An intelligent voice home appliance control system

Abstract

An intelligent voice home appliance control system includes a voice acquisition terminal, a voice processing terminal, an intelligent home appliance control and feedback terminal, a system monitoring and management terminal, and a GUI interaction terminal. The voice acquisition terminal is used to collect external user voice signals, the voice processing terminal is used to recognize received external voice and parse the user's voice commands, the intelligent home appliance control and feedback terminal is used to control home appliances and provide feedback to the user on the execution results of voice commands, the system monitoring and management terminal is used to monitor the status of the home appliance control system and manage user information, and the GUI interaction terminal is used to provide a user interaction interface and meet the user's personalized settings. The present invention proposes an improved dynamic spectrum optimization and adaptive voice recognition algorithm for user voice recognition, and an improved dynamic coordinated voice home appliance control algorithm for intelligent control of home appliances, providing a more optimized solution for an intelligent voice home appliance control system.

Description

Technical Field

[0001] The present invention relates to the fields of speech recognition and intelligent control, and in particular to an intelligent speech home appliance control system. Background Art

[0002] Speech recognition technology is a technology that converts user voice commands into executable commands, including sampling and quantizing analog voice signals, converting them into digital signals for computer processing, extracting key features that aid recognition from voice signals, such as Mel-frequency cepstral coefficients (MFCCs), using statistical models and machine learning algorithms to improve the accuracy of speech recognition, and using neural networks (such as convolutional neural networks (CNNs) and recurrent neural networks (RNNs)) for acoustic modeling and language modeling to process and understand the rules of human language to improve the accuracy of language models. It simulates the acoustic characteristics of speech signals to map speech features to corresponding phonemes and words, predict the probability distribution of word sequences, and help identify the order of words in continuous speech. These background technologies together constitute the basis of speech recognition technology in intelligent voice home appliance control systems, enabling them to accurately understand and respond to user voice commands, thereby promoting the development of intelligent home appliance control systems.

[0003] Intelligent control technology is a series of technologies used to achieve intelligent management and control of home appliances, including control theory and algorithm optimization technology. Linear and nonlinear control theory and PID controllers can optimize control strategies in actual systems. At the same time, optimization algorithms such as genetic algorithms, particle swarm optimization and simulated annealing are used to adjust and optimize the parameters of intelligent home appliance control. In addition, the flexible use of photoelectric detectors and sensors can improve the performance of monitoring home appliance control systems. These background technologies provide strong support for intelligent voice home appliance control systems, enabling them to achieve advanced automated control, personalized services and energy efficiency optimization. Summary of the Invention

[0004] In response to the above problems, the present invention aims to provide an intelligent voice home appliance control system.

[0005] To achieve the above objectives, the present invention provides the following technical solutions: an intelligent voice home appliance control system, comprising a voice acquisition terminal, a voice processing terminal, an intelligent home appliance control and feedback terminal, a system monitoring and management terminal, and a GUI interaction terminal. The voice acquisition terminal includes a voice acquisition module, which is used to acquire external user voice signals. The voice processing terminal includes a voice recognition module and a voice understanding module. The voice recognition module proposes an improved dynamic spectrum optimization and adaptive voice recognition algorithm to accurately recognize the received user voice. The voice understanding module is used to parse the user's voice instructions. The intelligent home appliance control and feedback terminal includes an intelligent home appliance control module, a voice feedback module, and a user confirmation module. The intelligent home appliance control module proposes an improved dynamic coordinated voice home appliance control algorithm to accurately control intelligent home appliances. The voice feedback module is used to provide users with feedback on the execution results of voice instructions. The user confirmation module is used to provide users with verification of the accuracy and completeness of voice instructions. The system monitoring and management terminal includes a status monitoring module and a user management module. The status monitoring module is used to monitor the status of the home appliance control system in real time. The user management module is used to manage user information. The GUI interaction terminal includes a user interface module, which is used to provide a user interaction interface and meet the user's personalized settings.

[0006] Furthermore, the voice collection module captures external user voice signals from different directions and positions using a microphone array with multiple microphone devices.

[0007] Furthermore, the speech recognition module proposes to improve the dynamic spectrum optimization and adaptive speech recognition algorithm to convert the received user speech signal into a text signal for accurate recognition.

