Method and Apparatus for Estimating Variability of Background Noise for Noise Suppression

• US 20140270249A1
• Filed: 07/25/2013
• Published: 09/18/2014
• Est. Priority Date: 03/12/2013
• Status: Abandoned Application

• Current Assignee: Google Technology Holdings LLC (Alphabet Inc.)
• Original Assignee: Motorola Mobility LLC (Lenovo Group Ltd.)
• Inventors: Jasiuk, Mark A., Ramabadran, Tenkasi V.
• Application Number:
  US13/950,980
• Publication Number:
  US 20140270249A1
• Time in Patent Office: Days
• US Class Current: 381/94.1
• CPC Class Codes: G10L 15/20 Speech recognition techniques specially adapted for robustness in adverse environments, e.g. in noise, of stress induced speech G10L21/02 takes precedence
• CPC Class Codes: G10L 21/0208 Noise filtering
• CPC Class Codes: G10L 21/0216 characterised by the method used for estimating noise

Abstract

An electronic device measures noise variability of background noise present in a sampled audio signal, and determines whether the measured noise variability is higher than a high threshold value or lower than a low threshold value. If the noise variability is determined to be higher than the high threshold value, the device categorizes the background noise as having a high degree of variability. If the noise variability is determined to be lower than the low threshold value, the device categorizes the background noise as having a low degree of variability. The high and low threshold values are between a high boundary point and a low boundary point. The high boundary point is based on an analysis of files including noises that exhibit a high degree of variability, and the low boundary point is based on an analysis of files including noises that exhibit a low degree of variability.

Claims

1. A method comprising:

measuring noise variability of background noise present in a sampled audio signal;

determining whether the measure of noise variability is higher than a high threshold value or lower than a low threshold value;

if the noise variability is determined to be higher than the high threshold value, categorizing the background noise as having a high degree of variability;

if the noise variability is determined to be lower than the low threshold value, categorizing the background noise as having a low degree of variability;

wherein the high threshold value and low threshold value are between a high boundary point and a low boundary point;

wherein the high boundary point is based on an analysis of a first data set including noises that exhibit a high degree of variability; and

wherein the low boundary point is based on an analysis of a second data set including noises that exhibit a low degree of variability.

2. The method of claim 1, further comprising:

if the background noise is categorized as having a high degree of variability, suppressing the background noise using a first noise suppression algorithm; and

if the background noise is categorized as having a low degree of variability, suppressing the background noise using a second noise suppression algorithm.

3. The method of claim 1, further comprising:

if the noise variability is determined to be between the low threshold value and the high threshold value, categorizing the background noise as having a degree of variability of a previous frame.

4. The method of claim 1, wherein the measuring of noise variability of the background noise comprises:

determining whether a frame including the background noise is a noise update frame;

if the frame is determined not to be a noise update frame, categorizing the background noise as having a degree of variability of a previous frame; and

if the frame is determined to be a noise update frame, determining whether the frame is part of a sequence of contiguous noise frames.

5. The method of claim 4, wherein the measuring of the noise variability of the background noise further comprises:

if the frame is determined not to be part of a sequence of contiguous noise frames, categorizing the background noise as having the degree of variability of the previous frame.

6. The method of claim 4, wherein if the frame is determined to be part of a sequence of contiguous noise frames, the measuring of the noise variability of the background noise further comprises:

determining a maximum value of smoothed channel noise and a minimum value of smoothed channel noise in the sequence of contiguous noise frames;

computing a smoothed maximum dB difference using the maximum value of smoothed channel noise and the minimum value of smoothed channel noise; and

calculating the noise variability of the background noise using a ratio of a difference between the smoothed maximum dB difference and the low boundary point to a difference between the high boundary point and the low boundary point.

7. The method of claim 1, wherein the measure of noise variability of the background noise is calculated using the following equation:

8. The method of claim 7, wherein the measure of noise variability of the background noise is calculated using the following equation:

9. The method of claim 1, wherein the measuring of noise variability of the background noise comprises:

measuring noise level of the background noise;

determining whether the measured noise level of the background noise is lower than a noise level threshold value;

if the noise level of the background noise is determined to be lower than the noise level threshold value, calculating a bias energy value;

adding the bias energy value to smoothed channel noise to generate modified smoothed channel noise; and

measuring the noise variability of the background noise using the modified smoothed channel noise.

