CN113225662A

CN113225662A - TWS earphone awakening test method with G-sensor

Info

Publication number: CN113225662A
Application number: CN202110593613.0A
Authority: CN
Inventors: 韦栋; 凌云; 刘述琨; 李前戎; 张东清
Original assignee: Hangzhou Nationalchip Science & Technology Co ltd
Current assignee: Hangzhou Guoxin Microelectronics Co.,Ltd.
Priority date: 2021-05-28
Filing date: 2021-05-28
Publication date: 2021-08-06
Anticipated expiration: 2041-05-28
Also published as: CN113225662B

Abstract

The invention discloses a TWS earphone awakening test method with a G-sensor. The traditional test method has large requirements on personnel, low efficiency and difficulty in quantifying test results. The method comprises the steps of firstly establishing a new test item for awakening the TWS earphone, measuring a sound pressure value at a microphone of the TWS earphone, and recording a numerical value of an acceleration sensor; and then calibrating the testing device, and testing the KWS awakening rate of the TWS earphone by using the testing device: the acceleration sensor detects the vibration frequency of the TWS earphone, the vibration frequency is compared with the vibration frequency collected by the microphone, if the vibration frequency is smaller than a threshold value, the earphone enters a KWS state, and the testing host counts the KWS awakening rate to obtain a testing result. The method simulates user vocal cord vibration, transmits vibration with the same frequency to the earphone acceleration sensor, enables the earphone to be judged as user operation, starts detection of subsequent awakening words to enable the earphone to be awakened correctly, can simulate the user to shout the awakening words, and completes test automatically.

Description

TWS earphone awakening test method with G-sensor

Technical Field

The invention belongs to the technical field of artificial intelligence, and relates to a TWS earphone awakening test method with a G-sensor (acceleration sensor).

Background

TWS headphones, True Wireless Stereo headphones, which means TWS is True Wireless Stereo, and the TWS technology is also based on the development of the bluetooth chip technology. According to the working principle, the mobile phone is connected with the main earphone, and then the main earphone is quickly connected with the auxiliary earphone in a wireless mode, so that the real wireless separation use of the left and right sound channels of the Bluetooth is realized.

G-sensor: a gradient sensor, an acceleration sensor, is a sensor capable of measuring acceleration. The damper is generally composed of a mass block, a damper, an elastic element, a sensitive element, an adjusting circuit and the like. In the acceleration process, the sensor obtains an acceleration value by measuring the inertial force borne by the mass block and utilizing Newton's second law. In recent years, the technology related to artificial intelligence voice recognition is rapidly developed, and in an artificial intelligence voice recognition device, a TWS earphone is one of the important parts.

Since the introduction of TWS earphones, the trend is rising all the way, and the increase in numbers from 918 ten thousand in 2016 to 1.2 million in 2019 highlights the importance of TWS earphones in the field of wearable devices.

In a range of speech recognition devices including TWS headsets, a significant problem is the contradiction between wake-up and false wake-up, and an increase in the wake-up capability of a speech recognition product often leads to an increase in false wake-up. An excellent speech recognition product, not only require to have a higher awakening rate and require a lower mistake awakening rate simultaneously, certainly improve the algorithm and also can well balance this contradiction, but some auxiliary means in some projects can also well solve this problem, especially add a G-sensor acceleration transducer in the TWS earphone, can well solve above-mentioned problem, the sound is produced through the vibration, the human voice is produced through the vibration of vocal cords, then add G-sensor acceleration transducer on the TWS earphone, detect whether the user is vocalized, and external noise is more difficult to influence G-sensor acceleration transducer, so can greatly reduce the interference of external noise, thus greatly reduce the probability of mistake awakening, really let the earphone only listen to your call.

Because the advantage that the G-sensor acceleration sensor is added on the TWS earphone can greatly reduce the false awakening, a plurality of manufacturers adopt the scheme, and the test work of the product in the development process comes along with the problem, but the traditional test mode has the following problems:

(1) the traditional method for shouting the awakening words and counting the awakening rate by manually carrying the TWS earphone is high in personnel demand and low in efficiency, and is difficult to quantify test results and only used for testing the experience.

