A method for training eloquence based on document editing and communication

Abstract

The present invention discloses an eloquence training method based on document editing and communication, the method comprising: obtaining a weight set by calculating the correlation between a target document and an eloquence dimension index set; obtaining a marked segment by marking each text segment of the target document; generating a to-do list according to the eloquence dimension index set, the marked segment and the weight set; constructing a first task according to the text information of voice data; generating a second task according to the interactive performance of real-time interaction with a user through an instant messaging function; and forming an eloquence training method by the to-do list, the first task and the second task. The present invention proposes an eloquence training method based on document editing and communication, which can solve the problem that it is difficult to generate a personalized eloquence training method that can perform real-time analysis and feedback based on practice documents and voice data by generating a to-do list and two different types of training plans to form an eloquence training method.

Description

Technical Field

[0001] The invention relates to the field of speech training, and in particular to an eloquence training method based on document editing and communication. Background Art

[0002] In daily life, in public speeches or business negotiations, the ability to persuade and influence others is crucial. Good eloquence can bring more opportunities. Moreover, communication is the cornerstone of building trust and relationships. People who are well-organized are usually more likely to gain respect and recognition from others. Therefore, having good eloquence can bring many benefits to personal life and career. Traditional eloquence training methods usually rely on real-time feedback from coaches or mentors, and provide static practice materials to help practice, or individuals consolidate and improve by repeatedly practicing pronunciation, intonation and speed.

[0003] However, traditional methods are unable to provide highly personalized training due to the lack of real-time analysis and feedback capabilities, which may result in users being unable to make full use of every practice opportunity. In addition, traditional methods are difficult to detect problems and take corrective measures in the first place, so they cannot help users monitor themselves and improve their eloquence skills in a timely manner. Summary of the invention

[0004] The present invention provides an eloquence training method based on document editing and communication, so as to solve the problem that it is difficult to generate a personalized eloquence training method capable of real-time analysis and feedback according to practice documents and voice data.

[0005] In order to solve the above problems, the present invention provides an eloquence training method based on document editing and communication, comprising:

[0006] Obtain target documents and user's voice data;

[0007] By calculating the correlation between the target document and a preset eloquence dimension indicator set, a weight set of the text segment set in the target document is obtained;

[0008] Marking each text segment in the target document according to the eloquence dimension score of each text segment to obtain a marked segment;

[0009] generating a to-do list according to the eloquence dimension indicator set, the marked segment and the text segment with the highest weight in the weight set;

[0010] Constructing a first task based on the textual representation, emotional information and body language information of the speech data;

[0011] Conducting real-time interaction with the user through a preset instant messaging function, and generating a second task according to the interaction performance and the eloquence dimension indicator set;

[0012] The to-do list, the first task and the second task constitute an eloquence training method.

[0013] In the present invention, the to-do list focuses on improving the user's eloquence training shortcomings according to the target document. Since the weight set is obtained through correlation analysis, the relevance of different contents can be measured, reflecting the training content that is most needed to be provided to the user in the target document. In addition, the to-do list is generated based on the marked fragments and the text fragments with the highest weights in the weight set, so the most beneficial training content for the user is extracted based on the target document; the first task focuses on improving the user's eloquence expression, which can help the user to effectively improve oral expression ability, facial expression ability and body expression ability; the second task focuses on generating tasks for training through feedback adjustment according to the user's real-time interactive data. This method of providing instant support and feedback helps to better understand eloquence problems to accelerate learning progress.

[0014] Compared with the prior art, the present invention forms an eloquence training method by generating a to-do list and two different types of training plans. It can generate the most valuable and targeted training content according to the target document, comprehensively improve the user's eloquence expression, and train the existing eloquence shortcomings according to the user's real-time training feedback. It can solve the problem of difficulty in generating a personalized eloquence training method that can perform real-time analysis and feedback based on practice documents and voice data.

[0015] As a preferred solution, by calculating the correlation between the target document and a preset eloquence dimension indicator set, a weight set of the text segment set in the target document is obtained, specifically:

[0016] Establishing an eloquence dimension weight matrix according to the target document and the eloquence dimension indicator set;

[0017] Extracting relevant features of each text segment and the eloquence dimension indicator set in the target document to obtain a relevant feature set;

[0018] Calculating the correlation between each text segment in the relevant feature set and the eloquence dimension indicator set to obtain a correlation function;

[0019] According to the weights of the eloquence dimension weight matrix, the correlation function and the number of text segments in the target document, a weight set of the text segment set in the target document is calculated.

[0020] In this preferred solution, since the correlation function can represent the degree of correlation between the vocabulary frequency, emotional expression and complexity between the text fragment and the eloquence dimension indicator set, the weight set established on this basis can reflect the training content that is most needed to be provided to the user in the target document; and, the weight set generated by combining the weights of the eloquence dimension weight matrix and the number of text fragments can provide the user with the best value practice content while measuring all the text fragments in the target document, which is helpful for the subsequent generation of a to-do list.

[0021] As a preferred solution, each text segment in the target document is marked according to its eloquence dimension score to obtain a marked segment, specifically:

[0022] Obtaining a speech clarity score by measuring the speech clarity of each text segment in the target document;

[0023] By evaluating the speech rate of each text segment, a speech rate control score is obtained;

[0024] By evaluating the emotional expression of each text segment, an emotional expression score is obtained;

[0025] By measuring the speech diversity of each of the text segments, a speech diversity score is obtained;

[0026] According to the speech clarity score, the speech rate control score, the emotional expression score, the speech diversity score and preset personalized needs, each text segment is marked to obtain the marked segment.

[0027] The marked segments in this preferred solution not only take into account a variety of practice measurement criteria such as speech clarity score, speech rate control score, emotional expression score and speech diversity score, but also incorporate personalized needs, which can truly provide users with the most suitable and most needed training benchmarks, and provide strong data support for subsequent calculations.

[0028] As a preferred solution, before generating a to-do list according to the eloquence dimension indicator set, the marked segment and the text segment with the highest weight in the weight set, the method further includes:

[0029] Extracting features from the marker fragments to obtain diversity features;

[0030] The diversity features are evaluated in eloquence dimension by using a preset formula set to obtain a speech clarity display score, a speech speed display score, an emotional expression display score, and a speech diversity display score;

[0031] According to the speech clarity display score, the speaking speed display score, the emotional expression display score and the speech diversity display score, each text segment in the target document is visually marked, and the visual marking results and the text segment with the highest weight in the weight set are highlighted in a preset manner.

[0032] This preferred solution can present the exercise content that needs attention to the user in the most intuitive way according to different scores through highlighting, helping users to easily identify and distinguish the direction of improvement. This visual marking method can enable users to escape from the boring and large-scale training data in the target document, grasp the key points of the exercise at the first time, and achieve good training results.