[0008] Furthermore, the dynamic spectrum optimization and adaptive speech recognition algorithm are improved as follows: first, the collected analog speech signal is converted into a digital speech signal, that is, Where D(m) is the digital voice signal at frequency index m in the frequency domain, m is the frequency component index in the frequency domain, s(k) is the analog voice signal at sampling point k of the original voice signal in the time domain, k is the sampling point index in the time domain, e (·) is an exponential function, K is the size of the sampling window, j is an imaginary unit, and then the digital speech signal is divided into multiple small segments. An improved Blackman window function is proposed to improve the windowing and framing process of the digital speech signal so that the signal spectrum has a lower sidelobe level, which can more clearly distinguish the signals of adjacent frequencies and further improve the performance of speech recognition. Among them, q i(t) is the value of the i-th signal sample at time point t after windowing, M is the length of the Blackman window, w[m] is the Blackman window function, D[mt] is the value of the digital speech signal at time point mt, a0 is the DC component of the Blackman window function, a1 is the decline rate control parameter of the Blackman window function, a2 is the edge smoothing parameter of the Blackman window function, α is the control parameter of the exponential decay, cos(·) is the cosine function, and then the processed speech signal is feature extracted. The correlation of the weighted input and output sequences is proposed to improve the feature extraction process to better distinguish the features of different speech, that is, Where μ is the mean of the signal sequence, P(t) is the speech feature output at time point t, f(X(t), Z(t)) is the input speech feature function, X(t) is the input signal sequence, Z(t) is the output signal sequence, n is the length of the input sequence, and w i is the weight coefficient of the correlation between the input and output sequences, and then a time adjustment coefficient is proposed to calculate the characteristic parameters of the speech signal to more flexibly adapt to the time changes of the speech signal and improve the expression ability of the feature, that is, Among them, FP(t) is the speech feature parameter output at time point t, λ is the time adjustment coefficient that controls the degree of influence between features at different time points, n is the length of the input sequence, and x i is the input sequence value of the i-th signal sample, z i is the value of the output sequence, and then a balance factor and a smoothing function are introduced to calculate the distance between the test template and the reference template to better handle speech templates of different lengths and reduce the distance calculation fluctuation caused by local alignment, that is, Among them, T k is the total distance between the test template and the reference template, K is the number of frames in the speech template, W(k) is the feature parameter vector of the test template in the kth frame, Y(x(k)) is the feature parameter vector of the reference template in the matched frame x(k), p[W(k),Y(x(k))] is the distance function between W(k) and Y(x(k)), δ is the balance factor for adjusting the effect of smoothing on the total distance, and d smooth(x(k), X(k)) is a smoothing function that reduces the fluctuation of distance calculation, x(k) is the kth frame in the reference template, and X(k) is the frame index in the reference template that is aligned with the kth frame of the test template. The improved dynamic spectrum optimization and adaptive speech recognition algorithm first proposes an improved Blackman window function for the windowing and framing process of the digital speech signal, so that the signal spectrum has a lower sidelobe level and can more clearly distinguish the signals of adjacent frequencies. Then, the correlation of the weighted input and output sequences is proposed to improve the feature extraction process to better distinguish the features of different speech. Finally, a balance factor and a smoothing function are introduced to calculate the distance between the test template and the reference template to better handle speech templates of different lengths and reduce the distance calculation fluctuation caused by local alignment, so as to accurately recognize the speech signal.

[0009] Furthermore, the speech understanding module converts the recognized speech text into structured commands that can be understood and executed by the home appliance control system through keyword matching.

[0010] Furthermore, the smart home appliance control module proposes to improve the dynamic coordination voice home appliance control algorithm to accurately control smart home appliances, directly interact with home appliances, and send control signals to execute control commands.

