Method and device for assisting foreign language learning based on AI large language model

Abstract

The invention provides a method and a device for assisting foreign language learning based on an AI large language model. The problems that foreign language learning resources are limited and individual requirements are difficult to meet are solved. The method specifically comprises the steps of obtaining a foreign language dialogue scene and foreign language voice output by a user, forming a Prompt, inputting the Prompt into an AI large language model, generating an answer through the AI large language model, and outputting voice of the answer; repeating the above steps to form multiple rounds of foreign language speech conversations and storing the conversations; evaluating the foreign language pronunciation of the user according to the user audio, and forming and outputting a foreign language pronunciation score of the user; and outputting a foreign language learning suggestion based on the foreign language pronunciation score of the user and the user audio, thereby improving the user experience.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a method and device for assisting foreign language learning based on an AI large language model. Background Art

[0002] With the acceleration of globalization, the importance of foreign language learning is becoming increasingly prominent. Foreign languages are not only a crucial tool for cross-cultural communication but also a critical skill for acquiring cutting-edge international knowledge and enhancing personal competitiveness. However, traditional foreign language learning methods face numerous challenges, such as limited learning resources, difficulty in personalizing learning progress, and a lack of authentic language environments. These factors make it difficult for many learners to effectively master a foreign language, resulting in suboptimal learning outcomes.

[0003] In recent years, the rapid development of artificial intelligence technology has brought new opportunities for foreign language learning. The emergence of AI-powered large language models, in particular, has provided new insights and approaches to foreign language learning. Large language models are deep learning models trained on massive amounts of text data, capable of understanding and generating natural language text. Their powerful language processing capabilities hold broad application prospects in foreign language learning.

[0004] The traditional foreign language teaching model, which primarily relies on teacher explanations and student practice, has several limitations. First, learning resources are relatively limited and struggle to meet the individual needs of different learners. Second, the learning schedule is typically set by the teacher, making it difficult to adapt to each student's level and learning speed. Furthermore, the lack of an authentic language environment is a major challenge in traditional foreign language learning, making it difficult for learners to flexibly apply what they have learned to real-world communication.

[0005] The emergence of AI-powered large language models offers a potential solution to these problems. Large language models can generate personalized learning content and practice materials based on learners' language proficiency, learning style, and interests. This personalized instruction not only increases learners' interest and motivation but also effectively improves learning outcomes.

[0006] In addition, the large language model can simulate a real language environment, providing learners with an immersive learning experience. By interacting with the AI model, learners can practice speaking and writing in a virtual environment, thereby better mastering the practical application of foreign languages.

[0007] In practical applications, AI big language models can be used in a variety of foreign language learning scenarios. For example, in vocabulary learning, big language models can intelligently recommend words and phrases that need review based on the learner's memory status. In grammar learning, big language models can generate example sentences that conform to grammatical rules and provide detailed analysis. In oral practice, big language models can engage in real-time conversations with learners, correcting pronunciation and grammatical errors.

[0008] The application of AI large language models not only improves learning efficiency and quality, but also provides richer and more diverse learning resources for foreign language learning. Learners can access personalized learning content anytime and anywhere through the internet, no longer restricted by traditional textbooks and classrooms.

[0009] With the continuous advancement of AI technology, the application of large language models in foreign language learning will become more extensive and in-depth. In the future, it is foreseeable that AI large language models will be combined with technologies such as virtual reality (VR) and augmented reality (AR) to create a more realistic language learning environment for learners. At the same time, AI large language models will also become more intelligent, better able to understand learners' needs and provide more precise personalized learning solutions.

[0010] In short, the AI big language model has brought unprecedented opportunities for foreign language learning. Its advantages in personalized teaching, immersive learning, and intelligent tutoring will greatly promote the development of foreign language learning. Summary of the Invention

[0011] The present invention provides a method and device for assisting foreign language learning based on an AI large language model to solve the problems of limited foreign language learning resources and difficulty in meeting personalized needs.

