Language memory card system with pronunciation error correction function

Abstract

The invention provides a language memory card system with a pronunciation error correction function, and relates to the technical field of language training. The language memory card system with the pronunciation error correction function comprises a user portrait and configuration system, a core learning and interaction system and a background management and optimization system which are bidirectionally linked to realize refined adaptation of dialects and crowd pronunciation and memory-pronunciation collaborative reinforcement. The user portrait and configuration system is used for collecting user basic information, completing a pronunciation baseline test, generating personalized configuration parameters and outputting the personalized configuration parameters to the core learning and interaction system; and the core learning and interaction system receives personalized configuration parameters. The system has the dialect and crowd refined adaptation capability, the pronunciation error correction precision is high, memory and pronunciation collaborative enhancement is achieved, personalized adjustment and dynamic optimization are supported, data safety is guaranteed, and the language learning efficiency and normalization are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of language training, in particular to a language memory card system with pronunciation correction function. BACKGROUND

[0002] In the context of increasingly frequent global communication, language learning has become an important demand in education, the workplace and daily communication, and pronunciation accuracy is a core component of language application ability. Non-native learners often face the problem of pronunciation correction difficulty in the process of language learning: on the one hand, learners in different regions are affected by the transfer of dialect pronunciation habits, and are prone to systematic pronunciation deviations, such as confusion between / l / and / r / among learners in some southern regions, and unclear flat and raised tongue pronunciation among learners in the north. Such deviations are deeply bound to the pronunciation system of the mother tongue, and conventional learning tools cannot accurately identify them. On the other hand, the pronunciation characteristics of special groups such as children and the elderly have unique features. Children's pronunciation is indistinct and unstable in volume, and the elderly's voice tone is shifted and their speech speed is slowed down due to aging of the vocal cords. Existing tools lack targeted pronunciation adaptation mechanisms, often resulting in overly strict or overly broad correction standards.

[0003] Among existing language learning tools, traditional memory card systems mainly focus on vocabulary meaning and spelling memory, lack pronunciation collection and correction functions, and cannot meet users' pronunciation training needs. Although conventional speech recognition tools can achieve basic pronunciation scoring, they use a unified standard model and do not consider dialect backgrounds and population differences, resulting in insufficient correction accuracy and easy misjudgment or omission. Some learning software with pronunciation functions only correct isolated pronunciation points and do not establish a coordination mechanism with memory algorithms. Users can obtain pronunciation feedback, but cannot consolidate the correction effect through memory reinforcement, and lack dynamic optimization of feature libraries and model support, making it difficult to adapt to diverse pronunciation scenarios and user needs. In addition, existing tools have shortcomings in data security protection and user individualized demand response. Some products do not encrypt sensitive data such as user voiceprints, or lack flexible correction parameter adjustment functions, further limiting the efficiency and experience of language learning. The existence of these problems makes it difficult for existing tools to achieve deep integration of pronunciation correction and memory reinforcement, and cannot provide accurate and efficient language learning solutions for learners of different backgrounds and types. Therefore, developing a language memory card system that can adapt to dialects and population characteristics and achieve memory-pronunciation collaborative reinforcement is an urgent need in the current language learning field. SUMMARY

[0004] TECHNICAL PROBLEM SOLVED To overcome the deficiencies of the prior art, the present application provides a language memory card system with pronunciation correction function, which solves the problems of insufficient dialect and population pronunciation adaptation and low correction accuracy of existing tools.

[0005] Technical solution To achieve the above object, the present application is implemented by the following technical solution: a language memory card system with pronunciation correction function, comprising a user portrait and configuration system, a core learning and interaction system, and a background management and optimization system, the three systems are bidirectionally linked to realize fine adaptation of dialect and crowd pronunciation and memory-pronunciation collaborative reinforcement; The user portrait and configuration system is used to collect user basic information, complete pronunciation baseline test and generate personalized configuration parameters, and output to the core learning and interaction system; The core learning and interaction system receives personalized configuration parameters, performs memory card pushing, pronunciation collection preprocessing, hierarchical pronunciation correction of dialect and crowd adaptation, personalized feedback guidance, and synchronously records learning data and feeds back to the background management and optimization system; The background management and optimization system receives feedback data from the core learning and interaction system, completes feature library updating, model iteration and system operation and maintenance, and outputs optimized feature library and model to reversely support the first two systems; The core learning and interaction system comprises a dialect and crowd adaptation pronunciation correction module, which is configured with a dialect-foreign language pronunciation transfer feature library and a crowd exclusive pronunciation model, can distinguish dialect-related bias, crowd characteristic bias and pure pronunciation error, and perform differentiated dynamic fault tolerance judgment.

