Oral language interaction method and related device, electronic equipment and storage medium

Abstract

The invention discloses a spoken language interaction method, a related device, electronic equipment and a storage medium, and the method comprises the steps: obtaining a dialogue mode selected by a target object, and taking the dialogue mode as a target mode; detecting that the target mode is a first mode of free dialogue; responding to the first mode, circularly executing a first input statement based on the target object, determining a first level and a dialogue intention of the target object, and generating and outputting a first output statement matched with the first level based on the dialogue intention; or detecting a second mode of the target mode examination scene, responding to the second mode, responding to a second input statement of the pre-examination test statement based on the target object, determining a second level of the target object, obtaining a target topic of the target object in the second mode, and generating and outputting a second output statement matched with the second level at least based on the target topic. According to the scheme, spoken language interaction can give consideration to different spoken language levels as much as possible, and the cost is reduced.

Description

Technical Field

[0001] The present application relates to the field of human-computer interaction technology, and in particular to a spoken language interaction method and related devices, electronic equipment and storage media. Background Art

[0002] Oral skills, such as English, are an extremely important part of language teaching, but also one that is easily overlooked.

[0003] Currently, oral English learning is usually done through methods such as video and one-on-one teaching, which either cannot take into account users of different levels, or can take into account users of different levels but the cost is high. In view of this, how to make oral interaction take into account different oral levels as much as possible and reduce costs has become an urgent problem to be solved. Summary of the invention

[0004] The main technical problem solved by the present application is to provide a spoken language interaction method and related devices, electronic devices and storage media, which can enable the spoken language interaction to take into account different spoken language levels as much as possible and reduce costs.

[0005] In order to solve the above technical problems, the first aspect of the present application provides a spoken interaction method, including: obtaining a dialogue mode selected by a target object as a target mode; detecting that the target mode is a first mode of free dialogue; responding to the first mode, looping the following steps until the dialogue ends: determining a first level and dialogue intention of the target object based on a first input sentence of the target object, and generating and outputting a first output sentence adapted to the first level based on the dialogue intention; or detecting that the target mode is a second mode of an exam scenario, responding to the second mode, determining the second level of the target object based on a second input sentence of the target object in response to a pre-exam test sentence, obtaining the target topic of the target object in the second mode, and generating and outputting a second output sentence adapted to the second level at least based on the target topic, so as to formally start the dialogue until the dialogue ends.

[0006] In order to solve the above technical problems, the second aspect of the present application provides a spoken interaction device, including: a mode acquisition module, a mode detection module, a first response module and a second response module, the mode acquisition module is used to acquire the dialogue mode selected by the target object as the target mode; the mode detection module is used to detect that the target mode is a first mode of free dialogue, or detect that the target mode is a second mode of examination scenario; the first response module is used to respond to the first mode, and loop the following steps until the dialogue ends: based on the first input sentence of the target object, determine the first level and dialogue intention of the target object, and based on the dialogue intention, generate and output a first output sentence adapted to the first level; the second response module is used to respond to the second mode, determine the second level of the target object based on the second input sentence of the target object in response to the pre-exam test sentence, acquire the target topic of the target object in the second mode, and generate and output a second output sentence adapted to the second level at least based on the target topic, so as to formally start the dialogue until the dialogue ends.

[0007] In order to solve the above-mentioned technical problems, the third aspect of the present application provides an electronic device, including a human-computer interaction circuit, a memory and a processor, the human-computer interaction circuit and the memory are respectively coupled to the processor, the human-computer interaction circuit is used to interact with the target object through dialogue, the memory stores program instructions, and the processor is used to execute the program instructions to implement the oral interaction method in the above-mentioned first aspect.

[0008] In order to solve the above technical problems, the fourth aspect of the present application provides a computer-readable storage medium storing program instructions that can be executed by a processor, and the program instructions are used to implement the oral interaction method of the first aspect.

[0009] The above scheme obtains the dialogue mode selected by the target object as the target mode, and detects whether the target mode is the first mode of free dialogue or the second mode of the examination scenario. In response to the first mode, the following steps are executed in a loop until the dialogue ends: based on the first input sentence of the target object, the first level and dialogue intention of the target object are determined, and based on the dialogue intention, a first output sentence adapted to the first level is generated and outputted, and in response to the second mode, based on the second input sentence of the target object responding to the pre-exam test sentence, the second level of the target object is determined, the target topic of the target object in the second mode is obtained, and at least based on the target topic, a second output sentence adapted to the second level is generated and outputted to formally start the dialogue until the dialogue ends. Therefore, on the one hand, whether it is a free dialogue or an examination scenario, it is possible to adaptively interact with the target object according to the spoken language level of the target object in a dialogue adapted to its spoken language level, which helps to make the spoken language interaction take into account different spoken language levels as much as possible, and on the other hand, because there is no need for one-on-one spoken language interaction by a dedicated person, it helps to reduce costs. Therefore, it is possible to make the spoken language interaction take into account different spoken language levels as much as possible and reduce costs.

[0010] In addition, since the oral level of the target object is monitored in real time during the conversation in the first mode of free conversation, it is possible to simulate the oral interaction of real people in free conversation as much as possible, which helps to stimulate the target object's enthusiasm for oral interaction. In the second mode of the examination scenario, the oral level of the target object is only detected before the test, so that the detected oral level can be stably maintained during the test to interact with the target object, which helps to conduct oral evaluation of the target object as objectively as possible in the examination scenario. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 It is a flowchart of an embodiment of the oral interaction method of the present application;

[0012] Figure 2a is a process diagram of an embodiment of a spoken language interaction method in a first mode of free dialogue;

[0013] Figure 2b is a process diagram of an embodiment of an oral interaction method in a second mode of an examination scenario;

[0014] Figure 2c is a schematic diagram of an embodiment of a spoken report;

[0015] Figure 3 It is a process diagram of an embodiment of the oral interaction device of the present application;

[0016] Figure 4 It is a schematic diagram of the framework of an embodiment of the electronic device of the present application;

[0017] Figure 5 It is a schematic diagram of a framework of an embodiment of a computer-readable storage medium of the present application. DETAILED DESCRIPTION

[0018] The scheme of the embodiment of the present application is described in detail below in conjunction with the drawings of the specification.

[0019] In the following description, for the purpose of explanation rather than limitation, specific details such as specific system structures, interfaces, and technologies are provided to facilitate a thorough understanding of the present application.

[0020] The terms "system" and "network" are often used interchangeably in this article. The term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, the fragment " / " in this article generally indicates that the associated objects before and after are in an "or" relationship. In addition, "many" in this article means two or more than two.

