AUDIO SIGNAL QUALITY ENHANCEMENT BASED ON QUANTITATIVE SNR ANALYSIS AND ADAPTIVE WIENER FILTERING

• US 20180277135A1
• Filed: 03/24/2017
• Published: 09/27/2018
• Est. Priority Date: 03/24/2017
• Status: Active Grant

• Current Assignee: Hyundai Motor Company (Hyundai Motor Group), Kia Motors Corporation (Kia Corp.)
• Original Assignee: Hyundai Motor Company (Hyundai Motor Group), Kia Motors Corporation (Kia Corp.)
• Inventors: Ali, Mahdi, Zhou, Xuan, Chabaan, Rakan, Jaber, Nabih, Lu, Meihui, Hua, Kun
• Granted Patent: US 10,224,053 B2
• Time in Patent Office: Days
• CPC Class Codes: G01R 29/26 Measuring noise figure; Measuring signal-to-noise ratio
• CPC Class Codes: G10L 21/0208 Noise filtering
• CPC Class Codes: G10L 21/0216 characterised by the method used for estimating noise
• CPC Class Codes: G10L 21/0232 Processing in the frequency domain
• CPC Class Codes: G10L 21/0364 for improving intelligibility
• CPC Class Codes: G10L 25/03 characterised by the type of extracted parameters
• CPC Class Codes: G10L 25/06 the extracted parameters being correlation coefficients
• CPC Class Codes: G10L 25/51 for comparison or discrimination
• CPC Class Codes: G10L 25/84 for discriminating voice from noise
• CPC Class Codes: H04R 3/04 for correcting frequency response

Abstract

An audio signal enhancement method includes: acquiring an audio signal; estimating a signal-to-noise ratio (SNR) of an audio frame of the audio signal; determining a SNR threshold for the audio frame; selecting an audio signal processing technique according to a comparison of the SNR threshold to the estimated SNR of the audio frame; filtering the audio frame using a Wiener filter applying the selected signal processing technique; and outputting the audio frame filtered using the Wiener filter applying the selected signal processing technique. A first-in, first-out (FIFO) signal processing technique is selected when the estimated SNR of the audio frame is less than the SNR threshold, and a log-energy voice activity detection (VAD) signal processing technique is selected when the estimated SNR of the audio frame is greater than the SNR threshold.

Claims

1. An audio signal enhancement method comprising:

acquiring an audio signal;

estimating a signal-to-noise ratio (SNR) of an audio frame of the audio signal;

determining a SNR threshold for the audio frame;

selecting an audio signal processing technique according to a comparison of the SNR threshold to the estimated SNR of the audio frame, wherein a first-in, first-out (FIFO) signal processing technique is selected when the estimated SNR of the audio frame is less than the SNR threshold, and a log-energy voice activity detection (VAD) signal processing technique is selected when the estimated SNR of the audio frame is greater than the SNR threshold;

filtering the audio frame using a Wiener filter applying the selected signal processing technique; and

outputting the audio frame filtered using the Wiener filter applying the selected signal processing technique.

2. The audio signal enhancement method of claim 1, wherein correlation coefficients of the FIFO signal processing technique and the log-energy VAD signal processing technique measure a correlation between a clean signal and respective output signals of the FIFO signal processing technique and the log-energy VAD signal processing technique.

3. The audio signal enhancement method of claim 2, further comprising:

calculating the correlation coefficients of the FIFO signal processing technique and the log-energy VAD signal processing technique, respectively.

4. The audio signal enhancement method of claim 2, wherein the SNR threshold is a SNR value at which the correlation coefficients of the FIFO signal processing technique and the log-energy VAD signal processing technique, respectively, are the same.

5. The audio signal enhancement method of claim 1, wherein the determining of the SNR threshold comprises:

estimating a noise level in an environment in which the audio signal was acquired; and

determining the SNR threshold based on the estimated noise level.

