A voice recognition method and a preparation method of a voice recognition sensor based on the same

Abstract

The application discloses a speech recognition method and a preparation method of a speech recognition sensor based on the speech recognition method, and applies to the technical field of speech sensors and recognition, and aims at solving the problem of poor noise resistance of an existing speech recognition sensor. The sensor perceives a rapid humidity change signal in a speech airflow, utilizes the difference of humidity change signals of different word speech airflows, and realizes speech recognition on different words. Since the humidity change signal is not sensitive to environmental noise, the speech recognition sensor has good noise resistance.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of speech recognition, and particularly relates to a sensor for speech recognition by using humidity change signals of speech airflow, a preparation method of the sensor and a speech recognition method. BACKGROUND

[0002] Traditional electromechanical speech sensors such as electrets and piezoelectric ceramics always face the challenge of environmental noise interference. In order to overcome the interference of noise, some new speech sensors are being developed. In 2014, Daeshik Kang et al. proposed a nanometer crack sensor in the document Ultrasensitive mechanical crack-based sensor inspired by the spider sensory system. The sensor is attached to the neck position and performs speech recognition by directly collecting the motion information of the neck skin. Since it does not need air conduction, it avoids environmental noise from air and has good noise resistance. However, this sensor needs to be close to the skin, and has poor comfort and mobility. SUMMARY

[0003] To solve the above technical problems, the present application proposes a speech recognition sensor and a recognition method based on humidity change signals of speech airflow to improve the noise resistance of speech recognition.

[0004] One of the schemes adopted by the present application is: a speech recognition method based on a speech recognition sensor structure comprising: a silicon substrate, a silicon dioxide layer above the silicon substrate, a first electrode and a second electrode on the upper surface of the silicon dioxide layer, a poly diallyl dimethyl ammonium chloride film on the upper surface of the silicon dioxide layer between the first electrode and the second electrode, a plurality of silicon dioxide microspheres wrapped by the poly diallyl dimethyl ammonium chloride film on the poly diallyl dimethyl ammonium chloride film, and a graphene quantum dot film on the upper surface of the poly diallyl dimethyl ammonium chloride film;

[0005] The recognition method is specifically:

[0006] A1, connecting the speech recognition sensor in series with a constant resistance;

[0007] A2, placing the speech recognition sensor near the oral cavity to perceive humidity change signals of speech airflow of different words;

[0008] A3, collecting the voltage signal across the constant resistance as the humidity change signal of the speech airflow of different words;

[0009] A4, using the difference of the humidity change signals of the speech airflow of different words to perform speech recognition on different words.

[0010] The upper surface of the first electrode and the second electrode further comprises: a poly diallyl dimethyl ammonium chloride film, a plurality of silica microspheres wrapped by the poly diallyl dimethyl ammonium chloride film on the poly diallyl dimethyl ammonium chloride film, and a graphene oxide quantum dot film on the upper surface of the poly diallyl dimethyl ammonium chloride film.

[0011] The first electrode and the second electrode are comb-shaped and staggered, that is, the second electrode is arranged between adjacent first electrodes, and the first electrode is arranged between adjacent second electrodes.

[0012] The distance between the adjacent first electrode and the second electrode is less than or equal to 20 microns, so as to obtain higher response sensitivity.

[0013] The particle size of the graphene oxide quantum dots is less than or equal to 20 nanometers, so as to obtain faster response speed.

[0014] The particle size of the silica microspheres is 10-10000 nanometers, so as to obtain better support effect.

[0015] In step A2, the voice recognition sensor is placed within a range of 50 cm from the front end of the oral cavity level position.

[0016] The fixed value resistance is between 1MΩ and 1000MΩ, so as to obtain an appropriate amplitude of output voltage.

