Intelligent device voice control method, device, equipment and storage medium

Abstract

The present application discloses a method, device, equipment and storage medium for voice control of intelligent devices, the method comprising: receiving and responding to the user's voice command to obtain the user's semantic information; inputting the user's semantic information into a trained semantic control model to obtain target control information, the semantic control model is trained based on the user's behavior habit sentence sample and the sample control information corresponding to the behavior habit sentence sample; based on the target control information, determining the home appliance to be adjusted and the adjustment parameter information corresponding to the home appliance to be adjusted, so as to control the home appliance to be adjusted according to the adjustment parameter information. The technical solution can automatically parse the user's voice command, and obtain the semantic control model based on the user's life habit training, and obtain the target control information, so as to intelligently control and adjust the home appliance to be adjusted to meet the user's needs, which greatly improves the user experience.

Description

Technical Field

[0001] The present invention generally relates to the technical field of smart devices, and specifically relates to a method, device, equipment and storage medium for voice control of smart devices. Background Art

[0002] With the development of smart home technology and Internet of Things technology, users and manufacturers are paying more and more attention to the intelligent interaction of home appliances. Smart controlled home appliances have emerged, and it is more convenient to operate smart devices with voice commands during normal use. In order to improve the convenience of voice control, how to control smart devices by voice is particularly important.

[0003] At present, smart devices in related technologies receive signals and respond to demands through perception modules to achieve intelligent control. However, this solution requires the voice control system in interactive devices such as panels and smart speakers to passively respond to the user's voice needs to achieve control. It is unable to understand the intention behind the user's voice, which makes the voice control flexibility of smart devices poor, resulting in a poor user experience. Summary of the invention

[0004] In view of the above-mentioned defects or deficiencies in the prior art, it is desirable to provide a method, apparatus, device and storage medium for voice control of an intelligent device.

[0005] In a first aspect, an embodiment of the present application provides a method for voice control of an intelligent device, the method comprising:

[0006] Receive and respond to the user's voice command to obtain the user's semantic information;

[0007] Inputting the semantic information of the user into a trained semantic control model to obtain target control information, wherein the semantic control model is trained based on the user's behavior habit sentence samples and sample control information corresponding to the behavior habit sentence samples;

[0008] Based on the target control information, the household electrical appliance to be adjusted and the adjustment parameter information corresponding to the household electrical appliance to be adjusted are determined, so as to control the household electrical appliance to be adjusted according to the adjustment parameter information.

[0009] In one embodiment, the user's semantic information is input into a trained semantic control model to obtain target control information, including:

[0010] Vectorizing the user's semantic information through a vectorization processing module in the semantic control model to obtain a feature vector;

[0011] Processing the feature vector through a feature extraction module in the semantic control model to obtain a semantic feature;

[0012] The semantic features are processed by a recognition module in the semantic control model to obtain target control information.

[0013] In one embodiment, based on the target control information, determining the household electrical appliance to be adjusted and the adjustment parameter information corresponding to the household electrical appliance to be adjusted includes:

[0014] Acquire a preset household appliance information database and search the household appliance corresponding to the target control information from the household appliance information database;

[0015] The household appliance corresponding to the target control information is used as the household appliance to be adjusted;

[0016] The current parameter information of the household appliance to be adjusted is obtained, and based on the current parameter information and the target control information, the adjustment parameter information corresponding to the household appliance to be adjusted is determined.

[0017] In one embodiment, searching the household appliance information database for the household appliance corresponding to the target control information includes:

[0018] Obtaining attribute information of all home appliances connected to the smart device control system;

[0019] Comparing the target control information and the home appliance attribute information according to a preset comparison rule;

[0020] The home appliance whose attribute information meets the comparison rule is regarded as the home appliance corresponding to the target control information.

[0021] In one embodiment, after obtaining the semantic information of the user, the method further includes:

[0022] Obtaining the user profile and behavior habit statement parameters of the user;

[0023] Based on the user portrait and the behavior habit statement parameters, target control information is determined.

[0024] In one embodiment, the training process of the semantic control model includes:

[0025] Acquire a user's behavior habit sentence sample and sample control information corresponding to the user's behavior habit sentence sample;

[0026] Dividing the user's behavior habit sentence samples into a training set and a validation set;

[0027] Using the training set to train the semantic control model to be constructed, to obtain the semantic control model to be verified;

[0028] The semantic control model to be verified in the verification set is optimized according to the loss function minimization based on the sample control information to obtain a semantic control model.

[0029] In one embodiment, the method further comprises:

[0030] After controlling the household appliance to be adjusted to be adjusted according to the adjustment parameter information, receiving and responding to the evaluation instruction of the user, and obtaining the evaluation information of the user;

[0031] When the user's evaluation information is used to indicate that the user is satisfied with the adjusted mode, the intelligent control ends.

[0032] In a second aspect, the present application provides a smart device voice control device, the device comprising:

[0033] An acquisition module, used to receive and respond to a user's voice command to acquire the user's semantic information;

[0034] A control recognition module, used for inputting the semantic information of the user into a trained semantic control model to obtain target control information, wherein the semantic control model is trained based on the user's behavior habit sentence samples and sample control information corresponding to the behavior habit sentence samples;

[0035] The control module is used to determine the household electrical appliance to be adjusted and the adjustment parameter information corresponding to the household electrical appliance to be adjusted based on the target control information, so as to control the household electrical appliance to be adjusted according to the adjustment parameter information.