[0011] Furthermore, the improved dynamic coordination voice home appliance control algorithm is as follows: Assume that the voice command has l executable parts, that is, M = (m1, m2, ..., m l ,), where M is a set of executable commands, m1 is the first executable command, m2 is the second executable command, and m l is the lth executable command, l is the number of executable parts, and the improved home appliance control objective function is Among them, PR v is the actual state of the vth command, J is the improved home appliance control objective function, w u1 is the adjustment weight of the difference between the original state and the target state related to the u-th command, w u2 is the adjustment weight of the state change smoothness related to the u-th command, w uv To ensure the weight coefficients of the u-th command and the v-th command are coordinated, PT u is the target state of the u-th command, PR u is the actual state of the u-th command, ΔPR u is the actual state change of the u-th command. To facilitate the solution, the improved home appliance control objective function is converted into a Lagrangian function through the Lagrangian multiplier method, that is, Where L is the Lagrange function, λ is the Lagrange multiplier, P totalis the total power state of the home appliance control system, and then a fuzzy neural network is proposed to train the Lagrangian function, that is, h(t)=h(t-1)+K P [e(t)-e(t-1)]+K I e(t)+K D [e(t)-2e(t-1)+e(t-2)], where h(t) is the controller output at time t, h(t-1) is the controller output at time t-1, and K P is the proportional coefficient, e(t) is the error signal at time t, e(t-1) is the error signal at time t-1, e(t-2) is the error signal at time t-2, K I is the integral coefficient, K D is the differential coefficient. The output of the output layer is used to adjust the parameters of home appliance control, that is, the proportional gain △K P , integral gain △K I and differential gain △K D , the output layer relationship is Where O(r) is the output of the output layer of the fuzzy neural network, NN is the number of neurons in the hidden layer, g is the index of the neurons in the hidden layer, and W rg is the connection weight from the gth neuron in the hidden layer to the rth neuron in the output layer, O(r) is the output of the hidden layer, so the proportional gain △K P The adjustment process is Among them, O(△K P ) is the new proportional gain △K after adjustment by the fuzzy neural network P , is the number of neurons from the gth neuron in the hidden layer to △K P Output weight, integral gain △K I The adjustment process is O(△K I ) is the new integral gain △K after adjustment by the fuzzy neural network I , is the number of neurons from the gth neuron in the hidden layer to △K I Output weight, differential gain △K D The adjustment process is O(△K D ) is the new differential gain △K after adjustment by the fuzzy neural network D , is the number of neurons from the gth neuron in the hidden layer to △K DThe output weight is used to adjust the parameters of home appliance control and improve the dynamic coordination voice home appliance control algorithm. First, the home appliance control objective function is improved to consider the execution differences between commands, the balance and consistency between the execution of different commands, and the weights between different commands can be customized according to the relative importance of commands according to specific application requirements. Then, the home appliance control parameters are trained and adjusted through the fuzzy neural network to effectively process the home appliance control process and achieve more accurate control of the intelligent voice home appliance control behavior.

[0012] Furthermore, the voice feedback module provides the user with immediate confirmation information, informing the user that the command has been received by the system, and provides the user with feedback on the execution result of the voice command, as well as the current status and operation result of the home appliance.

[0013] Furthermore, after feeding back the execution result and current system status of the home appliance control system to the user, the user confirmation module allows the user to confirm whether the voice command issued is executed correctly.

[0014] Furthermore, the status monitoring module is used to monitor the status of the home appliance control system in real time. It is responsible for real-time and continuous monitoring of the operating status of home appliances, identifying and reporting abnormal conditions in the system, and recording the operating data and status changes of home appliances, providing data support for system analysis, fault diagnosis and performance improvement.

[0015] Furthermore, the user management module is used to manage user information, ensure that only authorized users can access and control the home appliance system, authenticate identities through authentication mechanisms such as usernames and passwords, and store and manage basic user information.

[0016] Furthermore, the user interface module is used to provide a user interaction interface, giving users a visual display of system status, device information and operation results. When voice recognition is inaccurate and the environment is noisy, the user interface can serve as a supplement to voice interaction to help users convey instructions more accurately. At the same time, it allows users to adjust the interface layout, theme and settings of the home appliance control system according to their personal preferences to meet the user's personalized needs.

[0017] Compared with the prior art, the present invention has the following beneficial effects:

[0018] 1. An improved dynamic spectrum optimization and adaptive speech recognition algorithm is proposed to convert the received user voice signal into a text signal for accurate recognition. The innovation of the present invention lies in that the improved dynamic spectrum optimization and adaptive speech recognition algorithm first proposes to improve the windowing and framing process of the digital voice signal by improving the Blackman window function so that the signal spectrum has a lower sidelobe level and can more clearly distinguish the signals of adjacent frequencies. Then, the correlation of the weighted input and output sequences is proposed to improve the feature extraction process to better distinguish the characteristics of different voices. Finally, a balancing factor and a smoothing function are introduced to calculate the distance between the test template and the reference template to better process voice templates of different lengths and reduce the distance calculation fluctuation caused by local alignment, so as to accurately recognize the voice signal.

