Voice intention recognition method, device and electronic equipment

Abstract

The present invention discloses a method, device and electronic device for speech intention recognition. The method comprises: obtaining speech data and status data of a terminal device; inputting the speech data and status data into a multi-classification model to obtain an intention recognition result corresponding to the speech data, wherein the multi-classification model is trained based on multiple groups of sample data, the multiple groups of sample data include sample speech data and sample status data, and the intention corresponding to the sample speech data; and returning the intention recognition result to the terminal device. The present invention solves the technical problem of inaccurate speech intention recognition results in the related art.

Description

Technical Field

[0001] The present invention relates to the field of big data, and in particular to a method, device and electronic device for voice intention recognition. Background Art

[0002] In related technologies, semantic recognition is usually performed using means in the perceptual field or the cognitive field. The former uses computers to convert speech into text, and then combines it with natural language understanding, natural language generation, and speech synthesis technology to provide a speech-based human-computer interaction method; the latter uses semantic understanding and language generation for recognition. Since speech intent recognition in the vertical field of home appliances requires comprehensive consideration of multiple information, the above two commonly used methods have the problem of inaccurate speech intent recognition results.

[0003] Therefore, in the related art, there is a technical problem that the speech intent recognition result is inaccurate.

[0004] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0005] The embodiments of the present invention provide a method, device and electronic device for voice intent recognition, so as to at least solve the technical problem of inaccurate voice intent recognition results in the related art.

[0006] According to one aspect of an embodiment of the present invention, a method for voice intention recognition is provided, comprising: acquiring voice data and status data of the terminal device from a terminal device; inputting the voice data and the status data into a multi-classification model to obtain an intention recognition result corresponding to the voice data, wherein the multi-classification model is trained based on multiple groups of sample data, the multiple groups of sample data include sample voice data and sample status data, and the intention corresponding to the sample voice data; and returning the intention recognition result to the terminal device.

[0007] Optionally, obtaining the status data of the terminal device includes: obtaining a device identification of the terminal device and account information corresponding to the terminal device; and matching the status data of the terminal device based on the device identification and the account information.

[0008] Optionally, before inputting voice data and status data into a multi-classification model to obtain the intention recognition result corresponding to the voice data, it also includes: obtaining multiple groups of sample data, wherein the sample voice data included in the multiple groups of sample data includes keyword classification of voice, and the sample status data included in the multiple groups of sample data includes at least one of the following: time information, spatial information, and environmental information of the sample terminal device for receiving the sample voice data, the main control device information of the sample terminal device, the binding device information of the sample terminal device, and the five-tuple information of the sample account corresponding to the sample terminal device; the intention corresponding to the sample voice data includes: the operating device and operating instructions corresponding to the sample voice data; and using multiple groups of sample data for machine training to obtain a multi-classification model.

[0009] Optionally, the terminal device includes a smart speaker.

[0010] According to another aspect of an embodiment of the present invention, a method for voice intention recognition is also provided, including: collecting voice data; reporting the voice data and status data of a terminal device to a server, wherein the server is used to process the voice data and status data using a multi-classification model to obtain an intention recognition result corresponding to the voice data, the multi-classification model being trained based on multiple groups of sample data, the multiple groups of sample data including sample voice data and sample status data, and the intention corresponding to the sample voice data; receiving the intention recognition result returned by the server.

[0011] Optionally, the method further includes: when the intention recognition result includes a target operation device corresponding to the voice data, and a target operation instruction, sending the target operation instruction to the target operation device so that the target operation device executes the target operation instruction.

[0012] According to another aspect of an embodiment of the present invention, a speech intention recognition device is also provided, including: an acquisition module, used to acquire voice data and status data of the terminal device from a terminal device; a processing module, used to input the voice data and status data into a multi-classification model to obtain an intention recognition result corresponding to the voice data, wherein the multi-classification model is trained based on multiple groups of sample data, and the multiple groups of sample data include sample voice data and sample status data, as well as the intention corresponding to the sample voice data; and a return module, used to return the intention recognition result to the terminal device.