10. The method of claim 1, wherein the measuring of noise variability of the background noise comprises:

for a sequence of contiguous noise frames, computing an average frame energy using frame energies of the sequence of contiguous noise frames;

for each frame in the sequence of contiguous noise frames, subtracting the average frame energy from the frame energy to generate a frame energy difference;

subtracting the frame energy difference from corresponding channel noise energies to generate compensated channel noise energies; and

measuring the noise variability of the background noise using the compensated channel noise energies.

11. A device comprising:

a microphone that receives an audio signal;

a processor that is electrically coupled to the microphone, wherein the processor;

measures noise variability of background noise present in the audio signal;

determines whether the measure of noise variability is higher than a high threshold value or lower than a low threshold value;

if the noise variability is determined to be higher than the high threshold value, categorizes the background noise as having a high degree of variability; and

if the noise variability is determined to be lower than the low threshold value, categorizes the background noise as having a low degree of variability;

a memory that is electronically coupled to the processor, wherein the memory stores the high threshold value, the low threshold value, a high boundary point, and a low boundary point;

wherein the high threshold value and the low threshold value are between the high boundary point and the low boundary point;

wherein the high boundary point is based on an analysis of a first data set including noises that exhibit a high degree of variability; and

wherein the low boundary point is based on an analysis of a second data set including noises that exhibit a low degree of variability.

12. The device of claim 11, wherein the processor further:

suppresses the background noise using a first noise suppression algorithm, if the background noise is categorized as having a high degree of variability; and

suppresses the background noise using a second noise suppression algorithm, if the background noise is categorized as having a low degree of variability.

13. The device of claim 11, wherein if the noise variability is determined to be between the low threshold value and the high threshold value, the processor further categorizes the background noise as having a degree of variability of a previous frame.

14. The device of claim 11, wherein the processor further:

determines whether a frame including the background noise is a noise update frame;

if the frame is determined not to be a noise update frame, categorizes the background noise as having a degree of variability of a previous frame; and

if the frame is a noise update frame, determines whether the frame is part of a sequence of contiguous noise frames.

15. The device of claim 14, wherein if the frame is determined not to be part of a sequence of contiguous noise frames, the processor further categorizes the background noise as having a degree of variability of a previous frame.

16. The device of claim 14, wherein if the frame is part of a sequence of contiguous noise frames, the processor further:

determines a maximum value of smoothed channel noise and a minimum value of smoothed channel noise in the sequence of contiguous noise frames;

computes a smoothed maximum dB difference using the maximum value of smoothed channel noise and the minimum value of smoothed channel noise; and

calculates the noise variability of the background noise using a ratio of a difference between the smoothed maximum dB difference and the low boundary point to a difference between the high boundary point and the low boundary point.

17. The device of claim 11, wherein the processor measures the noise variability of the background noise using the following equation:

18. The device of claim 17, wherein the processor measures the noise variability of the background noise using the following equation:

19. The device of claim 11, wherein when the processor measures the noise variability of background noise present in the audio signal, the processor further:

measures noise level of the background noise;

determines whether the measured noise level of the background noise is lower than a noise level threshold value;

if the noise level of the background noise is determined to be lower than the noise level threshold value, calculates a bias energy value;

adds the bias energy value to smoothed channel noise to generate modified smoothed channel noise; and

measures the noise variability of the background noise using the modified smoothed channel noise.

20. The device of claim 11, wherein when the processor measures the noise variability of background noise in the audio signal, the processor further:

for a sequence of contiguous noise frames, computes an average frame energy using frame energies of the sequence of contiguous noise frame;

for each frame in the sequence of contiguous noise frames, subtracts the average frame energy from the frame energy to generate a frame energy difference;

subtracts the frame energy difference from corresponding channel noise energies of each frame to generate compensated channel noise energies; and

measures the noise variability of the background noise using the compensated channel noise energies.

21. A non-transitory computer readable storage medium having stored thereon a program executable by a computing processor to perform a method, the method comprising:

measuring noise variability of background noise present of a sampled audio signal;

determining whether the measure of noise variability is higher than a high threshold value or lower than a low threshold value;

if the noise variability is determined to be higher than the high threshold value, categorizing the background noise as having a high degree of variability;

if the noise variability is determined to be lower than the low threshold value, categorizing the background noise as having a low degree of variability;

wherein the high threshold value and low threshold value are between a high boundary point and a low boundary point;

wherein the high boundary point is based on an analysis of a first data set including noises that exhibit a high degree of variability; and

wherein the low boundary point is based on analysis of a second data set including noises that exhibit a low degree of variability.