A voice recognition control management method and system suitable for smart home appliances

Abstract

The present invention relates to the field of speech recognition control, and discloses a speech recognition control management method and system applicable to smart home appliances, comprising the following steps: performing networking configuration correction processing on all smart home appliances to obtain a smart home appliance networking control system, and performing audio processing on the collected environmental audio in the smart home appliance networking control system to obtain a target human voice audio, and finally performing audio recognition and control instruction conversion on the target human voice audio, so that the smart home appliance networking control system controls the networked qualified smart home appliances to operate in coordination according to the audio recognition results. The present invention can realize speech recognition control of smart home appliances, improve life efficiency in life, bring high efficiency and convenience to users, improve the quality of life, and also facilitate people with mobility difficulties, and reduce the danger of users using home appliances.

Description

Technical Field

[0001] The present invention relates to the field of speech recognition control, and in particular to a speech recognition control management method and system suitable for smart home appliances. Background Art

[0002] Smart home appliances refer to home appliances equipped with advanced sensors and Internet connections. These home appliances can be connected via wireless networks to achieve remote control, automated operation, and interconnection with other smart devices. Smart home appliances are designed to improve the convenience, efficiency, and comfort of life, while providing a personalized user experience. In traditional home appliances, home appliances need to be controlled by remote controls, or even by hands-on operation to start and run the home appliances. Home appliances with potential safety hazards, such as gas stoves and microwave ovens, may be dangerous if operated by hands-on. For people with limited mobility, such as the elderly and the disabled, operating home appliances by hands-on undoubtedly increases the potential safety hazards in their lives. If smart home appliances are controlled by voice and made to work, the potential hazards brought by home appliances in life can be reduced and life efficiency can be improved. Controlling smart home appliances through voice covers convenience, improving quality of life, providing auxiliary functions, promoting the construction of smart home ecosystems, and future development potential. With the advancement of technology and the popularization of applications, the application of voice recognition control in smart homes will become more extensive, further improving users' life experience and intelligence. Summary of the invention

[0003] The present invention overcomes the deficiencies of the prior art and provides a voice recognition control management method and system suitable for smart home appliances.

[0004] In order to achieve the above object, the technical solution adopted by the present invention is:

[0005] A first aspect of the present invention provides a voice recognition control management method applicable to smart home appliances, comprising the following steps:

[0006] Perform networking configuration and networking correction for all smart home appliances, and build a smart home appliance networking control system;

[0007] Controlling the qualified networked smart home appliances to perform audio preprocessing, audio segmentation processing and voiceprint analysis on the ambient audio to obtain the target human voice audio;

[0008] Perform audio recognition on the target noise-reduced human voice audio, and based on the audio recognition results, control the networked qualified smart appliances in the smart home appliance network control system to operate collaboratively.

[0009] Furthermore, in a preferred embodiment of the present invention, the networking configuration and networking correction of all smart home appliances to build a smart home appliance networking control system is specifically as follows:

[0010] The smart home appliance that needs to be controlled by voice recognition is marked as the target smart home appliance, and the networking module on the target smart home appliance is obtained and marked as the target networking module;

[0011] Acquire a network networking device, and based on the frequency band that the network networking device transmits to different target smart home appliances for network connection, mark it as a network frequency band;

[0012] Controlling the target networking modules of all target smart home appliances to receive the networking network frequency band, and after the target networking modules of the target smart home appliances receive the networking network frequency band, performing network tests on the target networking modules of different target smart home appliances, and determining whether the current network frequency band of the target networking modules of any target smart home appliance is not the networking network frequency band;

[0013] If not, the corresponding target smart home appliance is calibrated as a network-connected qualified smart home appliance;

[0014] If so, the target networking module of the target smart home appliance whose current network frequency band is not the networking network frequency band is marked as an abnormal networking module, and a big data network is obtained, and the firmware update program of the abnormal networking module is retrieved and output in the big data network to obtain a corrected networking module;

[0015] If the current network frequency band of the corrected networking module is still not the networking network frequency band, the target smart home appliance corresponding to the corrected networking module is marked as the abnormal networking smart home appliance, and after obtaining the receiving networking network frequency band, the external networking module whose current network frequency band is the real-time network frequency band is marked as the target external networking module;

[0016] Connecting the target external networking module to the abnormally networked smart home appliance to obtain a networked qualified smart home appliance;

[0017] Control all network-connected qualified smart home appliances to connect devices based on the network frequency band to obtain a smart home appliance network control system.

