Test method for simulating manual test of voice function

Abstract

The invention relates to the technical field of vehicle-mounted system testing, in particular to a testing method for simulating a manual voice testing function, which comprises the following steps of: by simulating a real driving condition, placing a simulation cockpit carrying a vehicle-mounted system in a simulated environment; the defects that a traditional manual test method is time-consuming and labor-consuming, and multiple test conditions are difficult to copy are avoided; an artificial mouth is used as a reference source to simulate real voice output, so that the consistency and repeatability of the test are ensured, and the accuracy of the test is improved; by means of simulating real driving conditions, simulating voice output by using an artificial mouth, installing a measuring microphone and a hand-free microphone, performing automatic analysis and anomaly detection, generating a detailed test report and the like, the problems that in the prior art, the test of a vehicle-mounted voice interaction system mainly depends on manpower, time and labor are consumed, and the test efficiency is high are solved. And a plurality of voice scenes and environment noise conditions are difficult to fully cover.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle-mounted system testing, and in particular to a testing method for simulating manual testing of voice functions. Background Art

[0002] The in-vehicle system refers to a series of electronic devices and software installed inside the car to provide vehicle information, entertainment functions and driving assistance functions. In the in-vehicle system, human-computer interaction is a module used to realize information exchange and control between the driver and the vehicle. With the development of intelligence, human-computer interaction can use voice interaction modules, allowing drivers to interact with the vehicle through voice commands. The driver's voice commands are converted into understandable commands through voice recognition technology, and then the system's feedback information is conveyed to the driver in the form of voice through speech synthesis technology. The voice interaction module can improve driving safety because the driver does not need to distract himself to touch the screen or buttons.

[0003] In the existing technology, the testing of in-vehicle voice interaction systems mainly relies on manual work, which is not only time-consuming and labor-intensive, but also difficult to fully cover a variety of voice scenarios and environmental noise conditions. Summary of the invention

[0004] The purpose of the present invention is to provide a testing method for simulating manual testing of voice functions, aiming to solve the technical problem in the prior art that the testing of the in-vehicle voice interaction system mainly relies on manual testing, which is not only time-consuming and labor-intensive, but also difficult to fully cover a variety of voice scenarios and environmental noise conditions.

[0005] To achieve the above object, the present invention adopts a test method for simulating manual testing of voice function, comprising the following steps:

[0006] Step 1: Simulate an environment of real driving conditions and place a simulated cockpit equipped with a vehicle system in the simulated environment;

[0007] Step 2: Install the measurement microphone and the hands-free microphone in the simulated cockpit, and connect the vehicle system to the communication device;

[0008] Step 3: The controller controls the artificial mouth to work. The artificial mouth is used as a reference source to simulate real speech output to the vehicle system. The vehicle system receives and processes the sound samples output by the artificial mouth to perform a speech interaction test.

[0009] Step 4: During the test, use professional recording and data analysis equipment to record the conversations recorded by the measurement microphone and the hands-free microphone, analyze the conversations, filter out abnormal interactive conversations, conduct in-depth analysis of abnormal data, and find out the defects of the vehicle system;

[0010] Step 5: Generate a test report based on the test results.

[0011] The specific method of simulating the environment of real driving conditions is as follows:

[0012] Using advanced virtual reality technology, a highly simulated driving environment is built in the laboratory;

[0013] Based on the collected noise data of real driving environment, the audio processing technology is used to generate noise similar to the real environment and play it in the simulated driving environment;

[0014] The noises played include engine sound, tire friction sound, wind sound, and the sound of people walking on the street.

[0015] When installing the measuring microphone and the hands-free microphone, the image processing or sensor technology is used to determine the precise position of the measuring equipment in the simulated cockpit;

[0016] At the same time, after installation, the measurement microphone and hands-free microphone need to be calibrated to ensure that they can accurately capture sound signals.

