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

Abstract

The invention relates to the technical field of computers, and discloses an intelligent equipment control method and device, computer equipment and a storage medium, at least one piece of audio data acquired by intelligent equipment is converted into a corresponding ASR text set, the ASR text set comprises at least one piece of ASR text corresponding to the at least one piece of audio data, and the accuracy of audio recognition is improved; on the basis, the at least one ASR text is respectively input into the preset filter to obtain the total intention information, so that the accuracy of speech recognition is improved, the calculation power occupation caused by predicting and recognizing a large number of ASR texts is avoided, and the recognition efficiency is improved; and finally, according to the total intention information, controlling the intelligent device to execute the instruction corresponding to the total intention information, thereby effectively reducing the calculation load of uninterruptedly obtaining the audio data, and improving the interaction speed of the intelligent device.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a smart device control method, device, computer equipment and storage medium. Background Art

[0002] Large language models have made very obvious progress in the field of human-computer interaction. With the widespread application of large language models, portable assistant scenarios have become more mature.

[0003] In the existing technology, the portable assistant needs to collect audio data for a long time and convert the audio data into ASR text. The ASR text is sent to the large model for semantic analysis and understanding to determine the user's intention and then complete the specific interaction task.

[0004] However, in the prior art, long-term collection of audio data will generate a large number of ASR results, resulting in the redundancy of a large amount of invalid data. Summary of the invention

[0005] In view of this, the present invention provides a smart device control method, device, computer device and storage medium to solve the problem of data redundancy caused by long-term collection of audio data by existing smart devices.

[0006] In a first aspect, the present invention provides a method for controlling an intelligent device, the method comprising:

[0007] Converting at least one piece of audio data acquired by the smart device into a corresponding ASR text set, wherein the ASR text set includes at least one piece of ASR text corresponding to the at least one piece of audio data;

[0008] Inputting at least one ASR text into preset filters respectively to obtain total intent information;

[0009] According to the overall intention information, the smart device is controlled to execute instructions corresponding to the overall intention information.

[0010] Beneficial effect: at least one audio data acquired by the smart device is converted into a corresponding ASR text set, wherein the ASR text set includes at least one ASR text corresponding to the at least one audio data, thereby improving the accuracy of audio recognition; on this basis, at least one ASR text is input into a preset filter respectively to obtain the overall intent information, thereby improving the accuracy of speech recognition, avoiding the computing power occupation caused by predicting and recognizing a large amount of ASR text, and improving the recognition efficiency; finally, according to the overall intent information, the smart device is controlled to execute the instructions corresponding to the overall intent information, thereby effectively reducing the computing load of continuously acquiring audio data and improving the interaction speed of the smart device.

[0011] In an optional implementation, the preset filter includes a classification model and a filter, and at least one ASR text is input into the preset filter to obtain the total intent information, which specifically includes:

[0012] Inputting a first ASR text into a classification model to determine a first intent classification, the first ASR text being any one of the at least one ASR text;

[0013] Determine the filter according to the first intent classification and the mapping relationship between the first intent classification and the filter;

[0014] Inputting the first ASR text into the filter to obtain first intent information;

[0015] The total intent information is determined according to the first intent information corresponding to each ASR text.

[0016] In an optional implementation, the filter includes a vertical model filter and a rule filter, the vertical model filter corresponds to the type information of the smart device, and the first ASR text is input into the filter to obtain the first intent information, which specifically includes:

[0017] Input the first ASR text into the vertical class small model filter to obtain the first initial intent sub-information;

[0018] When the first initial intent sub-information matches the preset output format of the smart device, determining the first initial intent sub-information as the first intent sub-information;

[0019] Inputting the first ASR text into a preset rule filter to obtain second intention sub-information;

[0020] The first intent information is determined according to the first intent sub-information, the second intent sub-information and a preset standard.

[0021] In an optional embodiment, the method further includes:

[0022] When the first initial intent sub-information does not match the preset output format of the smart device, the first ASR text is input into the preset large model for prediction to obtain the first intent sub-information.

[0023] In an optional implementation, determining the total intent information according to the first intent information corresponding to each ASR text specifically includes:

[0024] Filtering a subset of first intent information that matches the instruction rule corresponding to the smart device from all first intent information;

[0025] Total intent information is determined based on the first intent information subset.