[0013] According to another aspect of an embodiment of the present invention, a speech intention recognition device is also provided, including: a collection module for collecting speech data; a reporting module for reporting the speech data and the status data of the terminal device to a server, wherein the server is used to process the speech data and the status data using a multi-classification model to obtain an intention recognition result corresponding to the speech data, and the multi-classification model is trained based on multiple groups of sample data, and the multiple groups of sample data include sample speech data and sample status data, as well as the intention corresponding to the sample speech data; a receiving module for receiving the intention recognition result returned by the server.

[0014] According to another aspect of an embodiment of the present invention, there is also provided an electronic device, comprising: a processor; and a memory for storing processor executable instructions; wherein the processor is configured to execute instructions to implement any of the above-mentioned speech intent recognition methods.

[0015] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is also provided. When the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device can execute any of the above-mentioned speech intent recognition methods.

[0016] In an embodiment of the present invention, the voice data and state data obtained by the terminal device are input into a multi-classification model to perform voice intent recognition, and the intent recognition result is returned to the terminal device. Since not only the voice data itself but also the state data when the voice data is received are considered when performing voice intent recognition, the accuracy of voice intent recognition can be effectively improved based on rich information when the voice intent is recognized. In addition, since the multi-classification model has been fully trained with sample voice data, sample state data, and the intent corresponding to the sample voice data, wherein the sample voice data and the sample state data contain a variety of information in the current field, the multi-classification model can efficiently and accurately derive the intent corresponding to the input voice based on the input voice data and state data, thereby achieving the technical effect of efficiently and accurately identifying the intent corresponding to the voice, and thus solving the technical problem of inaccurate voice intent recognition results in related technologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0018] Figure 1 is a flow chart of a method for speech intention recognition according to an embodiment of the present invention;

[0019] Figure 2 is a flow chart of a second method for speech intention recognition according to an embodiment of the present invention;

[0020] Figure 3 It is a schematic diagram of a flow chart of an optional implementation mode of the present invention;

[0021] Figure 4 is a schematic diagram of a machine learning pre-training model according to an optional embodiment of the present invention;

[0022] Figure 5 is a structural block diagram of a speech intention recognition device 1 provided according to an embodiment of the present invention;

[0023] Figure 6 is a structural block diagram of a second speech intention recognition device provided according to an embodiment of the present invention;

[0024] Figure 7 2 is a schematic diagram of a speech intent recognition electronic device provided according to an embodiment of the present invention. DETAILED DESCRIPTION

[0025] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme 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 only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0026] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0027] First, some nouns or terms that appear in the description of the embodiments of the present application are subject to the following explanations:

[0028] Automatic Speech Recognition (ASR) is a technology that converts human speech into text.

[0029] Natural Language Processing (NLP) is a discipline that studies language issues in the interaction between humans and computers.

[0030] Natural Language Understanding (NLU), commonly known as human-computer dialogue, is a branch of artificial intelligence that studies the use of electronic computers to simulate the human language communication process, so that computers can understand and use the natural languages ​​of human society, such as Chinese and English, to achieve natural language communication between humans and computers.

[0031] Natural Language Generation (NLG) refers to the process of using natural language text to express the intentions a computer wants to achieve.

[0032] Network devices, a three-in-one product that combines physical parts, intelligent parts and connectors, can introduce the Internet of Things and human-computer dialogue.

[0033] XGboost, an optimized distributed gradient boosting library designed to be efficient, flexible, and portable.

[0034] The Internet of Things (IOT) refers to the use of various devices and technologies such as information sensors, radio frequency identification technology, global positioning system, infrared sensors, laser scanners, etc. to collect real-time information on any object or process that needs to be connected or interacted, including sound, light, heat, electricity, mechanics, chemistry, biology, location, etc., and to achieve ubiquitous connection between objects and objects, and between objects and people through various possible network accesses, and to realize intelligent perception, identification, and management of objects and processes.

[0035] HBase, a distributed, column-oriented open source database.

[0036] Kafka, an open source stream processing platform developed by the Apache Software Foundation, is written in Scala and Java. Kafka is a high-throughput distributed publish-subscribe messaging system that can handle all action stream data from consumers on a website.

