Method and apparatus for environmental noise compensation by determining a presence or an absence of an audio event

Patent metadata

Field	Value
Publication number	US-9711162-B2
Application number	US-201213539380-A
Country	US
Kind code	B2
Filing date	Jun 30, 2012
Priority date	Jul 5, 2011
Publication date	Jul 18, 2017
Grant date	Jul 18, 2017

Abstract

A method of environmental noise compensation a speech audio signal is provided that includes estimating a fast audio energy level and a slow audio energy level in an audio environment, wherein the speech audio signal is not part of the audio environment, and applying a gain to the speech audio signal to generate an environment compensated speech audio signal, wherein the gain is updated based on the estimated slow audio energy level when the estimated fast audio energy level is not indicative of an audio event in the audio environment and the estimated gain is not updated when the estimated fast audio energy level is indicative an audio event in the audio environment.

First claim

What is claimed is:

1. A method automatically performed by a system for environmental noise compensation of a first speech captured by a first audio capture device in a speech audio signal outside of an audio environment, the method comprising:
estimating a fast audio energy level and a slow audio energy level from an environment audio signal captured by a second audio capture device in the audio environment, wherein:
the fast audio energy level corresponds to a second speech captured in the audio environment, and
the slow audio energy level corresponds to an ambient noise captured in the audio environment;
applying a gain to the speech audio signal to generate an environment compensated speech audio signal;
determining either a presence or an absence of an audio event by comparing the fast audio energy level with the slow audio energy level against a predetermined energy threshold;
updating the gain based on the estimated slow audio energy level during the absence of the audio event; and
freezing the gain at a current level during the presence of the audio event.

2. The method of claim 1 , wherein estimating the fast audio energy level comprises estimating the fast audio energy level from the environment audio signal captured by the second audio capture device in the audio environment.

3. The method of claim 1 , wherein estimating the fast audio energy level comprises estimating the fast audio energy level based on a primary audio signal captured by a primary audio capture device in the audio environment and a secondary audio signal captured by a secondary audio capture device in the audio environment.

4. The method of claim 1 , wherein estimating the fast audio energy level comprises:
adapting the estimated slow audio energy level in proportion to a slow adaptation parameter when the estimated fast audio energy level is indicative of the audio event; and
adapting the estimated slow audio energy level in proportion to a fast adaptation parameter when the estimated fast audio energy level is not indicative of the audio event.

5. The method of claim 1 , further comprising:
filtering the speech audio signal;
normalizing the gain based on a maximum gain; and
wherein applying a gain comprises:
applying the gain to the filtered speech audio signal; and
mixing the gained, filtered speech audio signal and the speech audio signal in amounts proportional to the normalized gain to generate the environment compensated speech audio signal.

6. A method performed by a system for environmental noise compensation of a first speech captured by a first audio capture device in a speech audio signal outside an audio environment, the method comprising:
tracking a noise level and a speech level from samples of an environment audio signal captured by a second audio capture device in the audio environment, wherein:
the speech level corresponds to a second speech in the audio environment, and
the noise level corresponds to an ambient noise in the audio environment;

determining either a presence or an absence of the second speech in the audio environment by comparing the noise level and the speech level against a predetermined threshold;
updating a gain based on the noise level during the absence of the second speech;
freezing the gain at a current level during the presence of the second speech; and
applying the gain to the speech audio signal to generate an environment compensated speech audio signal.

7. The method of claim 6 , wherein tracking the noise level comprises estimating the noise level from the samples of the environment audio signal captured by the second audio capture device in the audio environment.

8. The method of claim 6 , wherein tracking the noise level comprises estimating the noise level based on a primary audio signal captured by a primary audio capture device in the audio environment and a secondary audio signal captured by a secondary audio capture device in the audio environment.

9. The method of claim 6 , wherein tracking the noise level comprises:
adapting the estimated noise level in proportion to a slow adaptation parameter when the second speech is present in the audio environment; and
adapting the estimated noise level in proportion to a fast adaptation parameter when the second speech is not present in the audio environment.

10. The method of claim 6 , further comprising:
filtering the speech audio signal;
normalizing the gain based on a maximum gain; and
wherein applying the gain comprises:
applying the gain to the filtered speech audio signal; and
mixing the gained, filtered speech audio signal and the speech audio signal in amounts proportional to the normalized gain to generate the environment compensated speech audio signal.

11. A digital system comprising:
a processor;
means for receiving an audio signal captured by a first audio capture device in an audio environment;
means for receiving a speech audio signal carrying a first speech captured by a second audio capture device outside of the audio environment; and
a non-transitory memory configured to store instructions that, when executed by the processor, cause the digital system to perform a method comprising:
estimating a fast audio energy level corresponding to a second speech captured in the audio environment;
estimating a slow audio energy level corresponding to an ambient noise captured in the audio environment;
applying a gain to the speech audio signal to generate an environment compensated speech audio signal;
determining either a presence or an absence of an audio event by comparing the fast audio energy level with the slow audio energy level against a predetermined energy threshold;
updating the gain based on the estimated slow audio energy level during the absence of the audio event; and
freezing the gain at a current level during the presence of the audio event.

12. The digital system of claim 11 , wherein the first audio capture device is configured to capture the environment audio signal, and wherein estimating the fast audio energy level comprises estimating the fast audio energy level and the slow audio energy level from the environment audio signal.

13. The digital system of claim 11 , wherein the first audio capture device includes a primary audio capture device configured to capture a primary audio signal and a secondary audio capture device configured to capture a secondary audio signal, and wherein estimating the fast audio energy level comprises estimating the fast audio energy level and the slow audio energy level based on the primary audio signal and the secondary audio signal.

14. The digital system of claim 11 , wherein estimating the fast audio energy level comprises:
adapting the estimated slow audio energy level in proportion to a slow adaptation parameter when the estimated fast audio energy level is indicative of an audio event; and
adapting the estimated slow audio energy level in proportion to a fast adaptation parameter when the estimated fast audio energy level is not indicative of an audio event.

15. The digital system of claim 11 , the method further comprising:
filtering the speech audio signal;
normalizing the gain based on a maximum gain; and
wherein applying a gain comprises:
applying the gain to the filtered speech audio signal; and

mixing the gained, filtered speech audio signal and the speech audio signal in amounts proportional to the normalized gain to generate the environment compensated speech audio signal.

Assignees

• Murthy Nitish Krishna
• Unno Takahiro
• Cole Edwin R
• Texas Instruments Inc

Inventors

• Murthy Nitish Krishna
• Unno Takahiro
• Cole Edwin R

Classifications

• G10L21/0208 Primary

  Noise filtering · CPC title

• H03G3/001

  Digital control of analog signals · CPC title

• G10L25/78

  Detection of presence or absence of voice signals (switching of direction of transmission by voice frequency in two-way loud-speaking telephone systems H04M9/10) · CPC title

• G10L25/84

  for discriminating voice from noise · CPC title

• G10L15/20

  Speech recognition techniques specially adapted for robustness in adverse environments, e.g. in noise, of stress induced speech (G10L21/02 takes precedence) · CPC title