Detection methods, voice interaction detection devices, equipment and non-volatile storage media

Abstract

This invention provides a detection method, a voice interaction detection device, an apparatus, and a non-volatile storage medium. The detection method includes: acquiring a recognition response detection command; controlling a voice playback module to play a first wake-up audio file; acquiring first response data generated by the voice interaction device based on the first wake-up audio file; determining whether the voice interaction device has been successfully woken up based on the first response data and the first response duration of the voice interaction device; after the voice interaction device is successfully woken up, controlling the voice playback module to play a skill command audio file, acquiring second response data generated based on the skill command audio file, and determining whether the voice interaction device's recognition response detection has been successful based on the second response data and the second response duration of the voice interaction device. The detection method of this invention solves the problems of low testing efficiency, high cost, and low reliability of detection results caused by the low degree of automation in existing voice interaction device response detection methods.

Description

Technical Field

[0001] This invention relates to the field of voice interaction, and more specifically, to a detection method, a voice interaction detection device, an apparatus, and a non-volatile storage medium. Background Technology

[0002] With the development of speech recognition, natural language processing, and speech synthesis technologies, voice interaction technology products, centered on speech recognition, processing, and synthesis, have permeated various fields such as home appliances, communications, automotive electronics, medical devices, home services, the consumer electronics market, and industry. Voice interaction products typically consist of voice interaction hardware devices and a voice interaction service system cloud platform, which connect via a network to provide users with customized voice interaction services. During the research and development and testing of voice interaction products, various responsiveness tests are conducted. The effectiveness of the voice response directly impacts the product's value. Therefore, the ability to promptly identify and resolve various voice response issues is crucial for shortening the development cycle, reducing subsequent production costs, and improving product quality and production efficiency.

[0003] Currently, the following technical issues exist regarding the response function and performance testing of voice devices involved in the research and development and testing of voice interaction products:

[0004] Existing testing methods rely on purely manual or semi-automated approaches to assess device responses. In this method, testers must constantly interact with the voice device, verbally reciting test sets of voice interaction commands (including wake-up and skill command test sets). This constant interaction with the device under test across various testing scenarios leads to an increased workload for testers as the number of test samples grows, increasing the probability of verbal errors. This results in low efficiency, high cost, and low reliability of test results through manual testing.

[0005] Existing testing methods do not automatically filter and extract response data in real time, nor do they perform error statistical analysis and judgment. Instead, they rely on manual judgment and statistical analysis of various errors. They cannot automatically determine various functional errors and performance defects in the test result output based on the actual response data, resulting in problems such as incomplete and unintuitive equipment test results and small test coverage. Summary of the Invention

[0006] The main objective of this invention is to provide a detection method, a voice interaction detection device, an equipment, and a non-volatile storage medium to solve the problems of low testing efficiency, high cost, and low reliability of detection results caused by the low degree of automation in the existing response detection methods for voice interaction devices.

[0007] To achieve the above objectives, according to a first aspect of the present invention, a method for detecting a voice interaction device is provided, comprising: acquiring a recognition response detection command; controlling a voice playback module to play a first wake-up audio file; acquiring first response data generated by the voice interaction device based on the first wake-up audio file; determining whether the voice interaction device has been successfully woken up based on the first response data and a first response duration of the voice interaction device; when the first response data contains first preset information and the first response duration is less than a first preset duration, the voice interaction device is successfully woken up; otherwise, the voice interaction device fails to wake up; after the voice interaction device is successfully woken up, controlling the voice playback module to play a skill command audio file, acquiring second response data generated by the voice interaction device based on the skill command audio file, and determining whether the voice interaction device recognition response detection is successful based on the second response data and a second response duration of the voice interaction device; wherein, when the second response data contains second preset information matching the skill command audio file and the second response duration is less than a second preset duration, the voice interaction device recognition response detection is determined to be successful; otherwise, the voice interaction device recognition response detection is determined to be unsuccessful.

[0008] Furthermore, when performing offline recognition response detection, the second response data is generated by the voice interaction device based on the skill command audio file.

[0009] Furthermore, when performing offline recognition response detection, the method for determining whether the voice interaction device recognition response detection is successful based on the second response data and the second response duration of the voice interaction device includes: when the second response data contains second preset information that is the same as the standard text corresponding to the skill command audio file and the second response duration is less than the second preset duration, the voice interaction device recognition response detection is determined to be successful; otherwise, the voice interaction device recognition response detection fails.

[0010] Furthermore, when performing online recognition response detection, the method for obtaining the second response data generated based on the skill command audio file sent by the voice interaction device includes: sending the skill command audio file to the voice interaction system cloud platform through the voice interaction device; generating the second response data based on the skill command audio file using the voice interaction system cloud platform, and sending the second response data to the voice interaction device, so that the voice interaction device sends the second response data.

[0011] Furthermore, the method of generating second response data based on the skill command audio file using the voice interaction system cloud platform and sending the second response data to the voice interaction device includes: converting the skill command audio file into text information through the ASR module of the voice interaction system cloud platform; performing semantic understanding on the text information and generating response text information through the NLU module of the voice interaction system cloud platform; converting the response text information into a response audio file through the TTS module of the voice interaction system cloud platform; and sending the text information, response text information, and response audio file to the voice interaction device through the voice interaction system cloud platform; wherein, the second response data includes text information, response text information, and response audio file.

[0012] Furthermore, when performing online recognition response detection, the method for determining whether the voice interaction device recognition response detection is successful based on the second response data and the second response duration of the voice interaction device includes: when the second response data contains semantically accurate second preset information corresponding to the skill command audio file and the second response duration is less than the second preset duration, the voice interaction device recognition response detection is successful; otherwise, the voice interaction device recognition response detection fails.

[0013] Furthermore, the detection method for voice interaction devices also includes: acquiring a voice command test set and classifying the voice command test set into domain categories; using a proximity algorithm to perform machine learning training on the domain category classification of the voice command test set and obtaining a trained domain model; when performing online recognition response detection, the method for determining whether the second response data contains semantically accurate second preset information corresponding to the skill command audio file includes: classifying the text in the second response data according to the domain model using a proximity algorithm; when the classification is correct, determining that the second response data contains semantically accurate second preset information corresponding to the skill command audio file; otherwise, the second response data does not contain semantically accurate second preset information corresponding to the skill command audio file.