[0033] As a preferred solution, a to-do list is generated according to the eloquence dimension indicator set, the marked segment and the text segment with the highest weight in the weight set, specifically:

[0034] According to the eloquence dimension index set, a comprehensive calculation is performed on the text segment in the marked segment and the text segment with the highest weight in the weight set to obtain a comprehensive eloquence dimension index set;

[0035] The comprehensive eloquence dimension indicator set is combined with a preset time factor set to generate a to-do priority set;

[0036] According to the to-do priority set, the text segments in the marked segment and the text segment with the highest weight in the weight set are prioritized to obtain the to-do list.

[0037] The to-do list in this preferred solution focuses on improving the user's shortcomings in eloquence training based on the target document; the to-do list is obtained by sorting the text fragments according to the priority of the weight, so that the user can first be exposed to the content that most needs to be practiced; and, by combining the time factor set, the practice content in the to-do list can be reasonably planned according to the user's personal preset time, so as to achieve the best training effect within a limited time.

[0038] As a preferred solution, the time factor set is specifically:

[0039] Obtain the user's speech target completion time;

[0040] According to the completion time of the speech target and the voice data, the practice time that the user needs to spend on the eloquence dimension indicator set when performing eloquence training is determined to obtain the time factor set.

[0041] Since the time factor set of this preferred scheme is calculated based on the speech target completion time and voice data, it can objectively measure the gap between the user's current speech control time and the target time, so as to develop the most suitable training content for the user in different dimensions of the eloquence dimension indicator set, so that the user's training progress is always in an appropriate state.

[0042] As a preferred solution, the first task is constructed according to the textual expression, emotional information and body information of the speech data, specifically:

[0043] According to the speech data, determining the initial weight of each eloquence dimension in the eloquence dimension set to obtain an initial weight set; wherein the eloquence dimension set includes clarity, speech speed, posture and language, expression and body language, emotional expression and grammatical accuracy;

[0044] Calculating a task weight set according to the initial weight set and the user's target score on the eloquence dimension set;

[0045] The first task is generated according to the eloquence dimension indicator set and in a manner of determining the priority by the task weights in the task weight set.

[0046] The first task of this preferred solution focuses on improving the user's eloquence and expression. By establishing the first task based on a set of eloquence dimensions including clarity, speaking speed, posture and language, facial expression and body language, emotional expression and grammatical accuracy, it can help users effectively improve their oral expression ability, facial expression ability and body expression ability.

[0047] As a preferred solution, according to the eloquence dimension indicator set, the first task is generated in a manner of determining the priority by the task weight in the task weight set, specifically:

[0048] According to the eloquence dimension indicator set and the speech data, generating a number of corresponding initial tasks on different dimensions of the eloquence dimension set;

[0049] Determining the priority of each task in the plurality of initial tasks according to the weight of each task in the task weight set;

[0050] The tasks in the plurality of initial tasks are sorted according to the priority levels, and the sorted plurality of initial tasks constitute the first task.

[0051] This preferred solution sorts each of the several initial tasks according to the priority level, so that the focus of the generated first task is clear, allowing users to access the content that needs to be trained and improved at the first time, which can improve training efficiency.

[0052] As a preferred solution, real-time interaction is performed with the user through a preset instant messaging function, and a second task is generated according to the interaction performance and the eloquence dimension indicator set, specifically:

[0053] Interacting with the user in real time through a preset instant messaging function to obtain real-time interaction data;

[0054] According to a preset mapping dimension indicator set, the real-time interactive data is subjected to eloquence performance quantification and eloquence performance feedback evaluation to obtain eloquence performance data and eloquence feedback suggestion data;

[0055] generating an eloquence improvement index according to the number of real-time interactions, the eloquence performance data and the eloquence feedback suggestion data;

[0056] The second task is generated according to the eloquence improvement index and the eloquence dimension indicator set.

[0057] The second task of this preferred solution focuses on generating tasks for training based on the user's real-time interactive data through feedback adjustment; wherein, the second task is generated based on the real-time interactive data with the user, so that the user's training progress can be understood in a timely manner; since the eloquence improvement index can reflect the user's progress in different eloquence dimensions, the limited training time can be used to train all eloquence shortcomings to the greatest extent, and improve the user's eloquence skills in a targeted manner. This method of providing instant support and feedback helps to better understand eloquence problems and accelerate learning progress.

[0058] As a preferred solution, the real-time interactive data is subjected to eloquence performance quantification and eloquence performance feedback evaluation according to a preset mapping dimension indicator set, to obtain eloquence performance data and eloquence feedback suggestion data, specifically:

[0059] According to the mapping dimension indicator set, the real-time interactive data is score-mapped according to the performance degree to obtain the eloquence performance data; wherein the mapping dimension indicator set includes speech speed, intonation, emotional expression and clarity;

[0060] The eloquence performance data is evaluated according to a preset score system to obtain the eloquence feedback suggestion data; wherein the eloquence feedback suggestion data includes the speed of speech, the monotony of intonation and the lack of emotional expression.

[0061] This preferred solution uses score mapping to enable the eloquence performance data to be used as an exact score to measure the user's eloquence performance; since the eloquence feedback suggestion data includes the speed of speech, the monotony of tone and the lack of emotional expression, the eloquence feedback suggestion data is not a single speech speed feedback data, allowing users to understand the current eloquence training effect from different angles.

[0062] The present invention also provides an eloquence training device based on document editing and communication, comprising an acquisition module, a weight module, a marking module, a list module, a construction module, a communication module and a combination module;

[0063] Wherein, the acquisition module is used to acquire the target document and the user's voice data;

[0064] The weight module is used to obtain a weight set of the text segment set in the target document by calculating the correlation between the target document and a preset eloquence dimension indicator set;

[0065] The marking module is used to mark each text segment in the target document according to the eloquence dimension score of each text segment to obtain a marked segment;

[0066] The list module is used to generate a to-do list according to the eloquence dimension indicator set, the marked segment and the text segment with the highest weight in the weight set;

[0067] The construction module is used to construct a first task according to the textual representation, emotional information and body information of the speech data;

[0068] The communication module is used to interact with the user in real time through a preset instant messaging function, and generate a second task according to the interaction performance and the eloquence dimension indicator set;

[0069] The combination module is used to form an eloquence training method by using the to-do list, the first task and the second task.