[0006] Preferably, the user portrait and configuration system comprises a user information management module, a pronunciation baseline test module and a personalized configuration module; the user information management module comprises a basic information input unit and an account privacy management unit, the basic information input unit is used to collect user learning goals, dialect background, crowd type, commonly used equipment and learning scene information, and the account privacy management unit is used to realize user registration and login, pronunciation data encrypted storage and data deletion functions; the pronunciation baseline test module comprises a test content pushing unit, a pronunciation sample collection unit and a baseline analysis unit, the test content pushing unit generates 5-10 words / sentences covering core pronunciation points, the pronunciation sample collection unit adapts to the pronunciation input needs of children and the elderly, and the baseline analysis unit determines dialect and crowd type in combination with user annotation information and audio features.

[0007] Preferably, the core learning and interaction system further comprises a memory card management module, a pronunciation collection and preprocessing module, a personalized feedback and correction module, and a memory and pronunciation collaborative iteration module; the memory card management module comprises a word library matching unit, a card content display unit and a self-defined card unit, the word library matching unit pushes corresponding test, daily or professional word library according to user learning goals, the card content display unit presents standard pronunciation audio, mouth shape demonstration and example sentences, and the self-defined card unit supports user to upload new words and edit card content.

[0008] Preferably, the dialect and population adaptation pronunciation correction module comprises a dialect feature library calling unit, a population-specific model calling unit, a hierarchical matching verification unit, and a dynamic fault tolerance determination unit; the dialect feature library calling unit matches the pronunciation deviation mapping table corresponding to the user's dialect, the population-specific model calling unit switches the pronunciation model specific to children, adults or the elderly, the hierarchical matching verification unit distinguishes between dialect-related deviations, population characteristic deviations and pure pronunciation errors through the cooperation of the feature library and the specific model, and the dynamic fault tolerance determination unit performs strict determination on core pronunciation points and differential relaxed fault tolerance on non-core deviations.

[0009] Preferably, the pronunciation collection and preprocessing module comprises a pronunciation guiding unit, a speech collection unit, and a noise filtering and feature extraction unit; the pronunciation guiding unit adapts pronunciation entry rules for different populations, the speech collection unit filters invalid sounds such as crying and noise, and the noise filtering and feature extraction unit extracts core pronunciation features such as syllables, stress, and tone after removing background noise and transmits them to the dialect and population adaptation pronunciation correction module.

[0010] Preferably, the individualized feedback and correction module comprises a correction result classification display unit, a correction auxiliary resource pushing unit, and a user preference adjustment unit; the correction result classification display unit clearly labels the types and reasons of dialect-related deviations, population characteristic deviations and pure pronunciation errors, the correction auxiliary resource pushing unit pushes comparative audio, mouth shape video and special training cards, and the user preference adjustment unit supports user manual switching of correction strictness or marking of deviations that do not need to be corrected.

[0011] Preferably, the memory and pronunciation collaborative iteration module comprises a learning data recording unit, a memory algorithm optimization unit, and a model dynamic updating unit; the learning data recording unit synchronously stores the user's word memory proficiency, pronunciation error type, error frequency and correction effect, the memory algorithm optimization unit incorporates the pronunciation error rate into the memory pushing weight, and preferentially pushes the words that are weak in both pronunciation and memory, and the model dynamic updating unit collects user pronunciation data and feeds it back to the background management and optimization system.

[0012] Preferably, the background management and optimization system comprises a feature library and model management module, a user feedback processing module and a system operation and maintenance module; the feature library and model management module comprises a dialect feature library maintenance unit, a crowd model training unit and a model iteration unit, for supplementing new dialect data and optimizing crowd-specific models; the user feedback processing module comprises a feedback collection unit, a feedback audit unit and a feedback landing unit, for receiving user error correction and reporting and landing feature library and model optimization; the system operation and maintenance module comprises a server monitoring unit, a function updating unit and a data security operation and maintenance unit, for ensuring smooth operation of the system and encryption security of the whole process data.