[0014] Furthermore, the method for determining whether the voice interaction device recognition response detection is successful based on the second response data and the second response duration of the voice interaction device includes: determining whether the voice interaction device recognition response detection is successful based on the text in the second response data and the second response duration of the voice interaction device; when the text is semantically accurate second preset information corresponding to the skill command audio file and the second response duration is less than the second preset duration, the voice interaction device recognition response detection is successful; otherwise, the voice interaction device recognition response detection fails; wherein, the text includes first domain text and first intent text; the detection method of the voice interaction device also includes: sending the text in the second response data to the voice interaction system cloud platform for secondary semantic understanding; dividing the text in the received second response data into second domain text and second intent text through the NLU module of the voice interaction system cloud platform; obtaining the second domain text and the second intent text, and comparing and verifying the first domain text and the second domain text, and comparing and verifying the first intent text and the second intent text; when the first domain text and the second domain text are the same and the first intent text and the second intent text are the same, the voice interaction device recognition response detection is successful; otherwise, the voice interaction device recognition response detection fails.

[0015] Furthermore, the detection method for voice interaction devices also includes: repeatedly executing the steps after obtaining the recognition response detection command multiple times; wherein, the voice playback module plays the same first wake-up audio file in a loop, and the voice playback module plays different skill command audio files in sequence; recording the recognition response detection results multiple times, and calculating the recognition rate of the current round of recognition response detection based on the multiple recognition response detection results.

[0016] Furthermore, the method for recording multiple identification response detection results and calculating the recognition rate of the current round of identification response detection based on the multiple identification response detection results includes: recording whether each identification response detection is a failure or a success, as well as the intermediate results of each test stage; and outputting the identification response detection results, the intermediate results of each test stage, and the recognition rate of the current round of identification response detection in the report; wherein, the intermediate results include the error type of identification response detection failure; when the first response data is never obtained, the identification response detection is recorded as failure and the error type is recorded as wake-up failure; when the first response data is obtained after the first preset time, the identification response detection is recorded as failure and the error type is recorded as wake-up too slow; when wake-up is successful but the second preset information is not received within the second preset time, the identification response detection is recorded as failure and the error type is recorded as no response.

[0017] Furthermore, the first response duration is from the end of playback of the first wake-up audio file to the acquisition of the first response data; and / or, the second response duration is from the end of playback of the skill command audio file to the acquisition of the second response data.

[0018] Furthermore, the detection method for the voice interaction device also includes: acquiring a wake-up response detection command; controlling the voice playback module to play a second wake-up audio file; acquiring third response data generated by the voice interaction device based on the second wake-up audio file; determining whether the voice interaction device has been successfully woken up based on the third response data and the third response duration of the voice interaction device; when the third response data contains third preset information and the third response duration is less than the third preset duration, the voice interaction device is successfully woken up; otherwise, the voice interaction device fails to wake up.

[0019] Furthermore, the detection method for voice interaction devices also includes: repeatedly executing the steps after obtaining the wake-up response detection command multiple times; wherein, the voice playback module plays the same second wake-up audio file in a loop; recording the wake-up response detection results multiple times, and calculating the wake-up rate of the current round of wake-up response detection based on the multiple wake-up response detection results.

[0020] Furthermore, the method for recording multiple wake-up response detection results and calculating the wake-up rate of this round of wake-up response detection based on the multiple wake-up response detection results includes: recording whether each wake-up response detection is a failure or a success, as well as the intermediate results of each test stage; and outputting the wake-up response detection results, the intermediate results of each test stage, and the recognition rate of this round of wake-up response detection in the report; wherein, the intermediate results include the error type of wake-up response detection failure; when the third response data is not obtained at all, the wake-up response detection is recorded as failure and the error type is recorded as wake-up failure; when the third response data is obtained after a third preset time, the wake-up response detection is recorded as failure and the error type is recorded as wake-up too slow.

[0021] Furthermore, the detection method for the voice interaction device also includes: while the voice interaction device generates third response data based on the second wake-up audio file, the voice interaction device plays a second response prompt tone.

[0022] Furthermore, the duration of the third response is from the end of playback of the second wake-up audio file to the acquisition of the third response data.

[0023] According to a second aspect of the present invention, a voice interaction detection device is provided, comprising: a first acquisition unit for acquiring a recognition response detection command; a first control unit for controlling a voice playback module to play a first wake-up audio file; a second acquisition unit for acquiring first response data generated by a voice interaction device based on the first wake-up audio file; a first judgment unit for judging whether the voice interaction device has been successfully woken up based on the first response data and a first response duration of the voice interaction device; a second control unit for controlling the voice playback module to play a skill command audio file; a third acquisition unit for acquiring second response data generated by the voice interaction device based on the skill command audio file; and a second judgment unit for judging whether the voice interaction device has successfully detected a recognition response based on the second response data and a second response duration of the voice interaction device.

[0024] According to a third aspect of the present invention, a voice interaction detection device is provided, comprising a processor and a memory, the memory for storing a computer program, and the processor for calling and running the computer program from the memory, such that the voice interaction detection device performs the above-described voice interaction device detection method.

[0025] According to a fourth aspect of the present invention, a non-volatile storage medium is provided, the non-volatile storage medium including a stored program, wherein, when the program is running, the device where the non-volatile storage medium is located is controlled to execute the above-described detection method for a voice interaction device.

[0026] The detection method for the voice interaction device of the present invention involves the following steps: After the detection terminal obtains a recognition response detection command, the detection terminal first controls the voice playback module to play a first wake-up audio file. The voice interaction device receives the first wake-up audio file, processes it, and generates first response data. The detection terminal obtains the first response data generated by the voice interaction device based on the first wake-up audio file. The detection terminal then determines whether the voice interaction device has been successfully woken up based on the first response data and the first response duration of the voice interaction device. If the first response data contains first preset information and the first response duration is less than the first preset duration, the detection terminal determines that the voice interaction device has been successfully woken up; otherwise, the detection terminal... The detection method first determines if the voice interaction device wake-up fails. Then, after successful wake-up, the detection terminal controls the voice playback module to play the skill command audio file. The detection terminal acquires the second response data generated by the voice interaction device based on the skill command audio file. The detection terminal then determines whether the voice interaction device's response recognition detection is successful based on the second response data and the second response duration of the voice interaction device. Specifically, if the second response data contains second preset information matching the skill command audio file and the second response duration is less than the second preset duration, the detection terminal determines that the voice interaction device's response recognition detection is successful; otherwise, the detection terminal determines that the voice interaction device's response recognition detection has failed. During the response recognition detection process, this method enables automatic playback of the first wake-up audio file and the skill command audio file, real-time acquisition, processing, and judgment of the first and second response data, and finally, obtaining the detection result. This significantly improves the detection efficiency of the voice interaction device and enhances the accuracy and reliability of the detection results. Attached Figure Description

[0027] The accompanying drawings, which form part of this application, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings:

[0028] Figure 1 A flowchart of the recognition response detection method for a voice interaction device according to the present invention is shown;

[0029] Figure 2 A flowchart illustrating an embodiment of the detection method for a voice interaction device according to the present invention is shown;

[0030] Figure 3 A schematic diagram of an embodiment of the voice interaction detection device according to the present invention is shown.