[0019] 2. An improved dynamic coordination voice appliance control algorithm is proposed to precisely control smart home appliances. The innovation of the present invention lies in that the improved dynamic coordination voice appliance control algorithm first improves the appliance control objective function to take into account the execution differences between commands, the balance and consistency between the execution of different commands, and the weights between different commands can be customized according to the relative importance of commands based on specific application requirements. Then, the appliance control parameters are trained and adjusted through a fuzzy neural network to effectively process the appliance control process and achieve more accurate control of the smart voice appliance control behavior. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The invention is further illustrated by the accompanying drawings, but the embodiments in the accompanying drawings do not constitute any limitation to the invention. A person skilled in the art can obtain other drawings based on the following drawings without making any creative effort.

[0021] Figure 1 It is a structural schematic diagram of the present invention. DETAILED DESCRIPTION

[0022] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0023] An intelligent voice home appliance control system includes a voice collection end, a voice processing end, an intelligent home appliance control and feedback end, a system monitoring and management end, and a GUI interaction end. The voice collection end includes a voice collection module, which is used to collect external user voice signals. The voice processing end includes a voice recognition module and a voice understanding module. The voice recognition module proposes to improve the dynamic spectrum optimization and adaptive voice recognition algorithm to accurately recognize the received user voice. The voice understanding module is used to parse the user's voice instructions. The intelligent home appliance control and feedback end includes an intelligent home appliance control module, a voice feedback module, and a user confirmation module. The intelligent home appliance control module proposes to improve the dynamic coordination voice home appliance control algorithm to accurately control intelligent home appliances. The voice feedback module is used to feedback the execution results of the voice instructions to the user. The user confirmation module is used to provide the user with verification of the accuracy and completeness of the voice instructions. The system monitoring and management end includes a status monitoring module and a user management module. The status monitoring module is used to monitor the status of the home appliance control system in real time. The user management module is used to manage user information. The GUI interaction end includes a user interface module, which is used to provide a user interaction interface and meet the user's personalized settings.

[0024] Preferably, the voice collection module captures external user voice signals from different directions and positions using a microphone array with multiple microphone devices.

[0025] Preferably, the speech recognition module proposes to improve the dynamic spectrum optimization and adaptive speech recognition algorithm to convert the received user speech signal into a text signal for accurate recognition.

[0026] Specifically, the improved dynamic spectrum optimization and adaptive speech recognition algorithm is as follows: First, the collected analog speech signal is converted into a digital speech signal, that is, Where D(m) is the digital voice signal at frequency index m in the frequency domain, m is the frequency component index in the frequency domain, s(k) is the analog voice signal at sampling point k of the original voice signal in the time domain, k is the sampling point index in the time domain, e (·) is an exponential function, K is the size of the sampling window, j is an imaginary unit, and then the digital speech signal is divided into multiple small segments. An improved Blackman window function is proposed to improve the windowing and framing process of the digital speech signal so that the signal spectrum has a lower sidelobe level, which can more clearly distinguish the signals of adjacent frequencies and further improve the performance of speech recognition. Among them, q i(t) is the value of the i-th signal sample at time point t after windowing, M is the length of the Blackman window, w[m] is the Blackman window function, D[mt] is the value of the digital speech signal at time point mt, a0 is the DC component of the Blackman window function, a1 is the decline rate control parameter of the Blackman window function, a2 is the edge smoothing parameter of the Blackman window function, α is the control parameter of the exponential decay, cos(·) is the cosine function, and then the processed speech signal is feature extracted. The correlation of the weighted input and output sequences is proposed to improve the feature extraction process to better distinguish the features of different speech, that is, Where μ is the mean of the signal sequence, P(t) is the speech feature output at time point t, f(X(t), Z(t)) is the input speech feature function, X(t) is the input signal sequence, Z(t) is the output signal sequence, n is the length of the input sequence, and w i is the weight coefficient of the correlation between the input and output sequences, and then a time adjustment coefficient is proposed to calculate the characteristic parameters of the speech signal to more flexibly adapt to the time changes of the speech signal and improve the expression ability of the feature, that is, Among them, FP(t) is the speech feature parameter output at time point t, λ is the time adjustment coefficient that controls the degree of influence between features at different time points, n is the length of the input sequence, and x i is the input sequence value of the i-th signal sample, z i is the value of the output sequence, and then a balance factor and a smoothing function are introduced to calculate the distance between the test template and the reference template to better handle speech templates of different lengths and reduce the distance calculation fluctuation caused by local alignment, that is, Among them, T k is the total distance between the test template and the reference template, K is the number of frames in the speech template, W(k) is the feature parameter vector of the test template in the kth frame, Y(x(k)) is the feature parameter vector of the reference template in the matched frame x(k), p[W(k),Y(x(k))] is the distance function between W(k) and Y(x(k)), δ is the balance factor for adjusting the effect of smoothing on the total distance, and d smooth(x(k), X(k)) is a smoothing function that reduces the distance calculation fluctuations caused by local alignment, x(k) is the kth frame in the reference template, and X(k) is the frame index in the reference template that is aligned with the kth frame of the test template. The improved dynamic spectrum optimization and adaptive speech recognition algorithm first proposes an improved Blackman window function for the windowing and framing process of the digital speech signal, so that the signal spectrum has a lower sidelobe level and can more clearly distinguish the signals of adjacent frequencies. Then, the correlation of the weighted input and output sequences is proposed to improve the feature extraction process to better distinguish the features of different speech. Finally, a balancing factor and a smoothing function are introduced to calculate the distance between the test template and the reference template to better handle speech templates of different lengths and reduce the distance calculation fluctuations caused by local alignment, so as to accurately recognize the speech signal.