[0017] The second technical solution of the present application is a preparation method of a voice recognition sensor, comprising:

[0018] S1, a silica microsphere solution with a concentration of 0.1-10.0 mg / ml is added with a poly diallyl dimethyl ammonium chloride solution with a concentration of 0.1-10.0 mg / ml, and ultrasonic treatment is performed for 2-5 hours to obtain a uniform mixed solution, and a layer of poly diallyl dimethyl ammonium chloride film is attached to the surface of the silica microspheres;

[0019] S2, a silica layer is deposited on the surface of a silicon substrate, and then a first electrode and a second electrode are formed on the surface of the silica layer by vacuum coating combined with photolithography process to obtain an electrode substrate;

[0020] S3, the electrode substrate obtained in step S2 is immersed in the mixed solution of step S1 for 10-20 minutes, so that a layer of poly diallyl dimethyl ammonium chloride film and a plurality of silica microspheres with a surface attached with poly diallyl dimethyl ammonium chloride film are attached to the surface of the silica layer between the first electrode and the second electrode;

[0021] S4, brushing the electrode substrate obtained in step S3 in deionized water for 1-5 minutes to remove the polydimethyl diallyl ammonium chloride and silica microspheres on the surface of the silica layer which are not firmly attached between the first electrode and the second electrode;

[0022] S5, placing the electrode substrate treated in step S4 in an oven at not higher than 50 DEG C for 1-2 hours for drying treatment;

[0023] S6, immersing the electrode substrate treated in step S5 in a graphene oxide quantum dot dispersion solution with a concentration of 0.2-12.0 mg / ml for 5-15 minutes to adsorb a layer of graphene oxide quantum dot film on the surface of the polydimethyl diallyl ammonium chloride film on the surface of the silica layer and the silica microspheres by electrostatic force;

[0024] S7, rinsing the electrode substrate obtained in step S6 in deionized water for 1-30 minutes to remove the graphene oxide quantum dots which are not firmly adsorbed on the surface of the support layer;

[0025] S8, placing the electrode substrate obtained in step S7 in an oven at not higher than 50 DEG C for 1-2 hours for drying treatment.

[0026] The speech recognition sensor provided by the application has the following advantages: the graphene oxide quantum dot sensitive film is adsorbed on the polydimethyl diallyl ammonium chloride film on the surface of the silica microspheres by electrostatic force, the adhesion of the graphene oxide quantum dot sensitive film is enhanced, the high-humidity resistance of the sensitive film is improved, compared with the planar sensitive structure, the adsorption on the surface of the microspheres not only significantly increases the surface area of the sensitive film, but also generates an additional adsorption force directed to the center of the sphere due to the surface tension, the response speed of the sensor is greatly improved, the sensor can sense the rapid humidity change signal in the speech airflow, different words can be recognized by using the difference between the humidity change signals of different word speech airflows, and the speech recognition sensor has good noise resistance because the humidity change signal is not sensitive to environmental noise. BRIEF DESCRIPTION OF DRAWINGS

[0027] Figure 1 is a structural schematic diagram of a speech recognition sensor of the application;

[0028] wherein, Figure 1 (a) is a sectional view of the speech recognition sensor, Figure 1 (b) is a top view of the speech recognition sensor, Figure 1 (c) is a structural schematic diagram of the speech recognition sensor as a whole;

[0029] Figure 2 is a schematic diagram of a speech recognition method of the application;

[0030] Figure 3 is a signal feature map of humidity change caused by voice airflow provided by the embodiment of the present application.

[0031] The figure mark: 1 is a silicon substrate, 2 is a silicon dioxide layer, 3 is a first electrode, 4 is a second electrode, 5 is a poly diallyl dimethyl ammonium chloride film, 6 is a silicon dioxide microsphere, 7 is a graphene oxide quantum dot film, 8 is a voice recognition sensor, 9 is a fixed resistor, and 10 is a wire. DETAILED DESCRIPTION

[0032] In order to facilitate those skilled in the art to understand the technical content of the present application, the content of the present application is further explained below in combination with the drawings.