(2) Before the TWS earphone with the G-sensor is awakened, the G-sensor is used for detecting vocal cord vibration and vibration frequency to judge whether a user operates or not, awakening words are played through a sound box, and then awakening rate is counted.

Disclosure of Invention

The invention aims to provide a TWS earphone wake-up test method with a G-sensor.

The method specifically comprises the following steps:

and (1) awakening a new testing item of the TWS earphone, connecting the testing personnel with the TWS earphone through mobile phone Bluetooth, wearing the TWS earphone on ears of the testing personnel, calling awakening words of the earphone by the testing personnel, simultaneously measuring a sound pressure value at a microphone of the TWS earphone through a sound pressure meter, and recording a value of the G-sensor acceleration sensor through the mobile phone.

And (2) calibrating the testing device, wherein the testing device comprises a sound box cavity, a loudspeaker is arranged on the top surface of the sound box cavity, passive radiation plates are arranged on two opposite side surfaces of the sound box cavity, an earphone fixing clamp is arranged at a position corresponding to the passive radiation plates, and a ventilation adjusting valve is arranged on the sound box cavity. The vibration generated by the loudspeaker is transmitted to the passive radiating plate through the air medium in the sound box cavity, so that the passive radiating plate generates vibration and drives the earphone fixed on the clamp to vibrate together, and finally the vibration is detected by the G-sensor acceleration sensor. The TWS headset wake performance data with the G-sensor may be tested by the device. The calibration method comprises the following steps:

placing the TWS earphone on an earphone clamp of a testing device, connecting a loudspeaker of the testing device with a testing host, playing a testing audio file recorded with the awakening words by the testing host, adjusting the output volume of the testing host, and measuring the sound pressure value at the microphone of the TWS earphone by using a sound pressure meter to enable the sound pressure value to reach the sound pressure value of the awakening words shout in the step (1);

and (3) controlling a loudspeaker of the testing device to play a wake-up word by the testing host, simultaneously checking the value of the G-sensor acceleration sensor returned by the TWS earphone on the mobile phone, and adjusting the ventilation adjusting valve on the testing device until the value of the G-sensor acceleration sensor returned by the TWS earphone displayed on the mobile phone is consistent with the value of the G-sensor acceleration sensor returned by the TWS earphone in the step (1), thereby completing the calibration of the testing device.

Step (3) testing the keyword detection (KWS) awakening rate of the TWS earphone with the G-sensor:

the loudspeaker of the testing device is connected with the testing host machine through an audio line, the testing host machine is connected with the mobile phone through a USB data line, and the mobile phone is connected with the TWS earphone through Bluetooth. The test host controls the loudspeaker to play a test audio file recorded with the voice algorithm awakening words, vibration generated by the loudspeaker is transmitted to a passive radiation plate of the test device through an air medium in the sound box cavity, the passive radiation plate drives an earphone on the clamp to vibrate together, and meanwhile, a TWS earphone microphone collects vibration frequency of the loudspeaker. The G-sensor acceleration sensor detects the vibration frequency of the TWS earphone, the vibration frequency is compared with the vibration frequency collected by a microphone of the TWS earphone, if the difference is smaller than a set threshold value, the TWS earphone enters a KWS state, and if the difference is larger than or equal to the set threshold value, no processing is performed; after the TWS earphone enters the KWS state, if a wake-up word is detected through a voice KWS algorithm, the TWS earphone wakes up successfully, if the wake-up word is not detected, the TWS earphone wakes up unsuccessfully, and information of successful or failed wake-up is transmitted back to the mobile phone through Bluetooth.

Step (4), the mobile phone sends information of successful or failed awakening to a test host, and the test host controls a loudspeaker to play a test audio file recorded with a voice algorithm awakening word for multiple times to complete a round of test; and the test host counts the KWS awakening rate, namely the ratio of the number of times of successful TWS earphone awakening to the number of times of awakening words played in the current round of test is obtained to obtain a test result.

The method of the invention has the following beneficial effects:

(1) the method can simulate the vocal cord vibration of a user when the awakening word is played, and conduct the vibration with the same frequency to the G-sensor acceleration sensor of the TWS ear to enable the earphone to judge as the user operation, and start the detection of the subsequent awakening word to enable the earphone to be awakened correctly.