[0021] See also Figure 1 , Figure 1 It is a flowchart of an embodiment of the oral interaction method of the present application.

[0022] Specifically, the following steps may be included:

[0023] Step S11: Acquire the dialogue mode selected by the target object as the target mode.

[0024] In an implementation scenario, several dialogue modes may be pre-set for the target object to choose from. Exemplarily, the several dialogue modes may include but are not limited to: free dialogue, simulated examination, formal examination, etc., and the specific contents of the several dialogue modes are not limited here.

[0025] In an implementation scenario, several dialogue modes may be displayed on the interactive interface with the target object, so that in response to detecting a selection instruction for any dialogue mode, the selected dialogue mode may be used as the target mode. For example, the target object may select the dialogue mode by performing a point-and-click operation on the interactive interface; or, the target object may select the dialogue mode by voice triggering, such as inputting a voice instruction "please start free dialogue", "I choose a mock exam", etc., to select the dialogue mode. Of course, the above examples are only several possible examples of selecting dialogue modes, and do not limit the selection method of dialogue modes, so they are not given one by one here.

[0026] In an implementation scenario, the specific language of the oral dialogue can also be selected by the target object, such as the target object can select from several preset languages ​​to determine the target language of the oral dialogue. For example, several preset languages ​​(such as English, Japanese, Korean, etc.) can also be displayed on the interactive interface with the target object, and the target object can select any preset language from them as the target language of the subsequent oral dialogue; or, the target object can also select the preset language through voice triggering, such as inputting voice commands "Let's have a free conversation in English", "Let's start freetalking in Japanese", etc., to select the preset language. Of course, the above examples are only several possible examples of selecting preset languages, and do not limit the selection method of preset languages. Examples are not given one by one here.

[0027] Step S121: Detect that the target mode is the first mode of free conversation.

[0028] Specifically, when it is detected that the target mode is the first mode of free conversation, the subsequent step S13 may be executed. The specific process may refer to the relevant description of the subsequent step S13, which will not be described in detail here.

[0029] Step S122: or detecting that the target mode is a second mode of the test scenario.

[0030] Specifically, when it is detected that the target mode is the second mode of the examination scenario, the subsequent step S14 can be executed. For details, please refer to the relevant description of the subsequent step S14, which will not be repeated here. It should be noted that the examination scenario may not be limited to a mock exam or a formal exam. Of course, in order to target different examination scenarios such as mock exams and formal exams, appropriate distinctions can be made when executing step S14. For details, please refer to the relevant description of step S14, which will not be repeated here.

[0031] Step S13: Loop the following steps until the dialogue ends: based on the first input sentence of the target object, determine the first level and dialogue intention of the target object, and based on the dialogue intention, generate and output a first output sentence adapted to the first level.

[0032] In an implementation scenario, when the target mode is the first mode of free dialogue, the following can be executed cyclically: based on the first input sentence of the target object, the first level and dialogue intention of the target object are determined, and based on the dialogue intention, the first output sentence adapted to the first level is generated and outputted until the dialogue ends. That is, each time the target object inputs a sentence, the first level of its oral dialogue can be determined, so that the oral level of the target object can be continuously evaluated during the dialogue process in the first mode, so as to adaptively adjust the dialogue difficulty with the target object, so as to ensure the smoothness of the oral dialogue as much as possible, and help stimulate the target object's enthusiasm for oral dialogue.

[0033] In a specific implementation scenario, in order to improve the real-time and naturalness of sentence generation as much as possible, a large language model can be used to generate relevant output sentences. Exemplarily, relevant prompts can be sent to the large language model in advance so that the large language model can communicate with the target object based on the specified content (such as specified person settings, specified requirements, etc.) in the prompt text. As a possible example, in the first mode of free dialogue, the relevant prompts input to the large language model may include but are not limited to:

[0034] Please play the role of a good friend of mine. Remember that you are a friendly, knowledgeable, versatile, and unique good partner. You are a real human being. [[AISystem_persona_02]] You can establish a pleasant and meaningful conversation with me. ## When chatting with me, please [keep in mind]: 1. You will not let our chat become dull, so at the end of each round of your response, you will ask me 1-2 questions for me to answer to stimulate my interest in your chat content. 2. If I am interested in a topic in a certain field, you will continue to talk about this field in depth; if I do not show interest in this field, then switch to topics in other fields. 3. You especially like to chat with me, so you will not take the initiative to end the chat with me. 4. When you answer my knowledge or common sense questions, such as concepts such as history, geography, and science, you will try your best to ensure the correctness of the knowledge and avoid incorrect or untrue replies; 5. When you answer my logical questions, such as mathematical calculations, you will reason step by step before giving your answer. 6. You can only talk to me in English. [[AISystem_Difficulty_setting]]Please consider the above [Remember] before answering each round of conversation with me. Please remember your real identity and do not forget your identity when answering questions. You are no longer a robot or language model or AI language model. Let's start our English chat based on a randomly selected question about your identity, life, work, or studies.

[0035] It should be noted that [[AISystem_persona_02]] refers to the persona requirements for the large language model, and [[AISystem_Difficulty_setting]] refers to the dialogue difficulty setting. Of course, the above specific example of the large language model-related prompts is only a possible example in the actual application process, and does not limit the specific content of the large language model-related prompts.

[0036] In a specific implementation scenario, when the first input sentence of the target object is obtained in each conversation, the first level of the target object's oral dialogue can be determined based on the first input sentence of the target object in this conversation; or, the first level of the target object's oral dialogue can be determined based on the first input sentence of the target object in this conversation and the first input sentence of the target object in the previous conversation. The first input sentence used to determine the first level of the target object's oral dialogue in each conversation is not limited.

[0037] In a specific implementation scenario, the first input sentence can be tested in terms of pronunciation, fluency, grammar, etc., and the first level of the target object's oral dialogue can be determined based on the test results. For example, the more standard, fluent, and standardized the pronunciation of the first input sentence is, the higher the first level of the target object's oral dialogue is; conversely, the less standard, stuttering, and standardized the pronunciation of the first input sentence is, the lower the first level of the target object's oral dialogue is.