[0027] Preferably, the speech understanding module converts the recognized speech text into structured commands that can be understood and executed by the home appliance control system through keyword matching.

[0028] Preferably, the smart home appliance control module proposes to improve the dynamic coordination voice home appliance control algorithm to accurately control smart home appliances, directly interact with home appliances, and send control signals to execute control commands, such as turning on / off devices and adjusting settings (such as temperature, brightness, and volume).

[0029] Specifically, the improved dynamic coordination voice home appliance control algorithm is as follows: Assume that the voice command has l executable parts, that is, M = (m1, m2, ..., m l ,), where M is a set of executable commands, m1 is the first executable command, m2 is the second executable command, and m l is the lth executable command, l is the number of executable parts, and the improved home appliance control objective function is Among them, PR v is the actual state of the vth command, J is the improved home appliance control objective function, w u1 is the adjustment weight of the difference between the original state and the target state related to the u-th command, w u2 is the adjustment weight of the state change smoothness related to the u-th command, w uv To ensure the weight coefficients of the u-th command and the v-th command are coordinated, PT u is the target state of the u-th command, PR u is the actual state of the u-th command, ΔPR u is the actual state change of the u-th command. To facilitate the solution, the improved home appliance control objective function is converted into a Lagrangian function through the Lagrangian multiplier method, that is, Where L is the Lagrange function, λ is the Lagrange multiplier, P total is the total power state of the home appliance control system, and then a fuzzy neural network is proposed to train the Lagrangian function, that is, h(t)=h(t-1)+K P [e(t)-e(t-1)]+K I e(t)+K D [e(t)-2e(t-1)+e(t-2)], where h(t) is the controller output at time t, h(t-1) is the controller output at time t-1, and K P is the proportional coefficient, e(t) is the error signal at time t, e(t-1) is the error signal at time t-1, e(t-2) is the error signal at time t-2, K I is the integral coefficient, K D is the differential coefficient. The output of the output layer is used to adjust the parameters of home appliance control, that is, the proportional gain △K P , integral gain △K I and differential gain △K D , the output layer relationship is Where O(r) is the output of the output layer of the fuzzy neural network, NN is the number of neurons in the hidden layer, g is the index of the neurons in the hidden layer, and W rg is the connection weight from the gth neuron in the hidden layer to the rth neuron in the output layer, O(r) is the output of the hidden layer, so the proportional gain △K P The adjustment process is Among them, O(△K P ) is the new proportional gain △K after adjustment by the fuzzy neural network P , is the number of neurons from the gth neuron in the hidden layer to △K P Output weight, integral gain △K I The adjustment process is O(△K I ) is the new integral gain △K after adjustment by the fuzzy neural network I , is the number of neurons from the gth neuron in the hidden layer to △K I Output weight, differential gain △K D The adjustment process is O(△K D ) is the new differential gain △K after adjustment by the fuzzy neural network D , is the number of neurons from the gth neuron in the hidden layer to △K DThe output weight is used to adjust the parameters of home appliance control and improve the dynamic coordination voice home appliance control algorithm. First, the home appliance control objective function is improved to consider the execution differences between commands, the balance and consistency between the execution of different commands, and the weights between different commands can be customized according to the relative importance of commands according to specific application requirements. Then, the home appliance control parameters are trained and adjusted through the fuzzy neural network to effectively process the home appliance control process and achieve more accurate control of the intelligent voice home appliance control behavior.

