National musical instrument playing action AI auxiliary teaching method and system

Abstract

The invention discloses a national musical instrument playing action AI auxiliary teaching method and system, relates to the technical field of national musical instrument teaching, and constructs a musical instrument acoustics-action association model, a dynamic model evolution mechanism and cross-modal error correction feedback through a deep coupling AI technology and national musical instrument characteristics. Based on pre-acquired data, a dynamic standard action model and a style quantification model are generated, embedded sensor data are integrated to evaluate student actions, personalized correction suggestions are generated, a teaching plan is optimized by applying cultural context-driven logic to obtain an updated teaching plan, and a quantitative evaluation report is finally generated by applying a progressive quantification model. The method relates to video analysis, sensor fusion and real-time intonation detection technologies, solves the problems that a traditional teaching model is poor in static adaptability and inaccurate in feedback, and solves the limitation problem existing in the current national musical instrument AI auxiliary teaching development feasibility analysis process.

Description

Technical Field

[0001] The present application relates to the technical field of teaching national musical instruments, and specifically to an AI-assisted teaching method and system for national musical instrument playing movements. Background Art

[0002] Traditional Chinese musical instruments have attracted much attention due to their profound historical heritage, unique artistic style and wide influence. Using AI technology to assist in teaching their playing movements will help learners better understand their cultural connotations and unique value.

[0003] Prior art, such as the invention patent application CN118588045A, discloses a guzheng training and teaching method and system based on sound recognition. This relates to the field of sound recognition. The guzheng training and teaching method based on sound recognition includes the following steps: S1. Data acquisition and preprocessing; S2. Feature extraction and analysis; S3. Model training and optimization; S4. System integration; S5. Real-time feedback and evaluation; and S6. Continuous improvement and optimization. The guzheng training and teaching system based on sound recognition includes an audio input module, a sound recognition module, a learning resource module, a user interface module, a data management module, an audio output module, and a management and settings module.

[0004] In response to the above scheme, the inventors of this application found that the above technology has at least the following technical problems: 1. There is currently a lack of a dynamic model evolution mechanism through standard eigenvectors. When the dynamic attenuation factor increases, it is initially biased towards the student's physiological data. There is also a lack of gradual convergence to the master standard when the dynamic attenuation factor decreases. It cannot solve the "one-size-fits-all" teaching problem, cannot improve the model adaptability, and cannot adjust standard movements for students with different finger lengths.

[0005] 2. Currently, there is a lack of an intelligent feedback generation mechanism driven by a joint loss function that combines movement differences with real-time pitch data, as well as a lack of dynamic adaptive teaching strategies based on genres, emotions, and cultural rules. This makes it impossible to jointly build an efficient, intelligent, and culturally profound music learning environment, and it is even more impossible to effectively solve the problems of hidden errors in traditional teaching that are difficult to detect, inaccurate error correction guidance, single and rigid teaching methods, and insufficient understanding of cultural connotations. At the same time, there is a lack of freedom to dynamically adjust teaching based on genres and the emotions of the repertoire, and without integrating cultural rules, it is impossible to enhance technical effects and avoid the rigidity of mechanical correction. Summary of the Invention

[0006] In response to the above-mentioned technical deficiencies, the purpose of this application is to provide an AI-assisted teaching method and system for traditional musical instrument playing movements.

[0007] In order to solve the above technical problems, the present application adopts the following technical solutions: In the first aspect, the present application provides an AI-assisted teaching method for national musical instrument playing movements, which method includes the following steps: S1. Embed the pre-acquired national musical instrument master video data and the data in the private library data of the intangible cultural heritage school into the instrument acoustic-motion association model and dynamic model to obtain a personalized standard motion model and a genre fingerprint style quantification model.

[0008] S2. Based on the personalized standard movement model, genre fingerprint style quantification model, student performance video, performance data and instrument information, movement evaluation and style matching are performed to obtain test data including the student's movement difference data, performance level and the most similar performance master.

[0009] S3. Based on the movement difference data, performance level and the most similar master performer, combined with cultural adaptation factors and cross-modal error correction feedback mechanism, generate personalized correction suggestions.

[0010] S4. Based on the personalized correction suggestions, cultural context driving factors and preset teaching objective data, optimize the teaching plan parameters to obtain an updated teaching plan.

[0011] S5. Input the updated teaching plan, movement difference data, and pitch data into the progress quantification model to generate an evaluation report.

[0012] Preferably, the method embeds the pre-acquired video data of national musical instrument masters and the data in the private library data of intangible cultural heritage schools into the musical instrument acoustic-motion association model and the dynamic model to obtain a personalized standard motion model and a genre fingerprint style quantification model, including: based on the pre-acquired video data of national musical instrument masters and the private library data of intangible cultural heritage schools, extracting the key action points in the video data of national musical instrument masters through OpenCV frame decomposition technology to obtain the master's standard feature vector, and at the same time initializing the master's standard feature vector to obtain a fully initialized master's standard feature vector that supports the dynamic model evolution mechanism.