[0037] Elasticsearch, a Lucene-based search server, provides a distributed multi-user full-text search engine based on a RESTful web interface. Elasticsearch is developed in Java and released as open source under the terms of the Apache license. It is a popular enterprise-level search engine.

[0038] Flink, a distributed computing framework, can quickly process data of any size.

[0039] Quintuple, a communication term, usually refers to IP address, source port, destination IP address, destination port, and transport layer protocol.

[0040] According to an embodiment of the present invention, a method embodiment of speech intent recognition is provided. It should be noted that the steps shown in the flowchart 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 flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0041] Figure 1 is a flow chart of a method for speech intention recognition according to an embodiment of the present invention. Figure 1 As shown, the method comprises the following steps:

[0042] Step S102, acquiring voice data from the terminal device and status data of the terminal device;

[0043] Step S104, inputting the voice data and the state data into a multi-classification model to obtain an intention recognition result corresponding to the voice data, wherein the multi-classification model is trained based on multiple groups of sample data, and the multiple groups of sample data include sample voice data and sample state data, and the intention corresponding to the sample voice data;

[0044] Step S106, returning the intent recognition result to the terminal device.

[0045] Through the above steps, the voice data and state data obtained by the terminal device are input into the multi-classification model to perform voice intent recognition, and the intent recognition result is returned to the terminal device. Since not only the voice data itself but also the state data when the voice data is received are considered when performing voice intent recognition, the accuracy of voice intent recognition can be effectively improved based on rich information when recognizing the voice intent. In addition, since the multi-classification model has been fully trained with sample voice data, sample state data, and the intent corresponding to the sample voice data, wherein the sample voice data and the sample state data contain a variety of information in the current field, the multi-classification model can efficiently and accurately derive the intent corresponding to the input voice based on the input voice data and state data, thereby achieving the technical effect of efficiently and accurately recognizing the intent corresponding to the voice, and thus solving the technical problem of inaccurate voice intent recognition results in related technologies.

[0046] As an optional embodiment, the voice data and state data acquired by the terminal device are input into a multi-classification model to perform voice intent recognition, and the intention recognition result is returned to the terminal device. Among them, the multi-classification model has been fully trained with sample voice data, sample state data, and the intent corresponding to the sample voice data, so that when the multi-classification model is used to predict the voice intent, it is not only efficient but also accurate. In addition, the sample voice data and sample state data can contain a variety of information in the current field. For example, the voice data can include device nouns or commonly used expression keywords in the current vertical field, and the state data can include the switch state of the current terminal device, indoor environment information, location information, etc. For example, the multi-classification model can accurately derive the intent corresponding to the input voice based on the input voice data and state data. For example, if there are keywords such as "black" or "dark" in the voice data, the time of receiving the voice in the state data is at night, and the switch state of the light is "off", it can be judged that the intention of the voice is "turn on the light". For another example, if there is a keyword "hot" in the voice data, the time of receiving the voice in the state data is summer, and the switch state of the air conditioner is "off", it can be judged that the intention of the voice is "turn on the air conditioner for cooling". This achieves the technical effect of efficiently and accurately identifying the corresponding intention based on voice, thereby solving the technical problem of inaccurate voice intention recognition results in related technologies and providing users with a better user experience.

[0047] It should be noted that, in actual applications, the recognition result of voice intent is not necessarily a single intent, and multiple intents can also be recognized in parallel based on the voice.

[0048] As an optional embodiment, the following method can be used to obtain the status data of the terminal device: obtain the device identification of the terminal device and the account information corresponding to the terminal device; match the status data of the terminal device based on the device identification and account information. By directly matching the status data of the terminal device according to the device identification and account information, the operation process of determining the status of the terminal device can be greatly simplified. For example, the terminal device only needs to report its corresponding device identification and account information to directly and accurately obtain the real-time status data of the terminal device. Among them, the status data can include multiple types, for example, it can include region, environment, room information, terminal device switch status, etc. Account information can also include multiple types, for example, it can include a list of terminal devices bound to the account or the account's operation preferences for each terminal device, etc. Since the above-mentioned status data and account information cover various categories of information that may be involved in the current vertical field, the status data and account information can provide a more comprehensive basis for judging intent recognition. It should be noted that the above-mentioned account information can be identification information bound to the terminal device, which can correspond to one user, multiple users, or one or more organizations.