[0036] In a third aspect, an embodiment of the present application provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the smart device voice control method of the first aspect described above is implemented.

[0037] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program is used to implement the smart device voice control method as described in the first aspect above.

[0038] The smart device voice control method, device, equipment and storage medium provided in the embodiments of the present application, by receiving and responding to the user's voice command, obtain the user's voice information, and input the user's semantic information into a trained semantic control model to obtain target control information, the semantic control model is trained based on the user's behavior habit sentence sample and the sample control information corresponding to the behavior habit sentence sample, and then based on the target control information, determine the home appliance to be adjusted and the adjustment parameter information corresponding to the home appliance to be adjusted, so as to control the home appliance to be adjusted according to the adjustment parameter information. Compared with the prior art, the technical solution can automatically parse the user's voice command, and actively understand the user's living habits to train to obtain a semantic control model, and obtain target control information, so as to intelligently control and adjust the home appliance to be adjusted to meet the user's needs, realize the autonomous control stage according to the user's voice command, and provide customized and non-sensitive services for specific users in specific scenarios, which greatly improves the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings:

[0040] Figure 1 A system architecture diagram of an application system for voice control of smart devices provided in an embodiment of the present application;

[0041] Figure 2 A flowchart of a method for voice control of an intelligent device provided in an embodiment of the present application;

[0042] Figure 3 A schematic diagram of a flow chart of a method for determining a household appliance to be adjusted and corresponding adjustment parameter information provided in an embodiment of the present application;

[0043] Figure 4 A schematic diagram of the structure of the voice control of an intelligent device provided in an embodiment of the present application;

[0044] Figure 5 A flowchart of a method for training a semantic control model provided in an embodiment of the present application;

[0045] Figure 6 A schematic diagram of the structure of a voice control device for an intelligent device provided in an embodiment of the present application;

[0046] Figure 7 A schematic diagram of the structure of a voice control device for a smart device provided in another embodiment of the present application;

[0047] Figure 8 A schematic diagram of the structure of a computer device shown in an embodiment of the present application. DETAILED DESCRIPTION

[0048] The present application will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the relevant inventions, rather than to limit the inventions. It should also be noted that, for ease of description, only the parts related to the invention are shown in the accompanying drawings.

[0049] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail with reference to the accompanying drawings and in combination with the embodiments. For ease of understanding, some technical terms involved in the embodiments of the present application are explained below:

[0050] Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology in computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a similar way to human intelligence. Artificial intelligence is to study the design principles and implementation methods of various intelligent machines so that machines have the functions of perception, reasoning and decision-making.

[0051] Artificial intelligence technology is a comprehensive discipline that covers a wide range of fields, including both hardware-level and software-level technologies. The basic technologies of artificial intelligence generally include sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics and other technologies. Artificial intelligence software mainly includes computer vision, speech processing technology, natural language technology, and machine learning / deep learning.

[0052] Machine Learning (ML) is a multi-disciplinary subject that involves probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory and other disciplines. It specializes in studying how computers simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to make computers intelligent. Its applications are spread across all areas of artificial intelligence. Machine learning and deep learning usually include artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and self-learning.

[0053] It is understandable that, because smart devices are more convenient to operate with voice commands during normal use, voice control of smart devices has been increasingly used in people's daily lives. At present, smart devices in related technologies receive signals and respond to needs through perception modules to achieve intelligent control. However, this solution requires the voice control system in interactive devices such as panels and smart speakers to passively respond to the user's voice needs to achieve control, and cannot understand the intention behind the user's voice, making the voice control flexibility of smart devices poor, which in turn leads to poor user experience.

[0054] Based on the above-mentioned defects, the embodiments of the present application provide a method, device, equipment and storage medium for voice control of smart devices. Compared with the prior art, the technical solution can automatically parse the user's voice commands, actively understand the user's living habits and train to obtain a semantic control model, and obtain target control information, so as to intelligently control and adjust the household appliances to be adjusted to meet user needs, realize the autonomous control stage according to the user's voice commands, and provide customized and seamless services for specific users in specific scenarios, which greatly improves the user experience.

[0055] The solutions provided in the embodiments of the present application involve technologies such as natural language processing and machine learning of artificial intelligence, which are specifically illustrated by the following embodiments.

[0056] Figure 1 : is an implementation environment architecture diagram of a smart device voice control method provided in an embodiment of the present application. Figure 1 As shown, the implementation environment architecture includes: a smart home appliance 100 and a smart device control system 200. The smart device control system 200 is connected to the smart home appliance 100.

[0057] Optionally, there may be more than one smart home appliance 100. The smart home appliance 100 may be an air conditioner, a television, a refrigerator, a washing machine, a temperature sensor, an electric fan, a movie projector, a smart curtain control device, a smart speaker, a smart toilet, a smart door lock, and the like.