[0070] As a preferred solution, the weight module includes a matrix unit, a feature set unit, a correlation unit and a weight set unit;

[0071] Wherein, the matrix unit is used to establish an eloquence dimension weight matrix according to the target document and the eloquence dimension indicator set;

[0072] The feature set unit is used to extract relevant features of each text segment and the eloquence dimension indicator set in the target document to obtain a relevant feature set;

[0073] The correlation unit is used to calculate the correlation between each text segment in the relevant feature set and the eloquence dimension indicator set to obtain a correlation function;

[0074] The weight set unit is used to calculate the weight set of the text segment set in the target document according to the weight of the eloquence dimension weight matrix, the correlation function and the number of text segments in the target document.

[0075] As a preferred solution, the marking module includes a score unit, a speech rate unit, a sentiment unit, a speech unit and a marking unit;

[0076] The score unit is used to obtain a speech clarity score by measuring the speech clarity of each text segment in the target document;

[0077] The speech rate unit is used to obtain a speech rate control score by evaluating the speech rate of each text segment;

[0078] The sentiment unit is used to obtain a sentiment expression score by evaluating the sentiment expression of each text segment;

[0079] The speech unit is used to obtain a speech diversity score by measuring the speech diversity of each text segment;

[0080] The marking unit is used to mark each text segment according to the speech clarity score, the speech rate control score, the emotional expression score, the speech diversity score and preset personalized needs to obtain the marked segment.

[0081] As a preferred solution, before the list module, it also includes a feature unit, an evaluation unit and a display unit;

[0082] Wherein, the feature unit is used to extract features from the marker fragments to obtain diversity features;

[0083] The evaluation unit is used to evaluate the eloquence dimension of the diversity features through a preset formula set to obtain a speech clarity display score, a speech speed display score, an emotional expression display score, and a speech diversity display score;

[0084] The display unit is used to visually mark each text segment in the target document according to the speech clarity display score, the speaking speed display score, the emotional expression display score and the speech diversity display score, and highlight the results of the visual marking and the text segment with the highest weight in the weight set in a preset manner.

[0085] As a preferred solution, the list module includes a comprehensive unit, a time unit and a sorting unit;

[0086] The comprehensive unit is used to perform comprehensive calculation on the text segments in the marked segments and the text segments with the highest weight in the weight set according to the eloquence dimension indicator set to obtain a comprehensive eloquence dimension indicator set;

[0087] The time unit is used to combine the comprehensive eloquence dimension indicator set with a preset time factor set to generate a to-do priority set;

[0088] The sorting unit is used to prioritize the text segments in the marked segments and the text segments with the highest weights in the weight set according to the to-do priority set to obtain the to-do list.

[0089] As a preferred solution, the time factor set is specifically:

[0090] Obtain the user's speech target completion time;

[0091] According to the completion time of the speech target and the voice data, the practice time that the user needs to spend on the eloquence dimension indicator set when performing eloquence training is determined to obtain the time factor set.

[0092] As a preferred solution, the building block includes an initial unit, a weight unit and a task unit;

[0093] The initial unit is used to determine the initial weight of each eloquence dimension in the eloquence dimension set according to the speech data to obtain an initial weight set; wherein the eloquence dimension set includes clarity, speech speed, posture and language, expression and body language, emotional expression and grammatical accuracy;

[0094] The weight unit is used to calculate the task weight set according to the initial weight set and the target score of the user on the eloquence dimension set;

[0095] The task unit is used to generate the first task according to the eloquence dimension indicator set and in a manner that the priority is determined by the task weight in the task weight set.

[0096] As a preferred solution, the task unit includes a first subunit, a second subunit and a third subunit;

[0097] The first subunit is used to generate a number of corresponding initial tasks on different dimensions of the eloquence dimension set according to the eloquence dimension indicator set and the voice data;

[0098] The second subunit is used to determine the priority of each task in the plurality of initial tasks according to the weight of each task in the task weight set;

[0099] The third subunit is used to sort each task in the plurality of initial tasks according to the priority level, and the sorted plurality of initial tasks constitute the first task.

[0100] As a preferred solution, the communication module includes an interaction unit, a data unit, a progress unit and a construction unit;

[0101] Wherein, the interactive unit is used to interact with the user in real time through a preset instant messaging function to obtain real-time interactive data;

[0102] The data unit is used to quantify the eloquence performance and evaluate the eloquence performance feedback of the real-time interactive data according to a preset mapping dimension indicator set, so as to obtain the eloquence performance data and the eloquence feedback suggestion data;

[0103] The improvement unit is used to generate an eloquence improvement index according to the number of real-time interactions, the eloquence performance data and the eloquence feedback suggestion data;

[0104] The construction unit is used to generate the second task according to the eloquence improvement index and the eloquence dimension indicator set.

[0105] As a preferred solution, the data unit includes a fourth subunit and a fifth subunit;

[0106] The fourth subunit is used to score-map the real-time interactive data according to the performance degree according to the mapping dimension indicator set to obtain the eloquence performance data; wherein the mapping dimension indicator set includes speech speed, intonation, emotional expression and clarity;

[0107] The fifth subunit is used to evaluate the eloquence performance data according to a preset score system to obtain the eloquence feedback suggestion data; wherein the eloquence feedback suggestion data includes the speed of speech, the monotony of tone and the lack of emotional expression.

[0108] The present invention also provides a storage medium, on which a computer program is stored. The computer program is called and executed by a computer to implement the above-mentioned eloquence training method based on document editing and communication. BRIEF DESCRIPTION OF THE DRAWINGS

[0109] Figure 1 It is a flowchart of an eloquence training method based on document editing and communication provided by an embodiment of the present invention;

[0110] Figure 2 It is a structural schematic diagram of an eloquence training device based on document editing and communication provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0111] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0112] In the description of the present application, it should be understood that the terms "first", "second", "third", "fourth" and "fifth" are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first", "second", "third", "fourth" and "fifth" may explicitly or implicitly include one or more of the features. In the description of the present application, unless otherwise specified, "several" means two or more.

[0113] The eloquence training method based on document editing and communication provided by the embodiment of the present invention is mainly used in speeches, public speeches, education, professional fields and other occasions where eloquence expression training is required.

[0114] Embodiment 1:

[0115] See also Figure 1 The embodiment of the present invention provides an eloquence training method based on document editing and communication, including S1 to S7, and the specific implementation steps are as follows:

[0116] S1. Obtain the target document and the user's voice data.

[0117] Step S1 of the embodiment of the present invention is specifically:

[0118] Get the target document D j and the user's voice data; wherein the target document refers to a document containing different eloquence training contents.

[0119] S2. By calculating the correlation between the target document and the preset eloquence dimension indicator set, a weight set of the text segment set in the target document is obtained.