[0049] As an optional embodiment, before inputting voice data and state data into a multi-classification model and obtaining the intention recognition result corresponding to the voice data, it is also possible to: obtain multiple groups of sample data, wherein the sample voice data included in the multiple groups of sample data includes keyword classification of voice, and the sample state data included in the multiple groups of sample data includes at least one of the following: time information, spatial information, environmental information of the sample terminal device receiving the sample voice data, the master control device information of the sample terminal device, the binding device information of the sample terminal device, and the five-tuple information of the sample account corresponding to the sample terminal device, and the intention corresponding to the sample voice data includes: the operation device and operation instruction corresponding to the sample voice data; use multiple groups of sample data for machine training to obtain a multi-classification model. Since the above multiple groups of sample data cover various categories of information that may be involved in the current vertical field, various factors that may generate the intention to operate the terminal device, and the context actions in the five-tuple information, the multi-classification model can be fully and comprehensively trained through the above multiple groups of sample data. In particular, the five-tuple information can coordinate the global behavior and sort it by time to distinguish each action, which guarantees the validity of the input data for the multi-classification model. At the same time, the actual operation performed by the user after the voice is uttered can be analyzed based on the state data, which can also be used as a training target marker for the multi-classification model or used for test results. In summary, after training with multiple sets of sample data, the multi-classification model can accurately determine the intention of the voice through voice data and state data.

[0050] As an optional embodiment, the terminal device may include multiple types, for example, a smart speaker. As an optional device of the terminal device, the smart speaker can not only obtain voice data and status data, or receive intent recognition results, but also realize further human-computer interaction according to the recognized intent, providing a better user experience.

[0051] Figure 2 is a flow chart of a second method for speech intention recognition according to an embodiment of the present invention. Figure 2 As shown, the method comprises the following steps:

[0052] Step S202, collecting voice data;

[0053] Step S204, reporting the voice data and the status data of the terminal device to the server, wherein the server is used to process the voice data and the status data using a multi-classification model to obtain an intention recognition result corresponding to the voice data, and the multi-classification model is trained based on multiple groups of sample data, and the multiple groups of sample data include sample voice data and sample status data, and the intention corresponding to the sample voice data;

[0054] Step S206, receiving the intent recognition result returned by the server.

[0055] Through the above steps, you only need to report the collected voice data and terminal status data to the server, and you can receive the intention recognition results returned by the server. Since the multi-classification model used for intent recognition is trained with multiple sets of sample data, and the multiple sets of sample data cover various categories of information that may be involved in the current vertical field, various factors that may generate the intention to operate the terminal device, and the true intention corresponding to the sample voice data, the intention recognition results obtained by the multi-classification model are accurate and reliable. Therefore, through the above steps, you can quickly obtain accurate results of voice intent recognition, thereby solving the technical problem of inaccurate voice intent recognition results in related technologies and providing users with a better user experience.

[0056] As an optional embodiment, when the intention recognition result includes the target operation device corresponding to the voice data and the target operation instruction, the target operation instruction is sent to the target operation device so that the target operation device executes the target operation instruction. Through the above operation, the corresponding target operation instruction can be executed on the target operation device according to the received intention recognition result, providing users with a more efficient and high-quality use experience.

[0057] Based on the above embodiments and optional embodiments, an optional implementation is provided, which is described in detail below.

[0058] In related technologies, speech intelligence is mainly divided into two fields, one is the perception field (ASR) and the other is the cognitive field (NLP). Among them, the perception field refers to the task of using computers to realize the automatic conversion of speech to text. In practical applications, speech recognition is usually combined with natural language understanding, natural language generation and speech synthesis technology to provide a natural and fluent human-computer interaction method based on speech. The cognitive field includes natural semantic understanding (NLU) and natural language generation (NLG). The industry usually uses third-party speech recognition engines in the field of language perception, but lacks mature solutions in the cognitive field. The industry segmentation knowledge is complex and rich, and it is difficult to unify. Therefore, conducting experimental research in vertical fields has become a way.