[0026] In a second aspect, the present invention provides a smart device control device, the device comprising:

[0027] An audio conversion module, used to convert at least one audio data acquired by the smart device into a corresponding ASR text set, where the ASR text set includes at least one ASR text corresponding to the at least one audio data;

[0028] A filtering module, used for inputting at least one ASR text into a preset filter to obtain total intent information;

[0029] The execution instruction module is used to control the smart device to execute the instructions corresponding to the general intention information according to the general intention information.

[0030] In an optional implementation, the preset filter includes a classification model and a filter, and the filtering module specifically includes:

[0031] an intention classification unit, configured to input a first ASR text into a classification model to determine a first intention classification, wherein the first ASR text is any one of the at least one ASR text;

[0032] A filter determination unit, configured to determine a filter according to the first intent classification and a mapping relationship between the first intent classification and the filter;

[0033] A filtering unit, configured to input the first ASR text into a filter to obtain first intent information;

[0034] The intention determination unit is used to determine the total intention information according to the first intention information corresponding to each ASR text.

[0035] In a third aspect, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the intelligent device control method of the first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0036] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the intelligent device control method of the first aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0038] Figure 1 is a flow chart of a smart device control method according to an embodiment of the present invention;

[0039] Figure 2 is a flow chart of another intelligent device control method according to an embodiment of the present invention;

[0040] Figure 3 is a schematic diagram of a smart device control method according to an embodiment of the present invention;

[0041] Figure 4 is a structural block diagram of an intelligent device control device according to an embodiment of the present invention;

[0042] Figure 5 It is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are 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 those skilled in the art without creative work are within the scope of protection of the present invention.

[0044] Large language models have made significant progress in the field of human-computer interaction. With the widespread application of large language models, portable assistant scenarios will become more mature. For portable assistant scenarios that require long-term voice pickup, it can be simply divided into three processes:

[0045] 1) The speech engine continuously picks up the sound and converts the audio data into ASR text;

[0046] 2) The ASR text is sent to the large model for semantic analysis and understanding;

[0047] 3) Execute the user intention parsed by the large model to complete specific interactive tasks, such as device control, etc.

[0048] In this process, since the sound pickup process 1) is constantly ongoing, a large number of ASR results will be generated. Many of the ASR results themselves are meaningless. Sending these meaningless ASRs to the large model for analysis will generate a large number of invalid QPS, increase the load and operating costs, and easily cause large model interaction delays.

[0049] In this embodiment, a smart device control method is provided. Figure 1is a flow chart of a smart device control method according to an embodiment of the present invention. It should be noted that the steps shown in the flow chart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flow chart, in some cases, the steps shown or described can be executed in a different order than that shown here.

[0050] like Figure 1 As shown, the process of the intelligent device control method includes the following steps:

[0051] Step S101: convert at least one audio data acquired by the smart device into a corresponding ASR text set.

[0052] The ASR text set includes at least one ASR text corresponding to at least one audio data.

[0053] Exemplarily, after acquiring at least one audio data, in order to avoid noise in the acquired audio data, the audio data can be preprocessed in advance, and the preprocessed audio data can be converted into ASR text. Specifically, the audio data is converted into text using automatic speech recognition (ASR) technology. The ASR algorithm converts an audio waveform or signal into a corresponding text representation. This usually involves multiple steps such as audio feature extraction, acoustic model and language model.

[0054] Step S102: input at least one ASR text into preset filters to obtain total intent information.

[0055] Exemplarily, after the ASR text data is converted from the audio data, the ASR text data is filtered by an input and output value filter to filter out meaningless ASR text. The filter here can be a rule-based filter, a template-based filter, or a machine learning method.

[0056] Specifically, the input ASR text is preprocessed as necessary, including removing irrelevant information, stemming, word segmentation, etc., in preparation for subsequent analysis and filtering. The filter model performs intent recognition based on the preprocessed ASR text. This can be done by matching keywords, phrases, or using more complex semantic analysis. In addition to identifying the intent of a single ASR text, the filter model can also consider the contextual relationship between texts. For example, combining multiple ASR texts to understand the user's complete intent or identify topic transitions in a conversation. Based on the analysis results of the filter model, the intent of multiple ASR texts is merged into total intent information. This can be a single intent identifier, an intent classification, or a detailed intent description.

[0057] In a preferred embodiment, the preset filter includes a classification model and a filter, such as Figure 2 As shown, the above step S102 includes:

[0058] Step S1021: input a first ASR text into a classification model to determine a first intent classification, wherein the first ASR text is any one of at least one ASR text.