[0120] In step S2 of the embodiment of the present invention, S2 includes S2.1 to S2.2; wherein S2.1 is a process of defining an initial eloquence dimension indicator set, and S2.2 is a process of obtaining a weight set of a text segment set in a target document, specifically:

[0121] S2.1. Define the initial eloquence dimension indicator set; the initial eloquence dimension indicator set includes speech clarity, speaking speed, intonation, emotional expression and expression skills, specifically:

[0122] ① Clarity of Expression: It is used to measure the clarity of a text and can be defined by calculating the number of complex words and the complexity of sentence structure in the text.

[0123] ②Pace: It is used to measure the speaking speed of a text and can be defined by calculating the average number of words or sentence length.

[0124] ③Tone: It is used to measure the tone of the text, which can be defined by sentiment analysis, for example, the distribution of positive, negative or neutral sentiment.

[0125] ④ Emotional Expression: It is used to measure the emotional expression ability of the text, which can be defined by the frequency and diversity of emotional vocabulary.

[0126] ⑤ Expression Techniques: used to measure the expression techniques of the text, which can be defined by vocabulary diversity, rhetoric and the use of metaphors.

[0127] Through machine learning models, natural language processing techniques (such as vocabulary analysis, grammar analysis, and sentiment analysis) are used to analyze the target document D according to the initial eloquence dimension indicator set. j The document content is analyzed and a set of eloquence dimension indicators is generated; the analysis process requires the use of NLP algorithms and preset libraries, such as NLTK and spaCy in Python;

[0128] Moreover, vocabulary analysis refers to identifying key words in a document to help determine clarity and professionalism; grammatical analysis refers to checking sentence structure and grammatical errors to improve fluency; and sentiment analysis refers to analyzing emotional expressions in text to assess the appropriateness and effectiveness of sentiment.

[0129] It should be noted that the initial set of eloquence dimension indicators is stored in the integrated eloquence dimension indicator database, which contains the definitions and relevant data of indicators such as clarity, fluency and emotional expression, and these indicators will be used to automatically identify and mark areas that need improvement based on the document content.

[0130] This embodiment introduces a variety of eloquence dimension indicators, which can comprehensively and accurately evaluate eloquence problems and help users improve their eloquence and expression skills in an all-round way.

[0131] S2.2. According to the target document D j and an eloquence dimension indicator set, establishing an eloquence dimension weight matrix by means of a training data set and using a machine learning algorithm; wherein the eloquence dimension weight matrix includes the correlation weights between the eloquence dimension indicator set and each text segment in the target document;

[0132] By document D j Each text segment T in i Perform natural language processing on the target document D j Extract each text segment T i The relevant features with the eloquence dimension indicator set are used to obtain the relevant feature set;

[0133] By calculating the relevant feature set for each text segment T i The correlation function F is obtained by comparing the correlation of each eloquence dimension index k in the eloquence dimension index set. k (T i );

[0134] According to the weight and correlation function F of the eloquence dimension weight matrix k (T i ) and the number of text fragments in the target document, and calculate the number of each tag M in the text fragment set i In the target document D j The weight M in ij , get the weight set;

[0135] Among them, the calculation method of each weight in the weight set is:

[0136]

[0137] Among them, M ij For the target document D j The mark M in i Weights, SpeechDimensionWeights ik is the weight of the i-th row and k-th column in the eloquence dimension weight matrix; F k (T i ) is the text segment T i The correlation function between the eloquence dimension index k in the eloquence dimension index set, for example, TF-IDF can be used to represent the correlation; D j1 For the target document D j The number of text snippets in .

[0138] In this embodiment S2.2, since the correlation function can represent the degree of correlation between the vocabulary frequency, emotional expression and complexity between the text fragment and the eloquence dimension indicator set, the weight set established on this basis can reflect the training content that is most needed to be provided to the user in the target document; and the weight set generated by combining the weights of the eloquence dimension weight matrix and the number of text fragments can provide the user with the most valuable practice content while measuring all the text fragments in the target document, which is helpful for the generation of the subsequent to-do list;

[0139] In addition, by learning the eloquence dimension weight matrix through training data sets and machine learning algorithms, more eloquence dimension indicators can be taken into account to ensure the comprehensiveness of the matrix; and the weights of document tags are calculated through the eloquence dimension weight matrix and correlation function, which can improve the accuracy and credibility of the calculation.

[0140] S3. Mark each text segment according to the eloquence dimension score of each text segment in the target document to obtain a marked segment.

[0141] In step S3 of the embodiment of the present invention, S3 includes S3.1 to S3.5; wherein S3.1 is a process of obtaining different scores, S3.2 is a process of marking text segments, S3.3 is a process of obtaining different display scores, S3.4 is a process of starting the marked display, and S3.5 is a process of verifying the effect;

[0142] Furthermore, S3.1-S3.2 is a complete process of marking a text segment, and S3.3-S3.4 is a complete process of displaying the mark;

[0143] Specifically:

[0144] S3.1. By measuring the target document D j The speech clarity of each text segment in T is obtained to obtain the speech clarity score; and the speech clarity score is used to measure the text segment T i the clarity of the rhetoric to identify areas for improvement;

[0145] By evaluating the speech rate of each text segment, a speech rate control score is obtained; and the speech rate control score is used to evaluate the speech rate of the text segment T i to identify whether the speaking rate needs to be changed;

[0146] By evaluating the emotional expression of each text segment, the emotional expression score is obtained; and the emotional expression score is used to evaluate the text segment T i emotional expression to determine whether emotional expression needs to be enhanced;

[0147] By measuring the speech diversity of each text segment, a speech diversity score is obtained; and the speech diversity score is used to measure the text segment T i diversity of rhetoric to determine whether there is a need for greater diversity;

[0148] Specifically:

[0149] ① Speech clarity scores are:

[0150]

[0151] Where N is the text segment T i The number of sentences in Conf j (T i ) is the clarity confidence score of the jth sentence.

[0152] ②Speech speed control score is:

[0153]

[0154] Among them, Speed j (Ti ) is the speaking speed score of the jth sentence.

[0155] ③Emotional expression scores are:

[0156]

[0157] Among them, Sentiment j (T i ) is the sentiment score of the jth sentence, which can be obtained through sentiment vocabulary analysis and sentiment intensity analysis.

[0158] ④The verbal diversity score is:

[0159]

[0160] Among them, Vocabulary (T i ) is the text segment T i The number of different words in TotalWords(T i ) is the text segment T i The total vocabulary in .

[0161] The marked segments in S3.1 of this embodiment take into account a variety of practice measurement criteria such as speech clarity score, speech rate control score, emotional expression score and speech diversity score, while adding personalized needs, which can truly provide users with the most suitable and most needed training benchmarks, and provide strong data support for subsequent calculations.