[0017] The specific method of using an artificial mouth as a reference source to simulate real speech output to the vehicle system is as follows:

[0018] First prepare a series of sound samples;

[0019] The sound sample is input into the sound signal parsing module, which is responsible for decomposing the sound signal into basic acoustic features;

[0020] Based on the analyzed acoustic features, control instructions are generated through the control algorithm to drive the mechanical structure of the artificial mouth to move;

[0021] The artificial mouth outputs sound signals according to the control instructions;

[0022] Among them, the sound samples are pre-recorded voice files or generated through text-to-speech synthesis technology, and the sound samples cover different volumes, speaking speeds, intonations, and dialects or accents.

[0023] Among them, the analysis steps of the sound signal analysis module are as follows:

[0024] Extract fundamental frequency information from sound signals using autocorrelation or Fourier transform methods;

[0025] Extract the amplitude envelope of the sound through low-pass filtering or envelope detection algorithm;

[0026] Analyze the formant characteristics of the sound using linear predictive coding or formant tracking algorithms;

[0027] The artificial mouth motion control algorithm is as follows:

[0028] Based on the physiological mechanism of human pronunciation, establish the movement model of artificial mouth;

[0029] Based on the analyzed acoustic features, the inverse kinematics algorithm or machine learning model is used to generate instructions for controlling the movement of the artificial mouth.

[0030] During the test, professional recording and data analysis equipment was used to record the conversations recorded by the measurement microphone and the hands-free microphone. The analysis of these conversations screened out the specific ways of abnormal interactive conversations:

[0031] Use professional recording equipment to capture high-quality audio signals to ensure the clarity and accuracy of the recording. Then record the conversation content and save it as an audio file. Then use data analysis equipment to pre-process the file data and extract key features. According to the preset anomaly definition and detection algorithm, perform anomaly detection on the extracted features to screen out abnormal interactive conversations.

[0032] Among them, the specific methods for conducting in-depth analysis of abnormal data and finding defects in the vehicle system are:

[0033] First, the abnormal data is cleaned and denoised;

[0034] Then, key features are extracted from the abnormal data after cleaning and denoising;

[0035] Then, based on the extracted features, machine learning algorithms or deep learning algorithms are used to locate and analyze defects;

[0036] Among them, the extracted key features include the spectral characteristics of the sound signal and the temporal characteristics of the system response.

[0037] Among them, the machine learning algorithm adopts one of decision tree, random forest, support vector machine, K-nearest neighbor algorithm or naive Bayes;

[0038] The deep learning algorithm uses one of the convolutional neural network, recurrent neural network, generative adversarial network, autoencoder and residual network.

[0039] Among them, the specific method of generating a test report according to the test results is:

[0040] Based on the test results and analysis report, the text and chart content of the test report are automatically generated, and the test report is classified and processed, and corresponding improvement suggestions or action plans are generated and fed back to relevant departments or teams so that they can optimize and improve products based on the test results.

[0041] After the test report is generated, the generated test report is reviewed to ensure the accuracy and completeness of the report content;

[0042] The report will then be revised and optimized as necessary based on the audit results.

[0043] A test method for simulating artificial testing of voice function of the present invention, when used specifically, first simulates an environment of real driving conditions, and places a simulated cockpit equipped with a vehicle-mounted system in the simulated environment; installs a measuring microphone and a hands-free microphone in the simulated cockpit, and connects the vehicle-mounted system with a communication device; controls the operation of an artificial mouth through a controller, and the artificial mouth is used as a reference source to simulate real voice output to the vehicle-mounted system, and the vehicle-mounted system receives and processes the sound samples output by the artificial mouth to perform a voice interaction test; during the test, uses professional recording and data analysis equipment to record the conversation contents recorded by the measuring microphone and the hands-free microphone, and analyzes the conversation contents to screen out abnormal interactive conversations, conducts in-depth analysis on the abnormal data, and finds out the defects of the vehicle-mounted system; generates a test report according to the test results;

[0044] The controller of the present invention can make the task running time interval controllable, the total task running time controllable, and the lower limit of the task running interval time is determined according to the product, and the upper limit is not fixed; the task running mode can switch to different gears for testing; the running time is not fixed, and the stability of all modules of the system can be tested. In this way, the technical problem in the prior art that the testing of the in-vehicle voice interaction system mainly relies on manual work, which is not only time-consuming and labor-intensive, but also difficult to fully cover a variety of voice scenarios and environmental noise conditions is solved.