[0059] In related technologies, AI faces the problems of difficulty in semantic understanding and lack of annotated data resources, especially in the field of voice intent recognition (NLU), where there is a technical bottleneck. Specifically, the user's fuzzy semantics cannot be accurately recognized, and it temporarily relies on fixed rule algorithms for judgment, which lacks logical associations, resulting in misjudgment of intent and user complaints. For example, when a user asks in a context without entity words such as "too cold" or "too dark", NLP is currently unable to accurately recognize the user's intent.

[0060] The optional implementation manner of the present invention solves the problem of inferring and judging through information such as the context, environment, time, user habits and preferences of voice inquiries, thereby improving the accuracy of user intent recognition and improving the user's smart scene experience.

[0061] The optional implementation mode of the present invention mainly studies semantic understanding, takes the daily language in the vertical field of home appliances as the basic corpus, combines the user's real-time home appliance status and key features such as language context, environment, and room location, conducts multi-classification model training, and outputs the types of home appliances and operating parameters that the user may operate.

[0062] The semantic understanding models in related technologies are based on rules and knowledge graphs, lacking information input such as context and environment; they also lack labeled data for verification; and their recognition accuracy is not high.

[0063] The optional implementation mode of the present invention uses large-scale corpus data and status data reported by network devices as training data, combines machine learning multi-classification model (XGboost) for model training, and combines manually annotated data for supervised feedback, which greatly improves the recall rate and precision rate of the model, and fills the gaps in the field of semantic understanding and the shortcomings of data raw materials of the AI ​​team. At the same time, the optional implementation mode of the present invention is a bridge link between AI and IOT, and can serve as a hub to connect the input end and the output end.

[0064] Figure 3 It is a schematic diagram of a flow chart of an optional embodiment of the present invention, such as Figure 3 As shown, optional embodiments of the present invention include:

[0065] (1) Architecture diagram (divided into four functional units: network device, voice, big data, and IoT), with Kafka as the real-time data stream transmission medium;

[0066] (2) The upstream input device reports status data, speech word vectors, and query time, space, and environment information;

[0067] (3) The midstream is the big data storage medium and processing equipment. The storage medium uses hbase or elasticsearch as the cache medium for real-time data streams; the processing equipment uses the Flink engine as a real-time computing engine to perform rule and logic processing; it also calls the offline pre-trained model to supplement the algorithm;

[0068] (4) The downstream output is the cloud network device operation command, including intent and slot information. The terminal output is parsed and expressed by the IOT for device operation instructions.

[0069] Figure 4 is a schematic diagram of a machine learning pre-training model according to an optional embodiment of the present invention, such as Figure 4 As shown, including:

[0070] (1) The AI ​​side serves as the data input source, inputting voice keyword classification, device ID, user ID, and voice query time;

[0071] (2) Big data matches the user's real-time location, environment, and room information based on the device ID and user ID, as well as the device's master control device information and the user's bound device list;

[0072] (3) At the same time, the big data will query the five-tuple behavior data (context behavior identifier) ​​in real time based on the user identifier;

[0073] (4) Integrate the offline and real-time data inputs (X1....XN) from (1) to (3), call the offline training model (XGBOOST) of the algorithm group, divide it into a training group and a test group, and the behavior tag (Y) is the real behavior identifier parsed according to the device reported status;

[0074] (5) The prediction result returned to the AI ​​is the selected device category, action, and device identification.

[0075] In an optional implementation of the present invention, the five-tuple behavior data coordinates the user's global behavior, sorts it by time, distinguishes the previous action from the next action, and inputs an effective data source for the classification model; the user's actual action identifier is deduced through the device reporting status as a behavior marker for the classification model. Among them, after experiments, the XGBOOST multi-classification model, as a supervised model under a multi-entropy state, has a significant training effect. The accuracy of classification selection is over 80%.

[0076] The adoption of the above optional implementation method effectively solves the problem that only certain links or certain categories of related technologies have intelligent control or interactive capabilities. Regarding related technologies, in the IoT ecosystem, only certain furniture equipment is included, and cross-category interconnection and interoperability are not established. Some have advantages in all categories of home furnishings and interconnection, but do not have high-level skills. The optional implementation method of the present invention can be applied to the whole house smart home, and at the same time, combined with AI to empower smart home, so that home appliances have a certain degree of memory, learning, and prediction capabilities, can perceive and identify user needs and intentions in real time, and provide users with considerate services and precise device control.