[0162] S3.2, marking each text segment according to the speech clarity score, speech speed control score, emotional expression score, speech diversity score and preset personalized needs to obtain a marked segment;

[0163] Among them, the marking method is adaptive marking triggering, which can adaptively adjust the weight distribution and the threshold of triggering marking according to the user's eloquence level and personalized needs to ensure that the user obtains the training suggestions that best meet their needs; and the weight distribution refers to adjusting the weight of the eloquence dimension indicators according to the user's preferences, and the threshold refers to the condition for determining when to trigger the eloquence marking. The threshold can be set by the user, or the threshold of triggering marking can be adjusted by the user according to the user's eloquence performance and progress;

[0164] And, based on the user's choice, by combining the eloquence dimension indicator set, it provides users with instant editing and tagging options;

[0165] Additionally, users can choose to trigger the tagging feature manually or activate it by clicking or selecting a specific section in the document.

[0166] S3.3, perform natural language processing, and convert the target document D j The text fragment T i Decompose into sentences, token segments and text segments T i Perform feature extraction to obtain diverse features;

[0167] The eloquence dimension of the diversity features is evaluated through a preset formula set to obtain the display scores of speech clarity, speech speed, emotional expression and speech diversity; and in the process of calculating each score, the range mapping of the indicator and the Sigmoid function are combined to ensure that the label range is between 0 and 1;

[0168] Among them, the speech clarity display score is used to present the text segment T i A mark of clarity of speech, ranging from 0 to 1, with 1 indicating the highest level of need for improvement;

[0169] The speaking rate display score is used to present the text segment T i The speech rate marker ranges from 0 to 1, with 1 indicating the highest level of improvement required;

[0170] The emotional expression display score is used to present the text segment T i The emotional expression mark ranges from 0 to 1, with 1 indicating the highest degree of improvement;

[0171] The speech diversity display score is used to present the text segment T i The rhetorical diversity mark ranges from 0 to 1, with 1 indicating the highest level of need for improvement;

[0172] Specifically:

[0173] ① Speech clarity score:

[0174]

[0175] Among them, μ C Clarity(T i ) is the text segment T i An indicator of speech clarity, ranging from 0 to 1, with 1 indicating complete clarity.

[0176] ②Speech speed display score is:

[0177]

[0178] Among them, μ P is the average value of the speaking speed index; Pace(T i ) is the text segment T iThe speaking speed index ranges from 0 to 1, where 1 indicates an appropriate speaking speed. In addition, an appropriate speaking speed needs to meet the following requirements: ensuring the clarity and fluency of the text, clearly presenting the context and purpose of the speech, ensuring that the audience can understand the content of the speech, and accurately emphasizing information and expressing emotions.

[0179] ③The emotional expression score is:

[0180]

[0181] Among them, μ E is the average value of the emotion expression index; Emotion(T i ) is the text segment T i The emotional expression index ranges from 0 to 1, with 1 indicating the highest emotional expression.

[0182] ④ The scores of verbal diversity are:

[0183]

[0184] Among them, μ D is the average value of the speech diversity index; Diversity(T i ) is the text segment T i An indicator of speech diversity, ranging from 0 to 1, with 1 indicating the highest diversity.

[0185] It should be noted that μ C , μ P , μ E and μ D is through the target document D j The calculation is performed on a large number of text fragments in the target document D j The overall clarity, speaking speed, emotional expression and verbal diversity of Chinese text segments.

[0186] S3.4, according to the speech clarity display score ClarityDisplay (T i )、Speech speed display score PaceDisplay(T i ), Emotion Display score EmotionDisplay(T i ) and DiversityDisplay(T i ), is the target document D j Visually mark each text segment in the text set, and highlight the visually marked results and the text segment with the highest weight in the weight set by presetting specific colors, highlighting and marking icons;

[0187] For example, clarity can be marked in blue, speech rate can be marked in red, and emotional expression can be marked in yellow to help users more intuitively understand where improvements are needed.

[0188] In this embodiment S3.4, by highlighting, the user can most intuitively present the exercise content that requires attention according to different scores, helping the user to easily identify and distinguish the direction of improvement. This visual marking method can enable the user to escape from the boring and large-scale training data in the target document, grasp the key points of the exercise at the first time, and achieve good training results.

[0189] S3.5. Verify the practicality of eloquence dimension marking and the effectiveness of helping users improve their eloquence by obtaining user feedback and regularly monitoring the effectiveness of eloquence training;

[0190] Also, conduct A / B testing to compare text with Eloquent Dimensions tags with text without tags to evaluate the tagging effect and user experience.

[0191] S4. Generate a to-do list based on the eloquence dimension indicator set, the marked fragments, and the text fragment with the highest weight in the weight set.

[0192] In step S4 of the embodiment of the present invention, S4 includes S4.1 to S4.19; wherein S4.1 is a process of obtaining a comprehensive eloquence dimension indicator set, and S4.2 is a process of obtaining a time factor set TimeToEvent (T i ), S4.3 is a process of obtaining a to-do priority set, S4.4 is a process of generating a to-do list, and S4.1-4.4 are a complete process of generating a to-do list;

[0193] S4.5 is a process of defining an eloquence dimension editing indicator set, S4.6 is a process of obtaining a current eloquence dimension score (CS), S4.7 is a process of obtaining an eloquence dimension score (ES), S4.8 is a process of generating user editing optimization suggestions, and S4.5-4.8 are a complete process of generating user editing optimization suggestions;

[0194] S4.9 is a process of obtaining an extended set of eloquence dimension indicators, S4.10 is a process of obtaining text analysis results, S4.11 is a process of obtaining eloquence dimension evaluation results, S4.12 is a process of obtaining context analysis results, S4.13 is a process of obtaining a comprehensive eloquence dimension score (CS), S4.14 is a process of generating eloquence optimization suggestions, and S4.9-4.14 are a complete process of generating eloquence optimization suggestions;

[0195] S4.15 is the process of defining the performance indicator set of the eloquence dimension, and S4.16 is the process of obtaining the comprehensive eloquence score (S total), S4.17 is a process of generating interactive tagging suggestions, S4.18 is a process of using an interactive interface to display information, S4.19 is a process of verifying the effect, and S4.15-4.18 are a complete process of interacting with the user on the interface;

[0196] Specifically:

[0197] S4.1. According to the eloquence dimension index set, the text segments in the marked segments and the text segments with the highest weight in the weight set are comprehensively calculated to obtain a comprehensive eloquence dimension index set; wherein the comprehensive eloquence dimension index P in the comprehensive eloquence dimension index set is i , used to represent the text segment T i The comprehensive eloquence dimension index ranges from 0 to 1, with 1 representing the best eloquence performance and 0 representing the worst;

[0198] Among them, the comprehensive eloquence dimension index P in the comprehensive eloquence dimension index set i for:

[0199] P i =α1·Clarity(T i )+α2·Pace(T i )+α3·Emotion(T i )+α4·Diversity(T i )+α5

[0200] ·Confidence(T i )

[0201] Among them, α1, α2, α3, α4 and α5 are weight coefficients used to balance the importance of different eloquence dimensions, and each coefficient corresponds to an eloquence dimension; the values ​​of these weight coefficients can be adjusted according to the relative importance of the eloquence dimensions, the user's eloquence training needs and preferences, so as to better reflect the user's eloquence training needs;

[0202] Clarity(T i ) represents the text segment T i Symbol for the Clarity Eloquence Dimension Index; The Clarity Eloquence Dimension Index is used to measure the clarity of the verbal expression in a text. The clarity of a text can be determined by vocabulary choice and sentence structure.

