the periods of speech pauses determined relative to the value of the dynamic noise overestimation factor for the periods of speech activity determined, wherein increasing the value of the dynamic noise overestimation factor enables the spectral weighting coefficients computed to increase suppression of the background noise relative to an amount of suppression of the background noise for the periods of speech activity determined.

13. The system of claim 11 , further including a spectrum estimator configured to estimate a power spectrum of the audio signal based on a transformation of the audio signal from a time domain to a frequency domain, wherein the soft speech detector is further configured to determine the speech detection result as a function of a combination of feature values determined in the time domain, frequency domain, or a combination thereof.

14. The system of claim 13 , wherein the combination of feature values includes kurtosis and at least one other feature value.

15. The system of claim 11 , wherein the background noise includes stationary and non-stationary noise components and further wherein changes in a power spectrum of the audio signal over a time interval are less for the stationary noise components than for the non-stationary noise components.

16. The system of claim 15 , wherein the noise suppressor is further configured to: compute, dynamically, a dynamic noise floor; and selectively lower the dynamic noise floor based on frequencies corresponding to the non-stationary noise components, wherein the spectral weighting coefficients are computed further based on the dynamic noise floor computed and selectively lowered.

17. A system configured to perform noise suppression of an audio signal, the audio signal including foreg

Assignees

• Cerence Operating Co

Inventors

• Graf Simon
• Herbig Tobias
• Buck Markus

Classifications

• G10L21/0216 Primary

  characterised by the method used for estimating noise · CPC title

• G10L25/93

  Discriminating between voiced and unvoiced parts of speech signals (G10L25/90 takes precedence) · CPC title

• G10L2021/02087

  the noise being separate speech, e.g. cocktail party · CPC title

• G10L21/0208

  Noise filtering · CPC title

• G10L25/21

  the extracted parameters being power information · CPC title