[0013] Acoustic mapping is defined to generate an instrument acoustic-action association model, and then timbre binding is performed to generate timbre features; based on the timbre features and the student's physiological reference vector, the dynamic model evolution mechanism is applied to update the standard feature vector to generate a personalized standard action model; based on the private library data of intangible cultural heritage genres, the genre characteristics are clustered and analyzed, and the intangible cultural heritage data is converted into quantitative indicators to generate a genre fingerprint style quantitative model.

[0014] Preferably, the musical instrument acoustic-motion association model and dynamic model evolution mechanism include: A1, through the formula of the musical instrument acoustic-motion association model Get the tone output value , Expressed as pressure intensity, Represented as an instrument material identifier, Expressed as the instrument material coefficient, Refers to the weight coefficient corresponding to the pressure intensity, Expressed as the correction factor corresponding to the timbre output value.

[0015] A2. Formula for the evolutionary mechanism through dynamic models Output standard feature vector ,in Expressed as the master standard eigenvector Represented as the student’s physiological reference vector, Expressed as a dynamic attenuation factor.

[0016] Preferably, the movement evaluation and style matching are performed to obtain test data containing the student's movement difference data, performance level and the most similar master performer, including: based on the instrument information, extracting the corresponding test performance repertoire from the preset repertoire database, the student completing the performance level test according to the test performance repertoire, and then obtaining the student's performance video data and performance data.

[0017] Based on the timestamp synchronization mechanism, the student's performance video data and performance data are temporally and spatially aligned to obtain a synchronized data set, which is input into a personalized standard movement model to output the student's movement difference data and performance level.

[0018] Based on the genre fingerprint style quantification model, the similarity between the student's playing style and the master's genre characteristics is matched to output the most similar performing master.

[0019] Preferably, the method generates personalized correction suggestions based on the movement difference data, performance level and the most similar master performers, combined with cultural adaptation factors and cross-modal error correction feedback mechanism, including: matching practice repertoire and practice frequency based on the movement difference data, performance level and the most similar master performers to formulate a preliminary learning plan, dynamically adjusting the teaching freedom based on the cultural context driving factor to obtain the teaching frequency; monitoring the student's execution process based on rhythm synchronization characteristics; performing joint loss calculation on movement differences and pitch deviations based on the cross-modal error correction feedback mechanism to generate personalized correction suggestions; and dynamically optimizing the suggestion weights based on the accumulation of historical learning data to improve feedback accuracy.

[0020] Preferably, the matching of practice repertoire and practice frequency to formulate a preliminary learning plan, dynamically adjusting the teaching freedom based on the cultural context driving factor to obtain the teaching frequency, includes: B1, according to the student's performance level, selecting the practice repertoire of the master's style that is most similar to the student from the preset repertoire library; according to the calculation formula The practice frequency is obtained, where Expressed as the practice adjustment coefficient, Expressed as the standard value of action difference data.

[0021] B2. According to the calculation formula The cultural context driving factors are obtained, among which Expressed as the genre coefficient, Expressed as the track emotion coefficient, and They are respectively expressed as the weight factor corresponding to the genre coefficient and the weight factor corresponding to the track emotion coefficient.

[0022] Preferably, the cross-modal error correction feedback mechanism includes: based on the loss function, defining the calculation formula of the cross-modal error correction feedback mechanism as ,when When generating visualization suggestions, sequence.

[0023] Preferably, the optimization of teaching plan parameters to obtain an updated teaching plan includes: dynamically adjusting and optimizing teaching plan parameters based on cultural context driving factors; applying a gradient descent algorithm to minimize a comprehensive loss function, with the optimization goal being to minimize expected movement differences and pitch deviations, and completing the optimization of teaching plan parameters to obtain an updated teaching plan.

[0024] Preferably, the updated teaching plan, movement difference data and intonation data are input into the progress quantification model to generate an evaluation report, including: based on the updated teaching plan, movement difference data and intonation data, the progress quantification model is applied, first analyzing the movement difference data and intonation data, and calculating the progress index by comparing the current movement difference with the initial movement difference and integrating the intonation data weight factor; based on the analysis results of the progress index and the movement difference data and intonation data, a structured report is generated; wherein the report includes: the progress index value, the movement improvement rate, the intonation stability score and a comparison chart with historical data.

[0025] In a second aspect, the present application provides an AI-assisted teaching system for performing movements of national musical instruments, including: a model acquisition module for embedding pre-acquired video data of national musical instrument masters and data from a private library of intangible cultural heritage schools into a musical instrument acoustic-motion association model and a dynamic model to obtain a personalized standard motion model and a school fingerprint style quantification model.