[0033] As Figure 1 (a) is a cross-sectional view of a voice recognition sensor, the voice recognition sensor of the present application comprises: a silicon substrate 1, a silicon dioxide layer 2 on the upper surface of the silicon substrate 1, a first electrode 3 and a second electrode 4 on the upper surface of the silicon dioxide layer 2, a poly diallyl dimethyl ammonium chloride film 5 on the surface of the silicon dioxide layer 2 between the first electrode 3 and the second electrode 4, a plurality of silicon dioxide microspheres 6 wrapped by the poly diallyl dimethyl ammonium chloride film 5 on the poly diallyl dimethyl ammonium chloride film 5, and a graphene oxide quantum dot film 7 on the surface of the poly diallyl dimethyl ammonium chloride film 5;

[0034] Figure 1 (b) is a top view of the voice recognition sensor.

[0035] As Figure 1 (c) is a schematic diagram of the overall structure of the voice recognition sensor, the electrode pattern in the electrode substrate is comb-shaped and staggered, that is, the second electrode 4 is arranged between adjacent first electrodes 3, and the first electrode 3 is arranged between adjacent second electrodes 4; the distance between the adjacent first electrode 3 and the second electrode 4 is less than or equal to 20 microns.

[0036] The particle size of the graphene oxide quantum dot is less than or equal to 20 nanometers.

[0037] The particle size of the silicon dioxide microsphere is 10-10000 nanometers.

[0038] As Figure 2 shown is a schematic diagram of the voice recognition method of the present application, the voice recognition method of the present application comprises the following steps: (1) connecting the voice recognition sensor 8 and the fixed resistor 9 in series; (2) placing the voice recognition sensor 8 near the oral cavity to perceive the humidity change signal of the airflow of different words; (3) according to the series voltage division principle, collecting the voltage signal U0 between the fixed resistor 8 as the humidity change signal of the airflow of different words; (4) using the difference of the humidity change signal of the airflow of different words to recognize the different words.

[0039] As shown in Figure 3 the signals of humidity change caused by the phonation airflow of the words "left", "five", "fly", and "zoo" are different from each other, so the present application can perform voice recognition by identifying the signal features of humidity change of the phonation airflow. Figure 3 The vertical coordinate Voltage represents voltage, and the horizontal coordinate Time represents time.

[0040] The voice recognition sensor 8 is within a distance of 50 cm from the oral cavity.

[0041] The fixed resistance 9 is between 1 MΩ and 1000 MΩ, and preferably between 20 MΩ and 50 MΩ.

[0042] The preparation method of the voice recognition sensor comprises the following steps:

[0043] S1. A solution of silica microspheres with a concentration of 0.1-10.0 mg / ml is added with a solution of polydiallyldimethylammonium chloride with a concentration of 0.1-10.0 mg / ml, and ultrasonic treatment is performed for 2-5 hours to obtain a uniform mixed solution and make the surface of the silica microspheres attached with a layer of polydiallyldimethylammonium chloride film;

[0044] S2. A layer of silica is deposited on the surface of a silicon substrate, and then a first electrode and a second electrode are formed on the surface of the silica layer by vacuum plating combined with photolithography to obtain an electrode substrate;

[0045] S3. The electrode substrate obtained in step S2 is immersed in the mixed solution described in step S1 for 10-20 minutes to make the surface of the silica layer between the first electrode and the second electrode attached with a layer of polydiallyldimethylammonium chloride film and a plurality of silica microspheres with a surface attached with a layer of polydiallyldimethylammonium chloride film;

[0046] S4. The electrode substrate obtained in step S3 is brushed in deionized water for 1-5 minutes to remove the polydiallyldimethylammonium chloride and silica microspheres that are not firmly attached to the surface of the silica layer between the first electrode and the second electrode;

[0047] S5. The electrode substrate obtained in step S4 is placed in an oven at a temperature not higher than 50°C for 1-2 hours for drying treatment;

[0048] S6. The electrode substrate treated in step S5 is immersed in a graphene oxide quantum dot dispersion liquid with a concentration of 0.2-12.0 mg / ml for 5-15 minutes to adsorb a layer of graphene oxide quantum dot film on the surface of the polydiallyldimethylammonium chloride film on the surface of the silica layer and the silica microspheres by electrostatic force.