[0059] Exemplarily, features are extracted from the first ASR text. These features can be text-based, such as keywords, phrases, syntactic structures, etc., or semantic-based, such as intent, emotion, etc. A training data set of known categories is used to train the classifier. During the training process, the classifier learns how to classify different texts into different intent categories based on the input features. The first ASR text is input into the trained classifier, and the classifier classifies the first ASR text into the corresponding intent category based on the extracted features and the learned classification rules. Based on the output of the classifier, the intent classification of the first ASR text is determined. This intent classification can be a specific category label or a confidence score for an intent classification. Specifically, intent classification can include commands, queries, small talk, and feedback, etc.

[0060] Step S1022: determine the filter according to the first intent classification and the mapping relationship between the first intent classification and the filter.

[0061] Exemplarily, different intent classifications correspond to different filters. For example, command-type intents can be identified after passing through the filter to obtain the output information and output format corresponding to the command-type filter, which is more conducive to the subsequent recognition of ASR text.

[0062] Step S1023: input the first ASR text into the filter to obtain first intent information.

[0063] Exemplarily, after the ASR text is input into the filter, the ASR text that meets the filter rules is determined as the corresponding intent information, and the ASR text that does not meet the filter rules is determined as no intent information, or the ASR text that meets the rules is set to 1 and the ASR text that does not meet the rules is set to 0. The ASR text that is subsequently set to 0 does not need to be input into the subsequent large model or the recognition model of the smart device for intent recognition again, thereby reducing the recognition of unnecessary ASR text and improving the recognition efficiency of audio data.

[0064] In addition, instead of filtering out directly, the intent information of the ASR text that does not meet the filter rules can be assigned an importance weight or other means can be used to reduce the impact of the ASR text that does not meet the filter rules on the final intent information.

[0065] In a preferred embodiment, the filter includes a vertical model filter and a rule filter, and the vertical model filter corresponds to the type information of the smart device. Step S1023 specifically includes:

[0066] Step a1: input the first ASR text into the vertical small model filter to obtain the first initial intent sub-information.

[0067] Step a2: when the first initial intent sub-information matches the preset output format of the smart device, determining the first initial intent sub-information as the first intent sub-information.

[0068] Exemplarily, the vertical category mini-model is a model corresponding to the type of smart device. Specifically, for example, if the smart device is a lighting lamp, the vertical category mini-model is a model trained according to various instructions of the lighting lamp. Therefore, inputting the first ASR text into the corresponding vertical category mini-model will help obtain more accurate intent information.

[0069] Based on the above embodiment, the method also includes: when the first initial intent sub-information does not match the preset output format of the smart device, the first ASR text is input into the preset large model for prediction to obtain the first intent sub-information.

[0070] This means that the corresponding vertical small model filter is not very consistent with the smart device type. Therefore, in this case, the corresponding ASR text can be directly input into the large model for recognition. The large model here is not limited to the small model of the smart device type. The large model has a wider range of use and the intent information obtained is more accurate.

[0071] Step a3: input the first ASR text into a preset rule filter to obtain second intention sub-information.

[0072] Step a4: determining the first intention information according to the first intention sub-information, the second intention sub-information and a preset standard.

[0073] Exemplarily, the preset rule filter may be a rule device of some instructions corresponding to the type of smart device, such as a regular expression, etc. After each ASR text passes through the vertical classification model and the rule regularizer, multiple intent sub-information corresponding to each ASR text is obtained, and the multiple intent sub-information is arbitrated and determined according to the preset standard to obtain the final first intent information. Here, the preset standard may be the weight information corresponding to each filter itself in the filter or the degree of conformity of the ASR text when passing through the filter, so as to determine the final first intent information.

[0074] Here, the first intent information is identified through multiple filters, which improves the accuracy of identification. At the same time, when the first intent information is meaningless or similar, the corresponding ASR text can be directly deleted, which improves the accuracy of identification.

[0075] Step S1024: determine the total intent information according to the first intent information corresponding to each ASR text.

[0076] Exemplarily, after determining the intent information corresponding to each ASR text, the meaningless ASR texts that do not conform to the control of the corresponding smart device are filtered out to obtain the total intent information corresponding to the valid ASR texts. The total intent information here may be determined by one ASR text or by multiple ASR texts, and there is no limit on the number of valid ASR texts.