[0012] In a first aspect, the present invention provides a method for assisting foreign language learning based on an AI large language model, specifically comprising the following steps:

[0013] Step S1: obtaining a foreign language dialogue scene;

[0014] Step S2: Obtain a foreign language voice request input by the user;

[0015] Step S3: Recognize the foreign language voice request output by the user to form a prompt word Prompt;

[0016] Step S4: input the prompt word Prompt into the AI large language model, generate an answer through the AI large language model, and output the voice of the answer;

[0017] Step S5: Repeat steps S2 - S4 to form multiple rounds (in this application, one round of conversation represents one user request and one answer generated by the AI large language model) of foreign language voice conversations, and save the user audio in the multiple rounds of conversations;

[0018] Step S6: Evaluate the user's foreign language pronunciation based on the user audio, and form and output the user's foreign language pronunciation score;

[0019] Step S7: Output suggestions for foreign language learning based on the user's foreign language pronunciation score and the user audio.

[0020] Preferably, in step S1, the conversation scenarios include scenarios such as oral foreign teachers, children's companionship, sentence correction, study abroad partners, simultaneous interpretation, text translation, English grammar, learning slang, memorizing words, naming, holiday blessings, sentence polishing, travel English, visa interviews, calling a car, dining, shopping, etc.

[0021] Preferably, in step S3, the foreign language voice request output by the user is recognized to form a prompt Prompt, which specifically includes the following steps:

[0022] Step S301: Process the user's foreign language voice request through speech-to-text (ASR) technology to form a request text;

[0023] Step S302: Filter sensitive words and normalize the request text to form a processed request text;

[0024] Step S303: Identify the keywords in the processed request text, and generate a prompt Prompt based on the keywords.

[0025] Preferably, in step S302, filtering sensitive words and normalizing the request text to form a processed request text specifically includes the following steps:

[0026] Step S302a: Filter sensitive words from the request text to form a first text; where the sensitive words include sensitive vocabulary such as violence and politically sensitive words;

[0027] Step S302b: Remove redundant characters from the first text to form a second text; where the redundant characters include characters such as spaces, line breaks, and special symbols;

[0028] Step S302c: Check for spelling mistakes in the second text (for example, "请帮我" is input as "请帮窝") to form a processed request text.

[0029] Preferably, in step S303, keywords of the processed request text are identified by using natural language processing (NLP) technology.

[0030] Preferably, in step S6, the evaluation includes relevance evaluation, fluency evaluation, vocabulary evaluation, grammar evaluation and pronunciation evaluation.

[0031] Preferably, the relevance evaluation is performed by recognizing the user audio and judging whether the user's answer conforms to the question-answering scenario through a large language model, and providing an evaluation result of relevance or irrelevance.

[0032] Preferably, the fluency evaluation calculates the speaking speed and gives the fluency evaluation result by identifying the duration, number of words, number of pause words, and number of empty pauses of the user data recording audio.

[0033] Preferably, the vocabulary assessment identifies the word content of the user data recording audio, maps the words to the CEFR (Common European Framework of Reference for Languages) grading system, analyzes the number of words contained in each level, the number of words occupied by the A2+ level, and gives the vocabulary assessment results.

[0034] Preferably, the grammar evaluation recognizes the text content of the recorded audio input by the user, analyzes the grammatical errors in the text expression by a large language model, and outputs the grammar evaluation result.

[0035] Preferably, the pronunciation evaluation is performed by identifying the content and pronunciation of each word in the user data recording audio, comparing the content and pronunciation with the standard pronunciation, and outputting the pronunciation evaluation result.

[0036] Among them, the fluency assessment, vocabulary assessment, grammar assessment and pronunciation assessment are all scored, with a score range of 1-9 points. The higher the score, the better the assessment result.

[0037] Preferably, the user's foreign language pronunciation score is obtained based on the fluency evaluation results, vocabulary evaluation results, grammar evaluation results and pronunciation evaluation results of the foreign language speech output by the user.

[0038] In a second aspect, the present invention further provides a device for assisting foreign language learning based on an AI large language model, specifically comprising the following modules:

[0039] A foreign language dialogue scene acquisition module is used to acquire foreign language dialogue scenes;

[0040] A foreign language voice acquisition module is used to acquire foreign language voice requests output by users;

[0041] A prompt word generation module, configured to recognize the foreign language voice request output by the user and generate a prompt word Prompt;

[0042] An answer generation module, configured to input the prompt word Prompt into the AI large language model, generate an answer through the AI large language model, and output the voice of the answer;

[0043] A multi-round dialogue generation module, configured to repeatedly execute the foreign language speech acquisition module, the prompt word generation module, and the answer generation module to form multiple rounds of foreign language speech dialogues and save the user audio in the multiple rounds of dialogues;

[0044] A foreign language pronunciation scoring module is used to evaluate the user's foreign language pronunciation based on the user audio, and form and output the user's foreign language pronunciation score;

[0045] The foreign language learning suggestion generating module is used to output foreign language learning suggestions based on the user's foreign language pronunciation score and the user audio.