[0018] Furthermore, in a preferred embodiment of the present invention, the control of the network-connected qualified smart home appliance performs audio preprocessing, audio segmentation processing and voiceprint analysis on the ambient audio to obtain the target human voice audio, specifically:

[0019] Obtain an audio recognition sensor of a qualified networked smart appliance, calibrate it as a target audio recognition sensor, obtain a manual of the qualified networked smart appliance, and based on the manual of the qualified networked smart appliance, obtain a maximum range of ambient audio collected by the qualified networked smart appliance, calibrate it as the maximum range of ambient audio collection;

[0020] Control the target audio recognition sensor to collect ambient audio in real time, collect ambient audio within the maximum range, and calibrate it as the target ambient audio;

[0021] Introducing an adaptive filter into a networked qualified smart home appliance, initializing filter parameters of the adaptive filter to obtain an initialized adaptive filter, and segmenting the target environment audio into short-time audio frames in the initialized adaptive filter;

[0022] In the initialized adaptive filter, the filter parameters of the adaptive filter are converged and trained based on the recursive least square method to obtain a trained adaptive filter, and all short-time audio frames are filtered based on the trained adaptive filter to obtain filtered short-time audio frames;

[0023] The filtered short-time audio frames of each frame are superimposed and reconstructed to obtain reconstructed audio, and the reconstructed audio is subjected to signal smoothing processing to obtain denoised audio, which is calibrated as denoised human voice audio;

[0024] The denoised human voice audio is segmented, and voiceprint analysis is performed on the denoised human voice audio after the audio segmentation. Based on the voiceprint analysis result, the target denoised human voice audio is obtained.

[0025] Further, in a preferred embodiment of the present invention, the denoised human voice audio is subjected to audio segmentation processing, the denoised human voice audio after audio segmentation is subjected to voiceprint analysis, and the target denoised human voice audio is obtained based on the voiceprint analysis result, specifically:

[0026] In the network-connected qualified smart home appliances, the voiceprint parameters of the person who can voice control the network-connected qualified smart home appliances are preset and calibrated as the target person's voiceprint parameters;

[0027] Introducing a GMM-UBM algorithm, constructing a GMM-UBM feature model based on the GMM-UBM algorithm, and converting the noise-reduced human voice audio into noise-reduced human voice audio feature parameters;

[0028] Importing the noise reduction human voice audio feature parameters into the GMM-UBM feature model, the GMM-UBM feature model performs character differentiation and audio segmentation on the noise reduction human voice audio feature parameters to obtain noise reduction sub-human voice audio;

[0029] Perform voiceprint analysis on all noise reduction sub-voice audios to obtain voiceprint parameters of the noise reduction sub-voice audios, calculate the Euclidean distance between all the noise reduction sub-voice audio voiceprint parameters and the target person's voiceprint parameters, and preset a first standard Euclidean distance interval;

[0030] If the Euclidean distance between the noise reduction sub-voice audio voiceprint parameters and the target person's voiceprint parameters is maintained within the first standard Euclidean distance interval, the corresponding noise reduction sub-voice audio voiceprint parameters are calibrated as the target noise reduction sub-voice audio voiceprint parameters, and the audio corresponding to the target noise reduction sub-voice audio voiceprint parameters is calibrated as the target noise reduction voice audio.

[0031] Furthermore, in a preferred embodiment of the present invention, the target noise-reduced human voice audio is subjected to audio recognition, and based on the audio recognition result, the qualified networked smart home appliances are controlled to operate in a smart home appliance network control system in a coordinated manner, specifically:

[0032] Presetting the wake-up human voice audio text information in the target audio recognition sensor, wherein the wake-up human voice audio text information is the text information that needs to be recognized to wake up the network-connected qualified smart home appliance and put the network-connected qualified smart home appliance into a working state;

[0033] Uploading the target noise-reduced human voice audio to a smart home appliance networking control system, and introducing a Viterbi algorithm into the smart home appliance networking control system, wherein the Viterbi algorithm can perform audio decoding on the target noise-reduced human voice audio and convert it into text information;

[0034] Based on the Viterbi algorithm, audio decoding is performed on the target denoised human voice audio to obtain text information of the target denoised human voice audio after parsing, and a text correction model is introduced to perform text correction processing on the text information of the target denoised human voice audio after parsing to obtain text information of the target denoised human voice audio;

[0035] In the smart home appliance network control system, text analysis is performed on the target noise reduction human voice audio text information, the Euclidean distance between the target noise reduction human voice audio text information and the wake-up human voice audio text information is calculated, and a second standard Euclidean distance interval is preset;

[0036] If the Euclidean distance between the target noise reduction human voice audio text information and the wake-up human voice audio text information is maintained within the second marked standard Euclidean distance interval, all networked qualified smart home appliances in the smart home appliance network control system are woken up;

[0037] When all the qualified smart home appliances in the smart home appliance networking control system are awakened, text analysis is performed on the target noise reduction human voice audio text information, and all the qualified smart home appliances are controlled to operate collaboratively based on the text analysis results.

[0038] Furthermore, in a preferred embodiment of the present invention, the text analysis of the target noise reduction human voice audio text information is performed, and all networked qualified smart home appliances are controlled to operate collaboratively based on the text analysis results, specifically:

[0039] Obtain all control instructions for controlling qualified networked smart home appliances in the smart home appliance network control system, mark them as home appliance control instructions, and obtain program codes of the home appliance control instructions, mark them as home appliance control instruction program codes;

[0040] Convert the home appliance control instruction program code into text information to obtain all text information that can control qualified networked smart home appliances and mark them as standard text information;

[0041] A big data network is introduced, and a text semantic comparison knowledge graph is retrieved based on the big data network, wherein the text semantic comparison knowledge graph describes all text information having the same semantics as the standard text information;

[0042] Based on the text semantic comparison knowledge graph, perform text semantic comparison on the standard text information and the denoised human voice audio text information to obtain the denoised human voice audio target text information;

[0043] The target text information of the noise reduction human voice audio is converted into a program code to obtain a target program code of the noise reduction human voice audio, and a target home appliance control instruction is obtained based on the target program code of the noise reduction human voice audio, wherein the target home appliance control instruction is a control instruction for the target noise reduction human voice audio to control the networked qualified smart home appliance;

[0044] When all the qualified smart home appliances in the smart home appliance networking control system are awakened, the smart home appliance networking control system controls the qualified smart home appliances to operate in a coordinated manner based on the target appliance control instructions.