[0049] S7, rinsing the electrode substrate obtained in step S6 in deionized water for 1-30 minutes to remove the graphene oxide quantum dots that are not firmly adsorbed to the surface of the support layer;

[0050] S8, placing the electrode substrate obtained in step S7 in an oven at a temperature not higher than 50°C for 1-2 hours for drying treatment.

[0051] In the process of preparing the sensor, the first electrode 3 and the second electrode 4 also have a polydiallydimethylammonium chloride film 5, a plurality of silica microspheres 6 wrapped by the polydiallydimethylammonium chloride film 5 on the polydiallydimethylammonium chloride film 5, and a graphene oxide quantum dot film 7 on the surface of the polydiallydimethylammonium chloride film 5 deposited on the upper surface; however, this part of structure on the surface of the electrode has less influence on the sensitivity of the sensor, and although this part of structure on the surface of the electrode is not shown in the present embodiment, it is clear to those skilled in the art according to the description of the present embodiment. Figure 1

[0052] Those skilled in the art will appreciate that the embodiments described herein are presented for the purpose of helping the reader understand the principles of the present application, and should be understood as not limiting the scope of protection of the present application to such specific recitations and embodiments. The present application can have various modifications and changes for those skilled in the art. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application.​

Claims

1. A speech recognition method, characterized in that, The voice recognition sensor structure includes: a silicon substrate, a silicon dioxide layer on top of the silicon substrate, a first electrode and a second electrode on the surface of the silicon dioxide layer, a polydiallyldimethylammonium chloride film on the surface of the silicon dioxide layer between the first electrode and the second electrode, multiple silicon dioxide microspheres wrapped by the polydiallyldimethylammonium chloride film on the polydiallyldimethylammonium chloride film, and a graphene oxide quantum dot film on the surface of the polydiallyldimethylammonium chloride film. The speech recognition method based on the speech recognition sensor includes: A1. Connect the voice recognition sensor in series with a fixed resistor; A2. Place the speech recognition sensor near the mouth to sense the humidity change signal of the airflow when speaking different words; A3. Collect the voltage signal across the fixed resistor as the humidity change signal of the speech airflow for different words; A4. Based on the differences in humidity change signals of airflow in the speech of different words, a speech recognition algorithm is used to perform speech recognition on different words.

2. The speech recognition method according to claim 1, characterized in that, The upper surfaces of the first electrode and the second electrode also include: a polydiallyl dimethyl ammonium chloride film, multiple silica microspheres wrapped by the polydiallyl dimethyl ammonium chloride film on the polydiallyl dimethyl ammonium chloride film, and a graphene oxide quantum dot film on the upper surface of the polydiallyl dimethyl ammonium chloride film.

3. The speech recognition method according to claim 1, characterized in that, The first electrode and the second electrode are arranged in a comb-like and staggered manner, that is, a second electrode is provided between adjacent first electrodes, and a first electrode is provided between adjacent second electrodes.

4. The speech recognition method according to claim 3, characterized in that, The distance between adjacent first and second electrodes is less than or equal to 20 micrometers.

5. The speech recognition method according to claim 1, characterized in that, The graphene oxide quantum dots have a particle size of less than or equal to 20 nanometers.

6. The speech recognition method according to claim 1, characterized in that, The silica microspheres have a particle size of 10–10,000 nanometers.

7. The speech recognition method according to claim 1, characterized in that, In step A2, the voice recognition sensor is placed within 50cm of the front of the oral cavity at the horizontal position.

8. The speech recognition method according to claim 1, characterized in that, The fixed resistor is between 1MΩ and 1000MΩ.

Images