[0026] The test data acquisition module performs movement evaluation and style matching based on the personalized standard movement model, genre fingerprint style quantification model, student performance video, performance data and instrument information to obtain test data containing the student's movement difference data, performance level and the most similar master performer.

[0027] The personalized correction suggestion generation module generates personalized correction suggestions based on the movement difference data, performance level and the most similar master performer, combined with cultural adaptation factors and cross-modal error correction feedback mechanism.

[0028] The teaching plan generation module optimizes the teaching plan parameters based on the personalized correction suggestions, cultural context driving factors and preset teaching goal data to obtain an updated teaching plan.

[0029] The evaluation report generating module is used to input the updated teaching plan, movement difference data and pitch data into the progress quantification model to generate an evaluation report.

[0030] The beneficial effects of the present application are: 1. The AI-assisted teaching method and system for national musical instrument playing movements provided by the present application, through deep coupling of AI technology and national musical instrument characteristics, constructs a musical instrument acoustic-motion association model, a dynamic model evolution mechanism and cross-modal error correction feedback, generates a dynamic standard motion model and a style quantification model based on pre-acquired data, integrates embedded sensor data to evaluate student movements, generates personalized correction suggestions, applies cultural context-driven logic to optimize the teaching plan to obtain an updated teaching plan, and applies a progressive quantification model to finally generate a quantitative evaluation report, involving video analysis, sensor fusion and real-time pitch detection technology, solves the problems of poor static adaptability and inaccurate feedback of traditional teaching models, and solves the limitations of the current feasibility analysis process of the development of AI-assisted teaching of national musical instruments.

[0031] 2. The instrument acoustic-action association model in this application converts action features and finger pressure into timbre features, ensuring that the evaluation process is bound to the physical characteristics of the instrument, such as detecting the timbre deviation of the guqin's "Yin Hua" fingering, avoiding the limitations of general video analysis; the dynamic model evolution mechanism uses standard feature vectors and initially favors the student's physiological data when the dynamic attenuation factor increases, and gradually converges to the master standard when the dynamic attenuation factor decreases, solving the "one-size-fits-all" teaching problem and improving the model's adaptability, such as adjusting standard movements for students with different finger lengths.

[0032] 3. This application combines movement differences with real-time pitch data, a loss function-driven intelligent feedback generation mechanism, and a dynamic adaptive teaching strategy based on genres, emotions, and cultural rules to jointly build an efficient, intelligent, and culturally profound music learning environment. It effectively solves the problems of hidden errors that are difficult to detect in traditional teaching, inaccurate error correction guidance, single and rigid teaching methods, and insufficient understanding of cultural connotations; it dynamically adjusts the teaching freedom based on genres and repertoire emotions, incorporates cultural rules, enhances technical effects, and avoids the rigidity of mechanical correction. For example, for the improvisational passages of Jiangnan Sizhu, priority will be given to retaining personalized processing that conforms to the aesthetics of the genre, and in practice, the repertoire corresponding to the cultural context driving factors will be used, which significantly improves the students' technical accuracy, artistic expression, music understanding, and learning efficiency, ultimately achieving better technical mastery and artistic expression. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0034] Figure 1 This is a flowchart of the steps for implementing the application method.

[0035] Figure 2 This is a schematic diagram of the system structure connection for this application. DETAILED DESCRIPTION

[0036] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0037] See also Figure 1 As shown, the present application provides, in the first aspect, an AI-assisted teaching method for playing movements of national musical instruments, including: S1, embedding pre-acquired video data of national musical instrument masters and data from the private library data of intangible cultural heritage schools into the musical instrument acoustic-motion association model and dynamic model to obtain a personalized standard motion model and a genre fingerprint style quantification model.

[0038] It should be noted that the private library data of intangible cultural heritage genres include instrument physical characteristic parameters, genre-specific performance characteristics, cultural rule metadata and master performance benchmark data.

[0039] It should be noted that the physical characteristic parameters of the musical instrument include the acoustic mapping parameters of the instrument material, such as the attribute values ​​of the erhu python skin, which are used to bind the action and timbre, and also include the instrument structure data, such as the guzheng string length and the pipa fret spacing, etc., which are used for calibration of the action standard model; the genre-specific performance characteristics include technical fingerprint data, such as the average frequency of the "yinhua" vibrato of the Guangling School of Guqin and the maximum and minimum amplitudes of the improvisational glissando of the Jiangnan Sizhu Erhu, etc., and also include style quantitative labels, such as the genre weight of Cantonese opera and the intensity score of the emotional label; cultural rule metadata includes improvisation freedom rules, such as the range of action deviation allowed in specific repertoires, and also includes emotional expression related data, such as the emotional mapping of the pressing intensity corresponding to the "Erquan Yingyue" repertoire; master performance benchmark data includes pre-extracted action feature vectors, such as the PCA dimensionality reduction results of the finger trajectories in the master performance videos, and also includes physiological adaptation parameters, such as the average fingering span of masters of different schools.