[0077] In a preferred embodiment, step S1024 specifically includes:

[0078] Step b1, selecting a first intent information subset that matches the instruction rule corresponding to the smart device from all the first intent information;

[0079] Step b2: determining the total intent information according to the first intent information subset.

[0080] Exemplarily, when there are still at least two intent information corresponding to ASR texts after passing through the filter, the intent information needs to satisfy the instructions that can be implemented by the smart device. Therefore, a subset of first intent information that matches the instruction rules corresponding to the smart device is screened from all first intent information, and finally the corresponding total intent information is determined based on the satisfied intent information subset.

[0081] Step S103: According to the general intention information, the smart device is controlled to execute instructions corresponding to the general intention information.

[0082] Exemplarily, after determining the overall intention information corresponding to the audio data, the smart device is controlled to complete the instructions for the overall intention information, thereby improving the control efficiency of the smart device.

[0083] It can be understood that the present invention does not impose specific limitations on smart devices. The smart devices mentioned in the present invention may be but are not limited to XR devices, smart phones, AR glasses, etc.

[0084] The following is an introduction to the implementation of the above embodiment using a specific example. Figure 3 As shown, taking the device control in the room as an example, there are two users chatting in the room, and three ASR results are generated (in actual scenarios, there is continuous sound pickup, and the number of ASR results generated is far more than three).

[0085] ASR1: The room is a bit stuffy.

[0086] ASR2: Let's go out for a walk. Okay.

[0087] ASR3: Turn off the lights.

[0088] In the solution in the prior art, these three ASRs will be sent to the big model for semantic understanding. But in fact, ASR1 and 2 are two completely meaningless semantic instructions for device control. In the embodiment of the present invention, through the vertical small model in ASRFilter (such as a small model specially trained for device control scenarios, or filtering by other filters such as rule engines, we will eventually filter out ASR1 and 2, leaving only ASR3 as the final result (or sending it to the big model service) for the next step of semantic understanding and task execution. Thereby effectively reducing the access load of the big model in the far-field uninterrupted sound pickup scenario, and after the original three requests are reduced to one, the interaction speed of the big model can also be effectively improved.

[0089] The smart device control method provided in this embodiment converts at least one audio data obtained by the smart device into a corresponding ASR text set, wherein the ASR text set includes at least one ASR text corresponding to the at least one audio data, thereby improving the accuracy of audio recognition; on this basis, at least one ASR text is input into a preset filter respectively to obtain overall intent information, thereby improving the accuracy of speech recognition, avoiding the computing power occupation caused by predictive recognition of a large amount of ASR text, and improving the recognition efficiency; finally, according to the overall intent information, the smart device is controlled to execute the instructions corresponding to the overall intent information, thereby effectively reducing the computing load of continuously obtaining audio data and improving the interaction speed of the smart device.

[0090] In this embodiment, a smart device control device is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware of a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0091] This embodiment provides a smart device control device, such as Figure 4 The device comprises:

[0092] An audio conversion module 401 is used to convert at least one audio data acquired by the smart device into a corresponding ASR text set, where the ASR text set includes at least one ASR text corresponding to the at least one audio data;

[0093] A filtering module 402, configured to input at least one ASR text into a preset filter to obtain total intent information;

[0094] The execution instruction module 403 is used to control the smart device to execute instructions corresponding to the general intention information according to the general intention information.

[0095] In an optional implementation, the preset filter includes a classification model and a filter, and the filtering module specifically includes:

[0096] an intention classification unit, configured to input a first ASR text into a classification model to determine a first intention classification, wherein the first ASR text is any one of the at least one ASR text;

[0097] A filter determination unit, configured to determine a filter according to the first intent classification and a mapping relationship between the first intent classification and the filter;

[0098] A filtering unit, configured to input the first ASR text into a filter to obtain first intent information;

[0099] The intention determination unit is used to determine the total intention information according to the first intention information corresponding to each ASR text.

[0100] In an optional implementation, the filter includes a vertical model filter and a rule filter, the vertical model filter corresponds to the type information of the smart device, and the filtering unit specifically includes:

[0101] A model filtering subunit, configured to input the first ASR text into a vertical class small model filter to obtain first initial intent sub-information;

[0102] a matching subunit, configured to determine that the first initial intent sub-information is the first intent sub-information when the first initial intent sub-information matches a preset output format of the smart device;

[0103] A rule filtering subunit, configured to input the first ASR text into a preset rule filter to obtain second intention sub-information;

[0104] The first intent information is determined according to the first intent sub-information, the second intent sub-information and a preset standard.