[0030] Preferably, the voice feedback module provides the user with immediate confirmation information, informing the user that his or her command has been received by the system, and provides the user with feedback on the execution results of the voice command, as well as the current status and operation results of the home appliance. If the voice recognition and understanding module fails to correctly parse and execute the user's command, the voice feedback module will notify the user and request a retry.

[0031] Preferably, after feeding back the execution result and current system status of the home appliance control system to the user, the user confirmation module allows the user to confirm whether the voice command issued is executed correctly.

[0032] Preferably, the status monitoring module is used to monitor the status of the home appliance control system in real time. It is responsible for real-time and continuous monitoring of the operating status of home appliances, identifying and reporting abnormal conditions in the system, helping the system optimize energy use, monitoring the energy consumption of home appliances, and recording the operating data and status changes of home appliances, providing data support for system analysis, fault diagnosis and performance improvement.

[0033] Preferably, the user management module is used to manage user information, ensure that only authorized users can access and control the home appliance system, authenticate users through authentication mechanisms such as username and password (such as fingerprint, facial recognition), store and manage users' basic information, monitor and record users' activities, and ensure the safe use of the system.

[0034] Preferably, the user interface module is used to provide a user interaction interface, providing users with a visual display of system status, device information and operation results. When voice recognition is inaccurate and the environment is noisy, the user interface can serve as a supplement to voice interaction to help users convey instructions more accurately. At the same time, it allows users to adjust the interface layout, theme and settings of the home appliance control system according to personal preferences to meet the user's personalized needs. For data such as energy consumption and frequency of use, the user interface can provide visual displays such as charts to help users better understand usage.

[0035] An intelligent voice home appliance control system is proposed, which is used to control and manage home appliances through voice commands. An intelligent voice home appliance control system is provided by integrating a voice acquisition end, a voice processing end, an intelligent home appliance control and feedback end, a system monitoring and management end, and a GUI interaction end. An improved dynamic spectrum optimization and adaptive speech recognition algorithm is proposed to convert the received user voice signal into a text signal for accurate recognition. The innovation of the present invention lies in that the improved dynamic spectrum optimization and adaptive speech recognition algorithm first proposes an improved Blackman window function for the windowing and framing process of the digital voice signal so that the signal spectrum has a lower sidelobe level and can more clearly distinguish signals of adjacent frequencies. Then, the correlation of weighted input and output sequences is proposed to improve the feature extraction process to better distinguish the features of different voices. Finally, a balance factor and a smoothing function are introduced to calculate the distance between the test template and the reference template to better handle voice templates of different lengths and reduce the distance calculation caused by local alignment. Fluctuations are detected in order to accurately identify voice signals, and an improved dynamic coordination voice appliance control algorithm is proposed to accurately control smart home appliances. The innovation of the present invention lies in that the improved dynamic coordination voice appliance control algorithm first improves the home appliance control objective function to consider the execution differences between commands, the balance and consistency between the execution of different commands, and the weights between different commands can be customized according to the relative importance of commands according to specific application requirements. Then, the home appliance control parameters are trained and adjusted through the fuzzy neural network to effectively process the home appliance control process and achieve more accurate control of the intelligent voice home appliance control behavior, effectively improve the working effect of an intelligent voice home appliance control system, provide more comprehensive and accurate technical support for the home appliance control system, and provide better decision support for scientific and efficient home appliance control systems. At the same time, the present invention relates to voice recognition and intelligent control technology, provides people with an efficient and convenient home appliance control system, and contributes important application value to the field of voice recognition and intelligent control.