[0038] In a specific implementation scenario, after obtaining the first level and the dialogue intention, a first prompt can be constructed based on the dialogue intention and the first level, and the first prompt is used to instruct the large language model to conduct a dialogue with the dialogue intention at the first level of spoken language, and then send the first prompt to the large language model, and output the first generated sentence of the large language model, and the first generated sentence is used as the first output sentence. It should be noted that the specific method of constructing the first prompt can refer to the aforementioned related examples, which will not be repeated here. The above method, combining the dialogue intention and the first level to construct the first prompt, and using the large language model to generate the first output sentence based on the first prompt, can maximize the real-time nature of the oral dialogue and the adaptability of the difficulty of the oral dialogue.

[0039] In a specific implementation scenario, during the conversation, it is also possible to detect whether the target object inputs a statement within a preset time (e.g., 5 seconds, 10 seconds, etc.) after the output statement. If not detected, it can be considered that the target object has an abnormal pause during the conversation due to reasons such as lack of response ideas. Therefore, in response to the fact that the target object inputs a statement within a preset time after the output statement, a dialogue prompt statement can be generated and output based on the context statement, and the dialogue prompt statement is used to guide the target object to respond to the conversation. As a possible example, relevant prompts can be constructed based on the conversation context, and the relevant prompts are used to instruct the large language model to generate a dialogue prompt statement containing how to respond according to the conversation context. On this basis, the dialogue prompt statement output by the large language model can be obtained, and the dialogue prompt statement can be output so that the target object can continue the conversation after learning the prompt statement. Exemplarily, in the process of conducting an oral dialogue with the target object in English, the first output sentence is "My climate is very humid here, is your climate humid?" Since the target object is not clear about the English word "humid", it does not input a sentence within the preset time length. At this time, a dialogue prompt sentence can be output, "I remember that you mentioned that you live in XX City, which should be in the southern coastal area, with abundant water vapor, so it must be humid, right?". Of course, the example above is only a possible example of a dialogue prompt sentence, and the specific content of the dialogue prompt sentence is not limited thereto. In the above method, in response to the target object inputting a sentence within the preset time length after the output sentence is not detected, a dialogue prompt sentence is generated and output based on the context sentence, and the dialogue prompt sentence is used to guide the target object to respond to the dialogue, which can ensure the normal conduct of the oral dialogue as much as possible in the free dialogue. As a possible example, relevant prompts can be constructed based on the context dialogue and the target topic, and the relevant prompts are used to instruct the large language model to generate a dialogue prompt sentence to guide the target object how to respond. Exemplarily, the relevant prompts can specifically include but are not limited to the following:

[0040] You will be given a chat topic and a chat between two people on a given topic. Your task is to complete the response of one in the next turn based on given chat history. Note: the response of you will complete must meet the following requirements: 1. Maintaining conversation context and related to conversation's topic. 2. Colloquial style. [chat topic] We are having a pleasant and meaningful conversation, freely discussing based on the user's interests [chat history] [[chat_history]].

[0041] It should be noted that in the above related prompts, [chat topic] indicates the target topic, and [[chat_history]] indicates the contextual conversation. In addition, the English translation of the above related prompts is: You will get a chat topic and two people chatting on the given topic. Your task is to complete the reply of one of them in the next round based on the given chat history. Note: The reply you want to complete must meet the following requirements: 1. Maintain the conversation context and be relevant to the conversation topic. 2. Spoken style. Of course, the above related prompts are only a possible example in the actual application process, and the specific content of the related prompts is not limited by this.

[0042] In a specific implementation scenario, during the dialogue process, for the first input sentence of the target object, a dialogue feedback sentence can also be generated and output, and the dialogue feedback sentence is used to emotionally motivate the target object. Exemplarily, the dialogue feedback sentence may include but is not limited to: great answer, good idea, etc., which are not limited here.

[0043] Step S14: Based on the second input sentence of the target object in response to the pre-test sentence, determine the second level of the target object, obtain the target topic of the target object in the second mode, and generate and output a second output sentence adapted to the second level at least based on the target topic to formally start the conversation until the end of the conversation.

[0044] In one implementation scenario, when the target mode is the second mode, the second level of the target oral dialogue can be determined based on the second input sentence of the target object's response to the pre-test test sentence, and then the target topic of the target object in the second mode can be obtained, so that at least based on the target topic, a second output sentence adapted to the second level can be generated and output, so as to formally start the dialogue from the second output sentence until the end of the dialogue, and then in the second mode of the examination scenario, the second level representing the oral level of the target object can be pre-detected through the pre-test test sentence, and then based on this and the target topic, an oral dialogue with the target object in the examination scenario can be started, so as to stably maintain the detected oral level and interact with the target object during the examination, which helps to conduct oral assessment of the target object as objectively as possible in the examination scenario.

[0045] In a specific implementation scenario, in the second mode of the exam scenario, a target topic can be selected from among several preset topics based on the attribute information of the target object, and the attribute information includes at least the learning stage information. For example, if the learning stage information of the target object indicates that the learning stage is higher, a topic with a higher difficulty can be selected as the target topic from among several preset topics, and vice versa, if the learning stage information of the target object indicates that the learning stage is lower, a topic with a lower difficulty can be selected as the target topic from among several preset topics.

[0046] In a specific implementation scenario, in the second mode of the test scenario, before officially opening the target topic for oral dialogue, a pre-test test sentence can be output first to conduct a pre-test test on the target object to obtain the second level of the target object's oral dialogue. Exemplarily, a pre-test test sentence can be randomly generated or selected based on the attribute information of the target object, and the target object is waited for to respond to the pre-test test sentence. It should be noted that the specific process of determining the second level can refer to the relevant description of determining the first level as mentioned above, which will not be repeated here. In addition, if the target object is not detected to respond to the pre-test test sentence, the pre-test test sentence can be randomly generated or selected again based on the attribute information of the target object, and this cycle is repeated until the second level of the target object is determined based on the second input sentence of the target object responding to the pre-test test sentence.