[0058] The above-mentioned smart device control system 200 may include a terminal, a server, and a microphone system. The microphone series includes multiple microphones and a microphone control center. The multiple microphones are distributed in various places of the home space corresponding to the smart device control system, and collect voice information in real time. The microphone control center is used to identify the voice information collected by multiple microphones in real time. Among them, AI applications based on natural language processing can be installed in the smart device control system 200. For example, the AI ​​application can be an application such as intelligent search and intelligent question and answer. The server can be a single server, or a server cluster composed of several servers, or the server 200 can include one or more virtualization platforms, or the server can be a cloud computing service center.

[0059] The smart device control system 200 and the smart home appliance 100 establish a communication connection through a wired or wireless network. Optionally, the wireless network or wired network uses standard communication technology and / or protocols. The network is usually the Internet, but it can also be any network, including but not limited to a local area network (LAN), a metropolitan area network (MAN), a wide area network (WAN), a mobile, wired or wireless network, a private network or any combination of a virtual private network.

[0060] The smart device control system 200 is used to collect the user's voice information, analyze it to obtain a recognition result, and then generate a control instruction according to the recognition result, and send the control instruction to the corresponding smart home appliance 100. The smart home appliance 100 is used to respond to the control instruction and perform the corresponding operation.

[0061] The semantic recognition method provided in the embodiments of the present application can be applied to different field scenarios such as smart homes, vehicle-mounted systems, and laboratories.

[0062] For ease of understanding and explanation, the following Figures 2 to 8 The smart device voice control method, apparatus, device and storage medium provided in the embodiments of the present application are described in detail.

[0063] Figure 2 The figure shows a flow chart of a method for voice control of an intelligent device according to an embodiment of the present application, which method can be executed by an intelligent device control system. Figure 2 As shown, the method includes:

[0064] S101, receiving and responding to a user's voice command to obtain the user's semantic information.

[0065] Specifically, the intelligent device control system includes multiple microphone series, each microphone system is used to collect the user's voice instructions, when the user needs to perform intelligent adjustment operations, he can speak to generate voice instructions, so that the intelligent device control system receives the user's voice instructions, and then responds to the user's voice instructions, and parses the user's voice instructions to obtain the user's semantic information. Among them, the user's voice instructions can include multiple characters or multiple words, or sentences or paragraphs composed of one or more words.

[0066] Optionally, when parsing the user's voice commands, a preset speech recognition algorithm can be used. For example, the voice commands may be preprocessed by first performing word segmentation on the voice commands and then removing repeated and redundant information, extra symbols or characters to obtain preprocessed text. Feature extraction is then performed on the preprocessed text, and recognition analysis is performed based on the information obtained through feature extraction to obtain semantic information of the user.

[0067] It should be noted that the above-mentioned speech recognition algorithm can be, for example, a model matching method, including vector quantization (VQ) and dynamic time warping (DTW); it can also be a probabilistic statistical method, including Gaussian mixture model (GMM) and hidden Markov model (HMM); it can also be a discriminator classification method, such as support vector machine (SVM), artificial neural network (ANN) and deep neural network (DNN), as well as a variety of combined algorithms.

[0068] Exemplarily, the user's voice command may be "turn on the air conditioner", "want to see the scenery", "the room is a bit cold", etc.

[0069] Optionally, after obtaining the user's semantic information, the user's user portrait and behavioral habit statement parameters may also be obtained, and based on the user portrait and behavioral habit statement parameters, the target control information may be determined.

[0070] It should be noted that the above user portrait refers to a labeled user model abstracted from information such as user demographic characteristics, behavioral operation data, online social activities and consumer behavior. It mainly uses the massive logs stored on the server and the large amount of data in the database for analysis and mining to "label" users, and "labels" are identifiers that can represent the characteristics of a certain dimension of the user. For specific label forms, please refer to the label that a website gives to one of the users. The above behavioral habit statement parameters may include the user's behavioral usage habits, such as the commonly used temperature of the air conditioner, the most commonly used lights, the lights that the user is accustomed to turning off, air purifiers, etc.

[0071] S102, inputting the user's semantic information into a trained semantic control model to obtain target control information, wherein the semantic control model is trained based on the user's behavior habit sentence samples and sample control information corresponding to the behavior habit sentence samples.

[0072] It should be noted that the above-mentioned semantic control model is a network structure model with semantic recognition and control capabilities learned by training on user behavior habit sentence samples. The scene recognition model takes the user's semantic information as input and outputs the target control information corresponding to the user's semantic information. It has the ability to semantically recognize and control the user's semantic information and is a neural network model that can predict semantic target control information. The semantic control model may include a multi-layer network structure, and the network structures at different layers process the input data differently and transmit the output results to the next network layer until the last network layer processes the data and obtains the output results.

[0073] The semantic control model may include: an input layer, a fusion layer, a character embedding layer, a convolutional layer, a fully connected layer, an output layer, etc. connected in series, and each layer has different functions.

[0074] Optionally, before inputting the user's semantic information into the trained semantic control model, the user's semantic information may be preprocessed, such as word segmentation, to determine the naming basis. The function of word segmentation is to effectively divide multiple characters in a sentence into one or more words. There are many specific ways of word segmentation, which can be based on mechanical matching, feature word library, constraint matrix, grammar analysis, etc. to determine one or more words.