[0045] The beneficial effects of the present invention are as follows:

[0046] 1. Significantly improve test efficiency and accuracy:

[0047] The present invention simulates real driving conditions and places a simulated cockpit equipped with a vehicle system in a simulated environment, thereby avoiding the disadvantages of traditional manual testing methods that are time-consuming and labor-intensive and difficult to replicate multiple test conditions.

[0048] Using an artificial mouth as a reference source to simulate real speech output ensures the consistency and repeatability of the test and improves the accuracy of the test.

[0049] 2. Comprehensive coverage of various speech scenarios and environmental noise conditions:

[0050] Installing a measurement microphone and a hands-free microphone in the simulated cockpit can simultaneously capture the voice interaction of the vehicle system in different positions (such as the driver's side and the passenger side), as well as its performance under different environmental noise conditions.

[0051] By adjusting the noise level in the simulated environment, the voice interaction performance of the vehicle system in various actual driving scenarios can be fully tested to ensure the comprehensiveness and practicality of the test.

[0052] 3. Automated analysis and anomaly detection:

[0053] Professional recording and data analysis equipment is used to record conversations, and algorithms are used to analyze the conversations. This can automatically screen out abnormal interactive conversations, conduct in-depth analysis of abnormal data, and identify defects in the vehicle system.

[0054] It not only improves the efficiency of the test, but also reduces the possibility of human error, making the test results more reliable.

[0055] 4. Generate detailed test report:

[0056] A test report is automatically generated based on the test results. The report content covers key information such as test conditions, test results, abnormal data and analysis conclusions.

[0057] The test report provides developers with valuable reference information, helping them quickly locate and fix defects in the vehicle system and improve the overall performance of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0059] Figure 1 It is a flow chart of the steps of the test method of simulating artificial testing of voice function of the present invention. DETAILED DESCRIPTION

[0060] Embodiments of the present invention are described in detail below. Examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, but should not be construed as limiting the present invention.

[0061] See also Figure 1 , Figure 1 It is a flow chart of the steps of the test method of simulating artificial testing of voice function of the present invention.

[0062] The present invention provides a testing method for simulating manual testing of voice functions, comprising the following steps:

[0063] S1. Simulate an environment of real driving conditions and place a simulated cockpit equipped with a vehicle system in the simulated environment;

[0064] For this specific implementation, the specific method of simulating the environment of real driving conditions is:

[0065] Using advanced virtual reality technology, a highly simulated driving environment is built in the laboratory;

[0066] Based on the collected noise data of real driving environment, the audio processing technology is used to generate noise similar to the real environment and play it in the simulated driving environment;

[0067] The noises played include engine sound, tire friction sound, wind sound, and the sound of people walking on the street.

[0068] S2. Install a measurement microphone and a hands-free microphone in the simulated cockpit, and connect the vehicle system to the communication device;

[0069] According to this specific embodiment, when installing the measuring microphone and the hands-free microphone, the precise position of the measuring device in the simulated cockpit is determined by using image processing or sensor technology;

[0070] At the same time, after installation, the measurement microphone and hands-free microphone need to be calibrated to ensure that they can accurately capture sound signals.

[0071] S3, the controller controls the artificial mouth to work, and the artificial mouth is used as a reference source to simulate the real voice output to the vehicle system, and the vehicle system receives and processes the sound samples output by the artificial mouth to perform voice interaction test;

[0072] For this specific implementation, the specific method of using an artificial mouth as a reference source to simulate real speech output to the vehicle-mounted system is:

[0073] First prepare a series of sound samples;

[0074] The sound sample is input into the sound signal parsing module, which is responsible for decomposing the sound signal into basic acoustic features;

[0075] Based on the analyzed acoustic features, control instructions are generated through the control algorithm to drive the mechanical structure of the artificial mouth to move;

[0076] The artificial mouth outputs sound signals according to the control instructions;

[0077] Among them, the sound samples are pre-recorded voice files or generated through text-to-speech synthesis technology, and the sound samples cover different volumes, speaking speeds, intonations, and dialects or accents.