[0031] The above figures include the following reference numerals:

[0032] 10. First acquisition unit; 20. First control unit; 30. Second acquisition unit; 40. First judgment unit; 50. Second control unit; 60. Third acquisition unit; 70. Second judgment unit. Detailed Implementation

[0033] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0034] It should be noted that the following detailed descriptions are illustrative and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0035] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0036] This invention provides a method for detecting voice interaction devices. Please refer to [the relevant documentation]. Figure 1 ,include:

[0037] Step S110: Obtain the recognition response detection instruction;

[0038] Step S120: Control the voice playback module to play the first wake-up audio file;

[0039] Step S130: Obtain the first response data generated by the voice interaction device based on the first wake-up audio file;

[0040] Step S140: Determine whether the voice interaction device has been successfully woken up based on the first response data and the first response duration of the voice interaction device; if the first response data contains the first preset information and the first response duration is less than the first preset duration, the voice interaction device is successfully woken up; otherwise, the voice interaction device fails to wake up.

[0041] Step S150: After the voice interaction device is successfully woken up, control the voice playback module to play the skill command audio file;

[0042] Step S160: Obtain the second response data generated based on the skill command audio file sent by the voice interaction device;

[0043] Step S170: Determine whether the voice interaction device recognition response detection is successful based on the second response data and the second response duration of the voice interaction device; wherein, when the second response data contains second preset information that matches the skill command audio file and the second response duration is less than the second preset duration, it is determined that the voice interaction device recognition response detection is successful; otherwise, it is determined that the voice interaction device recognition response detection fails.

[0044] The detection method for the voice interaction device of the present invention involves the following steps: After the detection terminal obtains a recognition response detection command, the detection terminal first controls the voice playback module to play a first wake-up audio file. The voice interaction device receives the first wake-up audio file, processes it, and generates first response data. The detection terminal obtains the first response data generated by the voice interaction device based on the first wake-up audio file. The detection terminal then determines whether the voice interaction device has been successfully woken up based on the first response data and the first response duration of the voice interaction device. If the first response data contains first preset information and the first response duration is less than the first preset duration, the detection terminal determines that the voice interaction device has been successfully woken up; otherwise, the detection terminal... The detection method first determines if the voice interaction device wake-up fails. Then, after successful wake-up, the detection terminal controls the voice playback module to play the skill command audio file. The detection terminal acquires the second response data generated by the voice interaction device based on the skill command audio file. The detection terminal then determines whether the voice interaction device's response recognition detection is successful based on the second response data and the second response duration of the voice interaction device. Specifically, if the second response data contains second preset information matching the skill command audio file and the second response duration is less than the second preset duration, the detection terminal determines that the voice interaction device's response recognition detection is successful; otherwise, the detection terminal determines that the voice interaction device's response recognition detection has failed. During the response recognition detection process, this method enables automatic playback of the first wake-up audio file and the skill command audio file, real-time acquisition, processing, and judgment of the first and second response data, and finally, obtaining the detection result. This significantly improves the detection efficiency of the voice interaction device and enhances the accuracy and reliability of the detection results.

[0045] As can be seen, this detection method solves the following technical problems: Existing detection methods are based on purely manual or semi-automatic methods to test the response of voice interaction devices. In this approach, the tester needs to maintain constant interaction with the voice interaction device, verbally reciting the voice interaction command test set. The tester must maintain constant interaction with the voice interaction device under test in various test scenarios. As the number of test samples increases, the workload of the tester also increases, and the probability of verbal recitation errors also increases. This results in low efficiency, high cost, and low reliability of test results through manual testing. The voice interaction command test set includes a wake-up voice command test set and a skill command voice command test set.

[0046] Specifically, the testing terminal communicates with the voice interaction device via a serial port.

[0047] Specifically, the testing terminal includes a voice playback module; or, the voice playback module is an external speaker, and the testing terminal controls the external speaker device to broadcast the test set audio and noise interference audio.

[0048] Specifically, during actual testing, testers can adjust the angle, height, and distance between the voice interaction device and the external speaker according to the actual test technical specifications and requirements. They can also control the volume of the broadcast audio and noise interference audio through the testing terminal, thereby setting up test environments such as quiet and noise-free, single-point noise, and multi-point noise.

[0049] In this embodiment, when performing offline recognition response detection, the second response data is generated by the voice interaction device based on the skill command audio file.

[0050] In this embodiment, when performing offline recognition response detection, the method for determining whether the voice interaction device recognition response detection is successful based on the second response data and the second response duration of the voice interaction device includes: when the second response data contains second preset information that is the same as the standard text corresponding to the skill command audio file and the second response duration is less than the second preset duration, the voice interaction device recognition response detection is determined to be successful; otherwise, the voice interaction device recognition response detection fails.

[0051] In specific implementation, when performing offline recognition response detection, if the second response data contains the same second preset information as the standard text corresponding to the skill command audio file and the second response duration is less than the second preset duration, the detection terminal determines that the voice interaction device recognition response detection is successful; otherwise, the detection terminal determines that the voice interaction device recognition response detection fails.

[0052] In this embodiment, when performing online recognition response detection, the method for obtaining the second response data generated based on the skill command audio file sent by the voice interaction device includes: sending the skill command audio file to the voice interaction system cloud platform through the voice interaction device; generating the second response data based on the skill command audio file using the voice interaction system cloud platform, and sending the second response data to the voice interaction device, so that the voice interaction device sends the second response data.

[0053] In practice, the detection terminal sends the skill command audio file to the voice interaction system cloud platform via a voice interaction device. The voice interaction system cloud platform then generates second response data based on the skill command audio file and sends this second response data to the voice interaction device, which in turn sends the second response data to the detection terminal. This setup improves the accuracy and reliability of the detection results.

[0054] In this embodiment, the method of generating second response data based on a skill command audio file using a voice interaction system cloud platform and sending the second response data to a voice interaction device includes: converting the skill command audio file into text information using the ASR module of the voice interaction system cloud platform; performing semantic understanding on the text information and generating response text information using the NLU module of the voice interaction system cloud platform; converting the response text information into a response audio file using the TTS module of the voice interaction system cloud platform; and sending the text information, response text information, and response audio file to the voice interaction device through the voice interaction system cloud platform; wherein, the second response data includes text information, response text information, and response audio file.