[0036] Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. An intelligent voice home appliance control system, characterized in that: It includes a voice collection end, a voice processing end, a smart home appliance control and feedback end, a system monitoring and management end and a GUI interaction end. The voice collection end includes a voice collection module, which is used to collect external user voice signals. The voice processing end includes a voice recognition module and a voice understanding module. The voice recognition module proposes to improve the dynamic spectrum optimization and adaptive voice recognition algorithm to accurately recognize the received user voice. The voice understanding module is used to parse the user's voice instructions. The smart home appliance control and feedback end includes a smart home appliance control module, a voice feedback module and a user confirmation module. The smart home appliance control module proposes to improve the dynamic coordination voice home appliance control algorithm to accurately control smart home appliances. The voice feedback module is used to feedback the execution results of the voice instructions to the user. The user confirmation module is used to provide users with verification of the accuracy and completeness of the voice instructions. The system monitoring and management end includes a status monitoring module and a user management module. The status monitoring module is used to monitor the status of the home appliance control system in real time. The user management module is used to manage user information. The GUI interaction end includes a user interface module, which is used to provide a user interaction interface and meet the user's personalized settings. The speech recognition module proposes an improved dynamic spectrum optimization and adaptive speech recognition algorithm to convert the received user speech signal into a text signal for accurate recognition; The improved dynamic spectrum optimization and adaptive speech recognition algorithm is as follows: First, the collected analog speech signal is converted into a digital speech signal, that is, 0≤m≤K-1, where D(m) is the digital voice signal at frequency index m in the frequency domain, m is the frequency component index in the frequency domain, s(k) is the analog voice signal at sampling point k of the original voice signal in the time domain, k is the sampling point index in the time domain, and e (·) is an exponential function, K is the size of the sampling window, j is an imaginary unit, and then the digital speech signal is divided into multiple small segments. An improved Blackman window function is proposed to improve the windowing and framing process of the digital speech signal so that the signal spectrum has a lower sidelobe level, which can more clearly distinguish the signals of adjacent frequencies and further improve the performance of speech recognition. Among them, q i (t) is the value of the i-th signal sample at time point t after windowing, M is the length of the Blackman window, w[m] is the Blackman window function, D[mt] is the value of the digital speech signal at time point mt, a0 is the DC component of the Blackman window function, a1 is the decline rate control parameter of the Blackman window function, a2 is the edge smoothing parameter of the Blackman window function, α is the control parameter of the exponential decay, cos(·) is the cosine function, and then the processed speech signal is feature extracted. The correlation of the weighted input and output sequences is proposed to improve the feature extraction process to better distinguish the features of different speech, that is, Where μ is the mean of the signal sequence, P(t) is the speech feature output at time point t, f(X(t), Z(t)) is the input speech feature function, X(t) is the input signal sequence, Z(t) is the output signal sequence, n is the length of the input sequence, and w i is the weight coefficient of the correlation between the input and output sequences, and then a time adjustment coefficient is proposed to calculate the characteristic parameters of the speech signal to more flexibly adapt to the time changes of the speech signal and improve the expression ability of the feature, that is, Among them, FP(t) is the speech feature parameter output at time point t, λ is the time adjustment coefficient that controls the degree of influence between features at different time points, n is the length of the input sequence, and x i is the input sequence value of the i-th signal sample, z i is the value of the output sequence, and then a balance factor and a smoothing function are introduced to calculate the distance between the test template and the reference template to better handle speech templates of different lengths and reduce the distance calculation fluctuation caused by local alignment, that is, Among them, T k is the total distance between the test template and the reference template, K is the number of frames in the speech template, W(k) is the feature parameter vector of the test template in the kth frame, Y(x(k)) is the feature parameter vector of the reference template in the matched frame x(k), p[W(k),Y(x(k))] is the distance function between W(k) and Y(x(k)), δ is the balance factor for adjusting the effect of smoothing on the total distance, and d smooth (x(k), X(k)) is a smoothing function that reduces the fluctuation of distance calculation, x(k) is the kth frame in the reference template, and X(k) is the frame index in the reference template that is aligned with the kth frame of the test template. The improved dynamic spectrum optimization and adaptive speech recognition algorithm first proposes an improved Blackman window function for the windowing and framing process of the digital speech signal, so that the signal spectrum has a lower sidelobe level and can more clearly distinguish the signals of adjacent frequencies. Then, the correlation of the weighted input and output sequences is proposed to improve the feature extraction process to better distinguish the features of different speech. Finally, a balance factor and a smoothing function are introduced to calculate the distance between the test template and the reference template to better handle speech templates of different lengths and reduce the distance calculation fluctuation caused by local alignment, so as to accurately recognize the speech signal.