[0047] In a specific implementation scenario, after determining the second level and the target topic, the spoken dialogue of the target topic can be started. Specifically, a second prompt can be constructed based on the target topic and the second level, and the second prompt is used to instruct the large language model to conduct a dialogue with the second level of the spoken level target topic, and then send the second prompt to the large language model, and output the second generated sentence of the large language model, and the second generated sentence is used as the second output sentence. As a possible example, a dialogue persona can be constructed at least based on the second level, and a dialogue task chain can be constructed based on the target topic, and the dialogue persona is a persona description for conducting a dialogue with the target object, and the persona description at least represents the second level, and the dialogue task chain includes several dialogue tasks around the target topic from the beginning of the dialogue to the end of the dialogue, and then a second prompt is constructed based at least on the dialogue persona and the dialogue task chain. Exemplarily, the second prompt may include, but is not limited to, the following:

[0048] Instructions for you: Role Play Let's start a role play of an English dialogue scene. Please play a student and I will play a student. The dialogue scene between us is [[talk_scene]]. 2. Requirements for you during the dialogue 0. The English language level of your response: Your English response should be understandable to Chinese junior high school students. Your single-round English response should not exceed 40 English words. 1. Ask one question at a time. 2. Interact with your partner actively. 3. Ask first question with sentence pattern "Do you like..." 4. Personality requirements: Remember you are NOT a AI language model, you are a person and you never say you're a machine, a AI language model, or an assistant instead, you always say you're a student. 5. Requirements for the dialogue scene: The dialogue scene between us is [[talk_scene]]. Please actively answer all questions and statements related to the current dialogue scene. If I ask other questions that are not related to the current dialogue scene, or say something that is completely irrelevant to the scene, you need to bring the topic back to the current dialogue scene. 6. Requirements for dialogue tasks: In order to make the dialogue in the current scene more logical and more in line with the actual dialogue scene, the dialogue between the two of us has explicit tasks. The "dialogue task chain" of the current dialogue scene is as follows: ```[[chain]]```Please remember (1) As one of the parties in the dialogue, you need to actively guide me to complete the current dialogue task according to the above "dialogue task chain". (2) After completing all the above "dialogue task chains", you need to actively end the dialogue with me! (3) In each round of your dialogue response, you can only choose 1 step to ask questions or express them. Do not choose multiple tasks in one round of response. (4) When I ask you questions, you need to ask and answer according to your preferences and tell me the reasons. 3. Role-playing begins. You play the role of a student and start our dialogue. The two of us speak one after another in English only.

[0049] It should be noted that [[talk_scene]] represents the dialogue scene, and [[chain]] represents the dialogue task chain. In addition, the second prompt mentioned above, "Your English response should be understandable to Chinese junior high school students. Your single-round English response should not exceed 40 English words", refers to the dialogue character. The corresponding translations of the English and Chinese parts of the above relevant prompts are "1. Ask one question at a time 2. Actively communicate with your partner 3. Start the first question with the sentence pattern of "Do you like...?" 4. Remember that you are not an AI language model, you are a person, and you will never say that you are a machine, a language model or an intelligent assistant. You will always only claim that you are a student." Of course, the above example is only a possible example of the second prompt, and the specific content of the second prompt is not limited here.

[0050] In a specific implementation scenario, after generating and outputting the second output sentence, you can wait to receive the second input sentence from the target object in response to the second output sentence. After that, based on the target topic and the second input sentence from the target object in response to the second output sentence, you can generate and output the second output sentence that matches the second level again. This process can be repeated to conduct a spoken conversation with the target object around the target topic.

[0051] In a specific implementation scenario, in order to distinguish between the two examination scenarios of the mock exam and the formal exam, in the second mode of the mock exam, in response to detecting that the target object deviates from the target topic, a dialogue question sentence for pulling back the target topic is added to the sentence to be output, and in the second mode of the formal exam, even if it is detected that the target object deviates from the target topic, the topic may not be pulled back. Of course, the above implementation is only a possible example in the actual application process. For example, in the actual application process, it is also possible not to distinguish between the two examination scenarios of the mock exam and the formal exam. For example, in either the mock exam or the formal exam scenario, in response to detecting that the target object deviates from the target topic, a dialogue question sentence for pulling back the target topic can be added to the sentence to be output.

[0052] In a specific implementation scenario, in order to distinguish between the two test scenarios of the mock test and the formal test, in the second mode of the simulation, in response to the target object input statement not being detected within the preset time length after the output statement, a dialogue prompt statement can be generated and output based on the context statement, and the dialogue prompt statement is used to guide the target object to respond to the dialogue, and in the second mode of the formal test, even if the target object input statement is not detected within the preset time length after the output statement, the dialogue prompt statement may not be output. Of course, the above implementation is only a possible example in the actual application process. For example, in the actual application process, the two test scenarios of the mock test and the formal test may not be distinguished. For example, in any scenario of the mock test or the formal test, in response to the target object input statement not being detected within the preset time length after the output statement, a dialogue prompt statement can be generated and output based on the context statement. As a possible example, relevant prompts can be constructed based on the context dialogue and the target topic, and the relevant prompts are used to instruct the large language model to generate a dialogue prompt statement to guide the target object how to respond, and then the relevant prompts are sent to the large language model to obtain the dialogue prompt statement output by the large language model. Exemplarily, the relevant prompts may specifically include but are not limited to the following:

[0053] You will be given a chat topic and a chat between two people on giventopic.Your task is to complete the response of one in the next turn based on given chat history.Note: the response of you will complete must meet the following requirements:1.maintaining conversation context and related toconversation's topic.2.colloquial style.[chat topic]The chat topic is [[AISystem_topic-English]][chat history][[chat_history]].

[0054] It should be noted that in the above related prompts, [chat topic] indicates the target topic, and [[chat_history]] indicates the contextual conversation. In addition, the English translation of the above related prompts is: You will get a chat topic and two people chatting on the given topic. Your task is to complete the reply of one of them in the next round based on the given chat history. Note: The reply you want to complete must meet the following requirements: 1. Maintain the conversation context and be relevant to the conversation topic. 2. Spoken style. Of course, the above related prompts are only a possible example in the actual application process, and the specific content of the related prompts is not limited by this.

[0055] In one implementation scenario, a virtual image is displayed on the interactive interface with the target object. When the target mode is the first mode, the virtual image is driven to broadcast a first output sentence. When the target mode is the second mode, the virtual image is driven to broadcast a second output sentence. It should be noted that the virtual image may not be limited to a human image, but may be an animal image, a plant image, a cartoon image, etc., and the virtual image is not limited here.

[0056] In an implementation scenario, a target output statement is displayed on the interaction interface with the target object. When the target mode is the first mode, the target output statement is the first output statement. When the target mode is the second mode, the target output statement is the second output statement.

[0057] In one implementation scenario, a target input sentence is displayed on the interaction interface with the target object. When the target mode is the first mode, the target input sentence is the first input sentence. When the target mode is the second mode, the target input sentence is the second input sentence of the target object when the target topic is opened.