[0040] It should be noted that PCA stands for principal component analysis, which is a mathematical dimensionality reduction method that uses orthogonal transformation to convert a series of possibly linearly correlated variables into a set of linearly uncorrelated new variables, also called principal components, so as to use the new variables to display the characteristics of the data in a smaller dimension.

[0041] In a specific example, the method embeds the pre-acquired video data of national musical instrument masters and the data in the private library data of intangible cultural heritage schools into the musical instrument acoustic-motion association model and the dynamic model to obtain a personalized standard motion model and a genre fingerprint style quantification model, including: based on the pre-acquired video data of national musical instrument masters and the private library data of intangible cultural heritage schools, extracting the key action points in the video data of national musical instrument masters through OpenCV frame decomposition technology to obtain the master's standard feature vector, and at the same time initializing the master's standard feature vector to obtain a fully initialized master's standard feature vector that supports the dynamic model evolution mechanism.

[0042] Acoustic mapping is defined to generate an instrument acoustic-action association model, and then timbre binding is performed to generate timbre features; based on the timbre features and the student's physiological reference vector, the dynamic model evolution mechanism is applied to update the standard feature vector to generate a personalized standard action model; based on the private library data of intangible cultural heritage genres, the genre characteristics are clustered and analyzed, and the intangible cultural heritage data is converted into quantitative indicators to generate a genre fingerprint style quantitative model.

[0043] It should be noted that the personalized standard movement model inputs the student's performance video and outputs the student's movement difference data and performance level; according to the calculation formula Obtaining action difference data , Represents the number corresponding to the eigenvector, , Expressed as the total number of eigenvectors, Indicates the student's eigenvector values, Indicates the master's eigenvector values; according to the calculation formula Determine the performance level ,in Indicates the pitch of the student's performance. Indicates the pitch of the master's performance. and They are respectively expressed as the weight factor corresponding to the difference data and the weight factor corresponding to the pitch.

[0044] It should be noted that , , .

[0045] It should be noted that the weight factors corresponding to the difference data and the weight factors corresponding to the pitch are obtained through factor analysis. First, the information of the difference data and pitch is condensed, and then the variance explanation rate after rotation is obtained. The weight is obtained by dividing the cumulative variance explanation rate.

[0046] It should be noted that factor analysis is a well-known technology. It is a multivariate statistical analysis method that starts from studying the internal dependencies of variables and reduces some variables with complex relationships to a few comprehensive factors; information concentration is expressed as calculating the median; the variance explanation rate is the amount of information extracted by the factor, and the variance explanation rate = characteristic root / total number of analysis items; the variance explanation rate after rotation is expressed as the variance explanation rate of the factor after maximum variance rotation.

[0047] It should be noted that the genre fingerprint style quantification model inputs the student's performance video and outputs the performance master who is most similar to the student's performance.

[0048] It should be noted that based on the cultural rule metadata, the teaching freedom is dynamically adjusted to obtain the teaching frequency to adapt to the playing rules of different schools; the student's physiological reference vector includes the vector eigenvalues ​​corresponding to height, weight, forearm length, upper arm length, length of each finger and the length of the knuckles corresponding to each finger, etc.

[0049] In a specific example, the musical instrument acoustic-motion association model and dynamic model evolution mechanism include: A1, through the formula of the musical instrument acoustic-motion association model Get the tone output value , Expressed as pressure intensity, Represented as an instrument material identifier, Expressed as the instrument material coefficient, Refers to the weight coefficient corresponding to the pressure intensity, Expressed as the correction factor corresponding to the timbre output value.

[0050] It should be noted that the timbre output value is a quantitative indicator of the instrument's timbre, which is used to bind the action to the acoustic characteristics. For example, the change in the bow pressure of the erhu leads to nonlinear changes in the timbre. The pressure intensity is the force applied during playing, which is collected by the pressure sensor. The instrument material identifier, such as python skin and wood, comes from the pre-acquired instrument information database. The instrument material coefficient is the weight coefficient of the material to the timbre mapping, which refers to the weight coefficient corresponding to the pressure intensity. , ensuring that the timbre increases exponentially with pressure. The correction factor corresponding to the timbre output value is a digital factor multiplied by the uncorrected measurement result to compensate for the error in the timbre acquisition calculation.

[0051] A2. Formula for the evolutionary mechanism through dynamic models Output standard feature vector ,in Expressed as the master standard eigenvector Represented as the student’s physiological reference vector, Expressed as a dynamic attenuation factor.

[0052] It should be noted that It is expressed as a dynamic attenuation factor with a preset initial value of 0.8. It decreases linearly to 0 with the number of exercises and is inversely proportional to the number of exercises. It is a control parameter for the model convergence rate.