[0046] Preferably, in the foreign language dialogue scene acquisition module, the dialogue scenes include oral foreign teachers, child companionship, sentence correction, study abroad companionship, simultaneous interpretation, text translation, English grammar, learning slang, memorizing words, naming, holiday greetings, sentence polishing, travel English, visa interviews, taxi hailing, dining, shopping and other scenes.

[0047] Preferably, the prompt word generation module specifically includes the following submodules:

[0048] A first submodule for generating prompt words is used to process the user's foreign language voice request through speech-to-text (ASR) technology to form a request text;

[0049] The second submodule of prompt word generation is used to filter and normalize the request text for sensitive words to form a processed request text;

[0050] The third submodule of prompt word generation is used to identify keywords of the processed request text and generate a prompt word Prompt according to the keywords.

[0051] Preferably, the prompt word generation second submodule specifically includes the following units:

[0052] The first unit is configured to filter the request text for sensitive words to form a first text; wherein the sensitive words include violence, politically sensitive words, and other sensitive words;

[0053] A second unit is configured to remove redundant characters from the first text to form a second text; wherein the redundant characters include spaces, line breaks, special symbols, and the like;

[0054] The third unit is configured to check the second text for spelling errors and form a processed request text.

[0055] Preferably, in the third submodule of prompt word generation, keywords of the processed request text are identified by natural language processing technology.

[0056] Preferably, in the foreign language pronunciation scoring module, the evaluation includes relevance evaluation, fluency evaluation, vocabulary evaluation, grammar evaluation and pronunciation evaluation.

[0057] Preferably, the relevance evaluation is performed by recognizing the user audio and judging whether the user's answer conforms to the question-answering scenario through a large language model, and providing an evaluation result of relevance or irrelevance.

[0058] Preferably, the fluency evaluation calculates the speaking speed and gives the fluency evaluation result by identifying the duration, number of words, number of pause words, and number of empty pauses of the user data recording audio.

[0059] Preferably, the vocabulary assessment identifies the word content of the user data recording audio, corresponds the words to the CEFR grading system, analyzes the number of words contained in each level, the number of words occupied by the A2+ level, and gives the vocabulary assessment results.

[0060] Preferably, the grammar evaluation recognizes the text content of the recorded audio input by the user, analyzes the grammatical errors in the text expression by a large language model, and outputs the grammar evaluation result.

[0061] Preferably, the pronunciation evaluation is performed by identifying the content and pronunciation of each word in the user data recording audio, comparing the content and pronunciation with the standard pronunciation, and outputting the pronunciation evaluation result.

[0062] Among them, the fluency assessment, vocabulary assessment, grammar assessment and pronunciation assessment are all scored, with a score range of 1-9 points. The higher the score, the better the assessment result.

[0063] Preferably, the user's foreign language pronunciation score is obtained based on the fluency evaluation results, vocabulary evaluation results, grammar evaluation results and pronunciation evaluation results of the foreign language speech output by the user.

[0064] In a third aspect, the present invention further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a method for assisting foreign language learning based on an AI large language model as described in any one of the first aspects of the present application.

[0065] In a fourth aspect, the present invention also provides an electronic device, comprising: a memory storing a computer program; and a processor communicatively connected to the memory, for executing, when calling the computer program, a method for assisting foreign language learning based on an AI large language model as described in any one of the first aspects of this application.

[0066] Compared with the prior art, the present invention has the following obvious outstanding substantial features and significant advantages:

[0067] The present invention provides a method and device for assisting foreign language learning based on an AI large language model, addressing the issues of limited foreign language learning resources and the difficulty in meeting personalized needs. Specifically, the method involves acquiring a foreign language conversation scene and the foreign language speech output by the user, forming a prompt word and inputting it into the AI large language model; generating an answer through the AI large language model and outputting the speech of the answer; repeating the above steps to form and save multiple rounds of foreign language speech conversations; evaluating the user's foreign language pronunciation based on the user audio, forming and outputting a foreign language pronunciation score for the user; and outputting foreign language learning suggestions based on the user's foreign language pronunciation score and the user audio, thereby improving the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] The accompanying drawings, which constitute part of the present invention, are provided to provide a further understanding of the present invention. 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 accompanying drawings:

[0069] Figure 1 This is a flow chart of a method for assisting foreign language learning based on an AI large language model according to a preferred embodiment of the present invention.

