Method for determining alcohol use by comparison of high-frequency signals in difference signal, and recording medium and device for implementing same

• US 9,916,845 B2
• Filed: 04/02/2014
• Issued: 03/13/2018
• Est. Priority Date: 03/28/2014
• Status: Active Grant

• Current Assignee: Foundation of Soongsil University-Industry Cooperation
• Original Assignee: Foundation of Soongsil University-Industry Cooperation
• Inventors: Bae, Myung Jin, Lee, Sang Gil, Bae, Seong Geon
• Primary Examiner(s): Serrou, Abdelali
• Application Number:
  US15/300,168
• Publication Number:
  US 20170140776A1
• Time in Patent Office: 1,441 Days
• CPC Class Codes: A61B 5/4803 Speech analysis specially adapted for diagnostic purposes
• CPC Class Codes: A61B 5/4845 Toxicology, e.g. by detection of alcohol, drug or toxic products A61B5/4848 takes precedence
• CPC Class Codes: A61B 5/7257 using Fourier transforms
• CPC Class Codes: A61B 5/7264 Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems neural networks per se G06N3/00; expert systems per se G06N5/00
• CPC Class Codes: A61B 7/00 Instruments for auscultation
• CPC Class Codes: G10L 25/15 the extracted parameters being formant information
• CPC Class Codes: G10L 25/21 the extracted parameters being power information
• CPC Class Codes: G10L 25/30 using neural networks
• CPC Class Codes: G10L 25/66 for extracting parameters related to health condition detecting or measuring for diagnostic purposes A61B5/00
• CPC Class Codes: G10L 25/93 Discriminating between voiced and unvoiced parts of speech signals G10L25/90 takes precedence
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems

Abstract

This method for determining alcohol use comprises the steps of: detecting an effective frame in an input audio signal; detecting a difference in signal within the original signal of the effective frame; performing a fast Fourier conversion on the difference signal to be transformed into a frequency domain; detecting high-frequency components within the difference signal subjected to the fast Fourier transform; and determining the state of alcohol use on the basis of a gradient difference between the high-frequency components. Accordingly, the present invention can identify the state and extent of alcohol use by a driver or an operator from a long distance and thus can prevent accidents caused by driving or operating vehicles and machines under the influence of alcohol.

Claims

1. A computer implemented method for determining whether alcohol is consumed by a person by using at least one processor for executing a specific computer software to perform steps comprising:

receiving an input voice signal from said person via a microphone sensor;

forming an effective frame by converting the input voice signal from said person;

forming a voice frame of the input voice signal from said person;

determining whether the formed voice frame corresponds to a voiced sound;

generating an original signal of the formed effective frame and a difference signal of the original signal, which is a difference between the original signal of the formed effective frame and a shift signal thereof;

converting the difference signal into a frequency domain signal by performing a fast Fourier transform algorithm on the difference signal;

detecting high-frequency components of the fast-Fourier-transformed difference signal; and

determining whether alcohol is consumed by said person based on slope differences among the high-frequency components; and

outputting a result of a determination whether alcohol is consumed by said person who is located in a remoted area.

2. The computer implemented method of claim 1, wherein the determining whether the formed voice frame corresponds to the voiced sound comprises:

extracting periodic characteristics of harmonics or features from the formed voice frame, and

determining whether the formed voice frame is from the voiced sound, an unvoiced sound, or background noise based on the extracted periodic characteristics of harmonics or features.

3. The computer implemented method of claim 2, wherein periodic characteristics of harmonics or features comprise root mean square energy (RMSE) or zero-crossing count (ZC) of a low-band voice signal energy area.

4. The computer implemented method of claim 2, wherein the determining whether the formed voice frame is from the voiced sound, an unvoiced sound, or background noise based on the extracted periodic characteristics of harmonics or features comprises use of neural network.

5. The computer implemented method of claim 1, wherein the generating the original signal of the formed effective frame and the difference signal of the original signal comprises:

generating the original signal S(n) of the formed effective frame;

generating the shift signal S(n−

1) by shifting the original signal S(n); and

generating the difference signal S(n)−

S(n−

1) between the original signal S(n) and the shift signal S(n−

1).