[0077] In summary, the optional implementation mode of the present invention uses an AI multi-classification selection model, combined with the context of user behavior and device status, as well as a whole-house home model, to accurately identify user intentions and provide high-level intelligent scene services.

[0078] According to an embodiment of the present invention, a device for implementing the above-mentioned speech intention recognition method 1 is also provided. Figure 5 is a structural block diagram of a speech intention recognition device according to an embodiment of the present invention. Figure 5 As shown, the device includes: an acquisition module 51, a processing module 52 and a return module 53. The device is described below.

[0079] An acquisition module 51 is used to acquire voice data and status data of the terminal device from the terminal device; a processing module 52 is connected to the acquisition module 51 and is used to input the voice data and status data into a multi-classification model to obtain an intention recognition result corresponding to the voice data, wherein the multi-classification model is trained based on multiple groups of sample data, and the multiple groups of sample data include sample voice data and sample status data, as well as the intention corresponding to the sample voice data; a return module 53 is connected to the processing module 52 and is used to return the intention recognition result to the terminal device.

[0080] As an optional embodiment, the acquisition module 51 includes: a first acquisition unit, used to acquire a device identification of a terminal device and account information corresponding to the terminal device; and a matching unit, used to match the status data of the terminal device based on the device identification and the account information.

[0081] As an optional embodiment, the above-mentioned device also includes: a second acquisition unit, used to acquire multiple groups of sample data, wherein the sample voice data included in the multiple groups of sample data include voice keyword classification, and the sample status data included in the multiple groups of sample data include at least one of the following: time information, spatial information, and environmental information of the sample terminal device for receiving the sample voice data, the main control device information of the sample terminal device, the binding device information of the sample terminal device, and the five-tuple information of the sample account corresponding to the sample terminal device; the intention corresponding to the sample voice data includes: the operating device and operating instructions corresponding to the sample voice data; a training unit, used to use multiple groups of sample data for machine training to obtain a multi-classification model.

[0082] As an optional embodiment, the terminal device includes a smart speaker.

[0083] According to an embodiment of the present invention, a device for implementing the second method for speech intention recognition is also provided. Figure 6 is a structural block diagram of a second speech intention recognition device according to an embodiment of the present invention. Figure 6 As shown, the device includes: a collection module 61, a reporting module 62 and a receiving module 63. The device is described below.

[0084] An acquisition module 61 is used to collect voice data; a reporting module 62 is connected to the acquisition module 61 and is used to report the voice data and the status data of the terminal device to the server, wherein the server is used to process the voice data and the status data using a multi-classification model to obtain an intention recognition result corresponding to the voice data, and the multi-classification model is trained based on multiple groups of sample data, and the multiple groups of sample data include sample voice data and sample status data, as well as the intention corresponding to the sample voice data; a receiving module 63 is connected to the reporting module 62 and is used to receive the intention recognition result returned by the server.

[0085] As an optional embodiment, the above-mentioned device also includes: an execution unit, which is used to send a target operation instruction to the target operation device when the intention recognition result includes: a target operation device corresponding to the voice data, and a target operation instruction, so that the target operation device executes the target operation instruction.

[0086] According to an embodiment of the present invention, an electronic device is also provided. Figure 7 is a schematic diagram of a speech intention recognition electronic device provided according to an embodiment of the present invention, such as Figure 7 As shown, the electronic device includes: a processor 702; a memory 704 for storing instructions executable by the processor, etc.

[0087] It should be noted that, in the embodiment of the present invention, the electronic device mentioned above may be a terminal device or a server.

[0088] The above-mentioned electronic device can execute the program code of the following steps in the voice intention recognition method of the application: obtaining voice data and status data of the terminal device from the terminal device; inputting the voice data and status data into a multi-classification model to obtain the intention recognition result corresponding to the voice data, wherein the multi-classification model is trained based on multiple groups of sample data, and the multiple groups of sample data include sample voice data and sample status data, as well as the intention corresponding to the sample voice data; and returning the intention recognition result to the terminal device.