[0013] Advantages The application provides a language memory card system with pronunciation error correction function. 1、The application provides a language memory card system with pronunciation error correction function, which can accurately identify systematic pronunciation deviations caused by dialects and typical pronunciation characteristics of different crowds by constructing a dialect-foreign language pronunciation transfer feature library and a child, adult and elderly specific pronunciation model, combining hierarchical matching verification and dynamic fault tolerance judgment mechanism, avoiding misjudgment and omission caused by traditional unified standard model. At the same time, the system supports user self-labeling of dialect and crowd information, and supports baseline test secondary verification, further strengthens the personalized adaptation effect, strictly corrects core pronunciation points, flexibly tolerates non-core deviations, ensures the standardization of pronunciation learning, avoids discouraging user learning enthusiasm, and provides accurate and adaptive pronunciation error correction solutions for learners of different backgrounds and types.

[0014] 2、The application provides a language memory card system with pronunciation error correction function, which deeply integrates pronunciation error correction data and memory algorithm, records pronunciation error types, error times and correction effects through a memory and pronunciation cooperative iteration module, includes pronunciation error rate in memory push weight, and preferentially pushes words with weak pronunciation and memory, realizes the synchronous promotion of "memory consolidation" and "pronunciation correction". At the same time, the system continuously collects user learning data through a background management and optimization system, dynamically updates the feature library and model, combines personalized feedback and special training resource push, forms a complete closed loop from learning, error correction to review, effectively shortens the period of user vocabulary memory and pronunciation correction, significantly improves the systematicness and efficiency of language learning, and solves the problem of disconnection between memory and pronunciation training in traditional tools. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 The figure is a system flowchart of the application. DETAILED DESCRIPTION

[0016] With reference to the accompanying drawings: clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative labor fall within the scope of the present application.

[0017] As shown in the above specific embodiments, the user is an adult learner with a Cantonese background, the learning goal is daily English communication, the commonly used device is a smart phone, and the learning scene covers quiet indoor and commuting. When the user first logs in to the system, the initialization process of the user portrait and configuration system is entered: the basic information input unit pops up the information collection interface, the user checks the dialect background as “Cantonese”, the crowd type as “adult”, the learning goal as “daily English”, the commonly used device as “smart phone”, and the commonly used scene as “quiet indoor + commuting”, the account privacy management unit synchronously completes the account registration, and automatically enables the AES encryption storage mode for the subsequent collected voice data, supporting the user to initiate data deletion application at any time through the system settings. Figure 1

[0018] After completing the basic information input, the pronunciation baseline test module starts the test process: the test content pushing unit generates 8 words such as “right”, “light”, “very”, “well” and the short sentence “Let’s go for a walk” for the core pronunciation points such as “l / r”, “n / l” and “v / w” that are easily confused by Cantonese users; the pronunciation sample collection unit adapts to the input rules of adult users, sets the default input duration to 3 seconds, and the user completes word-by-word and sentence-by-sentence reading according to the interface prompts, and the collection unit synchronously filters the slight commuting noise in the environment; the baseline analysis unit detects that the user has a dialect-related deviation of “r” sound biasing to “l” sound when pronouncing “right” in combination with the user's labeled Cantonese background and the collected audio features, further verifies and confirms that the user's dialect type is “Cantonese” and the crowd type is “adult”, and finally generates personalized parameters by the personalized configuration module, including the error correction strictness as “standard mode”, the key correction direction as “l / r / v / w pronunciation”, and the pronunciation error weight ratio in the memory algorithm as 30%, and pushes them to the core learning and interaction system.

[0019]

[0020] ​​After receiving the personalized parameters, the core learning and interaction system pushes the daily English communication vocabulary library through the vocabulary library matching unit of the memory card management module. The card content display unit presents the word phonetic symbols, standard pronunciation audio (supporting slow / normal speed switching), real person lip demonstration video, and two daily scene example sentences (such as "very good" combined with shopping scene dialogue) in each card. The user finds that the vocabulary library lacks coffee-related new words such as "barista", uploads the word through the self-defined card unit, and edits and adds phonetic symbols, pronunciation audio, and example sentence "Can I speak to the barista?". Subsequently, the system enters the pronunciation practice process: the pronunciation guide unit prompts the user to read the word first and then read the example sentence, the voice collection unit filters the traffic noise during the commute, the noise filtering and feature extraction unit extracts the syllable and stress features of the user's pronunciation, finds that the user's "very" "v" sound friction is insufficient, and transmits it to the dialect and crowd adaptation pronunciation correction module. The module matches the "v / w" confusion bias mapping table for Cantonese users through the dialect feature library calling unit, switches to the adult exclusive pronunciation model through the crowd exclusive model calling unit, determines that the bias is a dialect-related bias through the hierarchical matching verification unit, and executes strict judgment due to the "v" sound as the core pronunciation point; At the same time, it is detected that the user's "barista" has correct stress position, only slight deviation in tone, and is determined as a non-core bias, and executes relaxed fault tolerance.