[0075] After preprocessing the user's semantic information sample and obtaining the preprocessed result, it is then input into the trained semantic control model for recognition processing. Specifically, the user's semantic information can be vectorized through a vectorization processing module in the semantic control model to obtain a feature vector, and then the feature vector is processed through a feature extraction module in the semantic control model to obtain semantic features, and the semantic features are processed through a recognition module in the semantic control model to obtain target control information.

[0076] In an embodiment of the present application, the vectorization processing module is used to vectorize different word segments to obtain corresponding feature vectors, and then extract features from the feature vectors through a feature extraction module to obtain semantic features. The feature extraction module can convert abstract text into vectors operated by mathematical formulas, fully describing the character-level, word-level, sentence-level and even inter-sentence relationship features. The recognition module may include a fully connected layer and an activation function, which can classify the semantic features output by the feature extraction module to obtain target control information, which is used to characterize the control information required to control smart home appliances, and may include device identification and adjustment parameter information to be adjusted.

[0077] Among them, the vectorization processing module converts the semantic space relationship into a vector space relationship, that is, converts the semantic text into a vector that can be processed by a computer device.

[0078] Specifically, after obtaining the semantic information of the user and preprocessing it to obtain the characters or words of the semantic information of the user, the characters or words of the semantic information of the user can be input into the trained semantic control model, and the feature vector of the semantic information of the user can be obtained through the vectorization processing module, and then the feature extraction module is used to extract the features to obtain the semantic features, and the semantic features are processed through the fully connected layer in the recognition module to obtain the fully connected vector, and the fully connected vector is processed by the activation function to obtain the prediction result set corresponding to the semantic information of the user. The prediction result set includes multiple smart device identifications and corresponding adjustment parameter information label types, and the maximum value of the probability values ​​corresponding to the same label type in multiple prediction result sets is used as the device identification and adjustment parameter information to be adjusted. Optionally, the probability values ​​corresponding to the label types in the prediction result set can be sorted from large to small, and the maximum value is taken as the device identification and adjustment parameter information to be adjusted corresponding to the semantic information of the user.

[0079] Specifically, the recognition module may include but is not limited to a fully connected layer and an activation function. The fully connected layer may include one layer, or may also include multiple layers. The fully connected layer is mainly used to classify the user's semantic information.

[0080] Among them, the above activation function can be a softmax function. The role of the activation function is to add nonlinear factors. Because the linear model has insufficient expression ability, it can transform the input continuous real value into an output between 0 and 1.

[0081] S103: Based on the target control information, determine the home appliance to be adjusted and the adjustment parameter information corresponding to the home appliance to be adjusted, so as to control the home appliance to be adjusted according to the adjustment parameter information.

[0082] It should be noted that the intelligent device control system can obtain in advance the device identification and function parameters of all home appliances connected thereto, thereby building a home appliance information database.

[0083] Based on the above embodiment, a flowchart of a method for determining a household appliance to be adjusted and adjustment parameter information corresponding to the household appliance to be adjusted is also provided. Referring to FIG. 3 , the method includes:

[0084] S201: Obtain a preset household appliance information database and search for household appliances corresponding to target control information in the household appliance information database.

[0085] S202: The household electrical appliance corresponding to the target control information is used as the household electrical appliance to be adjusted.

[0086] Specifically, after obtaining the target control information, the attribute information of all household appliances connected to the smart device control system can be obtained, and then the target control information and the attribute information of the household appliances are compared and processed according to the preset comparison rules, and the household appliances whose attribute information meets the comparison rules are regarded as the household appliances corresponding to the target control information.

[0087] It should be noted that the home appliance attribute information may include the home appliance identification, functional parameters of the home appliance, usage methods of the home appliance, safety instructions and precautions of the home appliance, etc.

[0088] The above comparison rules can be customized according to actual needs. For example, the function parameters in the target control information can be obtained and the function parameters in the target control information can be compared with the attribute information of the home appliance.

[0089] S203: Acquire current parameter information of the household electrical appliance to be adjusted, and determine adjustment parameter information corresponding to the household electrical appliance to be adjusted based on the current parameter information and the target control information.

[0090] After the household appliance corresponding to the target control information is used as the household appliance to be adjusted, the current parameter information of the node device to be adjusted can be obtained. The current parameter information may include the functional parameters of the household appliance to be adjusted. Based on the current parameter information and the target control information, the adjustment parameter information corresponding to the household appliance to be adjusted is determined, for example, the current parameter information may be increased or decreased, or operations such as changing the mode, starting or stopping the device, etc.

[0091] For example, when the household appliance to be adjusted is an air conditioner, and the target control information obtained is to adjust the temperature to 26 degrees, the current parameter information of the air conditioner can be obtained to be the temperature is 28 degrees, then it can be determined that the adjustment parameter information corresponding to the air conditioner of the window household appliance to be adjusted is to lower the temperature by 2 degrees.