[0070] Figure 2 This is a schematic diagram of the structure of a device for assisting foreign language learning based on an AI large language model according to a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0071] The present invention provides a method and apparatus for assisting foreign language learning based on an AI large language model. To clarify the objectives, technical solutions, and effects of the present invention, the present invention is further described below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are intended only to illustrate the present invention and are not intended to limit the present invention.

[0072] It should be noted that the terms "first," "second," and the like in the specification and claims of the present invention and the accompanying 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 terms used in this manner are interchangeable under appropriate circumstances. In addition, the terms "including," "having," and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to these processes, methods, products, or apparatuses.

[0073] Example 1:

[0074] like Figure 1 As shown, the method for assisting foreign language learning based on an AI large language model described in this embodiment specifically includes the following steps:

[0075] Step S1, obtaining foreign language conversation scenarios; wherein the conversation scenarios include oral foreign language teaching, child companionship, sentence correction, study abroad companionship, simultaneous interpretation, text translation, English grammar, learning slang, memorizing words, naming, holiday greetings, sentence polishing, travel English, visa interviews, taxi hailing, dining, shopping and other scenarios.

[0076] Step S2: Obtain a foreign language voice request input by the user;

[0077] Step S3: Recognize the foreign language voice request output by the user to form a prompt word Prompt.

[0078] Optionally, the step S3 specifically includes the following steps S301 to S303.

[0079] Step S301: Process the user's foreign language voice request through speech-to-text (ASR) technology to form a request text.

[0080] Step S302: Filter sensitive words and normalize the request text to form a processed request text;

[0081] Preferably, in step S302, sensitive word filtering and normalization processing are performed on the request text to form a processed request text, which specifically includes the following steps:

[0082] Step S302a: filtering the request text for sensitive words to form a first text; wherein the sensitive words include violence, politically sensitive words, and other sensitive words;

[0083] Step S302b: removing redundant characters from the first text to form a second text; wherein the redundant characters include spaces, line breaks, special symbols, and the like;

[0084] Step S302c: Check for spelling mistakes in the second text (e.g., "请帮我" is input as "请帮窝"), and form the processed request text.

[0085] Step S303: Identify the keywords in the processed request text, and generate a prompt word Prompt based on the keywords. In this embodiment, natural language processing technology is used to identify the keywords in the processed request text.

[0086] Step S4: Input the prompt word Prompt into the AI large language model, and generate an answer through the AI large language model and output the voice of the answer; [[ID=⑨]] [[ID=⑩]]

[0087] [[ID=⑪]]Step S5: Repeat Step S2 - Step S4 to form multiple rounds of foreign language voice conversations, and save the user audio in the multiple rounds of conversations; [[ID=⑫]] [[ID=⑬]]

[0088] [[ID=⑭]]Step S6: Evaluate the foreign language pronunciation of the user based on the user audio, and form and output the foreign language pronunciation score of the user; [[ID=⑮]] [[ID=⑯]]

[0089] [[ID=⑰]]Among them, the evaluation includes relevance evaluation, fluency evaluation, vocabulary evaluation, grammar evaluation, and pronunciation evaluation. [[ID=⑱]] [[ID=⑲]]

[0090] [[ID=⑳]]The relevance evaluation determines whether the user's answer conforms to the Q&A scenario by identifying the user audio and using the large language model, and gives an evaluation result of relevant or irrelevant. [[ID=㉑]] [[ID=㉒]]

[0091] [[ID=㉓]]The fluency evaluation calculates the speaking speed by identifying the duration, number of words, number of pause words, and number of empty pauses in the user data recording audio, and gives a fluency evaluation result. The fluency evaluation result is given on a scale of 1 - 9 points in combination with the number of pauses. The faster the speaking speed and the fewer the pauses, the higher the score. The fluency evaluation criteria are shown in Table 1. [[ID=㉔]] [[ID=㉕]]

[0092] [[ID=㉖]]Table 1 Fluency Evaluation Criteria [[ID=㉗]] [[ID=㉘]]

[0093] [[ID=㉙]] [[ID=㉚]] [[ID=㉛]]

[0094] [[ID=㉜]] [[ID=㉝]] [[ID=㉞]]