[0053] In this application, the instrument acoustic-action association model converts action features and finger pressure strength into timbre features, ensuring that the evaluation process is bound to the physical characteristics of the instrument, such as detecting the timbre deviation of the guqin's "Yin Hua" fingering, avoiding the limitations of general video analysis; the dynamic model evolution mechanism uses standard feature vectors and initially favors the student's physiological data when the dynamic attenuation factor increases, and gradually converges to the master standard when the dynamic attenuation factor decreases, solving the "one-size-fits-all" teaching problem and improving the model's adaptability, such as adjusting standard movements for students with different finger lengths.

[0054] S2. Based on the personalized standard movement model, genre fingerprint style quantification model, student performance video, performance data and instrument information, movement evaluation and style matching are performed to obtain test data including the student's movement difference data, performance level and the most similar performance master.

[0055] It should be noted that the instrument information is customized by the students, and includes the instrument name, brand, size, shape, timbre and material, etc. The performance data includes the mechanical dynamic characteristics of the students' fingers when playing, and is collected by embedded sensors.

[0056] In a specific example, the movement evaluation and style matching are performed to obtain test data containing the student's movement difference data, performance level and the most similar performance master, including: based on the instrument information, extracting the corresponding test performance repertoire from the preset repertoire database, the student completing the performance level test according to the test performance repertoire, and then obtaining the student's performance video data and performance data.

[0057] Based on the timestamp synchronization mechanism, the student's performance video data and performance data are temporally and spatially aligned to obtain a synchronized data set, which is input into a personalized standard movement model to output the student's movement difference data and performance level.

[0058] Based on the genre fingerprint style quantification model, the similarity between the student's playing style and the master's genre characteristics is matched to output the most similar performing master.

[0059] It should be noted that the student performance video data: OpenCV is used to decompose the frame and extract the key action points; embedded sensor data, through the accelerometer, the sampling rate is 100Hz, to collect the finger bone micro-vibration signal; the student performance video data and performance data are time-space aligned, and a unified time stamp is added to the video frame and sensor data to ensure the accuracy. , synchronization is achieved through interpolation algorithm; the alignment formula is ,in is the time point in the timing sequence, ensuring that the action and mechanical signal are consistent in time and space.

[0060] It should be noted that the similarity matching is obtained based on the ratio of the student's performance data to the product of the student's performance data and the master's performance data.

[0061] S3. Based on the movement difference data, performance level and the most similar master performer, combined with cultural adaptation factors and cross-modal error correction feedback mechanism, generate personalized correction suggestions.

[0062] In a specific example, the method generates personalized correction suggestions based on the movement difference data, performance level and the most similar master performers, combined with cultural adaptation factors and cross-modal error correction feedback mechanisms, including: matching practice repertoire and practice frequency based on the movement difference data, performance level and the most similar master performers to formulate a preliminary learning plan; dynamically adjusting the teaching freedom based on the cultural context driving factor to obtain the teaching frequency; monitoring the student's execution process based on rhythm synchronization characteristics; performing a joint loss calculation on movement differences and pitch deviations based on the cross-modal error correction feedback mechanism to generate personalized correction suggestions; and dynamically optimizing the suggestion weights based on the accumulation of historical learning data to improve feedback accuracy.

[0063] It should be noted that based on the accumulation of historical learning data, the recommendation weights are dynamically optimized to improve feedback accuracy. The recommendation weights are optimized based on the mean and variance of the accumulated historical data.

[0064] In a specific example, the matching of practice repertoire and practice frequency to formulate a preliminary learning plan, dynamically adjusting the teaching freedom based on the cultural context driving factor to obtain the teaching frequency, includes: B1, according to the student's performance level, selecting the practice repertoire of the master's style that is most similar to the student from the preset repertoire library; according to the calculation formula The practice frequency is obtained, where Expressed as the practice adjustment coefficient, Expressed as the standard value of action difference data.

[0065] It should be noted that the practice adjustment coefficient is a preset value, which is initially set to 0.2.

[0066] It should be noted that the standard value of the action difference data is the square root of the master's eigenvector value multiplied by eighty percent.

[0067] B2. According to the calculation formula The cultural context driving factors are obtained, among which Expressed as the genre coefficient, Expressed as the track emotion coefficient, and They are respectively expressed as the weight factor corresponding to the genre coefficient and the weight factor corresponding to the track emotion coefficient.

[0068] It should be noted that when , allowing the movement deviation to expand by 20%, and the students' improvisation segments do not match the standard movements.

[0069] In a specific example, the cross-modal error correction feedback mechanism includes: based on the loss function, defining the calculation formula of the cross-modal error correction feedback mechanism as: ,when When generating visualization suggestions, sequence.

[0070] It should be noted that when the tone deviation is caused by insufficient pressure on the right hand, the visual suggestion is a video tutorial on how to use the right hand, and the degree of force used by the right hand is marked.