[0045] The second aspect of the present invention further provides a voice recognition control management system applicable to smart home appliances, the voice recognition control management system comprising a memory and a processor, the memory storing a voice recognition control management method, and when the voice recognition control management method is executed by the processor, the following steps are implemented:

[0046] Perform networking configuration and networking correction for all smart home appliances, and build a smart home appliance networking control system;

[0047] Controlling the qualified networked smart home appliances to perform audio preprocessing, audio segmentation processing and voiceprint analysis on the ambient audio to obtain the target human voice audio;

[0048] Perform audio recognition on the target noise-reduced human voice audio, and based on the audio recognition results, control the networked qualified smart appliances in the smart home appliance network control system to operate collaboratively.

[0049] The present invention solves the technical defects existing in the background technology, and has the following beneficial effects: all smart home appliances are subjected to networking configuration correction processing to obtain a smart home appliance networking control system, and the collected environmental audio is subjected to audio processing in the smart home appliance networking control system to obtain the target human voice audio, and finally the target human voice audio is subjected to audio recognition and control instruction conversion, so that the smart home appliance networking control system controls the networked qualified smart home appliances to operate in coordination according to the audio recognition results. The present invention can realize voice recognition control of smart home appliances, improve the living efficiency in life, bring high efficiency and convenience to users, improve the quality of life, and also facilitate people with mobility difficulties, reducing the danger of users using home appliances. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] 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, drawings of other embodiments can be obtained based on these drawings without paying creative work.

[0051] Figure 1 A flow chart showing a method for voice recognition control management applicable to smart home appliances is shown;

[0052] Figure 2 A flow chart of a method for performing audio recognition on target noise-reduced human voice audio is shown;

[0053] Figure 3 A program view of a voice recognition control management system suitable for smart home appliances is shown. DETAILED DESCRIPTION

[0054] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0055] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0056] Figure 1 A flow chart of a voice recognition control management method applicable to smart home appliances is shown, comprising the following steps:

[0057] S102: Perform networking configuration and networking correction on all smart home appliances to build a smart home appliance networking control system;

[0058] S104: controlling the qualified networked smart home appliance to perform audio preprocessing, audio segmentation processing, and voiceprint analysis on the ambient audio to obtain the target human voice audio;

[0059] S106: Performing audio recognition on the target noise-reduced human voice audio, and based on the audio recognition result, controlling the networked qualified smart home appliances to operate in a smart home appliance network control system in a coordinated manner.

[0060] Furthermore, in a preferred embodiment of the present invention, the networking configuration and networking correction of all smart home appliances to build a smart home appliance networking control system is specifically as follows:

[0061] The smart home appliance that needs to be controlled by voice recognition is marked as the target smart home appliance, and the networking module on the target smart home appliance is obtained and marked as the target networking module;

[0062] Acquire a network networking device, and based on the frequency band that the network networking device transmits to different target smart home appliances for network connection, mark it as a network frequency band;

[0063] Controlling the target networking modules of all target smart home appliances to receive the networking network frequency band, and after the target networking modules of the target smart home appliances receive the networking network frequency band, performing network tests on the target networking modules of different target smart home appliances, and determining whether the current network frequency band of the target networking modules of any target smart home appliance is not the networking network frequency band;

[0064] If not, the corresponding target smart home appliance is calibrated as a network-connected qualified smart home appliance;

[0065] If so, the target networking module of the target smart home appliance whose current network frequency band is not the networking network frequency band is marked as an abnormal networking module, and a big data network is obtained, and the firmware update program of the abnormal networking module is retrieved and output in the big data network to obtain a corrected networking module;

[0066] If the current network frequency band of the corrected networking module is still not the networking network frequency band, the target smart home appliance corresponding to the corrected networking module is marked as the abnormal networking smart home appliance, and after obtaining the receiving networking network frequency band, the external networking module whose current network frequency band is the real-time network frequency band is marked as the target external networking module;

[0067] Connecting the target external networking module to the abnormally networked smart home appliance to obtain a networked qualified smart home appliance;

[0068] Control all network-connected qualified smart home appliances to connect devices based on the network frequency band to obtain a smart home appliance network control system.