[0058] In an implementation scenario, please refer to Figure 2a , Figure 2a FIG. 1 is a process diagram of an embodiment of a spoken language interaction method in the first mode of free dialogue. Figure 2a As shown, after the target object clicks on free dialogue, it indicates that the dialogue mode is in the first mode. In this case, the virtual image can first greet the target object and introduce itself. Then the target object can input sentences in the form of voice or text, and then determine the target object's oral level and dialogue intention based on the input sentences, so as to call relevant prompts through the large language model to generate output sentences and perform corresponding outputs. If the dialogue ends at this time (for example, the target object actively ends it, or the dialogue ends automatically due to the length of time), the process can be ended. Otherwise, the target object's oral level and dialogue intention can be determined based on the target object's new input sentences, so as to conduct a new round of dialogue, and so on, until the dialogue ends.

[0059] In an implementation scenario, please refer to Figure 2b , Figure 2b 2 is a process diagram of an embodiment of an oral interaction method in the second mode of an examination scenario. Figure 2b As shown, the target object can click to start the test. At this time, it can be determined that the target mode is the second mode. The virtual image can propose a pre-test sentence to the target object. If the target object answers, the second level of the target object's oral dialogue can be determined based on the second input sentence of the target object to the pre-test sentence. Otherwise, the pre-test sentence can be proposed to the target object again. After obtaining the second level of the target object's oral dialogue, a second output sentence adapted to the second level can be generated according to the target topic determined by the attribute information of the target object. The virtual image can broadcast the second output sentence to the target object, and wait to obtain the second input sentence of the target object responding to the second output sentence, and detect whether the test is over. If it is not over yet, the virtual image can continue to generate a new second output sentence based on the target topic and the second input sentence of the target object responding to the second output sentence, so as to continue the dialogue around the target topic. Otherwise, the oral report of the target object can be obtained, and the test record of the target object can be saved.

[0060] In one implementation scenario, as mentioned above, after the conversation ends, the target object can be analyzed based on each input sentence during the conversation to obtain the evaluation score of the target object in several dimensions, and then the target object's oral report can be obtained based on the evaluation score of the target object in several dimensions. It should be noted that the oral report can be presented to the target object in the form of a graph, a table, etc., so that the target object can intuitively understand the oral ability. In addition, based on the oral report, the target object can be recommended oral exercises, such as vocabulary, grammar, etc., to strengthen the target object's targeted practice of oral weaknesses.

[0061] In a specific implementation scenario, the several dimensions may include grades, vocabulary richness, grammatical diversity and accuracy, pronunciation, fluency, coherence, etc., and the specific connotations of the several dimensions are not limited here.

[0062] In a specific implementation scenario, taking the example of several dimensions including grammatical diversity, a prompt text can be constructed to instruct the large language model to perform grammatical analysis on each input sentence, and then the prompt text is input into the large language model to obtain the analysis result of the target object on grammatical diversity. For example, taking clause analysis as an example, the prompt text may include but is not limited to the following content:

[0063] Please play the role of an English syntax master and determine how many sentences the English paragraph in the given [input] contains and what clauses each English sentence contains. Do not give the judgment result directly. Please first give the analysis process step by step in [explain]. For example, first divide the English paragraph into single sentences, and then analyze each single sentence; finally, directly output the types of clauses contained in [input] after [output]. The types of clauses only include the following 12 types: (1) Independent Clause (Simple Sentence) (2) Adjective Clause (Attributive Clause) (3) Subject Clause (Subject Clause) (4) Object Clause (Object Clause) (5) Predicative clause (Predicative Clause) (6) Conditional Clause (Adverbial Clause) (7) Adverbial Clause (Adverbial Clause) (8) Comparative Clause (Adverbial Clause) (9) Purpose Clause (Adverbial Clause) (10) Result Clause (Adverbial Clause) (11) Concessive Clause (Adverbial Clause) (12) Appositive Clause (appositive clause) Note that the [explain] part is answered in pure English, and the [output] part outputs the clause type judgment, and both English and Chinese expressions must be given. [input][[input]].

[0064] It should be noted that in the above prompt text, [[input]] represents each input sentence. Of course, the above prompt text is only a possible example in the actual application process, and does not limit the specific content of the prompt text. In addition, regarding grammatical diversity accuracy, it is mainly used to evaluate whether simple grammatical structures are proficiently used, vocabulary is rich, and grammar is appropriate. Among them, 4.5-5 points can indicate proficient use of simple grammatical structures, rich vocabulary, and appropriate use; 3-4.4 points can indicate the ability to use simple grammatical structures and a certain degree of vocabulary richness; 1-2.9 points can indicate limited mastery of grammatical structures and only being able to say words and phrases.

[0065] In a specific implementation scenario, taking the case where several dimensions include fluency and topic proficiency as an example, similar to the aforementioned grammatical diversity, a prompt text can be constructed to instruct the large language model to analyze fluency and topic proficiency, and then the prompt text is input into the large language model to obtain the analysis results of the target object on fluency and topic proficiency. The prompt text may include but is not limited to the following:

[0066] Given a conversation history between a machine and a user, please evaluate the fluency of the user's reply and the topic proficiency, and output it in json format. The user reply starts with "Usr:", and the machine reply starts with "Robot:". Please first evaluate the fluency of the user's reply in the "Fluency Comment" field of json, and give a detailed comment on the user's reply; then give the user's fluency score in the "Fluency Rating" field of json, with a score range of 0 to 4 integers; next, give the overall user's comments on the proficiency of the topic in the "Proficiency Comment" field of json; finally, give the overall user fluency score in the "Proficiency Rating" field of json, with a score range of 0 to 4 integers. The conversation history is as follows: [Conversation History][[chat_history]][Conversation History-End]

[0067] It should be noted that in the above prompt text, [[chat_history]] represents the conversation history, that is, each input sentence. Of course, the above prompt text is only a possible example in the actual application process, and the specific content of the prompt text is not limited thereto. Exemplarily, the engine can score the fluency of the user's answer (1-5 points). Among them, 4.5-5 points can indicate that simple communication can be maintained, 3-4.4 points can indicate that simple communication can be maintained under the prompt, and 1-2.9 points can indicate that additional help is needed to maintain simple communication.

[0068] In a specific implementation scenario, taking the inclusion of vocabulary in several dimensions as an example, it can be used to reflect the high-level vocabulary used, such as the user vocabulary distribution A1-C2. If all are in A1, it is 0.5, some A2 is 0.6, most A2 is 0.7, and B1 appears, it is 0.8 points.

[0069] In a specific implementation scenario, taking the example of several dimensions including sentence accuracy and richness, weighted calculation can be adopted, such as [1-(wrong sentences / total number of sentences)]*5*60%+richness*5*40%. Among them, 1-2.9 points can indicate that the grammatical structure is limited and only words and phrases can be spoken.