[0071] It should be noted that the stored Sequences are used to dynamically change the weights of items with high error rates, for example, increasing the weight by 0.1 when the tone deviates.

[0072] S4. Based on the personalized correction suggestions, cultural context driving factors and preset teaching objective data, optimize the teaching plan parameters to obtain an updated teaching plan.

[0073] It should be noted that the teaching plan parameters include practice intensity, target difficulty, time allocation and freedom threshold; among them, practice intensity includes the number of movement repetitions, target difficulty includes the complexity of the repertoire, time allocation includes the daily practice duration and freedom threshold includes the range of allowable movement deviation.

[0074] In a specific example, the optimization of teaching plan parameters to obtain an updated teaching plan includes: dynamically adjusting and optimizing teaching plan parameters based on cultural context driving factors; applying a gradient descent algorithm to minimize the comprehensive loss function, with the optimization goal being to minimize expected movement differences and pitch deviations, and completing the optimization of teaching plan parameters to obtain an updated teaching plan.

[0075] It should be noted that the teaching plan parameters are dynamically adjusted and optimized based on the cultural context driving factor. For example, when the genre weight increases, the freedom threshold is increased to accommodate genre-specific deviations; when the emotional tag increases, the difficulty is reduced to match the emotion of the piece, and the parameters are weighted and optimized. For example, if the personalized correction suggestion indicates "vibrato deviation" and the cultural context driving factor is greater than 0.6 (indicating a high freedom threshold), the number of repetitions is reduced, while the improvisation practice time is increased.

[0076] It should be noted that the gradient descent algorithm is used to minimize the comprehensive loss function, and the optimization goal is to minimize the expected action difference and pitch deviation; the formula is ,in Expressed as the vector corresponding to the teaching plan parameters, Represented as predicted action difference data, Expressed as pitch error, and They are respectively expressed as the weight factors corresponding to the predicted action difference data and the weight factors corresponding to the pitch error; through iterative calculation, the teaching plan parameters are adjusted to converge to the standard values; based on the optimized teaching plan parameters and cultural context driving factors, a teaching plan is generated.

[0077] It should be noted that the teaching plan includes specific instructions, such as selecting practice repertoire based on cultural context drivers and selecting additional practice content based on teaching plan parameters, such as increasing vibrato practice to 30 minutes per day with a permissible deviation of .

[0078] It should be noted that the weight factors corresponding to the predicted action difference data and the weight factors corresponding to the pitch error are obtained by factor analysis combined with instrument information.

[0079] S5. Input the updated teaching plan, movement difference data, and pitch data into the progress quantification model to generate an evaluation report.

[0080] In a specific example, the updated teaching plan, movement difference data and intonation data are input into a progress quantification model to generate an evaluation report, including: based on the updated teaching plan, movement difference data and intonation data, the progress quantification model is applied, first analyzing the movement difference data and intonation data, and calculating the progress index by comparing the current movement difference with the initial movement difference and integrating the intonation data weight factor; based on the analysis results of the progress index and the movement difference data and intonation data, a structured report is generated; wherein the report includes: the progress index value, the movement improvement rate, the intonation stability score and a comparison chart with historical data.

[0081] It should be noted that the progress index is calculated as follows: According to the calculation formula Progress Index ,in Expressed as a scaling factor, Expressed as The next action difference data, Expressed as pitch data coefficient.

[0082] It should be noted that the report is output in the form of an electronic document to guide subsequent teaching adjustments.

[0083] It should be noted that the scaling factor is based on a full score system, which is 100 here.

[0084] It should be noted that , the pitch data coefficient is preset based on the instrument type, such as Guzheng .

[0085] This application combines movement differences and real-time pitch data, a loss function-driven intelligent feedback generation mechanism, and a dynamic adaptive teaching strategy based on genre, emotion, and cultural rules to jointly build an efficient, intelligent, and culturally profound music learning environment. It effectively solves the problems of hidden errors that are difficult to detect in traditional teaching, inaccurate error correction guidance, single and rigid teaching methods, and insufficient understanding of cultural connotations; it dynamically adjusts the teaching freedom based on genre and repertoire emotion, incorporates cultural rules, enhances technical effects, and avoids the rigidity of mechanical correction. For example, for the improvisational passages of Jiangnan Sizhu, priority will be given to retaining personalized processing that conforms to the aesthetics of the genre, and in practice, the repertoire corresponding to the cultural context driving factors will be used, which significantly improves the students' technical accuracy, artistic expression, music understanding, and learning efficiency, and ultimately achieves better technical mastery and artistic expression effects.

[0086] See also Figure 2As shown, in the second aspect, the present application provides an AI-assisted teaching system for playing movements of national musical instruments, including: a model acquisition module for embedding pre-acquired video data of national musical instrument masters and data in the private library data of intangible cultural heritage schools into the musical instrument acoustic-motion association model and dynamic model to obtain a personalized standard motion model and a genre fingerprint style quantification model.