[0055] It should be noted that ASR stands for Automatic Speech Recognition; NLU stands for Natural Language Understanding; and TTS stands for Text To Speech.

[0056] Specifically, when conducting online detection (including online recognition response detection), the voice interaction device communicates with the cloud platform via the WebSocket protocol.

[0057] In this embodiment, when performing online recognition response detection, the method for determining whether the voice interaction device recognition response detection is successful based on the second response data and the second response duration of the voice interaction device includes: when the second response data contains semantically accurate second preset information corresponding to the skill command audio file and the second response duration is less than the second preset duration, the voice interaction device recognition response detection is successful; otherwise, the voice interaction device recognition response detection fails.

[0058] In this embodiment, the detection method for the voice interaction device further includes: acquiring a voice command test set and classifying the voice command test set into domain categories; using a proximity algorithm to perform machine learning training on the domain category classification of the voice command test set and obtaining a trained domain model; when performing online recognition response detection, the method for determining that the second response data contains semantically accurate second preset information corresponding to the skill command audio file includes: classifying the text in the second response data according to the domain model using a proximity algorithm; when the classification is correct, determining that the second response data contains semantically accurate second preset information corresponding to the skill command audio file; otherwise, the second response data does not contain semantically accurate second preset information corresponding to the skill command audio file.

[0059] Specifically, the voice command test set includes a first wake-up audio file and its corresponding response data, a second wake-up audio file and its corresponding response data, and a skill command audio file and its corresponding response data.

[0060] In specific implementation, the detection terminal acquires a voice command test set and classifies the voice command test set into domain categories; the detection terminal uses a proximity algorithm to perform machine learning training on the domain category classification of the voice command test set and obtains a trained domain model; when performing online recognition response detection, the method for determining whether the second response data contains semantically accurate second preset information corresponding to the skill command audio file includes: the detection terminal uses a proximity algorithm to classify the text in the second response data according to the domain model; when the classification is correct, the detection terminal determines that the second response data contains semantically accurate second preset information corresponding to the skill command audio file; otherwise, the detection terminal determines that the second response data does not contain semantically accurate second preset information corresponding to the skill command audio file.

[0061] In other embodiments, the method for determining whether the voice interaction device recognition response detection is successful based on the second response data and the second response duration of the voice interaction device includes: determining whether the voice interaction device recognition response detection is successful based on the text in the second response data and the second response duration of the voice interaction device; when the text is semantically accurate second preset information corresponding to the skill command audio file and the second response duration is less than the second preset duration, the voice interaction device recognition response detection is successful; otherwise, the voice interaction device recognition response detection fails; wherein, the text includes first domain text and first intent text; the detection method of the voice interaction device further includes: sending the text in the second response data to the voice interaction system cloud platform for secondary semantic understanding; dividing the text in the received second response data into second domain text and second intent text through the NLU module of the voice interaction system cloud platform; obtaining the second domain text and the second intent text, and comparing and verifying the first domain text and the second domain text, and comparing and verifying the first intent text and the second intent text; when the first domain text and the second domain text are the same and the first intent text and the second intent text are the same, the voice interaction device recognition response detection is successful; otherwise, the voice interaction device recognition response detection fails.

[0062] In practice, the detection terminal sends the text from the second response data to the voice interaction system cloud platform for secondary semantic understanding. The NLU module of the voice interaction system cloud platform then divides the received text from the second response data into second domain text and second intent text. The detection terminal acquires the second domain text and the second intent text, and compares and verifies both the first domain text and the second domain text, as well as the first intent text and the second intent text. If both the first domain text and the second domain text are identical, the detection terminal determines that the voice interaction device's recognition response detection is successful; otherwise, the detection terminal determines that the voice interaction device's recognition response detection has failed. This setup improves the accuracy and reliability of the detection results.

[0063] In this embodiment, the detection method for the voice interaction device further includes: repeatedly executing the step of acquiring recognition response detection instructions multiple times; wherein, the voice playback module loops and plays the same first wake-up audio file, and the voice playback module sequentially plays different skill command audio files; recording multiple recognition response detection results, and calculating the recognition rate of the current round of recognition response detection based on the multiple recognition response detection results. This setup enables standardized output of the detection results, greatly improving the detection efficiency of the voice device and enhancing the accuracy and reliability of the detection results.

[0064] Specifically, the method for recording multiple recognition response detection results and calculating the recognition rate of the current round of recognition response detection based on these results includes: recording whether each recognition response detection is a failure or success, as well as the intermediate results of each test stage; and outputting the recognition response detection results, intermediate results of each test stage, and the recognition rate of the current round of recognition response detection in the report; wherein, the intermediate results include the error type of recognition response detection failure; when the first response data is never obtained, the recognition response detection is recorded as a failure and the error type is recorded as wake-up failure; when the first response data is obtained after a first preset time period, the recognition response detection is recorded as a failure and the error type is recorded as wake-up too slow; when wake-up is successful but the second preset information is not received within a second preset time period, the recognition response detection is recorded as a failure and the error type is recorded as no response. This setting can standardize the output of detection results, greatly improve the detection efficiency of voice devices, and enhance the accuracy and reliability of detection results.

[0065] Specifically, when performing online recognition response detection, if the wake-up is successful but the second preset information with inaccurate semantics is received within the second preset time period, the recognition response detection is recorded as failed and the error type is recorded as semantic misunderstanding.

[0066] Specifically, the detection terminal determines the steps after repeatedly executing the acquisition and recognition response detection instruction, and the steps after acquiring and recognition response detection instruction refer to steps S120 to S170.

[0067] In this embodiment, the first response duration is from the end of playback of the first wake-up audio file to the acquisition of the first response data; and / or, the second response duration is from the end of playback of the skill command audio file to the acquisition of the second response data.

[0068] In this embodiment, the detection method for voice interaction devices further includes:

[0069] Step 210: Obtain the wake-up response detection command;

[0070] Step 220: Control the voice playback module to play the second wake-up audio file;

[0071] Step 230: Obtain the third response data generated by the voice interaction device based on the second wake-up audio file;

[0072] Step 240: Determine whether the voice interaction device has been successfully woken up based on the third response data and the third response duration of the voice interaction device; if the third response data contains third preset information and the third response duration is less than the third preset duration, the voice interaction device is successfully woken up; otherwise, the voice interaction device fails to wake up.