[0095] As shown in Figure 1, when the average effective word volume per minute exceeds 150 words, the score is 9 points, when the average effective word volume per minute is 130-150 words, the score is 8 points, when the average effective word volume per minute exceeds 110-130 words, the score is 7 points, when the average effective word volume per minute exceeds 90-110 words, the score is 6 points, when the average effective word volume per minute exceeds 80-90 words, the score is 5 points, when the average effective word volume per minute exceeds 70-80 words, the score is 4 points, when the average effective word volume per minute exceeds 60-70 words, the score is 3 points, when the average effective word volume per minute exceeds 50-60 words, the score is 2 points, and when the average effective word volume per minute is less than 50 words, the score is 1 point. Wherein, the effective words refer to words that are pronounced correctly, non-repeated, and non-filler words.

[0096] The vocabulary assessment identifies the word content of the user data recording audio, maps the words to the CEFR grading system, analyzes the number of words contained in each level and the number of words occupied by the A2+ level, and gives an assessment result between 1 and 9 points. The more advanced vocabulary is used, the higher the score. The vocabulary assessment standards are shown in Table 2.

[0097] Table 2 Vocabulary evaluation criteria

[0098]

[0099]

[0100] Among them, A2+ and B2+ mean that A2+ is an advanced version of A2, close to but not quite reaching B1. B2+ is an advanced version of B2, close to but not quite reaching C1.

[0101] The grammar assessment recognizes the text content of the recorded audio input by the user, analyzes the grammatical errors in the text expression using a large language model, and outputs a grammar assessment result between 1 and 9 points. The fewer grammatical errors, the higher the score. The grammar assessment criteria are shown in Table 3. In addition, modification suggestions can be provided.

[0102] Table 3 Syntax evaluation criteria

[0103]

[0104]

[0105] Among them, the proportion of speech errors is the proportion of sentences with grammatical errors in all dialogue sentences.

[0106] The pronunciation evaluation identifies the content and pronunciation of each word in the user data recording audio, compares it with the standard pronunciation, and marks the pronunciation of each word as excellent, good, and poor. The evaluation results are given between 1 and 9 points based on the proportion of words in the three levels. The higher the proportion of words with excellent pronunciation, the higher the score. The pronunciation evaluation standards are shown in Table 4.

[0107] Table 4 Pronunciation evaluation criteria

[0108]

[0109] In this embodiment, the user's foreign language pronunciation score is calculated based on the fluency, vocabulary, grammar, and pronunciation evaluation results of the foreign language speech output by the user. Specifically, the foreign language pronunciation score = (fluency score + vocabulary score + grammar score + pronunciation score) / 4, with the result rounded down.

[0110] Step S7: outputting foreign language learning suggestions based on the user's foreign language pronunciation score and the user audio.

[0111] Optionally, the method of assisting foreign language learning based on an AI large language model described in this embodiment can also realize functions such as dialogue content translation, dialogue content pronunciation, dialogue content grammar correction, and dialogue content collection. (1) Dialogue content translation: The user selects a dialogue content and can translate between Chinese and English content. Select a word and view the dictionary information of the word, including British and American pronunciation, part of speech, meaning, examples, etc.; (2) Dialogue content pronunciation: The user selects a dialogue content and can pronounce it aloud using the timbre of the current scene character. This facilitates user learning and comparison with their own pronunciation; (3) Dialogue content grammar correction: The user selects a dialogue content and can check whether the content has grammatical errors, where the errors are, and how to modify and correct them; (4) Dialogue content collection: The user can collect a dialogue content or a word, and after collection, it can be permanently saved and repeatedly viewed and studied.

[0112] Example 2:

[0113] like Figure 2 As shown, the device for assisting foreign language learning based on an AI large language model described in this embodiment specifically includes a foreign language dialogue scene acquisition module, a foreign language speech acquisition module, a prompt word generation module, an answer generation module, a multi-round dialogue generation module, a foreign language pronunciation scoring module, and a foreign language learning suggestion generation module.

[0114] The foreign language conversation scene acquisition module is used to acquire foreign language conversation scenes; wherein, the conversation scenes include oral foreign language teaching, child companionship, sentence correction, study abroad companionship, simultaneous interpretation, text translation, English grammar, learning slang, memorizing words, naming, holiday greetings, sentence polishing, travel English, visa interviews, taxi hailing, dining, shopping and other scenes.