[0069] It should be noted that smart home appliances are home appliances that can be operated through voice controllers. For example, if you say to adjust the temperature to 20 degrees to the air conditioner in the smart home appliance, the air conditioner will adjust accordingly according to the voice. Smart home appliances are usually matched. To control smart home appliances by voice, you first need to connect the smart home appliances to the network control system. The control system analyzes the voice and outputs the control instructions to the smart home appliances to make the smart home appliances run. Therefore, all smart home appliances need to be networked to form a network control system. To make all smart home appliances form a network control system, it is necessary to control the network frequency bands of all smart home appliances to be the same, so that all smart home appliances can be controlled uniformly. The network networking device, such as a router, can send a network frequency band for the target networking module in all smart home appliances to receive. Under normal circumstances, after the target networking module receives the network frequency band, its own network frequency band will be equal to the network frequency band, and the corresponding smart home appliance is a qualified smart home appliance for networking. If the target networking module receives the network frequency band, its own network frequency band is not the network frequency band, then it is judged that the networking module is abnormal and marked as an abnormal networking module. The reason for the abnormality of the networking module may be that the firmware in the module does not support the current networking network frequency band, so the firmware needs to be upgraded to ensure that the networking module supports the networking network frequency band. The firmware update program can be obtained from the big data network, which is equivalent to the Internet and stores various solutions. If the network frequency band in the networking module is still not the networking network frequency band after the firmware upgrade, it is determined that the networking module may be faulty. The method of externally connecting the networking module to the smart home appliance can be adopted to ensure that the smart home appliance can be controlled by voice recognition. Finally, all the qualified smart home appliances for networking are controlled to connect devices based on the networking network frequency band to obtain the smart home appliance networking control system.

[0070] Furthermore, in a preferred embodiment of the present invention, the control of the network-connected qualified smart home appliance performs audio preprocessing, audio segmentation processing and voiceprint analysis on the ambient audio to obtain the target human voice audio, specifically:

[0071] Obtain an audio recognition sensor of a qualified networked smart appliance, calibrate it as a target audio recognition sensor, obtain a manual of the qualified networked smart appliance, and based on the manual of the qualified networked smart appliance, obtain a maximum range of ambient audio collected by the qualified networked smart appliance, calibrate it as the maximum range of ambient audio collection;

[0072] Control the target audio recognition sensor to collect ambient audio in real time, collect ambient audio within the maximum range, and calibrate it as the target ambient audio;

[0073] Introducing an adaptive filter into a networked qualified smart home appliance, initializing filter parameters of the adaptive filter to obtain an initialized adaptive filter, and segmenting the target environment audio into short-time audio frames in the initialized adaptive filter;

[0074] In the initialized adaptive filter, the filter parameters of the adaptive filter are converged and trained based on the recursive least square method to obtain a trained adaptive filter, and all short-time audio frames are filtered based on the trained adaptive filter to obtain filtered short-time audio frames;

[0075] The filtered short-time audio frames of each frame are superimposed and reconstructed to obtain reconstructed audio, and the reconstructed audio is subjected to signal smoothing processing to obtain noise-reduced audio, which is calibrated as noise-reduced human voice audio.

[0076] It should be noted that there are sensors for collecting ambient audio in the qualified smart home appliances connected to the network, namely, target audio recognition sensors. Through the user manual of the smart home appliance, it can be known that the maximum range of the target audio recognition sensor to collect ambient audio, and the target audio recognition sensor collects ambient audio within the maximum range of ambient audio collection. After collecting the ambient audio, since the required audio is the voice emitted by a person, that is, human voice audio, it is necessary to pre-process the ambient audio, including audio noise reduction, denoising, etc., to obtain human voice audio. The pre-processing of the ambient audio can be performed by adaptive filtering to filter out the part of the ambient audio that does not belong to the human voice audio. Obtaining the initialized adaptive filter facilitates the training of the filtering parameters. The purpose of dividing the ambient audio into short-time audio frames is to improve the response speed and adaptability of the filter during the filtering process and improve the filtering effect. The filter parameters of the adaptive filter are trained by the recursive least squares method to converge, and the adaptive filter can be controlled to filter the audio other than the human voice audio, with a higher convergence speed. After filtering, the filtered short-time audio frame is obtained, and the filtered short-time audio frame is an audio frame of human voice audio. Finally, the filtered short-time audio frame is reconstructed to obtain the denoised audio signal. Since the reconstructed audio may have abnormal frequency amplitude or intermittent linearity, it is necessary to smooth the charged audio to reduce the above problems and obtain denoised human voice audio.

[0077] Further, in a preferred embodiment of the present invention, the denoised human voice audio is subjected to audio segmentation processing, the denoised human voice audio after audio segmentation is subjected to voiceprint analysis, and the target denoised human voice audio is obtained based on the voiceprint analysis result, specifically:

[0078] In the network-connected qualified smart home appliances, the voiceprint parameters of the person who can voice control the network-connected qualified smart home appliances are preset and calibrated as the target person's voiceprint parameters;

[0079] Introducing a GMM-UBM algorithm, constructing a GMM-UBM feature model based on the GMM-UBM algorithm, and converting the noise-reduced human voice audio into noise-reduced human voice audio feature parameters;

[0080] Importing the noise reduction human voice audio feature parameters into the GMM-UBM feature model, the GMM-UBM feature model performs character differentiation and audio segmentation on the noise reduction human voice audio feature parameters to obtain noise reduction sub-human voice audio;

[0081] Perform voiceprint analysis on all noise reduction sub-voice audios to obtain voiceprint parameters of the noise reduction sub-voice audios, calculate the Euclidean distance between all the noise reduction sub-voice audio voiceprint parameters and the target person's voiceprint parameters, and preset a first standard Euclidean distance interval;

[0082] If the Euclidean distance between the noise reduction sub-voice audio voiceprint parameters and the target person's voiceprint parameters is maintained within the first standard Euclidean distance interval, the corresponding noise reduction sub-voice audio voiceprint parameters are calibrated as the target noise reduction sub-voice audio voiceprint parameters, and the audio corresponding to the target noise reduction sub-voice audio voiceprint parameters is calibrated as the target noise reduction voice audio.