[0073] In specific implementation, the detection terminal acquires a wake-up response detection command; the detection terminal controls the voice playback module to play a second wake-up audio file; the detection terminal acquires third response data generated by the voice interaction device based on the second wake-up audio file; the detection terminal determines whether the voice interaction device has been successfully woken up based on the third response data and the third response duration of the voice interaction device; when the third response data contains third preset information and the third response duration is less than the third preset duration, the detection terminal determines that the voice interaction device has been successfully woken up; otherwise, the detection terminal determines that the voice interaction device has failed to wake up. During the wake-up response detection process, this detection method can achieve automatic playback of the second wake-up audio file, real-time acquisition, processing, and judgment of the third response data, and finally obtain the detection result, which can greatly improve the detection efficiency of voice interaction devices and enhance the accuracy and reliability of the detection results.

[0074] In this embodiment, the detection method for voice interaction devices further includes: repeatedly executing the step of acquiring wake-up response detection instructions multiple times; wherein, the voice playback module loops the same second wake-up audio file; recording multiple wake-up response detection results, and calculating the wake-up rate of the current round of wake-up response detection based on the multiple wake-up response detection results. This setup enables standardized output of the detection results, greatly improving the detection efficiency of the voice device and enhancing the accuracy and reliability of the detection results.

[0075] Specifically, the detection terminal repeatedly executes the steps after obtaining the wake-up response detection instruction multiple times. The steps after obtaining the wake-up response detection instruction refer to steps 220 to 240.

[0076] In this embodiment, the detection method for the voice interaction device further includes: while the voice interaction device generates third response data based on the second wake-up audio file, the voice interaction device plays a second response prompt tone.

[0077] In this embodiment, the third response duration is from the end of playback of the second wake-up audio file to the acquisition of the third response data.

[0078] Specifically, the method for recording multiple wake-up response detection results and calculating the wake-up rate of the current round of wake-up response detection based on these results includes: recording whether each wake-up response detection is successful or unsuccessful, as well as the intermediate results of each test stage; and outputting the wake-up response detection results, the intermediate results of each test stage, and the recognition rate of the current round of wake-up response detection in the report; wherein, the intermediate results include the error type of the wake-up response detection failure; when the third response data is not obtained at all, the wake-up response detection is recorded as failed and the error type is recorded as wake-up failure; when the third response data is obtained after a third preset time, the wake-up response detection is recorded as failed and the error type is recorded as wake-up too slow. This setting can standardize the output of detection results, greatly improve the detection efficiency of voice devices, and enhance the accuracy and reliability of detection results.

[0079] In this embodiment, a quiet, noise-free testing environment is set up, meaning only the test set audio is played. The external speaker and the voice interaction device are at the same level, and the distance between them is 1, 3, and 5 meters, respectively, for testing and outputting the results. With the external testing conditions set up, this embodiment follows... Figure 2 The detection method and process involve various tests, the details of which are described below:

[0080] When performing wake-up response testing, there is no distinction between offline and online states; testing can be performed in either state. The testing terminal plays the same second wake-up audio file 1000 times in loop playback mode. During each test, the built-in microphone array on the voice interaction device collects and processes the audio data being played. Starting from the moment the second wake-up audio file finishes playing, within the following 3 seconds (this time value can be set independently, i.e., a third preset duration), the testing terminal reads and processes the third response data sent by the voice interaction device in real time via serial port. If the third response data contains a string with the "WAKE UP TICK" identifier, the wake-up is successful. Simultaneously, the built-in speaker of the voice interaction device will play a "I'm here" sound as a second response prompt. The detection terminal uses the time difference between the end of the second wake-up audio file playback and the time of receiving the identifier string as the third response duration for this wake-up detection, and determines that the detection has passed. If no string with the "WAKE UP TICK" identifier is received within 3 seconds, the wake-up detection fails, with an error type of "wake-up failure." In this case, the third response duration is set to -1, indicating a missing string. Even if the identifier string is received after 3 seconds, the detection is still considered a failure, with an error type of "wake-up too slow." The second wake-up audio file is then played again for the next detection. When this round of detection ends, the detection terminal calculates the wake-up rate based on the ratio of successful wake-ups to the total number of wake-up detections in this round, and finally outputs a detailed detection report.

[0081] When performing offline recognition response detection, the detection terminal plays the same first wake-up audio file in loop mode and a video file containing 50 skill commands in sequential playback mode. Playing the first wake-up audio file and the skill command audio file separately constitutes one recognition response detection. During each detection, starting from the moment the first wake-up audio file finishes playing, the detection terminal reads and processes the first response data sent by the voice interaction device in real time via serial port within the following 3 seconds (this time value can be set independently, i.e., the first preset duration). If the first response data read contains the string "WAKE UP TICK", it indicates a successful wake-up. Simultaneously, the built-in speaker of the voice interaction device will play a "I'm here" sound as the first response prompt. The detection terminal calculates the first response duration for this test and determines a successful wake-up. If no string with the "WAKE UP TICK" identifier is received within 3 seconds, it indicates a wake-up failure, and the recognition detection is determined to have failed, with the error type "wake-up failure". In this case, the first response duration is set to -1, indicating a missing string. Even if the identifier string for this test response is received after 3 seconds, the recognition detection is still determined to have failed, with the error type "wake-up too slow". After 3 seconds, the skill command audio file is played. From the moment the skill command audio file finishes playing, within the next 5 seconds (this time value can be set independently, i.e., the second preset duration), assuming a successful wake-up, if the second response data read by the detection terminal contains "send", it indicates a successful wake-up. If the string identified by "result" is present in the data and the "asr_return" field is identical to the standard text corresponding to the skill command audio file used in this test, then the recognition response detection is considered successful. The detection terminal calculates the second response duration of the recognition response detection. If the identifier string is not received within 5 seconds, the second response duration is set to -1, indicating a missing string, and the recognition detection is considered a failure with an error type of "no response". The next recognition response detection is then performed. When the current round of detection ends, the detection terminal calculates the recognition rate of the current round of detection based on the ratio of the number of successful recognitions to the total number of recognition detections, and finally outputs a detailed detection report.