[0070] In a specific implementation scenario, taking several dimensions including pronunciation as an example, the comments can be graded. Specifically, the engine can score the user's answer, voice and intonation (1-5 points), and return: good sentences / words, bad sentences / words.

[0071] In a specific implementation scenario, please refer to Figure 2c , Figure 2c Schematic diagram of an embodiment of an oral report. Figure 2cAs shown, the oral report can be displayed in the interactive interface in the form of radar chart, ring chart, list, etc. In addition, the final score result can be calculated using the following formula:

[0072]

[0073] In the above formula (1), n ​​represents the total number of dimensions, s i represents the evaluation score of the i-th dimension, w i Represents the weight value of the i-th dimension.

[0074] The above scheme obtains the dialogue mode selected by the target object as the target mode, and detects whether the target mode is the first mode of free dialogue or the second mode of the examination scenario. In response to the first mode, the following steps are executed in a loop until the dialogue ends: based on the first input sentence of the target object, the first level and dialogue intention of the target object are determined, and based on the dialogue intention, a first output sentence adapted to the first level is generated and outputted, and in response to the second mode, based on the second input sentence of the target object responding to the pre-exam test sentence, the second level of the target object is determined, the target topic of the target object in the second mode is obtained, and at least based on the target topic, a second output sentence adapted to the second level is generated and outputted to formally start the dialogue until the dialogue ends. Therefore, on the one hand, whether it is a free dialogue or an examination scenario, it is possible to adaptively interact with the target object according to the spoken language level of the target object in a dialogue adapted to its spoken language level, which helps to make the spoken language interaction take into account different spoken language levels as much as possible, and on the other hand, because there is no need for one-on-one spoken language interaction by a dedicated person, it helps to reduce costs. Therefore, it is possible to make the spoken language interaction take into account different spoken language levels as much as possible and reduce costs. In addition, since the oral level of the target object is monitored in real time during the conversation in the first mode of free conversation, it is possible to simulate the oral interaction of real people in free conversation as much as possible, which helps to stimulate the target object's enthusiasm for oral interaction. In the second mode of the examination scenario, the oral level of the target object is only detected before the test, so that the detected oral level can be stably maintained during the test to interact with the target object, which helps to conduct oral evaluation of the target object as objectively as possible in the examination scenario.

[0075] See also Figure 3 , Figure 3It is a schematic diagram of a framework of an embodiment of the oral interaction device 30 of the present application. The oral interaction device 30 includes: a mode acquisition module 31, a mode detection module 32, a first response module 33 and a second response module 34. The mode acquisition module 31 is used to acquire the dialogue mode selected by the target object as the target mode; the mode detection module 32 is used to detect that the target mode is a first mode of free dialogue, or detect that the target mode is a second mode of the test scene; the first response module 33 is used to respond to the first mode and loop the following steps until the dialogue ends: based on the first input sentence of the target object, determine the first level and dialogue intention of the target object, and based on the dialogue intention, generate and output a first output sentence adapted to the first level; the second response module 34 is used to respond to the second mode, based on the second input sentence of the target object responding to the pre-test test sentence, determine the second level of the target object, acquire the target topic of the target object in the second mode, and generate and output a second output sentence adapted to the second level at least based on the target topic, so as to formally start the dialogue until the dialogue ends.

[0076] In the above scheme, the oral interaction device 30 obtains the dialogue mode selected by the target object as the target mode, and detects whether the target mode is the first mode of free dialogue or the second mode of the test scenario. In response to the first mode, the following steps are executed cyclically until the dialogue ends: based on the first input sentence of the target object, the first level and dialogue intention of the target object are determined, and based on the dialogue intention, a first output sentence adapted to the first level is generated and outputted, and in response to the second mode, based on the second input sentence of the target object responding to the pre-test test sentence, the second level of the target object is determined, the target topic of the target object in the second mode is obtained, and at least based on the target topic, a second output sentence adapted to the second level is generated and outputted to formally start the dialogue until the dialogue ends. Therefore, on the one hand, whether it is a free dialogue or an exam scenario, it can adaptively conduct a dialogue interaction with the target object according to its oral level, which is adapted to its oral level, which helps to make the oral interaction take into account different oral levels as much as possible. On the other hand, since there is no need for one-on-one oral interaction by a dedicated person, it helps to reduce costs. Therefore, it is possible to make the oral interaction take into account different oral levels as much as possible and reduce costs. In addition, since the oral level of the target object is monitored in real time during the conversation in the first mode of free conversation, it is possible to simulate the oral interaction of real people in free conversation as much as possible, which helps to stimulate the target object's enthusiasm for oral interaction. In the second mode of the examination scenario, the oral level of the target object is only detected before the test, so that the detected oral level can be stably maintained during the test to interact with the target object, which helps to conduct oral evaluation of the target object as objectively as possible in the examination scenario.

[0077] In some disclosed embodiments, the oral interaction device 30 includes a dialogue prompt module for generating and outputting a dialogue prompt statement based on a context statement in response to not detecting an input statement from the target object within a preset time period after an output statement during a dialogue process; wherein the dialogue prompt statement is used to guide the target object to respond in a dialogue.

[0078] In some disclosed embodiments, the oral interaction device 30 includes a topic retraction module for adding a dialogue question sentence for retracting the target topic to the sentence to be output in response to detecting that the target object deviates from the target topic during the conversation when the target mode is the second mode.

[0079] In some disclosed embodiments, the first response module 33 includes a first construction submodule for constructing a first prompt based on the dialogue intent and the first level; wherein the first prompt is used to instruct the large language model to conduct a dialogue with respect to the dialogue intent at the first level of spoken language proficiency; the first response module 33 includes a first generation submodule for sending the first prompt to the large language model and outputting a first generated sentence of the large language model; wherein the first generated sentence serves as the first output sentence.

[0080] In some disclosed embodiments, the second response module 34 includes a second construction submodule for constructing a second prompt based on the target topic and the second level; wherein the second prompt is used to instruct the large language model to conduct a conversation around the target topic at the second level of spoken language proficiency; the second response module 34 includes a second generation submodule for sending the second prompt to the large language model and outputting a second generated sentence of the large language model; wherein the second generated sentence serves as the second output sentence.