6. The computer implemented method of claim 5, wherein the difference signal S(n)−

S(n−

1) comprises a characteristic of highlighting high-frequency components.

7. The computer implemented method of claim 1, wherein the detecting of high-frequency components of the fast-Fourier-transformed difference signal comprises sequentially grouping the detected high-frequency components in sets of four.

8. The computer implemented method of claim 1, wherein the determining whether said person is drunk based on slope differences comprises:

generating frequency slopes of the high-frequency components;

measuring energy differences among the high-frequency components by computing slope differences among the high-frequency components;

calculating an average of the energy differences by adding up the measured energy differences; and

determining that said person is drunk when the calculated average of the energy differences is greater than a threshold and outputting a result of the determination.

9. The computer implemented method of claim 8, wherein each of the frequency slopes of the high-frequency components is a formant slope.

10. The computer implemented method of claim 9, wherein the measuring energy differences among the high-frequency components by computing slope differences among the high-frequency components comprises computing slope differences among neighboring high-frequency components.

11. A non-transitory computer-readable recording medium having the specific computer software recorded thereon for performing a method comprising:

receiving an input voice signal from said person via a microphone sensor;

forming an effective frame by converting the input voice signal from said person;

forming a voice frame of the input voice signal from said person;

determining whether the formed voice frame corresponds to a voiced sound;

generating an original signal of the formed effective frame and a difference signal of the original signal, which is a difference between the original signal of the formed effective frame and a shift signal thereof;

converting the difference signal into a frequency domain signal by performing a fast Fourier transform algorithm on the difference signal;

detecting high-frequency components of the fast-Fourier-transformed difference signal;

determining whether alcohol is consumed by said person based on slope differences among the high-frequency components; and

outputting a result of a determination whether alcohol is consumed by said person who is located in a remoted area.

12. The non-transitory computer-readable recording medium of claim 11, wherein the determining whether the formed voice frame corresponds to the voiced sound comprises:

extracting periodic characteristics of harmonics or features from the formed voice frame, and

determining whether the formed voice frame is from the voiced sound, an unvoiced sound, or background noise based on the extracted periodic characteristics of harmonics or features.

13. The non-transitory computer-readable recording medium of claim 12, wherein periodic characteristics of harmonics or features comprise root mean square energy (RMSE) or zero-crossing count (ZC) of a low-band voice signal energy area.

14. The non-transitory computer-readable recording medium of claim 12, wherein the determining whether the formed voice frame is from the voiced sound, an unvoiced sound, or background noise based on the extracted periodic characteristics of harmonics or features comprises use of neural network.

15. The non-transitory computer-readable recording medium of claim 11, wherein the generating the original signal of the formed effective frame and the difference signal of the original signal comprises:

generating the original signal S(n) of the formed effective frame;

generating the shift signal S(n−

1) by shifting the original signal S(n); and

generating the difference signal S(n)−

S(n−

1) between the original signal S(n) and the shift signal S(n−

1).

16. The non-transitory computer-readable recording medium of claim 15, wherein the difference signal S(n)−

S(n−

1) comprises a characteristic of highlighting high-frequency components.

17. The non-transitory computer-readable recording medium of claim 11, wherein the detecting of high-frequency components of the fast-Fourier-transformed difference signal comprises sequentially grouping the detected high-frequency components in sets of four.

18. The non-transitory computer-readable recording medium of claim 11, wherein the determining whether said person is drunk based on slope differences comprises:

generating frequency slopes of the high-frequency components;

measuring energy differences among the high-frequency components by computing slope differences among the high-frequency components;

calculating an average of the energy differences by adding up the measured energy differences; and

determining that said person is drunk when the calculated average of the energy differences is greater than a threshold and outputting a result of the determination.

19. The non-transitory computer-readable recording medium of claim 18, wherein each of the frequency slopes of the high-frequency components is a formant slope.

20. The non-transitory computer-readable recording medium of claim 19, wherein the measuring energy differences among the high-frequency components by computing slope differences among the high-frequency components comprises computing slope differences among neighboring high-frequency components.