[0087] The test data acquisition module performs movement evaluation and style matching based on the personalized standard movement model, genre fingerprint style quantification model, student performance video, performance data and instrument information to obtain test data containing the student's movement difference data, performance level and the most similar master performer.

[0088] The personalized correction suggestion generation module generates personalized correction suggestions based on the movement difference data, performance level and the most similar master performer, combined with cultural adaptation factors and cross-modal error correction feedback mechanism.

[0089] The teaching plan generation module optimizes the teaching plan parameters based on the personalized correction suggestions, cultural context driving factors and preset teaching goal data to obtain an updated teaching plan.

[0090] The evaluation report generating module is used to input the updated teaching plan, movement difference data and pitch data into the progress quantification model to generate an evaluation report.

[0091] The AI-assisted teaching method and system for folk musical instrument playing movements provided in this application deeply couples AI technology with the characteristics of folk musical instruments to construct a musical instrument acoustic-motion association model, a dynamic model evolution mechanism and cross-modal error correction feedback, generates a dynamic standard motion model and a style quantification model based on pre-acquired data, integrates embedded sensor data to evaluate students' movements, generates personalized correction suggestions, applies cultural context-driven logic to optimize the teaching plan to obtain an updated teaching plan, and applies a progressive quantification model to finally generate a quantitative evaluation report. It involves video analysis, sensor fusion and real-time pitch detection technology, solves the problems of poor static adaptability and inaccurate feedback of traditional teaching models, and solves the limitations of the current feasibility analysis process of the development of AI-assisted teaching of folk musical instruments.

[0092] In the several embodiments provided by the present invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For example, the module division is merely a logical function division, and other division methods may be used in actual implementation.

[0093] The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed across multiple network elements. Some or all of the modules may be selected to achieve the purpose of the solution of this embodiment according to actual needs.

[0094] In addition, the functional modules in various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or hardware plus software functional modules.

[0095] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0096] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to achieve optimal results.

[0097] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. AI-assisted teaching method and system for traditional musical instrument playing movements, characterized by: include: S1. Embed the pre-acquired video data of national instrument masters and the data from the private library of intangible cultural heritage schools into the instrument acoustic-motion association model and dynamic model to obtain a personalized standard motion model and a school fingerprint style quantification model; S2. Based on the personalized standard movement model, the genre fingerprint style quantification model, the student's performance video, performance data, and instrument information, perform movement evaluation and style matching to obtain test data including the student's movement difference data, performance level, and the most similar performance master; S3. Based on the movement difference data, performance level, and the most similar master performer, combined with cultural adaptation factors and a cross-modal error correction feedback mechanism, generate personalized correction suggestions; S4. Optimizing teaching plan parameters based on the personalized correction suggestions, cultural context driving factors, and preset teaching objective data to obtain an updated teaching plan; S5. Input the updated teaching plan, movement difference data, and pitch data into the progress quantification model to generate an evaluation report.

2. The AI-assisted teaching method and system for folk musical instrument playing movements according to claim 1 is characterized in that: The method of embedding the pre-acquired video data of national musical instrument masters and the data in the private library of intangible cultural heritage schools into the instrument acoustic-motion association model and dynamic model to obtain a personalized standard motion model and a school fingerprint style quantification model includes: Based on the pre-acquired video data of masters of national musical instruments and the private library data of intangible cultural heritage schools, the key action points in the video data of masters of national musical instruments are extracted through OpenCV frame decomposition technology to obtain the master's standard feature vector. At the same time, the master's standard feature vector is initialized to obtain the fully initialized master's standard feature vector that supports the dynamic model evolution mechanism; Acoustic mapping is defined to generate an instrument acoustic-action association model, and then timbre binding is performed to generate timbre features; based on the timbre features and the student's physiological reference vector, the dynamic model evolution mechanism is applied to update the standard feature vector to generate a personalized standard action model; based on the private library data of intangible cultural heritage genres, the genre characteristics are clustered and analyzed, and the intangible cultural heritage data is converted into quantitative indicators to generate a genre fingerprint style quantitative model.

3. The AI-assisted teaching method and system for folk musical instrument playing movements according to claim 2 is characterized in that: The musical instrument acoustic-motion correlation model and dynamic model evolution mechanism include: A1. Formula of the instrument acoustic-motion correlation model Get the tone output value , Expressed as pressure intensity, Represented as an instrument material identifier, Expressed as the instrument material coefficient, Refers to the weight coefficient corresponding to the pressure intensity, Expressed as the correction factor corresponding to the timbre output value; A2. Formula for the evolutionary mechanism through dynamic models Output standard feature vector ,in Expressed as the master standard eigenvector Represented as the student’s physiological reference vector, Expressed as a dynamic attenuation factor.