(2) The method can simulate the user to shout the awakening words, automatically complete the test, and greatly improve the test efficiency and the personnel investment.

Drawings

FIG. 1 is a schematic diagram of a test apparatus in the method of the present invention.

Detailed Description

The technical solution of the present invention is further illustrated by the following examples.

A TWS earphone awakening test method with a G-sensor is disclosed, and the test method is explained in detail by combining a test device. A TWS earphone awakening test method with a G-sensor comprises the steps of obtaining and recording calibration reference parameters of a test auxiliary device, calibrating the test auxiliary device, building a test environment according to a test standard, executing a test by a test host and automatically collecting a test report.

The method comprises the following specific steps:

and (1) for a new TWS earphone with a G-sensor to wake up a test item, firstly, a tester correctly wears a tested device (the TWS earphone with the G-sensor) on an ear, the earphone is correctly connected with a test mobile phone through Bluetooth, the tester uses normal volume to call out a wake-up word of the tested device, and simultaneously uses a sound pressure meter to measure and record a sound pressure value at an earphone microphone, and secondly, records a value of the G-sensor returned from the test mobile phone, so that the calibration reference parameter of the test auxiliary device is obtained and recorded.

The tested device is a novel TWS earphone with a G-sensor, has a voice awakening function, is an auxiliary testing mobile phone specially used for the test, and can also obtain acceleration real-time information and earphone awakening data returned by the Bluetooth earphone through the app on the testing mobile phone.

Step (2) calibrating the testing device, wherein fig. 1 is a schematic diagram of the testing device, and the testing device comprises a sound box cavity 1, a loudspeaker 2 is arranged on the top surface of the sound box cavity 1, passive radiation plates 3 are arranged on two opposite side surfaces of the sound box cavity 1, an earphone fixing clamp 4 is arranged at a position corresponding to the passive radiation plates 3, and a ventilation regulating valve 5 is arranged on the sound box cavity 1. The method is used for receiving the vibration when the awakening word is played and amplifying the vibration to generate the vibration similar to that of a loudspeaker to simulate the vibration of the sound produced when a user wears a headset, and the TWS headset with the G-sensor is mainly used for identifying the sound produced by the user. The vibration that speaker 2 phonation produced is transmitted to passive radiating plate 3 through the air medium in audio amplifier cavity 1 on, makes passive radiating plate 3 produce the vibration to drive the earphone of fixing on anchor clamps and vibrate together, finally detected by G-sensor acceleration sensor. The TWS headset wake performance data with the G-sensor may be tested by the device. The ventilation regulating valve 5 is used for controlling the air inflow of the sound box cavity 1 during audio playing, so that the vibration amplitude of the passive radiating plate is controlled, and the value of the G-sensor on the TWS earphone of the G-sensor is further influenced.

The specific operation is as follows: after the calibration reference parameter of the test auxiliary device is correctly taken in the step 1, the TWS earphone with the G-sensor is correctly placed on an earphone clamp of the test auxiliary device, a loudspeaker of the test auxiliary device is connected with a test host through an audio line, the test host plays a wakeup word, and a sound pressure meter is used for measuring a sound pressure value at a microphone of the test auxiliary device earphone. And (3) if the sound pressure value is not consistent with the sound pressure value in the calibration reference parameter of the test auxiliary device obtained in the step (1), adjusting the output volume of the played awakening word on the test host to enable the sound volume to be consistent with the measured sound volume, and completing calibration of the played sound pressure. And starting to calibrate the vibration amplitude on the premise of finishing the calibration of the playing sound pressure. And (3) continuously playing the awakening words by the test host to enable the speaker to continuously play the audio of the awakening words, observing whether the value of the G-sensor returned from the test mobile phone is consistent with the calibrated G-sensor reference parameter of the auxiliary device in the step (1), if not, adjusting the ventilation adjusting valve, adjusting the air inflow of the cavity of the sound box, adjusting the increase and decrease of the vibration amplitude, enabling the vibration of the ventilation adjusting valve to influence the value output by the G-sensor on the TWS earphone with the G-sensor, enabling the value of the ventilation adjusting valve to be consistent with the calibrated G-sensor reference parameter of the auxiliary device in the step (1), and completing the vibration amplitude calibration of the test auxiliary device.