[0083] It should be noted that the denoised human voice audio may be mixed with voices from different people, and the smart home appliance may only be able to recognize the voice of the householder, so it is necessary to perform audio segmentation on the denoised human voice audio and divide it into audios emitted by different tasks, that is, denoised sub-human voice audio. The GMM-UBM feature model is a Gaussian mixture model-universal background model, which can distinguish different speaking characters according to the audio and is a speaker separation algorithm. The denoised human voice audio can be distinguished and segmented by the GMM-UBM feature model, and before distinguishing and segmenting the characters, the denoised human voice audio needs to be parameterized to facilitate the GMM-UBM feature model recognition training. After obtaining the denoised sub-human voice audio, since there may be many speakers, there may be many denoised sub-human voice audios, and it is necessary to perform voiceprint analysis on the audio and select the householder's audio from the many denoised sub-human voice audios, that is, the target denoised human voice audio. The voiceprint parameters of the target person are recorded in advance in the smart home appliance, so the voiceprint parameters of the noise reduction sub-voice audio are obtained, and the Euclidean distance between all noise reduction sub-voice audio voiceprint parameters and the target person's voiceprint parameters is calculated. The Euclidean distance reflects the similarity between the data. The smaller the Euclidean distance, the higher the similarity between the data. The Euclidean distance interval is preset. If the Euclidean distance between the noise reduction sub-voice audio voiceprint parameters and the target person's voiceprint parameters is within the Euclidean established interval, it proves that the corresponding noise reduction sub-voice audio voiceprint parameters are the target noise reduction voice audio.

[0084] Figure 2 A flow chart of a method for performing audio recognition on target noise-reduced human voice audio is shown, comprising the following steps:

[0085] S202: Convert the target noise-reduced human voice audio into text information to obtain the target noise-reduced human voice audio text information;

[0086] S204: Calculate the Euclidean distance between the target noise reduction human voice audio text information and the wake-up human voice audio text information, and wake up all networked qualified smart home appliances in the smart home appliance network control system;

[0087] S206: Perform text analysis on the target noise reduction human voice audio text information, and control all networked qualified smart home appliances to operate collaboratively based on the text analysis results.

[0088] Furthermore, in a preferred embodiment of the present invention, the target noise reduction human voice audio is converted into text information to obtain the target noise reduction human voice audio text information, specifically:

[0089] Presetting the wake-up human voice audio text information in the target audio recognition sensor, wherein the wake-up human voice audio text information is the text information that needs to be recognized to wake up the network-connected qualified smart home appliance and put the network-connected qualified smart home appliance into a working state;

[0090] Uploading the target noise-reduced human voice audio to a smart home appliance networking control system, and introducing a Viterbi algorithm into the smart home appliance networking control system, wherein the Viterbi algorithm can perform audio decoding on the target noise-reduced human voice audio and convert it into text information;

[0091] Based on the Viterbi algorithm, audio decoding is performed on the target denoised human voice audio to obtain text information of the target denoised human voice audio after parsing, and a text correction model is introduced to perform text correction processing on the text information of the target denoised human voice audio after parsing to obtain text information of the target denoised human voice audio;

[0092] In the smart home appliance network control system, text analysis is performed on the target noise reduction human voice audio text information, the Euclidean distance between the target noise reduction human voice audio text information and the wake-up human voice audio text information is calculated, and a second standard Euclidean distance interval is preset;

[0093] If the Euclidean distance between the target noise reduction human voice audio text information and the wake-up human voice audio text information is maintained within the second marked standard Euclidean distance interval, all networked qualified smart home appliances in the smart home appliance networking control system are woken up.

[0094] It should be noted that the target smart home appliance information needs to receive the wake-up command before it can enter the working state. If it enters the working state directly, it may cause the smart home appliance to enter a chaotic working state. For example, if a character is chatting and mentions starting the air conditioner during the chat, the air conditioner, a smart home appliance, will start after recognizing the audio, but the target character's original intention is not to start the air conditioner, so it is necessary to set the wake-up voice audio text information, and control the smart home appliance to enter the working state after recognizing the wake-up voice text information. Audio recognition of the target noise reduction human voice audio requires converting the audio into text information, the purpose is to more efficiently and accurately determine the instructions of the target task by distinguishing the text information. Therefore, it is necessary to introduce the Viterbi algorithm to perform audio decoding on the target noise reduction human voice audio to obtain the text information of the audio. After decoding, since there may be decoding anomalies in the decoding process, resulting in missing text information, punctuation anomalies, etc., it is necessary to perform text correction on the text information of the audio obtained after decoding. Through a text correction model, such as a punctuation reply model, etc., the text information of the audio can be corrected to obtain the target noise reduction human voice audio text information. The Viterbi algorithm is an audio decoding and text information conversion algorithm. After obtaining the target noise reduction human voice audio text information, it is necessary to determine whether there is a wake-up human voice audio text information in the target noise reduction human voice audio text information. If it exists, the smart home appliance networking control system can be controlled to wake up the qualified smart home appliances. The judgment method is to calculate the Euclidean distance between the target noise reduction human voice audio text information and the wake-up human voice audio text information. If the Euclidean distance is maintained within the standard range, all qualified smart home appliances will be awakened. Since all qualified smart home appliances are connected to the smart home appliance networking control system, after the wake-up human voice audio text information is recognized, all qualified smart home appliances will be controlled to enter the running state and wait for voice control.