[0203] Pace(T i ) represents the text segment T i The symbol of the speed and eloquence dimension index; the speed and eloquence dimension index is used to measure the speed of speech expression in a text, which can be calculated based on the speed changes, pauses and stress in the text;

[0204] Emotion(Ti ) represents the text segment T i The symbol of the emotional expression eloquence dimension index; the emotional expression eloquence dimension index is used to measure the degree of emotional communication in the text, which can be calculated by the emotional vocabulary, tone and emotional intensity in the text;

[0205] Diversity(T i ) represents the text segment T i The symbol of the diversity eloquence dimension index; the diversity eloquence dimension index is used to measure the degree of diversity of words and expressions in a text, which can be calculated through vocabulary diversity, changes in sentence structure, etc.

[0206] Confidence(T i ) represents the text segment T i The symbol of the confidence eloquence dimension index; the confidence eloquence dimension index is used to measure the confidence level of verbal expression in a text, which can be calculated through the confident vocabulary, tone and confidence level in the text; and the confidence level can be specifically evaluated through the stability of speech speed and tone, clarity of speech, body language and facial expressions, word choice and user feedback.

[0207] S4.2, obtaining the user's speech target completion time and speech date;

[0208] According to the completion time of the speech goal, the speech date and the voice data, determine the practice time that the user needs to spend on the eloquence dimension indicator set when conducting eloquence training, and obtain the time factor set;

[0209] Among them, the time factor set is used to measure the time that users need to spend to complete relevant exercises before the formal speech or eloquence expression event; each time factor in the time factor set is expressed as TimeToEvent (T i ), and TimeToEvent(T i ) and the text fragment T i Corresponding.

[0210] S4.3. Combine the comprehensive eloquence dimension indicator set with the preset time factor set to generate a to-do priority set; among which, the to-do priority Priority (T i ) is the comprehensive eloquence dimension index P i The function takes into account factors such as time urgency and is used to represent the text segment T i The priority of the to-do items, ranging from 0 to 1, with 1 being the highest priority;

[0211] Among them, the priority of the to-do item in the to-do priority set is Priority (T i )for:

[0212] Priority(T i )=f(P i ,TimeToEvent(T i ))

[0213] Among them, f(P i ,TimeToEvent(T i )) is a function used to calculate the priority of a to-do item, which accepts the eloquence dimension indicator P i and each time factor in the time factor set TimeToEvent(T i ) as input, and the specific form of the function can be adjusted according to the system design and user needs. The function is used to comprehensively consider the eloquence performance and time urgency to determine the priority; TimeToEvent(T i ) represents each time factor in the time factor set.

[0214] Since the time factor set S4.3 of this embodiment is calculated based on the speech target completion time and voice data, it can objectively measure the gap between the user's current speech control time and the target time, so that the user can better consider the user's time constraints and speech dates in different dimensions of the eloquence dimension indicator set, formulate the most suitable training content for the user, and ensure that the user's training progress is always in an appropriate state.

[0215] S4.4, according to the to-do priority set, prioritize the text segments in the marked segment and the text segments with the highest weight in the weight set to obtain a to-do list;

[0216] Each to-do item in the to-do list includes a related text segment T i , and suggested directions for improvement in the eloquence dimension; for example, asking users to practice improving clarity, controlling speech speed, or enhancing emotional expression;

[0217] Furthermore, the to-do list is used to present the generated to-do items to the user for viewing and management.

[0218] The to-do list in S4.1-S4.4 of this embodiment focuses on improving the user's shortcomings in eloquence training based on the target document; wherein, the to-do list is obtained by sorting the text fragments according to the priority of the weight, so that the user can first be exposed to the content that most needs to be practiced; and, by combining the time factor set, the practice content in the to-do list can be reasonably planned according to the user's personal preset time, so as to achieve the best training effect within a limited time.

[0219] S4.5. Define the editing index set of eloquence dimension; the editing index set of eloquence dimension includes speech clarity, speaking speed, intonation, emotional expression, body language and word choice, specifically:

[0220] ① Speech clarity (C): used to indicate the clarity of the text, ranging from 0 to 100.

[0221] ②Speed ​​(S): used to indicate the speaking speed of the text, ranging from 0 to 100.

[0222] ③ Tone (T): used to indicate the tone of the text; the range is -100 to 100, negative numbers indicate low tones, and positive numbers indicate high tones.

[0223] ④ Emotional expression (E): used to indicate the emotional expression of the text; the range is -100 to 100, negative numbers represent negative emotions, and positive numbers represent positive emotions.

[0224] ⑤Body language (B): used to indicate the accuracy of body language and expressions in the text, ranging from 0 to 100.

[0225] ⑥Word Choice (W): used to indicate whether the word choice of the text is appropriate, ranging from 0 to 100.

[0226] S4.6, calculating the current eloquence dimension score (CS) of each to-do item in the to-do list according to the text content of the to-do list and the eloquence dimension editing indicator set;

[0227] Among them, the current eloquence dimension score is:

[0228] CS=(C1+S1+T1+E1+B1+W1) / 6

[0229] Among them, C1, S1, T1, E1, B1 and W1 are respectively the speech clarity index, speaking speed index, intonation index, emotional expression index, body language index and word selection index of each to-do list before editing; and the above indicators are calculated based on the indicators in the eloquence dimension editing indicator set for the text content of the to-do list.

[0230] S4.7, obtaining the results of the user's edits to the to-do list, including speech clarity, speech speed, intonation, emotional expression, body language, and word choice;

[0231] According to the text content in the editing result, the eloquence dimension score (ES) of the user after editing is calculated;

[0232] Among them, the scores of eloquence dimension are:

[0233] ES=(C2+S2+T2+E2+B2+W2) / 6

[0234] Among them, C2, S2, T2, E2, B2 and W2 respectively represent the speech clarity index, speaking speed index, intonation index, emotional expression index, body language index and word choice index of each to-do item in the to-do list after editing.