[0092] For further information, see Figure 4 As shown, it is possible to receive and respond to a user's voice command 3-1, parse the user's voice command 3-1, obtain the user's semantic information 3-2, input the user's semantic information 3-2 into a trained semantic control model 3-3, and obtain target control information 3-4; based on the target control information 3-4, determine the home appliance to be adjusted and the adjustment parameter information 3-5 corresponding to the home appliance to be adjusted, so as to control the home appliance to be adjusted according to the adjustment parameter information.

[0093] For example, when the user needs to control, he can say "the room is so cold", and then the intelligent device control system receives and responds to the user's voice command, obtains the user's semantic information "the room is cold", and inputs it into the trained semantic control model to obtain the target control information, which can be, for example, adjusting the temperature, and then based on the target control information, determine the home appliance to be adjusted and the adjustment parameter information corresponding to the home appliance to be adjusted, and control the home appliance to be adjusted according to the adjustment parameter information. For example, the home appliance to be adjusted is an air conditioner, and then according to the room temperature and the room equipment status, the room air conditioner temperature is automatically increased, or the heating in the room is turned on.

[0094] For another example, when the user's voice command is "turn on the light", although the user did not say which light to turn on or which light to turn on, the above method can be used to determine the target control information based on the user's previous living habits, and automatically check in the light that the user uses most often in the current room; when the user's voice command is "turn on the air conditioner", when the user says to turn on the air conditioner, the target control information will be determined, and then the air conditioner in the current room will be automatically turned on and adjusted to the user's favorite temperature or mode according to the target control information; when the user says that he wants to see the scenery, that is, the user's voice command is "I want to see a scenery", the target control information will be determined, and the curtains in the current room will be automatically opened; when the user When the voice command is "goodbye", the target control information will be determined, and then when the user leaves home, the lights, TV, air conditioner, etc. will be automatically turned off, the camera will be started, the gateway alert mode will be turned on, the sweeping robot will start working and respond according to the user's daily habits when going out; when the user's voice command is "I'm back", when the user returns home, the smart device control system will automatically turn on the user's most frequently used lights, turn on the air conditioner, and turn off the camera; when the user's voice command is "good night", that is, when the user is ready to rest, the lights will be turned off according to the user's habits, the curtains will be closed, the air conditioner will be set to the sleeping temperature, and the air purifier will enter the sleep mode.

[0095] The smart device control system of this embodiment can be personalized according to user behavior habits, actively understand user living habits, recommend related smart device operations, recommend related scenarios, and accumulate sensor perception data and train a semantic control model through deep learning to convert multi-type data into effective information, thereby achieving autonomous control and providing customized, seamless services for specific users in specific scenarios.

[0096] Furthermore, the intelligent device control system can also receive and respond to the user's evaluation instructions and obtain the user's evaluation information after controlling the home appliance to be adjusted according to the adjustment parameter information; when the user's evaluation information is used to indicate that the user is satisfied with the adjusted mode, the intelligent control ends. When the user's evaluation information is used to indicate that the user is not satisfied with the adjusted mode, the user's voice instructions are continued to be obtained, the user's semantic information is parsed, and the target control information is obtained through the trained semantic control model, and then the home appliance to be adjusted and the adjustment parameter information corresponding to the home appliance to be adjusted are determined according to the target control information, so as to control the home appliance to be adjusted according to the adjustment parameter information until the user's evaluation content is satisfactory.

[0097] The smart device voice control method provided in the embodiment of the present application receives and responds to the user's voice command, obtains the user's voice information, and inputs the user's semantic information into a trained semantic control model to obtain target control information. The semantic control model is trained based on the user's behavior habit sentence sample and the sample control information corresponding to the behavior habit sentence sample, and then based on the target control information, determines the home appliance to be adjusted and the adjustment parameter information corresponding to the home appliance to be adjusted, so as to control the home appliance to be adjusted according to the adjustment parameter information. Compared with the prior art, the technical solution can automatically parse the user's voice command, and actively understand the user's living habits to train to obtain a semantic control model, and obtain target control information, so as to intelligently control and adjust the home appliance to be adjusted to meet the user's needs, realize the autonomous control stage according to the user's voice command, and provide customized and non-sensitive services for specific users in specific scenarios, which greatly improves the user experience.

[0098] In one embodiment, the above embodiment includes a pre-trained semantic control model when determining the target control information. The following is a description of the training process of the semantic control model. Please refer to Figure 5 , the method may include:

[0099] S301. Obtain a user's behavior habit sentence sample and sample control information corresponding to the user's behavior habit sentence sample.

[0100] S302: Divide the user's behavior habit sentence samples into a training set and a validation set.

[0101] It should be noted that the above-mentioned user's behavior habit sentence sample may be multiple or one, wherein each behavior habit sentence sample may include at least one character or word. For example, the behavior habit sentence sample may include multiple characters or words, which may be some commonly used standard sentences extracted from a certain scene.

[0102] Specifically, after obtaining the user's behavioral habit sentence samples, the user's behavioral habit sentence samples can be randomly divided into a training set and a verification set according to a certain ratio, wherein the training set is used to train the initial semantic control model to obtain a trained semantic control model, and the verification set is used to verify the trained semantic control model to verify the performance of the semantic control model.