2. The intelligent voice home appliance control system according to claim 1, characterized in that: The voice acquisition module captures external user voice signals from different directions and positions through a microphone array with multiple microphone devices.

3. The intelligent voice home appliance control system according to claim 1, characterized in that: The speech understanding module converts the recognized speech text into structured commands that can be understood and executed by the home appliance control system through keyword matching.

4. The intelligent voice home appliance control system according to claim 1, characterized in that: The smart home appliance control module proposes to improve the dynamic coordination voice home appliance control algorithm to accurately control smart home appliances, directly interact with home appliances, and send control signals to execute control commands.

5. The intelligent voice home appliance control system according to claim 4, characterized in that: The improved dynamic coordination voice home appliance control algorithm is as follows: Assume that the voice command has l executable parts, that is, M = (m1, m2, ..., m l ,), where M is a set of executable commands, m1 is the first executable command, m2 is the second executable command, and m l is the lth executable command, l is the number of executable parts, and the improved home appliance control objective function is Among them, J is the improved home appliance control objective function, w u1 is the adjustment weight of the difference between the original state and the target state related to the u-th command, w u2 is the adjustment weight of the state change smoothness related to the u-th command, w uv To ensure the weight coefficients of the u-th command and the v-th command are coordinated, PT u is the target state of the u-th command, PR u is the actual status of the u-th command, PR v is the actual state of the vth command, ΔPR u is the actual state change of the u-th command. To facilitate the solution, the improved home appliance control objective function is converted into a Lagrangian function through the Lagrangian multiplier method, that is, Where L is the Lagrange function, λ is the Lagrange multiplier, P total is the total power state of the home appliance control system, and then a fuzzy neural network is proposed to train the Lagrangian function, that is, h(t)=h(t-1)+K P [e(t)-e(t-1)]+K I e(t)+K D [e(t)-2e(t-1)+e(t-2)], where h(t) is the controller output at time t, h(t-1) is the controller output at time t-1, and K P is the proportional coefficient, e(t) is the error signal at time t, e(t-1) is the error signal at time t-1, e(t-2) is the error signal at time t-2, K I is the integral coefficient, K D is the differential coefficient. The output of the output layer is used to adjust the parameters of home appliance control, that is, the proportional gain △K P , integral gain △K I and differential gain △K D , the output layer relationship is Where O(r) is the output of the output layer of the fuzzy neural network, NN is the number of neurons in the hidden layer, g is the index of the neurons in the hidden layer, and W rg is the connection weight from the gth neuron in the hidden layer to the rth neuron in the output layer, O(r) is the output of the hidden layer, so the proportional gain △K P The adjustment process is Among them, O(△K P ) is the new proportional gain after adjustment by the fuzzy neural network is the number of neurons from the gth neuron in the hidden layer to △K P Output weight, integral gain △K I The adjustment process is O(△K I ) is the new integral gain △K after adjustment by the fuzzy neural network I , is the number of neurons from the gth neuron in the hidden layer to △K I Output weight, differential gain △K D The adjustment process is is the new differential gain △K after adjustment by the fuzzy neural network D , is the number of neurons from the gth neuron in the hidden layer to △K D The output weight is used to adjust the parameters of home appliance control and improve the dynamic coordination voice home appliance control algorithm. First, the home appliance control objective function is improved to consider the execution differences between commands, the balance and consistency between the execution of different commands, and the weights between different commands can be customized according to the relative importance of commands according to specific application requirements. Then, the home appliance control parameters are trained and adjusted through the fuzzy neural network to effectively process the home appliance control process and achieve more accurate control of the intelligent voice home appliance control behavior.

6. The intelligent voice home appliance control system according to claim 1, characterized in that: The voice feedback module provides the user with immediate confirmation information, informing the user that their command has been received by the system, and provides the user with feedback on the execution results of the voice command, as well as the current status and operation results of the home appliance.

7. The intelligent voice home appliance control system according to claim 1, characterized in that: After feeding back the execution result and current system status of the home appliance control system to the user, the user confirmation module allows the user to confirm whether the voice command issued is executed correctly.

8. The intelligent voice home appliance control system according to claim 1, characterized in that: The status monitoring module is used to monitor the status of the home appliance control system in real time. It is responsible for real-time and continuous monitoring of the operating status of home appliances, identifying and reporting abnormal conditions in the system, and recording the operating data and status changes of home appliances, providing data support for system analysis, fault diagnosis and performance improvement.

Images