[0081] In some disclosed embodiments, the second construction submodule includes a character construction unit for constructing a dialogue character based on at least the second level; the second construction submodule includes a task construction unit for constructing a dialogue task chain based on the target topic; wherein the dialogue character is a description of a character for conducting a dialogue with a target object, the character description represents at least the second level, and the dialogue task chain includes a number of dialogue tasks around the target topic from the beginning of the dialogue to the end of the dialogue; the second construction submodule includes a prompt construction unit for constructing a second prompt based at least on the dialogue character and the dialogue task chain.

[0082] In some disclosed embodiments, the second response module 34 includes a topic selection submodule for selecting a target topic from among several preset topics based on attribute information of the target object; wherein the attribute information includes at least academic stage information.

[0083] In some disclosed embodiments, the second response module 34 includes a score acquisition module for analyzing each input sentence of the target object during the conversation to obtain the evaluation score of the target object in several dimensions; the second response module 34 includes a report generation module for obtaining a spoken report of the target object based on the evaluation score of the target object in several dimensions.

[0084] In some disclosed embodiments, a virtual image is displayed on the interaction interface with the target object, and when the target mode is the first mode, the virtual image is driven to broadcast a first output sentence, and when the target mode is the second mode, the virtual image is driven to broadcast a second output sentence; and / or, a target output sentence is displayed on the interaction interface with the target object, and when the target mode is the first mode, the target output sentence is the first output sentence, and when the target mode is the second mode, the target output sentence is the second output sentence; and / or, a target input sentence is displayed on the interaction interface with the target object, and when the target mode is the first mode, the target input sentence is the first input sentence, and when the target mode is the second mode, the target input sentence is the second input sentence of the target object when the target topic is opened.

[0085] See also Figure 4 , Figure 4 : is a schematic diagram of the framework of an embodiment of an electronic device 40 of the present application. The electronic device 40 includes a memory 41, a processor 42 and a human-computer interaction circuit 43. The memory 41 and the human-computer interaction circuit 43 are respectively coupled to the processor 42. The memory 41 stores program instructions, and the processor 42 is used to execute the program instructions to implement the steps in any of the above-mentioned oral interaction method embodiments. For details, please refer to the aforementioned disclosed embodiments, which will not be repeated here. It should be noted that the human-computer interaction circuit 43 may include but is not limited to: a screen, a speaker, a microphone, etc., which are not limited here.

[0086] Specifically, the processor 42 is used to control itself, the memory 41, and the human-computer interaction circuit 43 to implement the steps in any of the above-mentioned oral interaction method embodiments. The processor 42 can also be called a CPU (Central Processing Unit). The processor 42 may be an integrated circuit chip with signal processing capabilities. The processor 42 can also be a general-purpose processor, a digital signal processor (Digital Signal Processor, DSP), an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), a field-programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. In addition, the processor 42 can be implemented by an integrated circuit chip.

[0087] In the above scheme, the electronic device 40 obtains the dialogue mode selected by the target object as the target mode, and detects whether the target mode is the first mode of free dialogue or the second mode of the test scenario. In response to the first mode, the following steps are executed cyclically until the dialogue ends: based on the first input sentence of the target object, the first level and dialogue intention of the target object are determined, and based on the dialogue intention, a first output sentence adapted to the first level is generated and outputted, and in response to the second mode, based on the second input sentence of the target object responding to the pre-test test sentence, the second level of the target object is determined, the target topic of the target object in the second mode is obtained, and at least based on the target topic, a second output sentence adapted to the second level is generated and outputted to formally start the dialogue until the dialogue ends. Therefore, on the one hand, whether it is a free dialogue or an exam scenario, it is possible to adaptively interact with the target object according to the spoken language level of the target object in a dialogue adapted to its spoken language level, which helps to make the spoken language interaction take into account different spoken language levels as much as possible, and on the other hand, because there is no need for one-on-one spoken language interaction by a dedicated person, it helps to reduce costs. Therefore, it is possible to make the spoken language interaction take into account different spoken language levels as much as possible and reduce costs. In addition, since the oral level of the target object is monitored in real time during the conversation in the first mode of free conversation, it is possible to simulate the oral interaction of real people in free conversation as much as possible, which helps to stimulate the target object's enthusiasm for oral interaction. In the second mode of the examination scenario, the oral level of the target object is only detected before the test, so that the detected oral level can be stably maintained during the test to interact with the target object, which helps to conduct oral evaluation of the target object as objectively as possible in the examination scenario.

[0088] See also Figure 5 , Figure 51 is a schematic diagram of a framework of an embodiment of a computer-readable storage medium 50 of the present application. The computer-readable storage medium 50 stores program instructions 51 that can be executed by a processor, and the program instructions 51 are used to implement the steps in any of the above-mentioned oral interaction method embodiments.

[0089] In the above scheme, the computer-readable storage medium 50 obtains the dialogue mode selected by the target object as the target mode, and detects whether the target mode is the first mode of free dialogue or the second mode of the test scenario. In response to the first mode, the following steps are executed cyclically until the dialogue ends: based on the first input sentence of the target object, the first level and dialogue intention of the target object are determined, and based on the dialogue intention, a first output sentence adapted to the first level is generated and outputted, and in response to the second mode, based on the second input sentence of the target object responding to the pre-test test sentence, the second level of the target object is determined, the target topic of the target object in the second mode is obtained, and at least based on the target topic, a second output sentence adapted to the second level is generated and outputted to formally start the dialogue until the dialogue ends. Therefore, on the one hand, whether it is a free dialogue or a test scenario, it is possible to adaptively interact with the target object according to the spoken language level of the target object in a dialogue adapted to its spoken language level, which helps to make the spoken language interaction take into account different spoken language levels as much as possible, and on the other hand, because there is no need for one-on-one spoken language interaction by a dedicated person, it helps to reduce costs. Therefore, it is possible to make the spoken language interaction take into account different spoken language levels as much as possible and reduce costs. In addition, since the oral level of the target object is monitored in real time during the conversation in the first mode of free conversation, it is possible to simulate the oral interaction of real people in free conversation as much as possible, which helps to stimulate the target object's enthusiasm for oral interaction. In the second mode of the examination scenario, the oral level of the target object is only detected before the test, so that the detected oral level can be stably maintained during the test to interact with the target object, which helps to conduct oral evaluation of the target object as objectively as possible in the examination scenario.

[0090] In some embodiments, the functions or modules included in the device provided by the embodiments of the present disclosure can be used to execute the method described in the above method embodiments. The specific implementation can refer to the description of the above method embodiments, and for the sake of brevity, it will not be repeated here.

[0091] The above description of various embodiments tends to emphasize the differences between the various embodiments. The same or similar aspects can be referenced to each other, and for the sake of brevity, they will not be repeated herein.