[0103] S303: Use the training set to train the semantic control model to be constructed to obtain the semantic control model to be verified.

[0104] S304 , using the semantic control model to be verified in the verification set, optimizing the semantic control model to be verified according to the minimization of the loss function based on the sample control information to obtain the semantic control model.

[0105] After dividing the user's behavior habit sentence samples into a training set and a verification set, the training set is input into the semantic control model to be constructed, which includes multiple connected vectorization processing modules, feature extraction modules and recognition modules. The training set can be first processed by the feature vectorization processing module to obtain an initial word vector, and the initial word vector is input into the feature extraction module in the semantic control model to be constructed to obtain the corresponding result, and the result is input into the recognition module to obtain the output result of the user's behavior habit sentence sample. The training set is used to train the vectorization processing module, feature extraction module and recognition module to be constructed to obtain the vectorization processing module, feature extraction module and recognition module to be verified.

[0106] In the process of training the semantic control model, the computer device uses the vectorization processing module, feature extraction module and recognition module to be verified in the verification set, and optimizes the vectorization processing module, feature extraction module and recognition module to be verified according to the minimization of the loss function to obtain the vectorization processing module, feature extraction module and recognition module. According to the difference between the result obtained in the semantic control model to be verified by inputting the verification set and the sample control information corresponding to the user's behavior habit statement sample, the parameters in the semantic control model to be constructed are updated to achieve the purpose of training the semantic control model, wherein the sample control information corresponding to the above-mentioned user's behavior habit statement sample can be the result obtained when the user's behavior habit statement sample is intelligently adjusted manually, and can include the device to be adjusted and the adjustment parameters corresponding to the user's behavior habit statement sample.

[0107] Optionally, the updating of the parameters in the semantic control model to be verified may be the updating of matrix parameters such as the weight matrix and the bias matrix in the semantic control model to be constructed. The weight matrix and the bias matrix include but are not limited to the matrix parameters in the self-attention layer, the feedforward network layer, and the fully connected layer in the semantic control model to be verified.

[0108] In the embodiment of the present application, a loss function can be used to calculate the loss value of the result and sample control information obtained in the semantic control model to be verified by inputting the verification set, so as to update the parameters in the semantic control model to be verified. Optionally, the loss function can use a cross entropy loss function, a normalized cross entropy loss function,

[0109] Among them, when the parameters in the semantic control model to be verified are updated by the loss function, when it is determined according to the loss function that the semantic control model to be verified has not converged, the parameters in the model are adjusted so that the semantic control model to be verified converges, thereby obtaining the semantic control model. The convergence of the semantic control to be verified may mean that the difference between the output result of the semantic control model to be verified for the verification set and the sample control information corresponding to the user's behavior habit sentence sample is less than a preset threshold, or the rate of change of the difference between the output result and the sample control information of the training data approaches a lower value. When the calculated loss function is small, or the difference between the loss function output from the previous round of iteration approaches 0, the semantic control model to be verified is considered to have converged.

[0110] In this embodiment, by training the semantic control model, the user's semantic information can be recognized and processed through the trained semantic control model, so that the target control information can be accurately obtained, thereby improving the semantic recognition efficiency and saving development costs.

[0111] on the other hand, Figure 6The structure diagram of a smart device voice control device provided in an embodiment of the present application is shown in FIG. The device may be a device in a terminal or a server, such as Figure 6 As shown, the device 700 includes:

[0112] The acquisition module 710 is used to receive and respond to the user's voice command to acquire the user's semantic information;

[0113] The control identification module 720 is used to input the user's semantic information into a trained semantic control model to obtain target control information. The semantic control model is trained based on the user's behavior habit sentence samples and the sample control information corresponding to the behavior habit sentence samples;

[0114] The control module 730 is used to determine the home appliance to be adjusted and the adjustment parameter information corresponding to the home appliance to be adjusted based on the target control information, so as to control the home appliance to be adjusted to perform adjustment according to the adjustment parameter information.

[0115] Optional, see Figure 7 As shown, the control identification module 720 includes:

[0116] A vectorization processing unit 721 is used to vectorize the user's semantic information through a vectorization processing module in the semantic control model to obtain a feature vector;

[0117] A feature extraction unit 722, used to process the feature vector through a feature extraction module in the semantic control model to obtain a semantic feature;

[0118] The recognition unit 723 is used to process the semantic features through the recognition module in the semantic control model to obtain the target control information.

[0119] Optionally, the control module 730 is specifically used to:

[0120] Acquire a preset home appliance information database and search the home appliance corresponding to the target control information in the home appliance information database;

[0121] The household appliance corresponding to the target control information is used as the household appliance to be adjusted;

[0122] The current parameter information of the household appliance to be adjusted is obtained, and based on the current parameter information and the target control information, the adjustment parameter information corresponding to the household appliance to be adjusted is determined.

[0123] Optionally, the control module 730 is further used to:

[0124] Obtain the property information of all home appliances connected to the smart device control system;

[0125] Compare the target control information and the home appliance attribute information according to the preset comparison rules;

[0126] The home appliance whose attribute information meets the comparison rule is regarded as the home appliance corresponding to the target control information.