[0082] When performing online recognition response detection, the rules for playing audio on the detection terminal are the same as for offline recognition response detection. In this embodiment, 100 skill command audio files are played for detection. During each detection, starting from the moment the first wake-up audio file finishes playing, within the following 3 seconds (this time value can be set independently, i.e., the first preset duration), the detection terminal reads and processes the first response data sent by the voice interaction device in real time via the serial port. If the read first response data contains a string with the identifier "WAKE UP TICK", it indicates that the wake-up is successful. At the same time, the built-in speaker of the voice interaction device will also play the sound "I am here" as the first response prompt tone. The detection terminal calculates the first response duration of this test and determines that the wake-up is successful. If no string with "WAKE UP TICK" is received within 3 seconds... If the string "TICK" indicates a wake-up failure, the recognition and detection are deemed a failure, with the error type being "wake-up failure". The initial response duration is set to -1, indicating no response. Even if the response string is received after 3 seconds, the recognition and detection are still deemed a failure, with the error type being "wake-up too slow". After 3 seconds, the skill command audio file is played. From the moment the audio file finishes playing, within the next 5 seconds (this time can be set independently, i.e., the second preset duration), assuming a successful wake-up, the voice interaction device will send the recorded and processed skill command audio file to the voice interaction service system cloud platform via a network request for processing. The cloud-based ASR module converts the skill command sound information into text (i.e., text information). The converted text is then semantically understood by the NLU module, which provides reasonable response information (i.e., response text information). The TTS module converts the response text contained in the NLU output response information into sound and stores it in an audio file (i.e., response audio file). Finally, the cloud platform integrates the ASR-recognized text, NLU output information, and TTS-synthesized audio resource link address (i.e., second response data) and sends it to the voice interaction device. The voice interaction device parses and processes the received second response data and sends it to the detection terminal via serial port. Table 1 lists the identifier response data (text in the second response data) corresponding to some skill command audio files for the voice interaction device. If the identifier response data read by the detection terminal contains "send",... If the string identified by "result" is used, the text field within it is extracted for semantic accuracy determination. If the determination result is incorrect, the detection fails with the error type "semantic misunderstanding". The detection terminal calculates the second response time for the recognition response detection. If the identified string is not received within 5 seconds, the second response time is set to -1, indicating a missing string, and the recognition detection is determined to have failed with the error type "no response". The next recognition response detection is then performed. When the current round of detection ends, the detection terminal calculates the recognition rate of the current round of detection and finally outputs a detailed detection report.

[0083] Table 1: Examples of Identifier Response Data for Some Skill Command Audio Files

[0084]

[0085] In specific implementation, when performing online semantic accuracy detection of the recognition response, the determination method is as follows: The detection terminal pre-classifies the voice command test set into domain categories. The domain is used to identify the business domain corresponding to the voice command. Then, based on the known semantic response reference attribute word sample set of the domain, the coordinate values ​​are quantified. Taking the voice command "What is the weather like in Zhuhai today?" as an example, this voice command belongs to the climate domain, and it is calibrated as the origin of the coordinates (x0, y0). Table 2 shows the coordinate calibration values ​​and classification results of the reference attribute words in the semantic understanding response of this command. Multi-dimensional coordinate quantization can be performed according to the actual calibration algorithm. For all calibrated attribute words, the distance between them and the origin of the coordinates is used as the quantization value. As a standard, when the distance value meets condition d i If the value is less than or equal to D, where D is a preset threshold, then the attribute is considered correctly categorized (belonging to the climate domain). That is, the closer it is to the origin, the more it belongs to the domain's attribute words; otherwise, it is considered incorrectly categorized. Subsequently, the KNN nearest neighbor algorithm is used to train the model on the sample database according to the above rules to obtain the trained domain model. The semantic response text is then subjected to attribute word extraction and coordinate labeling, and the domain's KNN model is used for classification. The accuracy of the semantic understanding is determined based on the correctness of the classification.

[0086] Table 2

[0087]

[0088] In practice, the NLU module divides the response result into domain and intent during semantic understanding. After the detection terminal extracts the text containing the "asr_return" field from the second response data, it separately requests the NLU module of the cloud platform to perform secondary semantic understanding. The obtained request response result is then compared and verified with the identifier response data. If the values ​​of the "domain" field and the "intent" field are the same, the recognition response detection is considered to have passed; otherwise, the recognition response detection is considered to have failed, and the error type is "semantic understanding error".

[0089] In this embodiment, the detection method implemented by the detection terminal is achieved through GUI host computer software programming using the Python programming language and its third-party libraries.

[0090] This application enables automatic playback of voice interaction test commands, real-time acquisition, processing, and judgment of test response content, provision of error type suggestions, and calculation of indicators such as wake-up rate, recognition rate, and response time. Finally, it standardizes and outputs the test results, significantly improving the testing efficiency of voice devices and enhancing the accuracy and reliability of the test results. Existing testing methods do not automate the real-time filtering and extraction of response data or perform error statistical analysis and judgment. Instead, they rely on manual judgment and statistical analysis of various errors. They cannot automatically determine various functional errors and performance defects in the test result output based on actual response data, resulting in incomplete and inconspicuous device test results and limited testing coverage.

[0091] This invention also provides a voice interaction detection device, please refer to... Figure 3 The voice interaction detection device includes:

[0092] The first acquisition unit 10 is used to acquire the recognition response detection command;

[0093] The first control unit 20 is used to control the voice playback module to play the first wake-up audio file;

[0094] The second acquisition unit 30 is used to acquire the first response data generated by the voice interaction device based on the first wake-up audio file;

[0095] The first judgment unit 40 is used to determine whether the voice interaction device has been successfully woken up based on the first response data and the first response duration of the voice interaction device.

[0096] The second control unit 50 is used to control the voice playback module to play skill command audio files;

[0097] The third acquisition unit 60 is used to acquire the second response data generated based on the skill command audio file sent by the voice interaction device;

[0098] The second judgment unit 70 is used to determine whether the voice interaction device recognition response detection is successful based on the second response data and the second response duration of the voice interaction device.

[0099] In this voice interaction detection device, a first acquisition unit 10 is used to acquire a recognition response detection command; a first control unit 20 is used to control the voice playback module to play a first wake-up audio file; a second acquisition unit 30 is used to acquire first response data generated by the voice interaction device based on the first wake-up audio file; a first judgment unit 40 is used to judge whether the voice interaction device has been successfully woken up based on the first response data and the first response duration of the voice interaction device; when the first response data contains first preset information and the first response duration is less than the first preset duration, the voice interaction device is successfully woken up, otherwise the voice interaction device fails to wake up; a second control unit 50 is used to control the voice playback module to play a skill command audio file; a third acquisition unit 60 is used to acquire second response data generated by the voice interaction device based on the skill command audio file; a second judgment unit 70 is used to judge whether the voice interaction device's recognition response detection is successful based on the second response data and the second response duration of the voice interaction device; wherein, when the second response data contains second preset information that matches the skill command audio file and the second response duration is less than the second preset duration, the voice interaction device's recognition response detection is judged to be successful, otherwise the voice interaction device's recognition response detection is judged to be failed. During the recognition and response detection process, this voice interaction detection device can automatically play the first wake-up audio file and the skill command audio file, collect the first response data and the second response data in real time, process and judge them, and finally obtain the detection results. This can greatly improve the detection efficiency of voice interaction devices and enhance the accuracy and reliability of the detection results.