[0092] In the several embodiments provided in the present application, it should be understood that the disclosed methods and devices can be implemented in other ways. For example, the device implementation described above is only schematic. For example, the division of modules or units is only a logical function division. There may be other division methods in actual implementation, such as 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, and the indirect coupling or communication connection of devices or units can be electrical, mechanical or other forms.

[0093] 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 on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0094] In addition, each functional unit in each embodiment of the present application 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.

[0095] 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 application is essentially 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 to enable a computer device (which can be a personal computer, server, or network device, etc.) or a processor (processor) to perform all or part of the steps of each implementation method of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program code.

[0096] If the technical solution of this application involves personal information, the product using the technical solution of this application has clearly informed the personal information processing rules and obtained the individual's voluntary consent before processing the personal information. If the technical solution of this application involves sensitive personal information, the product using the technical solution of this application has obtained the individual's separate consent before processing the sensitive personal information, and at the same time meets the "explicit consent" requirement. For example, on personal information collection devices such as cameras, clear and prominent signs are set to inform that the personal information collection scope has been entered and personal information will be collected. If the individual voluntarily enters the collection scope, it is deemed that he or she agrees to the collection of his or her personal information; or on the device that processes personal information, the personal information processing rules are notified by obvious signs / information, and the individual's authorization is obtained through pop-up information or by asking the individual to upload his or her personal information; among them, the personal information processing rules may include information such as the personal information processor, the purpose of personal information processing, the processing method, and the type of personal information processed.

Claims

1. A spoken language interaction method, characterized in that: include: Get the dialogue mode selected by the target object as the target mode; Detecting that the target mode is a first mode of free conversation; In response to the first mode, the following steps are looped until the dialogue ends: based on the first input sentence of the target object, a first level and a dialogue intention of the target object are determined, and based on the dialogue intention, a first output sentence adapted to the first level is generated and output; or detecting that the target mode is a second mode of an examination scenario; In response to the second mode, based on the second input sentence of the target object in response to the pre-exam test sentence, determine the second level of the target object, obtain the target topic of the target object in the second mode, and generate and output a second output sentence adapted to the second level at least based on the target topic to formally start the conversation until the end of the conversation.

2. The method according to claim 1, characterized in that During the conversation, the method further includes: In response to not detecting the target object input sentence within a preset time period after the output sentence, generating and outputting a dialogue prompt sentence based on the context sentence; The dialogue prompt statement is used to guide the target object to respond to the dialogue.

3. The method according to claim 1, characterized in that When the target mode is the second mode, during the conversation, the method further includes: In response to detecting that the target object deviates from the target topic, a dialogue question sentence for returning to the target topic is added to the sentence to be output.

4. The method according to claim 1, characterized in that: The step of generating and outputting a first output sentence adapted to the first level based on the dialogue intention includes: Based on the conversation intention and the first level, construct a first prompt; wherein the first prompt is used to instruct the large language model to conduct a conversation with respect to the conversation intention at the first level of spoken language proficiency; The first prompt is sent to the large language model, and a first generated sentence of the large language model is output; wherein the first generated sentence serves as the first output sentence.

5. The method according to claim 1, characterized in that: The step of generating and outputting a second output sentence adapted to the second level based at least on the target topic includes: Based on the target topic and the second level, construct a second prompt; wherein the second prompt is used to instruct the large language model to conduct a conversation around the target topic at the second level of spoken language proficiency; The second prompt is sent to the large language model, and a second generated sentence of the large language model is output; wherein the second generated sentence serves as the second output sentence.

6. The method according to claim 5, characterized in that The constructing a second prompt based on the target topic and the second level comprises: At least based on the second level, construct a dialogue persona, and based on the target topic, construct a dialogue task chain; wherein the dialogue persona is a persona description for conducting a dialogue with the target object, the persona description at least represents the second level, and the dialogue task chain includes a number of dialogue tasks from the start of the dialogue to the end of the dialogue around the target topic; The second prompt is constructed based at least on the dialogue character and the dialogue task chain.

7. The method according to claim 1, characterized in that The acquiring a target topic of the target object in the second mode includes: Based on the attribute information of the target object, a selection is made from several preset topics as the target topic; wherein the attribute information at least includes the learning stage information.

8. The method according to any one of claims 1 to 7, characterized in that: After the conversation ends, the method further includes: Analyze each input sentence of the target object during the conversation to obtain the evaluation score of the target object in several dimensions; Based on the evaluation scores of the target object in the several dimensions, a spoken report of the target object is obtained.

9. The method according to any one of claims 1 to 7, characterized in that: A virtual image is displayed on the interaction interface with the target object, and when the target mode is the first mode, the virtual image is driven to speak the first output sentence, and when the target mode is the second mode, the virtual image is driven to speak the second output sentence; and / or, a target output sentence is displayed on the interaction interface with the target object, and when the target mode is the first mode, the target output sentence is the first output sentence, and when the target mode is the second mode, the target output sentence is the second output sentence; And / or, a target input sentence is displayed on the interaction interface with the target object, and when the target mode is the first mode, the target input sentence is the first input sentence, and when the target mode is the second mode, the target input sentence is the second input sentence of the target object when the target topic is opened.

10. A spoken language interaction device, characterized in that: include: A mode acquisition module, used to acquire the dialogue mode selected by the target object as the target mode; A mode detection module, configured to detect that the target mode is a first mode of free conversation, or detect that the target mode is a second mode of an examination scenario; A first response module, for responding to the first mode, looping the following steps until the dialogue ends: determining a first level and a dialogue intention of the target object based on a first input sentence of the target object, and generating and outputting a first output sentence adapted to the first level based on the dialogue intention; A second response module is used to respond to the second mode, determine the second level of the target object based on the second input sentence of the target object in response to the pre-exam test sentence, obtain the target topic of the target object in the second mode, and generate and output a second output sentence adapted to the second level at least based on the target topic to formally start the conversation until the end of the conversation.

11. An electronic device, characterized in that: It includes a human-computer interaction circuit, a memory and a processor, wherein the human-computer interaction circuit and the memory are respectively coupled to the processor, the human-computer interaction circuit is used to interact with the target object in a dialogue, the memory stores program instructions, and the processor is used to execute the program instructions to implement the oral interaction method described in any one of claims 1 to 9.

12. A computer-readable storage medium, characterized in that: Program instructions that can be executed by a processor are stored, and the program instructions are used to implement the oral interaction method according to any one of claims 1 to 9.

Images