[0089] Among them, the memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the speech intention recognition method and device in the embodiment of the present invention. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, that is, realizing the above-mentioned speech intention recognition method. The memory may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include a memory remotely arranged relative to the processor, and these remote memories can be connected to the computer terminal 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.

[0090] The processor can call the information and application programs stored in the memory through the transmission device to perform the following steps: obtain voice data and status data from the terminal device; input the voice data and status data into a multi-classification model to obtain the intention recognition result corresponding to the voice data, wherein the multi-classification model is trained based on multiple groups of sample data, and the multiple groups of sample data include sample voice data and sample status data, as well as the intention corresponding to the sample voice data; and return the intention recognition result to the terminal device.

[0091] Optionally, the processor may also execute program code of the following steps: obtaining a device identification of the terminal device and account information corresponding to the terminal device; and matching status data of the terminal device based on the device identification and account information.

[0092] Optionally, the processor may also execute the program code of the following steps: obtaining multiple groups of sample data, wherein the sample voice data included in the multiple groups of sample data include keyword classification of voice, and the sample status data included in the multiple groups of sample data include at least one of the following: time information, spatial information, and environmental information of the sample terminal device receiving the sample voice data, the main control device information of the sample terminal device, the binding device information of the sample terminal device, and the five-tuple information of the sample account corresponding to the sample terminal device, and the intention corresponding to the sample voice data includes: the operating device and operating instructions corresponding to the sample voice data; using multiple groups of sample data for machine training to obtain a multi-classification model.

[0093] Optionally, the processor may also execute program code of the following steps: the terminal device includes a smart speaker.

[0094] Optionally, the processor may also execute the program code of the following steps: collecting voice data; reporting the voice data and the status data of the terminal device to the server, wherein the server is used to process the voice data and the status data using a multi-classification model to obtain an intention recognition result corresponding to the voice data, the multi-classification model being trained based on multiple groups of sample data, the multiple groups of sample data including sample voice data and sample status data, and the intention corresponding to the sample voice data; receiving the intention recognition result returned by the server.

[0095] Optionally, the processor may also execute the following program code: when the intention recognition result includes a target operation device corresponding to the voice data, and a target operation instruction, send the target operation instruction to the target operation device so that the target operation device executes the target operation instruction.

[0096] According to an embodiment of the present invention, a computer-readable storage medium is also provided. When the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device can execute the voice intention recognition method of any one of the above embodiments.

[0097] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0098] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0099] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0100] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed over multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0101] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0102] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

[0103] The above are only preferred embodiments of the present invention. It should be pointed out that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for speech intent recognition, It is characterized in that include: Acquiring voice data from a terminal device and status data of the terminal device; Inputting the voice data and the state data into a multi-classification model to obtain an intention recognition result corresponding to the voice data, wherein the multi-classification model is trained based on multiple groups of sample data, the multiple groups of sample data include sample voice data and sample state data, and the intention corresponding to the sample voice data; Returning the intention recognition result to the terminal device; Wherein, before inputting the voice data and the status data into a multi-classification model to obtain the intention recognition result corresponding to the voice data, it also includes: acquiring the multiple groups of sample data, wherein the sample voice data included in the multiple groups of sample data includes keyword classification of voice, and the sample status data included in the multiple groups of sample data includes: time information, spatial information, and environmental information of the sample terminal device receiving the sample voice data, the main control device information of the sample terminal device, the binding device information of the sample terminal device, and the five-tuple information of the sample account corresponding to the sample terminal device, and the intention corresponding to the sample voice data includes: the operating device and operating instructions corresponding to the sample voice data; using the multiple groups of sample data for machine training to obtain the multi-classification model, the five-tuple information includes: IP address, source port, destination IP address, destination port and transport layer protocol.

2. The method according to claim 1, It is characterized in that The acquiring the status data of the terminal device comprises: Obtaining a device identification of the terminal device and account information corresponding to the terminal device; Based on the device identification and the account information, the status data of the terminal device is matched.

3. The method according to any one of claims 1 to 2, It is characterized in that The terminal device includes a smart speaker.