[0095] Furthermore, in a preferred embodiment of the present invention, the text analysis of the target noise reduction human voice audio text information is performed, and all networked qualified smart home appliances are controlled to operate collaboratively based on the text analysis results, specifically:

[0096] Obtain all control instructions for controlling qualified networked smart home appliances in the smart home appliance network control system, mark them as home appliance control instructions, and obtain program codes of the home appliance control instructions, mark them as home appliance control instruction program codes;

[0097] Convert the home appliance control instruction program code into text information to obtain all text information that can control qualified networked smart home appliances and mark them as standard text information;

[0098] A big data network is introduced, and a text semantic comparison knowledge graph is retrieved based on the big data network, wherein the text semantic comparison knowledge graph describes all text information having the same semantics as the standard text information;

[0099] Based on the text semantic comparison knowledge graph, perform text semantic comparison on the standard text information and the denoised human voice audio text information to obtain the denoised human voice audio target text information;

[0100] The target text information of the noise reduction human voice audio is converted into a program code to obtain a target program code of the noise reduction human voice audio, and a target home appliance control instruction is obtained based on the target program code of the noise reduction human voice audio, wherein the target home appliance control instruction is a control instruction for the target noise reduction human voice audio to control the networked qualified smart home appliance;

[0101] When all the qualified smart home appliances in the smart home appliance networking control system are awakened, the smart home appliance networking control system controls the qualified smart home appliances to operate in a coordinated manner based on the target appliance control instructions.

[0102] It should be noted that in the smart home appliance networking control system, there are multiple home appliance control instructions, and different home appliance control instructions are output to smart home appliances, which have different control effects on smart home appliances. The home appliance control instructions are converted from text information, so the text information corresponding to the home appliance control instructions can be obtained through the program coding of the home appliance control instructions. For example, if the light in the living room needs to be turned on, the control instructions are divided into three parts: turn on, living room, and light. Turn on, living room, and light are text information. When the noise reduction human voice audio text information has these three text information, turn on, living room, and light, or similar text information, the smart home appliance networking control system will control the networked qualified smart home appliances to execute the corresponding command. Since the voice emitted by the person may be different from the standard, for example, the standard text information is light, and the voice emitted by the target person may be a light tube, in order to improve the recognition efficiency and optimize the user experience, the noise reduction human voice audio text information and the standard text information can be semantically compared to identify the text in the noise reduction human voice audio text information that has the same semantics as the standard text information. The semantic comparison between texts can be performed through the text semantic comparison knowledge graph, and all approximate and similar texts of the standard text information are recorded in the text semantic comparison knowledge graph. After the text semantic comparison, the noise reduction human voice audio target text information is obtained. The smart home appliance networking control system can analyze the noise reduction human voice audio target text information, convert it into program code, and generate target home appliance control instructions. In the smart home appliance networking control system, according to the target home appliance control instruction, all networked qualified smart home appliances are controlled to operate in coordination. Taking the lights as an example, there may be multiple lights in the living room. After the target person says to turn on the lights in the living room, the smart home appliance networking control system converts the audio into text information and converts it into the target home appliance control instruction, which will turn on all the lights in the living room. When the target person adds an attributive, such as turning on the chandelier in the living room, the light turning process will be accurately performed to realize the voice recognition control of smart home appliances.

[0103] like Figure 3 As shown, the second aspect of the present invention further provides a voice recognition control management system applicable to smart home appliances, the voice recognition control management system comprises a memory 31 and a processor 32, the memory 31 stores a voice recognition control management method, and when the voice recognition control management method is executed by the processor 32, the following steps are implemented:

[0104] Perform networking configuration and networking correction for all smart home appliances, and build a smart home appliance networking control system;

[0105] Controlling qualified networked smart home appliances to perform audio preprocessing, audio segmentation processing and voiceprint analysis on ambient audio to obtain target human voice audio;

[0106] Perform audio recognition on the target noise-reduced human voice audio, and based on the audio recognition results, control the networked qualified smart home appliances in the smart home appliance network control system to operate collaboratively.