[0021] The error correction result classification display unit of the personalized feedback and correction module labels "[Dialect Influence]'very' 'v' sound friction is insufficient, easy to confuse 'w' sound", and the correction auxiliary resource pushing unit pushes "v / w" pronunciation comparison audio, lip close-up video, and 3 groups of special training cards (including "very-well" "visit-wisit" comparison exercises); The user thinks that the tone deviation of "barista" does not affect communication, and marks the deviation "no correction" through the preference adjustment unit, and the system synchronously records the preference. The learning data recording unit of the memory and pronunciation collaborative iteration module stores "very" pronunciation error 1 time, "barista" pronunciation without core error, "right" memory proficiency level "fuzzy" and other data, and the memory algorithm optimization unit lists "very" (pronunciation error + memory fuzzy) as high-priority review vocabulary, and "barista" (pronunciation without core error + memory proficiency) as low-priority review vocabulary. The next round of learning prioritizes the push of "very" special training cards; The model dynamic updating unit collects user "v" sound pronunciation data and feeds back to the background management and optimization system.

[0022] After the feature library and model management module of the background management and optimization system receives the data, the dialect feature library maintenance unit supplements the deviation details of the "v" sound pronunciation of the user "v" to the Cantonese feature library, the population model training unit includes the pronunciation sample of the adult user into the training data set of the adult exclusive model, and the model iteration unit optimizes the recognition threshold of the "v / w" sound; the user finds that the pronunciation of "light" is misjudged by the system in subsequent use, submits a "correct error judgment" report through the feedback collection unit, the feedback audit unit verifies that the misjudgment is a labeling error of a certain deviation mapping table in the dialect feature library, the feedback landing unit updates the mapping table, and the model recognition logic is optimized synchronously; the server monitoring unit of the system operation and maintenance module monitors the voice collection delay in the commuting scene in real time, the function update unit iteratively adds a "commuting mode" noise filtering algorithm, and the data security operation and maintenance unit regularly performs security detection on the voiceprint data stored by the user in an encrypted manner, to ensure that the data is not illegally accessed. Through the above process, the system realizes closed-loop operation from personalized adaptation, learning interaction to background optimization, and continuously improves the pronunciation error correction accuracy and language learning efficiency.

[0023] Although embodiments of the present application have been shown and described, it is to be understood that various modifications, substitutions, replacements and changes can be made to these embodiments without departing from the principles and spirit of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. A language memory card system with pronunciation correction function, characterized in that, It includes a user profiling and configuration system, a core learning and interaction system, and a backend management and optimization system. The user profiling and configuration system, the core learning and interaction system, and the backend management and optimization system work together to achieve refined adaptation of dialects and pronunciation of the population and memory-pronunciation collaborative reinforcement. The user profiling and configuration system is used to collect basic user information, complete pronunciation baseline tests, generate personalized configuration parameters, and output them to the core learning and interaction system. The core learning and interaction system receives personalized configuration parameters, performs memory card push, pronunciation acquisition preprocessing, dialect and population-adapted layered pronunciation correction, personalized feedback guidance, and simultaneously records learning data and feeds it back to the background management and optimization system. The back-end management and optimization system receives feedback data from the core learning and interaction system, completes feature library updates, model iterations and system maintenance, and outputs optimized feature libraries and models to support the first two systems. The core learning and interaction system includes a dialect and population-adapted pronunciation correction module. This module is equipped with a dialect-foreign language pronunciation transfer feature library and a population-specific pronunciation model. It can distinguish between dialect-related deviations, population feature deviations, and pure pronunciation errors, and perform differentiated dynamic error-tolerant judgment.