[0235] S4.8, calculating the difference between the eloquence dimension score after the user edits and the eloquence dimension score before the user edits, and obtaining the influence value (IE);

[0236] Generate user editing optimization suggestions based on the influence value (IE) and eloquence dimension editing indicator set; specifically: if the influence value is positive, encourage the user to continue similar editing; if the influence value is negative, provide improvement suggestions, such as improving the accuracy of body language or improving vocabulary selection;

[0237] The impact value is:

[0238] IE=ES-CS

[0239] Among them, ES is the eloquence dimension score, and CS is the current eloquence dimension score.

[0240] S4.9. Expand and redefine the initial eloquence dimension indicator set to obtain an extended set of eloquence dimension indicators; the extended set of eloquence dimension indicators includes speech clarity, speaking speed, intonation, emotional expression, discourse logic and expression confidence, specifically:

[0241] ① Speech clarity (C): used to indicate the clarity of the text, ranging from 0 to 100.

[0242] ②Speed ​​(S): used to indicate the speaking speed of the text, ranging from 0 to 100.

[0243] ③ Tone (T): used to indicate the tone of the text; the range is -100 to 100, negative numbers indicate low tones, and positive numbers indicate high tones.

[0244] ④ Emotional expression (E): used to indicate the emotional expression of the text; the range is -100 to 100, negative numbers represent negative emotions, and positive numbers represent positive emotions.

[0245] ⑤Discourse logic (L): used to indicate the logic of the text, ranging from 0 to 100.

[0246] ⑥ Expression confidence (CF): used to indicate the expression confidence of the text, ranging from 0 to 100.

[0247] S4.10. For target document D j Perform natural language processing (NLP) analysis to identify key elements in the text, such as topics, sentiment, keywords, and logical structure, and obtain text analysis results.

[0248] S4.11. For each eloquence dimension indicator in the eloquence dimension indicator extension set, analyze the target document D according to the text analysis results. j The elements related to the indicator are used to obtain the evaluation results of the eloquence dimension;

[0249] For example, speech clarity may be related to sentence structure and vocabulary, speaking rate may be related to sentence length and pauses, intonation may be related to modal particles and pitch changes, emotional expression may be related to emotional vocabulary and emotional emotions, discourse logic may be related to sentence connection and argument expression, and expression confidence may be related to assertive vocabulary and tone.

[0250] S4.12. According to the target document D j Conduct context analysis on the content of the preceding and following texts to obtain the context analysis results;

[0251] Specifically: based on the relevance and logical relationship of text elements, context weighting is applied to each eloquence dimension indicator in the extended set of eloquence dimension indicators, and the context analysis result is obtained according to the context weighted result analysis; wherein, the weight of the context weighting is determined according to the importance of the context elements.

[0252] S4.13. Obtain a comprehensive eloquence dimension score (CS) by contextually weighting the comprehensive eloquence dimension indicators;

[0253] Among them, the comprehensive eloquence dimension score is:

[0254] CS=(C*WC+S*WS+T*WT+E*WE+L*WL+CF*WCF) / (WC+WS+WT+WE

[0255] +WL+WCF)

[0256] Among them, C, S, T, E, L and CF respectively represent the speech clarity index, speaking speed index, intonation index, emotional expression index, discourse logic index and expression confidence index of the current text in the eloquence dimension evaluation results, and the above indicators correspond one-to-one to the indicators in the eloquence dimension indicator extension set; WC, WS, WT, WE, WL and WCF respectively represent the context weighting of the corresponding indicators.

[0257] S4.14. Generate eloquence optimization suggestions based on the comprehensive eloquence dimension score (CS) and context analysis results; for example, if the discourse logic score is lower than the preset value, it is recommended that the user improve the argument connection method and expression logic;

[0258] Furthermore, users can view the eloquence optimization suggestions and choose whether to accept them.

[0259] In this embodiment S4.10-S4.14, through context analysis, the parts of the document that require special attention are identified, and interactive eloquence optimization suggestions are provided, which help users to conduct eloquence training more focused and efficient, so that users can better understand the areas that need improvement, improve the quality of expression, and provide more targeted improvement suggestions.

[0260] S4.15. Define the eloquence dimension performance indicator set; the eloquence dimension performance indicator set includes language fluency, emotional coherence, tone adjustment and topic consistency, specifically:

[0261] ① Language Fluency (F): used to measure the fluency and coherence of the text; the range is 0 to 100, where 100 means the most fluent.

[0262] ② Emotional coherence (CE): used to consider the consistency of emotional expression in the text; the range is 0 to 100.

[0263] Where 100 means the most consistent.

[0264] ③ Tone Adjustment (MT): used to detect the tone in the text; the range is -100 to 100, negative numbers represent negative tone, and positive numbers represent positive tone.

[0265] ④ Thematic consistency (CT): used to consider the consistency of each part of the text with the theme; the range is 0 to 100, where 100 means the most consistent.

[0266] S4.16, calculate the comprehensive eloquence score (S total );

[0267] Among them, the comprehensive eloquence score is:

[0268] S total =w1C+w2S+w3T+w4E+w5L+w6CF+w7F+w8CE+w9MT+w 10 CT

[0269] Among them, w1 to w 10 They are the weights of each eloquence dimension in the eloquence dimension performance index set, and can be adjusted according to user needs.

[0270] S4.17, according to the comprehensive eloquence score (S total ) determines the importance level of the suggestions; texts with lower scores will receive more suggestions for improvement;

[0271] According to the set of eloquence dimension performance indicators, relevant interactive tagging suggestions are generated for each indicator. For example, if the emotional coherence (CE) score is low, it is recommended that the user maintain consistent emotional expression in the text; and the user can view the interactive tagging suggestions and choose whether to accept them;

[0272] According to the comprehensive eloquence score (S total ), optionally generating goal settings to optimize interactive tagging suggestions; for example, if a user's rating is low, setting achievable goals to help the user improve over time.

[0273] S4.18. Develop a user interactive interface using front-end development technology and back-end development technology, and integrate the recommendations with the user interactive interface to display the recommendations, goals, and user progress;

[0274] In addition, users can activate the tagging function and edit to-do items, as well as view tags, suggestions and their own improvement progress, and choose which suggestions to implement, set goals, and then track the corresponding improvements;

[0275] Among them, front-end development refers to the use of front-end technologies such as HTML, CSS, and JavaScript to create interactive user interfaces; back-end development refers to the use of server-side scripting languages ​​(such as Python and Node.js, etc.) to process user requests and data exchanges;

[0276] Among them, suggestions are made through a preset core algorithm and are associated with eloquence dimension indicators; for example, after the user marks a piece of text, similar practice content will be recommended to help the user concentrate on practicing specific eloquence skills. Users can view these suggestions on the interactive interface and perform corresponding exercises.