And (3) after the step (1) and the step (2) are completed, a test environment can be built according to the test standard of the test item. And testing the environmental noise playing device, the noise playing position, the noise playing angle, the noise audio file and the noise playing volume, and waking up the word playing device, the waking up word playing position, the waking up word playing angle and the waking up word playing volume. Keyword detection (KWS) wake-up rate for TWS headphones with G-sensor:

Step (4), the mobile phone sends information of successful or failed awakening to a test host, and the test host controls a loudspeaker to play a test audio file recorded with a voice algorithm awakening word for multiple times to complete a round of test; and the test host counts the KWS awakening rate, namely the ratio of the number of times of successful awakening of the TWS earphone to the number of times of playing awakening words in the current round of test, generates a test report and completes the TWS earphone awakening test with the G-sensor.

Claims

1. A TWS earphone wake-up test method with a G-sensor is characterized by comprising the following steps:

waking up a new test item of the TWS earphone, connecting the TWS earphone with a tester through mobile phone Bluetooth, wearing the TWS earphone on the ear of the tester, calling out a wake-up word of the earphone by the tester, measuring a sound pressure value at the microphone of the TWS earphone through a sound pressure meter, and recording a value of a G-sensor acceleration sensor through the mobile phone;

calibrating a testing device, wherein the testing device comprises a sound box cavity, a loudspeaker is arranged on the top surface of the sound box cavity, passive radiation plates are arranged on two opposite side surfaces of the sound box cavity, an earphone fixing clamp is arranged at a position corresponding to the passive radiation plates, and a ventilation adjusting valve is arranged on the sound box cavity; the vibration generated by the sounding of the loudspeaker is transmitted to the passive radiation plate through an air medium in the cavity of the sound box, so that the passive radiation plate generates vibration and drives an earphone fixed on the clamp to vibrate together, and the vibration is finally detected by the G-sensor acceleration sensor; the TWS earphone awakening performance data with the G-sensor can be tested by the device; the calibration method comprises the following steps:

the testing host controls a loudspeaker of the testing device to play a wake-up word, meanwhile, the value of the G-sensor acceleration sensor returned by the TWS earphone on the mobile phone is checked, a ventilation adjusting valve on the testing device is adjusted until the value of the G-sensor acceleration sensor returned by the TWS earphone displayed on the mobile phone is consistent with the value of the G-sensor acceleration sensor returned by the TWS earphone in the step (1), and calibration of the testing device is completed;

and (3) testing the keyword detection awakening rate of the TWS earphone with the G-sensor:

connecting a loudspeaker of the testing device with a testing host machine through an audio line, connecting the testing host machine with a mobile phone through a USB data line, and connecting the mobile phone with a TWS earphone through Bluetooth;

the testing host controls the loudspeaker to play a testing audio file recorded with the voice algorithm awakening words, vibration generated by the loudspeaker drives the earphone on the clamp to vibrate together through the passive radiating plate, and meanwhile, the TWS earphone microphone collects the vibration frequency of the loudspeaker; the method comprises the steps that a G-sensor acceleration sensor detects the vibration frequency of a TWS earphone, the vibration frequency is compared with the vibration frequency collected by a microphone of the TWS earphone, if the difference is smaller than a set threshold value, the TWS earphone enters a keyword detection state, and if the difference is larger than or equal to the set threshold value, no processing is performed; after the TWS earphone enters a keyword detection state, if a wake-up word is detected through a voice keyword detection algorithm, the TWS earphone wakes up successfully, if the wake-up word is not detected, the TWS earphone wakes up unsuccessfully, and information of successful or failed wake-up is transmitted back to the mobile phone through Bluetooth;

step (4), the mobile phone sends information of successful or failed awakening to a test host, and the test host controls a loudspeaker to play a test audio file recorded with a voice algorithm awakening word for multiple times to complete a round of test; and the test host counts the keyword detection awakening rate, namely the ratio of the number of times that the TWS earphone is awakened successfully to the number of times that the TWS earphone plays awakening words in the current test is obtained, so as to obtain a test result.