2. The language memory card system with pronunciation correction function according to claim 1, characterized in that, The user profiling and configuration system includes a user information management module, a pronunciation baseline test module, and a personalized configuration module. The user information management module includes a basic information input unit and an account privacy management unit. The basic information input unit is used to collect information on user learning goals, dialect background, population type, commonly used devices, and learning scenarios. The account privacy management unit is used to implement user registration and login, encrypted storage of pronunciation data, and data deletion functions. The pronunciation baseline testing module includes a test content push unit, a pronunciation sample collection unit, and a baseline analysis unit. The test content push unit generates 5-10 words / short phrases covering core pronunciation points. The pronunciation sample collection unit adapts to the pronunciation input needs of children and the elderly. The baseline analysis unit combines user annotation information and audio features to determine dialect and population type.

3. The language memory card system with pronunciation correction function according to claim 1, characterized in that, The core learning and interaction system also includes a memory card management module, a pronunciation acquisition and preprocessing module, a personalized feedback and correction module, and a memory and pronunciation collaborative iteration module. The memory card management module includes a vocabulary matching unit, a card content display unit, and a custom card unit. The vocabulary matching unit pushes corresponding test-taking, daily, or professional vocabulary according to the user's learning goals. The card content display unit presents standard pronunciation audio, mouth shape demonstrations, and example sentences. The custom card unit allows users to upload new words and edit card content.

4. The language memory card system with pronunciation correction function according to claim 1, characterized in that, The dialect and population-adapted pronunciation correction module includes a dialect feature library calling unit, a population-specific model calling unit, a hierarchical matching verification unit, and a dynamic error tolerance judgment unit. The dialect feature library calling unit matches the pronunciation deviation mapping table corresponding to the user's dialect. The population-specific model calling unit switches between pronunciation models specific to children, adults, or the elderly. The hierarchical matching verification unit distinguishes between dialect-related deviations, population-specific deviations, and pure pronunciation errors through the collaboration of the feature library and the specific model. The dynamic error tolerance judgment unit performs strict judgment on core pronunciation points and performs differentiated tolerance for non-core deviations.

5. The language memory card system with pronunciation correction function according to claim 3, characterized in that, The speech acquisition and preprocessing module includes a speech guidance unit, a speech acquisition unit, and a noise filtering and feature extraction unit; The pronunciation guidance unit adapts to the pronunciation input rules of different groups of people, the voice acquisition unit filters out crying, noise and invalid sounds, and the noise filtering and feature extraction unit removes background noise, extracts the core pronunciation features of syllables, stress and tone, and transmits them to the dialect and group-adapted pronunciation correction module.

6. The language memory card system with pronunciation correction function according to claim 3, characterized in that, The personalized feedback and correction module includes a correction result classification and display unit, a correction auxiliary resource push unit, and a user preference adjustment unit. The correction result classification and display unit clearly marks the type and cause of dialect-related deviations, population characteristic deviations, and pure pronunciation errors. The correction auxiliary resource push unit pushes comparison audio, lip-reading videos, and special training cards. The user preference adjustment unit allows users to manually switch the correction strictness or mark deviations that do not need to be corrected.

7. The language memory card system with pronunciation correction function according to claim 3, characterized in that, The memory and pronunciation collaborative iteration module includes a learning data recording unit, a memory algorithm optimization unit, and a model dynamic update unit. The learning data recording unit synchronously stores the user's word memorization proficiency, pronunciation error type, error count, and correction effect. The memory algorithm optimization unit incorporates the pronunciation error rate into the memory push weight, prioritizing the push of words with both weak pronunciation and memory. The model dynamic update unit collects user pronunciation data and feeds it back to the background management and optimization system.

8. The language memory card system with pronunciation correction function according to claim 1, characterized in that, The backend management and optimization system includes a feature library and model management module, a user feedback processing module, and a system operation and maintenance module. The feature library and model management module includes a dialect feature library maintenance unit, a population model training unit, and a model iteration unit, used to supplement and add new dialect data and optimize population-specific models. The user feedback processing module includes a feedback collection unit, a feedback review unit, and a feedback implementation unit, used to receive user error correction reports and implement them in feature library and model optimization. The system operation and maintenance module includes a server monitoring unit, a function update unit, and a data security operation and maintenance unit, used to ensure smooth system operation and end-to-end data encryption security.

Images