[0078] The analysis steps of the sound signal analysis module are as follows:

[0079] Extract fundamental frequency information from sound signals using autocorrelation or Fourier transform methods;

[0080] Extract the amplitude envelope of the sound through low-pass filtering or envelope detection algorithm;

[0081] Analyze the formant characteristics of the sound using linear predictive coding or formant tracking algorithms;

[0082] The artificial mouth motion control algorithm is as follows:

[0083] Based on the physiological mechanism of human pronunciation, establish the movement model of artificial mouth;

[0084] Based on the analyzed acoustic features, the inverse kinematics algorithm or machine learning model is used to generate instructions for controlling the movement of the artificial mouth.

[0085] S4. During the test, professional recording and data analysis equipment is used to record the conversations recorded by the measurement microphone and the hands-free microphone. The abnormal interactive conversations are screened out through analysis of the conversations, and the abnormal data is deeply analyzed to find out the defects of the vehicle system.

[0086] For this specific implementation method, during the test, professional recording and data analysis equipment is used to record the conversation content recorded by the measurement microphone and the hands-free microphone, and the specific mode of abnormal interactive dialogue is screened out by analyzing the content of these conversations:

[0087] Use professional recording equipment to capture high-quality audio signals to ensure the clarity and accuracy of the recording. Then record the conversation content and save it as an audio file. Then use data analysis equipment to pre-process the file data and extract key features. According to the preset anomaly definition and detection algorithm, perform anomaly detection on the extracted features to screen out abnormal interactive conversations.

[0088] The specific methods for conducting in-depth analysis of abnormal data and finding defects in the vehicle system are as follows:

[0089] First, the abnormal data is cleaned and denoised;

[0090] Then, key features are extracted from the abnormal data after cleaning and denoising;

[0091] Then, based on the extracted features, machine learning algorithms or deep learning algorithms are used to locate and analyze defects.

[0092] Among them, the extracted key features include the spectral characteristics of the sound signal and the temporal characteristics of the system response.

[0093] The machine learning algorithm uses one of the decision trees, random forests, support vector machines, K-nearest neighbor algorithms, or naive Bayes;

[0094] The deep learning algorithm uses one of the convolutional neural network, recurrent neural network, generative adversarial network, autoencoder and residual network.

[0095] S5. Generate a test report based on the test results.

[0096] For this specific implementation, the specific method of generating a test report based on the test results is:

[0097] According to the test results and analysis reports, the text and chart content of the test report are automatically generated, and the test report is classified and processed, and the corresponding improvement suggestions or action plans are generated and fed back to the relevant departments or teams so that they can optimize and improve the product according to the test results. After the test report is generated, the generated test report is reviewed to ensure the accuracy and completeness of the report content;

[0098] The report will then be revised and optimized as necessary based on the audit results.

[0099] This ensures that the various functions of the system are functioning properly, such as answering / ending calls, switching calls, adjusting volume, etc. Test whether the emergency call function can be activated correctly when needed;

[0100] A test method for simulating artificial testing of voice functions of the present invention is used. When used specifically, firstly, an environment of real driving conditions is simulated, and a simulated cockpit equipped with a vehicle-mounted system is placed in the simulated environment; a measuring microphone and a hands-free microphone are installed in the simulated cockpit, and the vehicle-mounted system is connected to a communication device; an artificial mouth is controlled by a controller, and the artificial mouth is used as a reference source to simulate real voice output to the vehicle-mounted system, and the vehicle-mounted system receives and processes the sound samples output by the artificial mouth to perform a voice interaction test; during the test, professional recording and data analysis equipment is used to record the conversation contents recorded by the measuring microphone and the hands-free microphone, and the conversation contents are analyzed to screen out abnormal interactive conversations, and the abnormal data is deeply analyzed to find out the defects of the vehicle-mounted system; a test report is generated according to the test results;

[0101] The controller of the present invention can make the task running time interval controllable, the total task running time controllable, and the lower limit of the task running interval time is determined according to the product, and the upper limit is not fixed; the task running mode can switch to different gears for testing; the running time is not fixed, and the stability of all modules of the system can be tested. In this way, the technical problem in the prior art that the testing of the in-vehicle voice interaction system mainly relies on manual work, which is not only time-consuming and labor-intensive, but also difficult to fully cover a variety of voice scenarios and environmental noise conditions is solved.