[0127] Optionally, the above device is specifically used for:

[0128] Obtain the user's user profile and behavior habit statement parameters;

[0129] Determine target control information based on user portraits and behavioral habit statement parameters.

[0130] Optionally, the training process of the semantic control model includes:

[0131] Acquire a user's behavior habit sentence sample and sample control information corresponding to the user's behavior habit sentence sample;

[0132] Divide the user's behavior habit sentence samples into a training set and a validation set;

[0133] The semantic control model to be constructed is trained using the training set to obtain the semantic control model to be verified;

[0134] By using the semantic control model to be verified in the verification set, the semantic control model to be verified is optimized according to the minimization of the loss function based on the sample control information to obtain the semantic control model.

[0135] Optionally, the above device is also used for:

[0136] After controlling the home appliance to be adjusted according to the adjustment parameter information, receiving and responding to the user's evaluation instruction, and obtaining the user's evaluation information;

[0137] When the user's evaluation information is used to indicate that the user is satisfied with the adjusted mode, the intelligent control ends.

[0138] It can be understood that the functions of each functional module of the smart device voice control apparatus of this embodiment can be specifically implemented according to the method in the above method embodiment, and its specific implementation process can refer to the relevant description of the above method embodiment, which will not be repeated here.

[0139] To sum up, the smart device voice control device provided in the embodiment of the present application can automatically parse the user's voice commands, actively understand the user's living habits and train to obtain a semantic control model, and obtain target control information, so as to intelligently control and adjust the household appliances to be adjusted to meet user needs, and realize the autonomous control stage according to the user's voice commands, providing customized and seamless services for specific users in specific scenarios, thereby greatly improving the user experience.

[0140] On the other hand, the computer device provided in the embodiment of the present application includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the above-mentioned smart device voice control method is implemented.

[0141] Reference below Figure 8 , Figure 8 A schematic diagram of the structure of a computer system of a terminal device according to an embodiment of the present application.

[0142] like Figure 8 As shown, the computer system 300 includes a central processing unit (CPU) 301, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 302 or a program loaded from a storage part 303 to a random access memory (RAM) 303. In the RAM 303, various programs and data required for the operation of the system 300 are also stored. The CPU 301, the ROM 302, and the RAM 303 are connected to each other through a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0143] The following components are connected to the I / O interface 305: an input section 306 including a keyboard, a mouse, etc.; an output section 307 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 308 including a hard disk, etc.; and a communication section 309 including a network interface card such as a LAN card, a modem, etc. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. A removable medium 311, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 310 as needed, so that a computer program read therefrom is installed into the storage section 308 as needed.

[0144] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a machine-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 303, and / or installed from the removable medium 311. When the computer program is executed by the central processing unit (CPU) 301, the above-mentioned functions defined in the system of the present application are executed.

[0145] It should be noted that the computer-readable medium shown in the present application may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0146] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each box in the flow chart or block diagram can represent a module, a program segment or a part of the code, and the aforementioned module, program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0147] The units or modules involved in the embodiments of the present application may be implemented in software or hardware. The units or modules described may also be provided in a processor, for example, may be described as: a processor comprising: an acquisition module, a control identification module and a control module. Among them, the names of these units or modules do not constitute a limitation on the units or modules themselves in certain circumstances, for example, the acquisition module may also be described as "used to receive and respond to the user's voice instructions and acquire the user's semantic information".

[0148] As another aspect, the present application further provides a computer-readable storage medium, which may be included in the electronic device described in the above embodiment; or may exist independently without being assembled into the electronic device. The above computer-readable storage medium stores one or more programs, and when the above programs are used by one or more processors to execute the smart device voice control method described in the present application:

[0149] Receive and respond to the user's voice command to obtain the user's semantic information;

[0150] Inputting the semantic information of the user into a trained semantic control model to obtain target control information, wherein the semantic control model is trained based on the user's behavior habit sentence samples and sample control information corresponding to the behavior habit sentence samples;

[0151] Based on the target control information, the household electrical appliance to be adjusted and the adjustment parameter information corresponding to the household electrical appliance to be adjusted are determined, so as to control the household electrical appliance to be adjusted according to the adjustment parameter information.

[0152] The smart device voice control method, device, equipment and storage medium provided in the embodiments of the present application, by receiving and responding to the user's voice command, obtain the user's voice information, and input the user's semantic information into a trained semantic control model to obtain target control information, the semantic control model is trained based on the user's behavior habit sentence sample and the sample control information corresponding to the behavior habit sentence sample, and then based on the target control information, determine the home appliance to be adjusted and the adjustment parameter information corresponding to the home appliance to be adjusted, so as to control the home appliance to be adjusted according to the adjustment parameter information. Compared with the prior art, the technical solution can automatically parse the user's voice command, and actively understand the user's living habits to train to obtain a semantic control model, and obtain target control information, so as to intelligently control and adjust the home appliance to be adjusted to meet the user's needs, realize the autonomous control stage according to the user's voice command, and provide customized and non-sensitive services for specific users in specific scenarios, which greatly improves the user experience.

[0153] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the inventive concept. For example, the above features are replaced with (but not limited to) technical features with similar functions disclosed in the present application.