[0100] The present invention also provides a voice interaction detection device, including a processor and a memory, wherein the memory is used to store a computer program, and the processor is used to call and run the computer program from the memory, so that the voice interaction detection device performs the voice interaction device detection method in the above embodiments.

[0101] The present invention also provides a non-volatile storage medium, which includes a stored program, wherein, when the program is running, it controls the device where the non-volatile storage medium is located to execute the voice interaction device detection method in the above embodiments.

[0102] Specifically, the aforementioned storage medium is used to store program instructions that perform the following functions, thereby achieving the following functions:

[0103] The system acquires a recognition response detection command; controls the voice playback module to play a first wake-up audio file; acquires first response data generated by the voice interaction device based on the first wake-up audio file; determines whether the voice interaction device has been successfully woken up based on the first response data and the first response duration of the voice interaction device; if the first response data contains first preset information and the first response duration is less than the first preset duration, the voice interaction device is successfully woken up; otherwise, the voice interaction device fails to wake up; after the voice interaction device is successfully woken up, controls the voice playback module to play a skill command audio file; acquires second response data generated by the voice interaction device based on the skill command audio file; determines whether the voice interaction device's recognition response detection is successful based on the second response data and the second response duration of the voice interaction device; wherein, if the second response data contains second preset information matching the skill command audio file and the second response duration is less than the second preset duration, the voice interaction device's recognition response detection is successful; otherwise, the voice interaction device's recognition response detection is failed.

[0104] As can be seen from the above description, the embodiments of the present invention achieve the following technical effects:

[0105] The detection method for the voice interaction device of the present invention involves the following steps: After the detection terminal obtains a recognition response detection command, the detection terminal first controls the voice playback module to play a first wake-up audio file. The voice interaction device receives the first wake-up audio file, processes it, and generates first response data. The detection terminal obtains the first response data generated by the voice interaction device based on the first wake-up audio file. The detection terminal then determines whether the voice interaction device has been successfully woken up based on the first response data and the first response duration of the voice interaction device. If the first response data contains first preset information and the first response duration is less than the first preset duration, the detection terminal determines that the voice interaction device has been successfully woken up; otherwise, the detection terminal... The detection method first determines if the voice interaction device wake-up fails. Then, after successful wake-up, the detection terminal controls the voice playback module to play the skill command audio file. The detection terminal acquires the second response data generated by the voice interaction device based on the skill command audio file. The detection terminal then determines whether the voice interaction device's response recognition detection is successful based on the second response data and the second response duration of the voice interaction device. Specifically, if the second response data contains second preset information matching the skill command audio file and the second response duration is less than the second preset duration, the detection terminal determines that the voice interaction device's response recognition detection is successful; otherwise, the detection terminal determines that the voice interaction device's response recognition detection has failed. During the response recognition detection process, this method enables automatic playback of the first wake-up audio file and the skill command audio file, real-time acquisition, processing, and judgment of the first and second response data, and finally, obtaining the detection result. This significantly improves the detection efficiency of the voice interaction device and enhances the accuracy and reliability of the detection results.

[0106] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented, for example, in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0107] For ease of description, spatial relative terms such as "above," "on top of," "on the upper surface of," "above," etc., are used herein to describe the spatial positional relationship of a device or feature as shown in the figures to other devices or features. It should be understood that spatial relative terms are intended to encompass different orientations in use or operation beyond the orientation of the device as described in the figures. For example, if the device in the figures were inverted, a device described as "above" or "on top of" other devices or structures would subsequently be positioned as "below" or "under" other devices or structures. Thus, the exemplary term "above" can include both "above" and "below." The device may also be positioned in other different ways (rotated 90 degrees or in other orientations), and the spatial relative descriptions used herein will be interpreted accordingly.

[0108] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for detecting a voice interaction device, characterized in that, include: Obtain the recognition response detection command; Control the voice playback module to play the first wake-up audio file; Obtain the first response data generated by the voice interaction device based on the first wake-up audio file; The voice interaction device is successfully woken up based on the first response data and the first response duration of the voice interaction device. When the first response data contains the first preset information and the first response duration is less than the first preset duration, the voice interaction device is successfully woken up; otherwise, the voice interaction device fails to wake up. After the voice interaction device is successfully woken up, the voice playback module is controlled to play the skill command audio file, and the second response data generated by the voice interaction device based on the skill command audio file is obtained. Based on the second response data and the second response duration of the voice interaction device, it is determined whether the voice interaction device recognition response detection is successful. Wherein, when the second response data contains second preset information that matches the skill command audio file and the second response duration is less than the second preset duration, it is determined that the voice interaction device recognition response detection is successful; otherwise, it is determined that the voice interaction device recognition response detection fails. When performing online recognition response detection, the method for determining whether the voice interaction device recognition response detection is successful based on the second response data and the second response duration of the voice interaction device includes: when the second response data contains semantically accurate second preset information corresponding to the skill command audio file and the second response duration is less than the second preset duration, the voice interaction device recognition response detection is successful; otherwise, the voice interaction device recognition response detection fails. The detection method for the voice interaction device further includes: acquiring a voice command test set and classifying the voice command test set into domain categories; using a proximity algorithm to perform machine learning training on the domain category classification of the voice command test set and obtaining a trained domain model; when performing online recognition response detection, the method for determining that the second response data contains semantically accurate second preset information corresponding to the skill command audio file includes: classifying the text in the second response data according to the domain model using the proximity algorithm; when the classification is correct, determining that the second response data contains semantically accurate second preset information corresponding to the skill command audio file; otherwise, the second response data does not contain semantically accurate second preset information corresponding to the skill command audio file.

2. The detection method for a voice interaction device according to claim 1, characterized in that, When performing offline recognition response detection, the second response data is generated by the voice interaction device based on the skill command audio file.

3. The detection method for a voice interaction device according to claim 1, characterized in that, When performing offline recognition response detection, the method for determining whether the voice interaction device's recognition response detection is successful based on the second response data and the second response duration of the voice interaction device includes: When the second response data contains the same second preset information as the standard text corresponding to the skill command audio file and the second response duration is less than the second preset duration, it is determined that the voice interaction device recognition response detection is successful; otherwise, the voice interaction device recognition response detection fails.