[0102] The beneficial effects of the present invention are as follows:

[0103] 1. Significantly improve test efficiency and accuracy:

[0104] The present invention simulates real driving conditions and places a simulated cockpit equipped with a vehicle system in a simulated environment, thereby avoiding the disadvantages of traditional manual testing methods that are time-consuming and labor-intensive and difficult to replicate multiple test conditions.

[0105] Using an artificial mouth as a reference source to simulate real speech output ensures the consistency and repeatability of the test and improves the accuracy of the test.

[0106] 2. Comprehensive coverage of various speech scenarios and environmental noise conditions:

[0107] Installing a measurement microphone and a hands-free microphone in the simulated cockpit can simultaneously capture the voice interaction of the vehicle system in different positions (such as the driver's side and the passenger side), as well as its performance under different environmental noise conditions.

[0108] By adjusting the noise level in the simulated environment, the voice interaction performance of the vehicle system in various actual driving scenarios can be fully tested to ensure the comprehensiveness and practicality of the test.

[0109] 3. Automated analysis and anomaly detection:

[0110] Professional recording and data analysis equipment is used to record conversations, and algorithms are used to analyze the conversations. This can automatically screen out abnormal interactive conversations, conduct in-depth analysis of abnormal data, and identify defects in the vehicle system.

[0111] It not only improves the efficiency of the test, but also reduces the possibility of human error, making the test results more reliable.

[0112] 4. Generate detailed test report:

[0113] A test report is automatically generated based on the test results. The report content covers key information such as test conditions, test results, abnormal data and analysis conclusions.

[0114] The test report provides developers with valuable reference information, helping them quickly locate and fix defects in the vehicle system and improve the overall performance of the system.

[0115] In summary, the present invention effectively solves the problems existing in the testing process of the in-vehicle voice interaction system by simulating real driving conditions, using an artificial mouth to simulate voice output, installing a measurement microphone and a hands-free microphone, automating analysis and anomaly detection, and generating detailed test reports, thereby improving the efficiency and accuracy of the test.

[0116] At the same time, the present invention not only significantly improves the efficiency and accuracy of the vehicle-mounted voice interaction system test by introducing a variety of advanced algorithms, but also simplifies the test process and improves the repeatability and reliability of the test through the automatic configuration of the controller.

[0117] What is disclosed above is only a preferred embodiment of the present invention, and it certainly cannot be used to limit the scope of rights of the present invention. Ordinary technicians in this field can understand that all or part of the processes of the above embodiment and equivalent changes made according to the claims of the present invention still fall within the scope of the invention.

Claims

1. A test method for simulating manual testing of voice functions, characterized in that: The steps include: Step 1: Simulate an environment of real driving conditions and place a simulated cockpit equipped with a vehicle system in the simulated environment; Step 2: Install the measurement microphone and the hands-free microphone in the simulated cockpit, and connect the vehicle system to the communication device; Step 3: The controller controls the artificial mouth to work. The artificial mouth is used as a reference source to simulate real speech output to the vehicle system. The vehicle system receives and processes the sound samples output by the artificial mouth to perform a speech interaction test. Step 4: During the test, use professional recording and data analysis equipment to record the conversations recorded by the measurement microphone and the hands-free microphone, analyze the conversations, filter out abnormal interactive conversations, conduct in-depth analysis of abnormal data, and find out the defects of the vehicle system; Step 5: Generate a test report based on the test results.