[0115] The foreign language voice acquisition module is used to obtain the foreign language voice request input by the user.

[0116] The prompt word generation module is used to recognize the foreign language voice request output by the user to form a prompt word Prompt.

[0117] The prompt word generation module specifically includes a prompt word generation first submodule, a prompt word generation second submodule and a prompt word generation third submodule.

[0118] The prompt word generation first submodule is used to process the user's foreign language voice request through speech-to-text (ASR) technology to form a request text.

[0119] The second submodule of prompt word generation is used to filter and normalize sensitive words on the request text to form a processed request text.

[0120] The prompt word generates a second submodule, which specifically includes a first unit, a second unit and a third unit.

[0121] The first unit is configured to filter the request text for sensitive words to form a first text; wherein the sensitive words include violence, politically sensitive words, and other sensitive words;

[0122] A second unit is configured to remove redundant characters from the first text to form a second text; wherein the redundant characters include spaces, line breaks, special symbols, and the like;

[0123] The third unit is configured to check the second text for spelling errors and form a processed request text.

[0124] The third submodule of prompt word generation is used to identify keywords of the processed request text and generate prompt words according to the keywords, wherein the keywords of the processed request text are identified by natural language processing technology.

[0125] The answer generation module is used to input the prompt word Prompt into the AI large language model, generate an answer through the AI large language model and output the voice of the answer.

[0126] The multi-round dialogue generation module is used to repeatedly execute the foreign language voice acquisition module, the prompt word generation module and the answer generation module to form multiple rounds of foreign language voice dialogues and save the user audio in the multiple rounds of dialogues.

[0127] The foreign language pronunciation scoring module is used to evaluate the user's foreign language pronunciation based on the user audio, and form and output the user's foreign language pronunciation score.

[0128] The assessment includes relevance assessment, fluency assessment, vocabulary assessment, grammar assessment and pronunciation assessment.

[0129] The relevance evaluation recognizes the user's audio and uses a large language model to determine whether the user's answer is consistent with the question-answering scenario, and gives an evaluation result of whether it is relevant or not.

[0130] The fluency evaluation is performed by identifying the duration, number of words, number of pause words, and number of empty pauses of the user data recording audio, calculating the speaking speed and giving the fluency evaluation result.

[0131] The vocabulary assessment identifies the word content of the user data recording audio, maps the words to the CEFR grading system, analyzes the number of words contained in each level, the number of words occupied by the A2+ level, and provides the vocabulary assessment results.

[0132] The grammar evaluation recognizes the text content of the recorded audio input by the user, analyzes the grammatical errors in the text expression by a large language model, and outputs the grammar evaluation result.

[0133] The pronunciation evaluation is performed by identifying the content and pronunciation of each word in the user data recording audio, comparing it with the standard pronunciation, and outputting the pronunciation evaluation result.

[0134] Among them, the fluency assessment, vocabulary assessment, grammar assessment and pronunciation assessment are all scored, with a score range of 1-9 points. The higher the score, the better the assessment result.

[0135] The user's foreign language pronunciation score is obtained based on the fluency evaluation results, vocabulary evaluation results, grammar evaluation results and pronunciation evaluation results of the foreign language speech output by the user.

[0136] The foreign language learning suggestion generating module is used to output foreign language learning suggestions based on the user's foreign language pronunciation score and the user audio.

[0137] While the specific embodiments of the present invention have been described in detail above, these are merely exemplary and the present invention is not limited thereto. For those skilled in the art, any equivalent modifications and substitutions to the present invention are also within the scope of the present invention. Therefore, any equivalent changes and modifications made without departing from the spirit and scope of the present invention are intended to be encompassed within the scope of the present invention.

Claims

1. A method for assisting foreign language learning based on an AI large language model, characterized in that: The specific steps include: Step S1: obtaining a foreign language dialogue scene; Step S2: Obtain a foreign language voice request input by the user; Step S3: Recognize the foreign language voice request output by the user to form a prompt word Prompt; Step S4: input the prompt word Prompt into the AI large language model, generate an answer through the AI large language model, and output the voice of the answer; Step S5: Repeat steps S2 to S4 to form multiple rounds of foreign language voice dialogues, and save the user audio in the multiple rounds of dialogues; Step S6: Evaluate the user's foreign language pronunciation based on the user audio, and generate and output a foreign language pronunciation score for the user; Step S7: outputting foreign language learning suggestions based on the user's foreign language pronunciation score and the user audio.