Claims

1. A method for voice control of an intelligent device, characterized in that: Applied to intelligent equipment control systems, including: Receive and respond to the user's voice command to obtain the user's semantic information; Inputting the semantic information of the user into a trained semantic control model to obtain a user portrait and behavior habit statement parameters of the user; obtaining target control information based on the user portrait and behavior habit statement parameters, wherein the semantic control model is trained based on the behavior habit statement samples of the user and the sample control information corresponding to the behavior habit statement samples; Based on the target control information, determining the household electrical appliance to be adjusted and the adjustment parameter information corresponding to the household electrical appliance to be adjusted, so as to control the household electrical appliance to be adjusted according to the adjustment parameter information; The smart device control system includes a microphone system, which includes multiple microphones and a microphone control center. The multiple microphones are distributed in various places of the home space corresponding to the smart device control system. The microphone system is used to collect user voice commands; Based on the target control information, determining the household electrical appliance to be adjusted and adjustment parameter information corresponding to the household electrical appliance to be adjusted includes: Acquire a preset household appliance information database and search the household appliance corresponding to the target control information from the household appliance information database; The home appliance corresponding to the target control information is used as the home appliance to be adjusted; wherein the home appliance to be adjusted is the home appliance in the current room where the microphone that collects the voice command is located; Acquire current parameter information of the household appliance to be adjusted, and determine adjustment parameter information corresponding to the household appliance to be adjusted based on the current parameter information and the target control information; Wherein, searching the household appliance information database for the household appliance corresponding to the target control information includes: Acquire attribute information of all home appliances connected to the smart device control system; the attribute information of home appliances includes home appliance identification, functional parameters of home appliances, usage methods of home appliances, safety instructions and precautions of home appliances; Comparing the target control information and the home appliance attribute information according to a preset comparison rule; The home appliance whose attribute information meets the comparison rule is regarded as the home appliance corresponding to the target control information.

2. The method according to claim 1, characterized in that Inputting the user's semantic information into the trained semantic control model to obtain target control information includes: Vectorizing the user's semantic information through a vectorization processing module in the semantic control model to obtain a feature vector; Processing the feature vector through a feature extraction module in the semantic control model to obtain a semantic feature; The semantic features are processed by a recognition module in the semantic control model to obtain target control information.

3. The method according to claim 1, characterized in that: The training process of the semantic control model includes: Acquire a user's behavior habit sentence sample and sample control information corresponding to the user's behavior habit sentence sample; Dividing the user's behavior habit sentence samples into a training set and a validation set; Using the training set to train the semantic control model to be constructed, to obtain the semantic control model to be verified; The semantic control model to be verified in the verification set is optimized according to the loss function minimization based on the sample control information to obtain a semantic control model.

4. The method according to claim 1, characterized in that: The method further comprises: After controlling the home appliance to be adjusted according to the adjustment parameter information, receiving and responding to the user's evaluation instruction, and obtaining the user's evaluation information; When the user's evaluation information is used to indicate that the user is satisfied with the adjusted mode, the intelligent control ends.

5. A voice control device for an intelligent device, characterized in that: The device comprises: An acquisition module, used to receive and respond to a user's voice command to acquire the user's semantic information; A control recognition module is used to input the semantic information of the user into a trained semantic control model to obtain a user portrait and behavior habit statement parameters of the user; based on the user portrait and behavior habit statement parameters, target control information is obtained, and the semantic control model is trained based on the behavior habit statement samples of the user and the sample control information corresponding to the behavior habit statement samples; A control module, configured to determine the household electrical appliance to be adjusted and adjustment parameter information corresponding to the household electrical appliance to be adjusted based on the target control information, so as to control the household electrical appliance to be adjusted according to the adjustment parameter information; The smart device control system includes a microphone system, which includes multiple microphones and a microphone control center. The multiple microphones are distributed in various places of the home space corresponding to the smart device control system. The microphone system is used to collect user voice commands; The control module is also used to: Acquire a preset household appliance information database and search the household appliance corresponding to the target control information from the household appliance information database; The home appliance corresponding to the target control information is used as the home appliance to be adjusted; wherein the home appliance to be adjusted is the home appliance in the current room where the microphone that collects the voice command is located; Acquire current parameter information of the household appliance to be adjusted, and determine adjustment parameter information corresponding to the household appliance to be adjusted based on the current parameter information and the target control information; Wherein, searching the household appliance information database for the household appliance corresponding to the target control information includes: Acquire attribute information of all home appliances connected to the smart device control system; the attribute information of home appliances includes home appliance identification, functional parameters of home appliances, usage methods of home appliances, safety instructions and precautions of home appliances; Comparing the target control information and the home appliance attribute information according to a preset comparison rule; The home appliance whose attribute information meets the comparison rule is regarded as the home appliance corresponding to the target control information.

6. A computer device, characterized in that: The computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor is configured to implement the smart device voice control method according to any one of claims 1 to 4 when executing the program.

7. A computer-readable storage medium having a computer program stored thereon, wherein the computer program is used to implement the smart device voice control method according to any one of claims 1 to 4.

Images