2. The method for simulating artificial testing of voice function according to claim 1, characterized in that: The specific methods of simulating the environment of real driving conditions are: Using advanced virtual reality technology, a highly simulated driving environment is built in the laboratory; Based on the collected noise data of real driving environment, the audio processing technology is used to generate noise similar to the real environment and play it in the simulated driving environment; The noises played include engine sound, tire friction sound, wind sound, and the sound of people walking on the street.

3. The test method for simulating artificial testing of voice function as claimed in claim 2, characterized in that: When installing the measurement microphone and the hands-free microphone, use image processing or sensor technology to determine the exact position of the measurement equipment in the simulated cockpit; At the same time, after installation, the measurement microphone and hands-free microphone need to be calibrated to ensure that they can accurately capture sound signals.

4. The method for simulating artificial testing of voice function as claimed in claim 3, characterized in that: The specific method of using an artificial mouth as a reference source to simulate real speech output to the vehicle system is as follows: First prepare a series of sound samples; The sound sample is input into the sound signal parsing module, which is responsible for decomposing the sound signal into basic acoustic features; Based on the analyzed acoustic features, control instructions are generated through the control algorithm to drive the mechanical structure of the artificial mouth to move; The artificial mouth outputs sound signals according to the control instructions; Among them, the sound samples are pre-recorded voice files or generated through text-to-speech synthesis technology, and the sound samples cover different volumes, speaking speeds, intonations, and dialects or accents.

5. The method for simulating artificial testing of voice function as claimed in claim 4, characterized in that: The analysis steps of the sound signal analysis module are as follows: Extract fundamental frequency information from sound signals using autocorrelation or Fourier transform methods; Extract the amplitude envelope of the sound through low-pass filtering or envelope detection algorithm; Analyze the formant characteristics of the sound using linear predictive coding or formant tracking algorithms; The artificial mouth motion control algorithm is as follows: Based on the physiological mechanism of human pronunciation, establish the movement model of artificial mouth; Based on the analyzed acoustic features, the inverse kinematics algorithm or machine learning model is used to generate instructions for controlling the movement of the artificial mouth.

6. The test method for simulating artificial testing of voice function according to claim 5, characterized in that: During the test, professional recording and data analysis equipment was used to record the conversations recorded by the measurement microphone and the hands-free microphone. The analysis of these conversations screened out the specific ways of abnormal interactive conversations: Use professional recording equipment to capture high-quality audio signals to ensure the clarity and accuracy of the recording. Then record the conversation content and save it as an audio file. Then use data analysis equipment to pre-process the file data and extract key features. According to the preset anomaly definition and detection algorithm, perform anomaly detection on the extracted features to screen out abnormal interactive conversations.

7. The method for simulating artificial testing of voice function according to claim 6, characterized in that: The specific methods for conducting in-depth analysis of abnormal data and finding defects in the vehicle system are as follows: First, the abnormal data is cleaned and denoised; Then, key features are extracted from the abnormal data after cleaning and denoising; Then, based on the extracted features, machine learning algorithms or deep learning algorithms are used to locate and analyze defects; Among them, the extracted key features include the spectral characteristics of the sound signal and the temporal characteristics of the system response.

8. The method for simulating artificial testing of voice function according to claim 7, characterized in that: The machine learning algorithm uses one of the decision trees, random forests, support vector machines, K-nearest neighbor algorithms, or naive Bayes; The deep learning algorithm uses one of the convolutional neural network, recurrent neural network, generative adversarial network, autoencoder and residual network.

9. The method for simulating artificial testing of voice function according to claim 8, characterized in that: The specific method of generating a test report based on the test results is as follows: Based on the test results and analysis report, the text and chart content of the test report are automatically generated, and the test report is classified and processed, and corresponding improvement suggestions or action plans are generated and fed back to relevant departments or teams so that they can optimize and improve products based on the test results.

10. The test method for simulating artificial testing of voice function according to claim 9, characterized in that: After generating the test report, review the generated test report to ensure the accuracy and completeness of the report content; The report will then be revised and optimized as necessary based on the audit results.

Images