2. The method of assisting foreign language learning based on an AI large language model according to claim 1, characterized in that: In step S1, the dialogue scenarios include oral English teaching, child companionship, sentence correction, study abroad companionship, simultaneous interpretation, text translation, English grammar, learning slang, memorizing words, naming, holiday greetings, sentence polishing, travel English, visa interview, taxi hailing, dining, and shopping scenarios.

3. The method for assisting foreign language learning based on an AI large language model according to claim 1, characterized in that: In step S3, the foreign language voice request output by the user is recognized to form a prompt word Prompt, which specifically includes the following steps: Step S301: Processing the user's foreign language voice request through speech-to-text (ASR) technology to generate a request text; Step S302: Filter sensitive words and normalize the request text to form a processed request text; Step S303: Identify keywords in the processed request text and generate a prompt word Prompt according to the keywords.

4. The method of assisting foreign language learning based on an AI large language model according to claim 3, characterized in that: In step S302, sensitive word filtering and normalization processing are performed on the request text to form a processed request text, which specifically includes the following steps: Step S302a: filtering the request text for sensitive words to form a first text; wherein the sensitive words include violence, politically sensitive words, and other sensitive words; Step S302b: removing redundant characters from the first text to form a second text; wherein the redundant characters include spaces, line breaks, special symbols, and the like; Step S302c: Check the second text for spelling errors to form a processed request text.

5. The method of assisting foreign language learning based on an AI large language model according to claim 3, characterized in that: In step S303, keywords of the processed request text are identified using natural language processing technology.

6. The method of assisting foreign language learning based on an AI large language model according to claim 1, characterized in that: In step S6, the evaluation includes topic relevance evaluation, fluency evaluation, vocabulary evaluation, grammar evaluation and pronunciation evaluation; The on-topic evaluation is performed by recognizing the user's audio and judging whether the user's answer is consistent with the question-answering scenario through a large language model, and providing an on-topic or off-topic evaluation result; The fluency assessment is performed by identifying the duration, number of words, number of pause words, and number of empty pauses of the user data recording audio, calculating the speaking speed and giving the fluency assessment result; The vocabulary assessment identifies the word content of the user data recording audio, maps the words to the CEFR grading system, analyzes the number of words contained in each level, the number of words occupied by the A2+ level, and provides the vocabulary assessment results; The grammar evaluation recognizes the text content of the recorded audio input by the user, analyzes the grammatical errors in the text expression by a large language model, and outputs the grammar evaluation result; The pronunciation evaluation is performed by identifying the content and pronunciation of each word in the user data recording audio, comparing it with the standard pronunciation, and outputting the pronunciation evaluation result; Among them, the fluency assessment, vocabulary assessment, grammar assessment and pronunciation assessment are all scored, with a score range of 1-9 points. The higher the score, the better the assessment result.

7. The method of assisting foreign language learning based on an AI large language model according to claim 6, characterized in that: The user's foreign language pronunciation score is obtained based on the fluency evaluation results, vocabulary evaluation results, grammar evaluation results and pronunciation evaluation results of the foreign language speech output by the user.

8. A device for assisting foreign language learning based on an AI large language model, characterized in that: Specifically, it includes the following modules: A foreign language dialogue scene acquisition module is used to acquire foreign language dialogue scenes; A foreign language voice acquisition module is used to acquire foreign language voice requests output by users; A prompt word generation module, configured to recognize the foreign language voice request output by the user and generate a prompt word Prompt; An answer generation module, configured to input the prompt word Prompt into the AI large language model, generate an answer through the AI large language model, and output the voice of the answer; A multi-round dialogue generation module, configured to repeatedly execute the foreign language speech acquisition module, the prompt word generation module, and the answer generation module to form multiple rounds of foreign language speech dialogues and save the user audio in the multiple rounds of dialogues; A foreign language pronunciation scoring module is used to evaluate the user's foreign language pronunciation based on the user audio, and form and output the user's foreign language pronunciation score; The foreign language learning suggestion generating module is used to output foreign language learning suggestions based on the user's foreign language pronunciation score and the user audio.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, which, when executed by a processor, implements a method for assisting foreign language learning based on an AI large language model as described in any one of claims 1 to 7.

10. An electronic device, characterized in that: The method comprises a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for assisting foreign language learning based on an AI large language model as described in any one of claims 1 to 7 is implemented.

Images