4. A method for recognizing speech intent, It is characterized in that include: Collect voice data; The voice data and the status data of the terminal device are reported to the server, wherein the server is used to process the voice data and the status data using a multi-classification model to obtain an intention recognition result corresponding to the voice data, and the multi-classification model is trained based on multiple groups of sample data, and the multiple groups of sample data include sample voice data and sample status data, as well as the intention corresponding to the sample voice data; before the voice data and the status data are input into the multi-classification model to obtain the intention recognition result corresponding to the voice data, the server is also used to obtain the multiple groups of sample data, wherein the multiple groups of sample data include sample voice data The sample state data includes the keyword classification of the voice, the sample terminal device receiving the sample voice data, the spatial information, the environmental information, the master device information of the sample terminal device, the binding device information of the sample terminal device, and the five-tuple information of the sample account corresponding to the sample terminal device, and the intention corresponding to the sample voice data includes: the operation device and operation instruction corresponding to the sample voice data; the multiple groups of sample data are used for machine training to obtain the multi-classification model, and the five-tuple information includes: IP address, source port, destination IP address, destination port and transport layer protocol; Receive the intent recognition result returned by the server.

5. The method according to claim 4, It is characterized in that The method further comprises: In a case where the intention recognition result includes: a target operation device corresponding to the voice data, and a target operation instruction, the target operation instruction is sent to the target operation device so that the target operation device executes the target operation instruction.

6. A speech intention recognition device, It is characterized in that include: An acquisition module, used to acquire voice data from a terminal device and status data of the terminal device; A processing module, used for inputting the voice data and the state data into a multi-classification model to obtain an intention recognition result corresponding to the voice data, wherein the multi-classification model is trained based on multiple groups of sample data, the multiple groups of sample data include sample voice data and sample state data, and the intention corresponding to the sample voice data; A returning module, used for returning the intention recognition result to the terminal device; The device further includes: a second acquisition unit for acquiring multiple groups of sample data, wherein the sample voice data included in the multiple groups of sample data include keyword classification of voice, and the sample status data included in the multiple groups of sample data include: time information, spatial information, environmental information of the sample terminal device receiving the sample voice data, the main control device information of the sample terminal device, the binding device information of the sample terminal device, and the five-tuple information of the sample account corresponding to the sample terminal device, and the intention corresponding to the sample voice data includes: the operating device and operating instructions corresponding to the sample voice data; a training unit for using multiple groups of sample data for machine training to obtain a multi-classification model, and the five-tuple information includes: IP address, source port, destination IP address, destination port and transport layer protocol.

7. A speech intention recognition device, It is characterized in that include: A collection module, used for collecting voice data; A reporting module is used to report the voice data and the status data of the terminal device to the server, wherein the server is used to process the voice data and the status data using a multi-classification model to obtain an intention recognition result corresponding to the voice data, and the multi-classification model is trained based on multiple groups of sample data, and the multiple groups of sample data include sample voice data and sample status data, and the intention corresponding to the sample voice data; before inputting the voice data and the status data into the multi-classification model to obtain the intention recognition result corresponding to the voice data, the server is also used to obtain the multiple groups of sample data, wherein the multiple groups of sample data include sample The voice data includes keyword classification of the voice, the sample status data included in the multiple groups of sample data include: time information, spatial information, environmental information of the sample terminal device receiving the sample voice data, the master control device information of the sample terminal device, the binding device information of the sample terminal device, and the five-tuple information of the sample account corresponding to the sample terminal device, the intention corresponding to the sample voice data includes: the operation device and operation instruction corresponding to the sample voice data; the multiple groups of sample data are used for machine training to obtain the multi-classification model, and the five-tuple information includes: IP address, source port, destination IP address, destination port and transport layer protocol; A receiving module is used to receive the intention recognition result returned by the server.

8. An electronic device, It is characterized in that include: processor; a memory for storing instructions executable by the processor; The processor is configured to execute the instructions to implement the speech intention recognition method as described in any one of claims 1 to 5.

9. A computer-readable storage medium, It is characterized in that When the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the voice intention recognition method as described in any one of claims 1 to 5.

Images