[0107] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A voice recognition control management method suitable for smart home appliances, characterized in that: The following steps are involved: Perform networking configuration and networking correction for all smart home appliances, and build a smart home appliance networking control system; Controlling qualified networked smart home appliances to perform audio preprocessing, audio segmentation processing and voiceprint analysis on ambient audio to obtain target human voice audio; Perform audio recognition on the target noise-reduced human voice audio, and based on the audio recognition results, control the networked qualified smart home appliances in the smart home appliance network control system to operate in coordination; The networking configuration and networking correction of all smart home appliances to build a smart home appliance networking control system is specifically as follows: The smart home appliance that needs to be controlled by voice recognition is marked as the target smart home appliance, and the networking module on the target smart home appliance is obtained and marked as the target networking module; Acquire a network networking device, and based on the frequency band that the network networking device transmits to different target smart home appliances for network connection, mark it as a network frequency band; Controlling the target networking modules of all target smart home appliances to receive the networking network frequency band, and after the target networking modules of the target smart home appliances receive the networking network frequency band, performing network tests on the target networking modules of different target smart home appliances, and determining whether the current network frequency band of the target networking modules of any target smart home appliance is not the networking network frequency band; If not, the corresponding target smart home appliance is calibrated as a network-connected qualified smart home appliance; If so, the target networking module of the target smart home appliance whose current network frequency band is not the networking network frequency band is marked as an abnormal networking module, and a big data network is obtained, and the firmware update program of the abnormal networking module is retrieved and output in the big data network to obtain a corrected networking module; If the current network frequency band of the corrected networking module is still not the networking network frequency band, the target smart home appliance corresponding to the corrected networking module is marked as the abnormal networking smart home appliance, and after obtaining the receiving networking network frequency band, the external networking module whose current network frequency band is the real-time network frequency band is marked as the target external networking module; Connecting the target external networking module to the abnormally networked smart home appliance to obtain a networked qualified smart home appliance; Control all network-connected qualified smart home appliances to connect devices based on the network frequency band to obtain a smart home appliance network control system.

2. A voice recognition control management method applicable to smart home appliances according to claim 1, characterized in that: The control of the qualified networked smart home appliances to perform audio preprocessing, audio segmentation processing and voiceprint analysis on the ambient audio to obtain the target human voice audio is specifically as follows: Obtain an audio recognition sensor of a qualified networked smart appliance, calibrate it as a target audio recognition sensor, obtain a manual of the qualified networked smart appliance, and based on the manual of the qualified networked smart appliance, obtain a maximum range of ambient audio collected by the qualified networked smart appliance, calibrate it as the maximum range of ambient audio collection; Control the target audio recognition sensor to collect ambient audio in real time, collect ambient audio within the maximum range, and calibrate it as the target ambient audio; Introducing an adaptive filter into a networked qualified smart home appliance, initializing filter parameters of the adaptive filter to obtain an initialized adaptive filter, and segmenting the target environment audio into short-time audio frames in the initialized adaptive filter; In the initialized adaptive filter, the filter parameters of the adaptive filter are converged and trained based on the recursive least square method to obtain a trained adaptive filter, and all short-time audio frames are filtered based on the trained adaptive filter to obtain filtered short-time audio frames; The filtered short-time audio frames of each frame are superimposed and reconstructed to obtain reconstructed audio, and the reconstructed audio is subjected to signal smoothing processing to obtain denoised audio, which is calibrated as denoised human voice audio; The denoised human voice audio is segmented, and voiceprint analysis is performed on the denoised human voice audio after the audio segmentation. Based on the voiceprint analysis result, the target denoised human voice audio is obtained.

3. A voice recognition control management method applicable to smart home appliances according to claim 2, characterized in that: The audio segmentation processing is performed on the noise-reduced human voice audio, the voiceprint analysis is performed on the noise-reduced human voice audio after the audio segmentation, and the target noise-reduced human voice audio is obtained based on the voiceprint analysis result, specifically: In the network-connected qualified smart home appliances, the voiceprint parameters of the person who can voice control the network-connected qualified smart home appliances are preset and calibrated as the target person's voiceprint parameters; Introducing a GMM-UBM algorithm, constructing a GMM-UBM feature model based on the GMM-UBM algorithm, and converting the noise-reduced human voice audio into noise-reduced human voice audio feature parameters; Importing the noise reduction human voice audio feature parameters into the GMM-UBM feature model, the GMM-UBM feature model performs character differentiation and audio segmentation on the noise reduction human voice audio feature parameters to obtain noise reduction sub-human voice audio; Perform voiceprint analysis on all noise reduction sub-voice audios to obtain voiceprint parameters of the noise reduction sub-voice audios, calculate the Euclidean distance between all the noise reduction sub-voice audio voiceprint parameters and the target person's voiceprint parameters, and preset a first standard Euclidean distance interval; If the Euclidean distance between the noise reduction sub-voice audio voiceprint parameters and the target person's voiceprint parameters is maintained within the first standard Euclidean distance interval, the corresponding noise reduction sub-voice audio voiceprint parameters are calibrated as the target noise reduction sub-voice audio voiceprint parameters, and the audio corresponding to the target noise reduction sub-voice audio voiceprint parameters is calibrated as the target noise reduction voice audio.