4. The AI-assisted teaching method and system for folk musical instrument playing movements according to claim 3 is characterized in that: The movement assessment and style matching are performed to obtain test data including the student's movement difference data, performance level, and the most similar performance master, including: Based on the instrument information, the corresponding test performance repertoire is extracted from the preset repertoire database, and the student completes the performance level test according to the test performance repertoire, thereby obtaining the student's performance video data and performance data; Based on the timestamp synchronization mechanism, the student's performance video data and performance data are temporally and spatially aligned to obtain a synchronized data set, which is input into the personalized standard movement model to output the student's movement difference data and performance level; Based on the genre fingerprint style quantification model, the similarity between the student's playing style and the master's genre characteristics is matched to output the most similar performing master.

5. The AI-assisted teaching method and system for folk musical instrument playing movements according to claim 4 is characterized in that: The method generates personalized correction suggestions based on the movement difference data, performance level, and the most similar master performer, combined with cultural adaptation factors and a cross-modal error correction feedback mechanism, including: Based on movement difference data, performance level and the most similar master performers, the practice repertoire and practice frequency are matched to formulate a preliminary learning plan. Based on the cultural context driving factor, the teaching freedom is dynamically adjusted to obtain the teaching frequency; the student's execution process is monitored based on the rhythm synchronization characteristics; based on the cross-modal error correction feedback mechanism, the joint loss of movement differences and pitch deviations is calculated to generate personalized correction suggestions; based on the accumulation of historical learning data, the suggestion weights are dynamically optimized to improve feedback accuracy.

6. The AI-assisted teaching method and system for folk musical instrument playing movements according to claim 5 is characterized in that: The matching of practice repertoire and practice frequency to formulate a preliminary learning plan, and the dynamic adjustment of teaching freedom based on cultural context driving factors to obtain the teaching frequency, include: B1. Based on the student's performance level, select the practice repertoire in the preset repertoire library that is most similar to the student's performance style; according to the calculation formula The practice frequency is obtained, where Expressed as the practice adjustment coefficient, Expressed as the standard value of action difference data; B2. According to the calculation formula The cultural context driving factors are obtained, among which Expressed as the genre coefficient, Expressed as the track emotion coefficient, and They are respectively expressed as the weight factor corresponding to the genre coefficient and the weight factor corresponding to the track emotion coefficient.

7. The AI-assisted teaching method and system for folk musical instrument playing movements according to claim 6 is characterized in that: The cross-modal error correction feedback mechanism includes: Based on the loss function, the calculation formula of the cross-modal error correction feedback mechanism is defined as ,when When generating visualization suggestions, sequence.

8. The AI-assisted teaching method and system for folk musical instrument playing movements according to claim 7 is characterized in that: Optimizing the teaching plan parameters to obtain an updated teaching plan includes: Based on the cultural context driving factors, the teaching plan parameters are dynamically adjusted and optimized; the gradient descent algorithm is applied to minimize the comprehensive loss function. The optimization goal is to minimize the expected movement difference and pitch deviation, and the teaching plan parameters are optimized to obtain an updated teaching plan.

9. The AI-assisted teaching method and system for folk musical instrument playing movements according to claim 8 is characterized in that: The updated teaching plan, movement difference data, and pitch data are input into the progress quantification model to generate an evaluation report, including: Based on the updated teaching plan, movement difference data, and pitch data, the progress quantification model is applied. First, the movement difference data and pitch data are analyzed. By comparing the current movement difference with the initial movement difference and integrating the pitch data weight factor, the progress index is calculated. Based on the analysis results of the progress index and the movement difference data and pitch data, a structured report is generated. The report includes: progress index value, movement improvement rate, pitch stability score and comparison chart with historical data.

10. An AI-assisted teaching system for performing the traditional Chinese musical instrument playing movements according to any one of claims 1 to 9, characterized in that: include: The model acquisition module is used to embed pre-acquired video data of national instrument masters and data from the private library of intangible cultural heritage genres into the instrument acoustic-motion association model and dynamic model to obtain a personalized standard motion model and genre fingerprint style quantification model; A test data acquisition module performs movement evaluation and style matching based on the personalized standard movement model, the genre fingerprint style quantification model, the student's performance video, performance data, and instrument information to obtain test data including the student's movement difference data, performance level, and the most similar performance master; A personalized correction suggestion generation module generates personalized correction suggestions based on the movement difference data, performance level, and the most similar master performers, combined with cultural adaptation factors and a cross-modal error correction feedback mechanism; A teaching plan generation module, which optimizes teaching plan parameters based on the personalized correction suggestions, cultural context driving factors and preset teaching goal data to obtain an updated teaching plan; The evaluation report generating module is used to input the updated teaching plan, movement difference data and pitch data into the progress quantification model to generate an evaluation report.

Images