6. The audio signal enhancement method of claim 5, wherein estimating of the noise level comprises:

determining one or more environmental conditions present when the audio signal is acquired; and

estimating the noise level based on the one or more environmental conditions.

7. The audio signal enhancement method of claim 6, wherein the one or more environmental conditions include one or more of a vehicle speed, a fan speed, a weather condition, whether a vehicle window is open, revolutions per minute (RPM) of an engine, and a volume of media being played.

8. The audio signal enhancement method of claim 5, wherein estimating of the noise level comprises:

estimating the noise level using the FIFO signal processing technique.

9. The audio signal enhancement method of claim 5, further comprising:

referencing a look-up table to determine the SNR threshold based on the estimated noise level.

10. The audio signal enhancement method of claim 9, further comprising:

measuring SNR thresholds across a plurality of noise conditions; and

generating the look-up table using the measured SNR thresholds across the plurality of noise levels.

11. The audio signal enhancement method of claim 1, wherein the SNR threshold varies according to noise conditions in an environment in which the audio signal was acquired.

12. The audio signal enhancement method of claim 1, wherein the estimating of the SNR comprises:

estimating the SNR of the audio frame using the FIFO signal processing technique.

13. The audio signal enhancement method of claim 1, further comprising:

dividing the acquired audio signal into a plurality of audio frames,

wherein the audio frame is one of the plurality of audio frames.

14. The audio signal enhancement method of claim 13, wherein the steps of estimating the SNR, determining the SNR threshold, selecting the audio signal processing technique, filtering the audio frame using the Wiener filter applying the selected signal processing technique, and outputting the audio frame using the Wiener filter applying the selected signal processing technique are performed for each of the plurality of audio frames.

15. The audio signal enhancement method of claim 1, wherein the acquired audio signal includes a combination of noise and speech.

16. The audio signal enhancement method of claim 1, wherein the outputted filtered audio frame includes speech from which noise present in the audio frame is removed.

17. The audio signal enhancement method of claim 1, further comprising:

converting the audio frame into a frequency domain using a fast Fourier transform (FFT) before the filtering of the audio frame.

18. The audio signal enhancement method of claim 1, wherein the audio signal is acquired via Bluetooth.

19. The audio signal enhancement method of claim 1, wherein the audio signal is acquired in a vehicle.

20. An audio signal enhancement apparatus comprising:

an audio acquisition device acquiring an audio signal in a vehicle; and

a control unit equipped in the vehicle configured to;

estimate a signal-to-noise ratio (SNR) of an audio frame of the audio signal;

determine a SNR threshold for the audio frame;

select an audio signal processing technique according to a comparison of the SNR threshold to the estimated SNR of the audio frame, wherein a first-in, first-out (FIFO) signal processing technique is selected when the estimated SNR of the audio frame is less than the SNR threshold, and a log-energy voice activity detection (VAD) signal processing technique is selected when the estimated SNR of the audio frame is greater than the SNR threshold;

filter the audio frame using a Wiener filter applying the selected signal processing technique; and

output the audio frame filtered using the Wiener filter applying the selected signal processing technique.

21. A non-transitory computer readable medium containing program instructions for performing an audio signal enhancement method, the computer readable medium comprising:

program instructions that estimate a signal-to-noise ratio (SNR) of an audio frame of an acquired audio signal;

program instructions that determine a SNR threshold for the audio frame;

program instructions that select an audio signal processing technique according to a comparison of the SNR threshold to the estimated SNR of the audio frame, wherein a first-in, first-out (FIFO) signal processing technique is selected when the estimated SNR of the audio frame is less than the SNR threshold, and a log-energy voice activity detection (VAD) signal processing technique is selected when the estimated SNR of the audio frame is greater than the SNR threshold;

program instructions that filter the audio frame using a Wiener filter applying the selected signal processing technique; and

program instructions that output the audio frame filtered using the Wiener filter applying the selected signal processing technique.