4. According to the voice recognition control management method applicable to smart home appliances as described in claim 1, it is characterized in that: The audio recognition of the target noise reduction human voice audio is performed, and based on the audio recognition result, the qualified networked smart home appliances are controlled in the smart home appliance network control system to operate in a coordinated manner, specifically: Presetting the wake-up human voice audio text information in the target audio recognition sensor, wherein the wake-up human voice audio text information is the text information that needs to be recognized to wake up the network-connected qualified smart home appliance and put the network-connected qualified smart home appliance into a working state; Uploading the target noise-reduced human voice audio to a smart home appliance networking control system, and introducing a Viterbi algorithm into the smart home appliance networking control system, wherein the Viterbi algorithm can perform audio decoding on the target noise-reduced human voice audio and convert it into text information; Based on the Viterbi algorithm, audio decoding is performed on the target denoised human voice audio to obtain text information of the target denoised human voice audio after parsing, and a text correction model is introduced to perform text correction processing on the text information of the target denoised human voice audio after parsing to obtain text information of the target denoised human voice audio; In the smart home appliance network control system, text analysis is performed on the target noise reduction human voice audio text information, the Euclidean distance between the target noise reduction human voice audio text information and the wake-up human voice audio text information is calculated, and a second standard Euclidean distance interval is preset; If the Euclidean distance between the target noise reduction human voice audio text information and the wake-up human voice audio text information is maintained within the second marked standard Euclidean distance interval, all networked qualified smart home appliances in the smart home appliance network control system are woken up; When all the qualified smart home appliances in the smart home appliance networking control system are awakened, text analysis is performed on the target noise reduction human voice audio text information, and all the qualified smart home appliances are controlled to operate collaboratively based on the text analysis results.

5. A voice recognition control management method applicable to smart home appliances according to claim 4, characterized in that: The text analysis of the target noise reduction human voice audio text information is performed, and based on the text analysis result, all networked qualified smart home appliances are controlled to operate in coordination, specifically: Obtain all control instructions for controlling qualified networked smart home appliances in the smart home appliance network control system, mark them as home appliance control instructions, and obtain program codes of the home appliance control instructions, mark them as home appliance control instruction program codes; Convert the home appliance control instruction program code into text information to obtain all text information that can control qualified networked smart home appliances and mark them as standard text information; A big data network is introduced, and a text semantic comparison knowledge graph is retrieved based on the big data network, wherein the text semantic comparison knowledge graph describes all text information having the same semantics as the standard text information; Based on the text semantic comparison knowledge graph, perform text semantic comparison on the standard text information and the denoised human voice audio text information to obtain the denoised human voice audio target text information; The target text information of the noise reduction human voice audio is converted into a program code to obtain a target program code of the noise reduction human voice audio, and a target home appliance control instruction is obtained based on the target program code of the noise reduction human voice audio, wherein the target home appliance control instruction is a control instruction for the target noise reduction human voice audio to control a networked qualified smart home appliance; When all the qualified smart home appliances in the smart home appliance networking control system are awakened, the smart home appliance networking control system controls the qualified smart home appliances to operate in a coordinated manner based on the target appliance control instructions.

6. A voice recognition control management system suitable for smart home appliances, characterized in that: The speech recognition control management system comprises a memory and a processor, wherein a speech recognition control management method program is stored in the memory, and when the speech recognition control management method program is executed by the processor, the speech recognition control management method steps as described in any one of claims 1 to 5 are implemented: Perform networking configuration and networking correction for all smart home appliances, and build a smart home appliance networking control system; Controlling qualified networked smart home appliances to perform audio preprocessing, audio segmentation processing and voiceprint analysis on ambient audio to obtain target human voice audio; Perform audio recognition on the target noise-reduced human voice audio, and based on the audio recognition results, control the networked qualified smart home appliances in the smart home appliance network control system to operate in coordination; The networking configuration and networking correction of all smart home appliances to build a smart home appliance networking control system is specifically as follows: The smart home appliance that needs to be controlled by voice recognition is marked as the target smart home appliance, and the networking module on the target smart home appliance is obtained and marked as the target networking module; Acquire a network networking device, and based on the frequency band that the network networking device transmits to different target smart home appliances for network connection, mark it as a network frequency band; Controlling the target networking modules of all target smart home appliances to receive the networking network frequency band, and after the target networking modules of the target smart home appliances receive the networking network frequency band, performing network tests on the target networking modules of different target smart home appliances, and determining whether the current network frequency band of the target networking modules of any target smart home appliance is not the networking network frequency band; If not, the corresponding target smart home appliance is calibrated as a network-connected qualified smart home appliance; If so, the target networking module of the target smart home appliance whose current network frequency band is not the networking network frequency band is marked as an abnormal networking module, and a big data network is obtained, and the firmware update program of the abnormal networking module is retrieved and output in the big data network to obtain a corrected networking module; If the current network frequency band of the corrected networking module is still not the networking network frequency band, the target smart home appliance corresponding to the corrected networking module is marked as the abnormal networking smart home appliance, and after obtaining the receiving networking network frequency band, the external networking module whose current network frequency band is the real-time network frequency band is marked as the target external networking module; Connecting the target external networking module to the abnormally networked smart home appliance to obtain a networked qualified smart home appliance; Control all network-connected qualified smart home appliances to connect devices based on the network frequency band to obtain a smart home appliance network control system.

Images