[0105] In an optional embodiment, the device is used to:

[0106] When the first initial intent sub-information does not match the preset output format of the smart device, the first ASR text is input into the preset large model for prediction to obtain the first intent sub-information.

[0107] In an optional implementation, determining the intention unit specifically includes:

[0108] A matching subunit, configured to select a subset of first intent information that matches the instruction rule corresponding to the smart device from all first intent information;

[0109] The intention determination subunit is used to determine the total intention information according to the first intention information subset.

[0110] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0111] The intelligent device control apparatus in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0112] The embodiment of the present invention also provides a computer device having the above Figure 4 The smart device control device is displayed.

[0113] See also Figure 5 , Figure 5 is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, such as Figure 5 As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 5 A processor 10 is taken as an example.

[0114] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0115] The memory 20 stores instructions executable by at least one processor 10, so that at least one processor 10 executes the method shown in the above embodiment.

[0116] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0117] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0118] The computer device further comprises a communication interface 30 for the computer device to communicate with other devices or a communication network.

[0119] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0120] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A method for controlling an intelligent device, characterized in that: The method comprises: Converting at least one piece of audio data acquired by the smart device into a corresponding ASR text set, wherein the ASR text set includes at least one piece of ASR text corresponding to the at least one piece of audio data; Inputting the at least one ASR text into preset filters respectively to obtain overall intent information; According to the general intention information, the smart device is controlled to execute instructions corresponding to the general intention information.

2. The method according to claim 1, characterized in that The preset filter includes a classification model and a filter, and the at least one ASR text is input into the preset filter to obtain the total intent information, specifically including: Inputting a first ASR text into the classification model to determine a first intent classification, wherein the first ASR text is any one of the at least one ASR text; determining the filter according to the first intent classification and a mapping relationship between the first intent classification and the filter; Inputting the first ASR text into the filter to obtain the first intent information; The total intent information is determined according to the first intent information corresponding to each ASR text.

3. The method according to claim 2, characterized in that The filter includes a vertical category small model filter and a rule filter, and the vertical category small model filter corresponds to the type information of the smart device.

4. The method according to claim 3, characterized in that: The step of inputting the first ASR text into the filter to obtain the first intent information specifically includes: Inputting the first ASR text into the vertical category small model filter to obtain first initial intent sub-information; When the first initial intent sub-information matches the preset output format of the smart device, determining that the first initial intent sub-information is the first intent sub-information; Inputting the first ASR text into a preset rule filter to obtain second intent sub-information; The first intent information is determined according to the first intent sub-information, the second intent sub-information and a preset standard.

5. The method according to claim 4, characterized in that The method further comprises: When the first initial intent sub-information does not match the preset output format of the smart device, the first ASR text is input into a preset large model for prediction to obtain the first intent sub-information.

6. The method according to claim 5, characterized in that The determining the overall intent information according to the first intent information corresponding to each ASR text specifically includes: Filtering a subset of first intent information that matches the instruction rule corresponding to the smart device from all first intent information; The overall intent information is determined based on the first intent information subset.

7. A smart device control device, characterized in that: The device comprises: An audio conversion module, used to convert at least one audio data acquired by the smart device into a corresponding ASR text set, wherein the ASR text set includes at least one ASR text corresponding to the at least one audio data; A filtering module, used for inputting the at least one ASR text into a preset filter to obtain the total intent information; The execution instruction module is used to control the smart device to execute the instruction corresponding to the general intention information according to the general intention information.

8. The device according to claim 7, characterized in that The preset filter includes a classification model and a filter, and the filtering module specifically includes: an intention classification unit, configured to input a first ASR text into the classification model to determine a first intention classification, wherein the first ASR text is any one of the at least one ASR text; a filter determination unit, configured to determine the filter according to the first intent classification and a mapping relationship between the first intent classification and the filter; a filtering unit, configured to input the first ASR text into the filter to obtain the first intent information; The intention determination unit is used to determine the total intention information according to the first intention information corresponding to each ASR text.

9. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the intelligent device control method according to any one of claims 1 to 6 by executing the computer instructions.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the intelligent device control method according to any one of claims 1 to 6.

Images