4. The detection method for a voice interaction device according to claim 1, characterized in that, When performing online recognition response detection, the method for obtaining the second response data generated based on the skill command audio file sent by the voice interaction device includes: The skill command audio file is sent to the voice interaction system cloud platform via the voice interaction device. The cloud platform of the voice interaction system generates the second response data based on the skill command audio file, and sends the second response data to the voice interaction device so that the voice interaction device sends the second response data.

5. The detection method for a voice interaction device according to claim 4, characterized in that, The method for generating the second response data based on the skill command audio file using the cloud platform of the voice interaction system, and sending the second response data to the voice interaction device includes: The ASR module of the voice interaction system cloud platform converts the skill command audio file into text information. The NLU module of the voice interaction system cloud platform performs semantic understanding on the text information and generates response text information. The response text information is converted into a response audio file through the TTS module of the cloud platform of the voice interaction system; The voice interaction system cloud platform sends the text information, the response text information, and the response audio file to the voice interaction device; wherein, the second response data includes the text information, the response text information, and the response audio file.

6. The detection method for a voice interaction device according to claim 1, characterized in that, The method for determining whether the voice interaction device's recognition response detection is successful based on the second response data and the second response duration of the voice interaction device includes: The voice interaction device recognizes and detects whether the response is successful based on the text in the second response data and the second response duration of the voice interaction device. When the text is semantically accurate second preset information corresponding to the skill command audio file and the second response duration is less than the second preset duration, the voice interaction device recognizes and detects the response successfully; otherwise, the voice interaction device recognizes and detects the response fails. The text includes first domain text and first intent text. The detection method for the voice interaction device also includes: The text in the second response data is sent to the voice interaction system cloud platform for secondary semantic understanding; The NLU module of the voice interaction system cloud platform divides the text in the received second response data into second domain text and second intent text. Obtain the second domain text and the second intent text, and compare and verify the first domain text and the second domain text, and compare and verify the first intent text and the second intent text; When the first domain text and the second domain text are the same, and the first intent text and the second intent text are the same, the voice interaction device recognition response detection is successful; otherwise, the voice interaction device recognition response detection fails.

7. The detection method for a voice interaction device according to claim 1, characterized in that, The detection method for the voice interaction device also includes: The steps following the acquisition and recognition response detection command are repeated multiple times; wherein, the voice playback module plays the same first wake-up audio file in a loop, and the voice playback module plays different skill command audio files sequentially; Record the results of multiple recognition response detections, and calculate the recognition rate of the current round of recognition response detection based on the results of multiple recognition response detections.

8. The detection method for a voice interaction device according to claim 7, characterized in that, The method for recording multiple recognition response detection results and calculating the recognition rate of the current round of recognition response detection based on the multiple recognition response detection results includes: Record whether each recognition response detection is successful or unsuccessful, as well as the intermediate results of each test stage; and output the recognition response detection results, intermediate results of each test stage, and recognition rate of this round of recognition response detection in the report; The intermediate results include: identifying the error type of response detection failure; when the first response data is not obtained, recording the response detection failure and the error type as wake-up failure; when the first response data is obtained after the first preset time, recording the response detection failure and the error type as wake-up too slow; when wake-up is successful but the second preset information is not received within the second preset time, recording the response detection failure and the error type as no response.

9. The detection method for a voice interaction device according to claim 1, characterized in that, The first response duration is from the end of playback of the first wake-up audio file to the acquisition of the first response data; And / or, the second response duration is from the end of playback of the skill command audio file to the acquisition of the second response data.

10. The detection method for a voice interaction device according to claim 1, characterized in that, The detection method for the voice interaction device also includes: Obtain wake-up response detection command; Control the voice playback module to play the second wake-up audio file; Obtain the third response data generated by the voice interaction device based on the second wake-up audio file; The voice interaction device is successfully woken up based on the third response data and the third response duration of the voice interaction device. When the third response data contains third preset information and the third response duration is less than the third preset duration, the voice interaction device is successfully woken up; otherwise, the voice interaction device fails to wake up.

11. The detection method for a voice interaction device according to claim 10, characterized in that, The detection method for the voice interaction device also includes: The steps following the acquisition of the wake-up response detection command are repeated multiple times; wherein, the voice playback module plays the same second wake-up audio file in a loop: Record the wake-up response detection results multiple times, and calculate the wake-up rate of this round of wake-up response detection based on the multiple wake-up response detection results.

12. The detection method for a voice interaction device according to claim 11, characterized in that, The method for recording multiple wake-up response detection results and calculating the wake-up rate of the current round of wake-up response detection based on the multiple wake-up response detection results includes: Record whether each wake-up response detection is successful or unsuccessful, as well as the intermediate results of each test stage; and output the wake-up response detection results, the intermediate results of each test stage, and the recognition rate of this round of wake-up response detection in the report; The intermediate results include the error type of wake-up response detection failure; when the third response data is not obtained, wake-up response detection failure is recorded and the error type is wake-up failure; when the third response data is obtained after the third preset time, wake-up response detection failure is recorded and the error type is wake-up too slow.

13. The detection method for a voice interaction device according to claim 10, characterized in that, The detection method for the voice interaction device also includes: While the voice interaction device generates the third response data based on the second wake-up audio file, the voice interaction device plays the second response prompt tone.

14. The detection method for a voice interaction device according to claim 10, characterized in that, The duration of the third response is from the end of playback of the second wake-up audio file to the acquisition of the third response data.

15. A voice interaction detection device, characterized in that, The detection method applicable to the voice interaction device according to any one of claims 1 to 14, wherein the voice interaction detection device comprises: The first acquisition unit is used to acquire the recognition response detection command; The first control unit is used to control the voice playback module to play the first wake-up audio file; The second acquisition unit is used to acquire the first response data generated by the voice interaction device based on the first wake-up audio file; The first judgment unit is used to determine whether the voice interaction device has been successfully woken up based on the first response data and the first response duration of the voice interaction device. The second control unit is used to control the voice playback module to play skill command audio files; The third acquisition unit is used to acquire the second response data generated based on the skill command audio file sent by the voice interaction device; The second judgment unit is used to determine whether the voice interaction device recognition response detection is successful based on the second response data and the second response duration of the voice interaction device.

16. A voice interaction detection device, characterized in that, The device includes a processor and a memory, the memory being used to store a computer program, and the processor being used to call and run the computer program from the memory, causing the voice interaction detection device to perform the voice interaction device detection method according to any one of claims 1 to 14.

17. A non-volatile storage medium, characterized in that, The non-volatile storage medium includes a stored program, wherein, when the program is executed, the device containing the non-volatile storage medium is controlled to perform the detection method of the voice interaction device according to any one of claims 1 to 14.

Images