[0277] It should be noted that, in the above according to the target document D j In the process of generating tags and suggestions for the eloquence dimension indicator set, the eloquence dimension indicator extension set, and the eloquence dimension performance indicator set, machine learning algorithms are used. Specifically, the training models in the machine learning algorithms are used to predict the eloquence dimension indicators, and these prediction models are optimized based on the existing training data. These algorithms will generate tags and suggestions based on different document types and eloquence dimension indicators. The specific technical means include:

[0278] ① Feature engineering: Convert text data into features that can be used by machine learning algorithms, such as word frequency and grammatical structure.

[0279] ② Training model: Use existing data sets to train machine learning models, such as recurrent neural networks (RNNs) or deep learning models, to learn the complex relationship between text fragments and eloquence dimensions to predict eloquence dimension indicators.

[0280] ③Model evaluation and optimization: Evaluate the performance of machine learning models through methods such as cross-validation, and adjust and optimize them according to user requirements.

[0281] In this embodiment S4.18, the marking and suggestion functions can be activated at any time to obtain instant feedback on eloquence expression. This real-time interactive method helps users correct problems, improve performance, and provide a more efficient eloquence training experience.

[0282] S4.19. The results of the above training programs shall be verified through user satisfaction surveys, eloquence skill assessments, data analysis and comparative studies, specifically:

[0283] 1. User satisfaction survey: Conduct user satisfaction surveys regularly to collect user feedback and suggestions to understand user satisfaction and experience with the training program. User feedback will be used to improve the training program.

[0284] 2. Oratory skills assessment: The degree of improvement is assessed by regularly evaluating the user's speech or eloquence performance before and after using the training program. This can be achieved by comparing the user's eloquence skill score or actual performance.

[0285] 3. Data analysis: Analyze the user's markings, suggestions, and practice data in the training program to evaluate the user's learning progress and improvement; for example, you can track the number of exercises completed by the user and the improvement in the eloquence dimension.

[0286] 4. Comparative study: compare with traditional eloquence training methods to prove the originality and effectiveness of the eloquence training program in this embodiment; and the objects of comparison may include indicators such as user improvement speed, learning effect and satisfaction.

[0287] S5. Construct the first task based on the textual representation, emotional information and body language information of the speech data.

[0288] In step S5 of the embodiment of the present invention, S5 includes S5.1 to S5.4; wherein S5.1 is a process of obtaining an initial weight set and a score of an eloquence dimension, S5.2 is a process of obtaining a task weight set, S5.3 is a process of generating a first task, and S5.4 is a process of performing effect verification, specifically:

[0289] S5.1. Use natural language processing and speech recognition engines to convert voice data recorded in real time through a microphone into text form, and use natural language processing technology to identify the emotional polarity and emotional intensity of the speech text in the voice data to obtain text conversion results; the text conversion results include sentiment analysis results and speech content; the natural language processing and speech recognition engines can choose Google's Speech-to-Text or Microsoft's Azure SpeechService; the natural language processing technology can choose a sentiment vocabulary library and sentiment analysis algorithm, such as VADER sentiment analyzer;

[0290] According to the transformation results, the initial weights of each eloquence dimension in the eloquence dimension set are determined to obtain the initial weight set; the eloquence dimension set includes clarity (C), speaking speed (P), posture and language (B), expression and body language (F), emotional expression (E) and grammatical accuracy (G); the weights in the initial weight set include W C , W P , W B , W F , W E and W G , and, as indicated by the letter subscripts of each weight, each weight in the initial weight set and each eloquence dimension in the eloquence dimension set are in one-to-one correspondence;

[0291] Based on the hybrid method of acoustic analysis, text analysis and sentiment analysis, the text conversion results are calculated according to the eloquence dimension set using a preset mathematical model to obtain the score of each eloquence dimension, including the clarity score C Score , speaking speed score P Score , Posture and Language Score B Score , Expression and body language score F Score , Emotional expression score E Score and grammatical accuracy score G Score ;

[0292] S5.2. Calculate the task weight set based on the initial weight set and the user's target score on the eloquence dimension set; the task weight set includes the clarity weight Task WeightC , Speech rate weight Task WeightP , Posture and Language Weight Task WeightB , facial expression and body language weight Task WeightF , Emotional expression weight Task WeightE and grammatical accuracy weight Task WeightG , specifically:

[0293] Task WeightC =W C ΔC;

[0294] Task WeightP =W P ΔP;

[0295] Task WeightB =W B ΔB;

[0296] Task WeightF =W F ΔF;

[0297] Task WeightE =W E ΔE;

[0298] Task WeightG =W G ΔG

[0299] And, for the parameters ΔC, ΔP, ΔB, ΔF, ΔE and ΔG:

[0300] ΔC=C Tanget ―C Score ;

[0301] ΔP=P Tanget ―P Score ;

[0302] ΔB=B Target ―B Score ;

[0303] ΔF=F Target ―F Score ;

[0304] ΔE=E Target ―E Score ;

[0305] ΔG=G Tanget ―G Score

[0306] Among them, C Tanget , P Tanget , B Target 、F Target 、E Target and G Tanget is the user's target score in each eloquence dimension in each eloquence dimension set. These target scores can be set by the user or according to the suggestions of the eloquence training program; W C , W P , W B , W F , W E and W Gare all weights in the initial weight set; parameters ΔC, ΔP, ΔB, ΔF, ΔE, and ΔG represent the score differences of each eloquence dimension in the eloquence dimension set, C Score , P Score , B Score 、F Score 、E Score and G Score They are clarity score, speaking speed score, posture and language score, expression and body language score, emotional expression score and grammatical accuracy score.

[0307] S5.3. Generate a number of corresponding initial tasks on different dimensions of the eloquence dimension set according to the eloquence dimension set and the voice data; specifically: use supervised learning, reinforcement learning or generative adversarial networks (GANs) and other machine learning techniques to generate a number of corresponding initial tasks on different dimensions of the eloquence dimension set according to the user's eloquence dimension set, text conversion results (the results include sentiment analysis results and speech content) and the user's relevant training data; and the number of initial tasks include specific suggestions, examples and practice methods, and the user can choose whether to accept these suggestions later and then start practicing the speech or eloquence expression task;

[0308] Determine the priority of each task in the initial tasks according to the weight of each task in the task weight set;

[0309] Sorting each of the plurality of initial tasks according to the priority, and forming a first task from the sorted plurality of initial tasks;

[0310] In addition, detailed suggestions for each training task in the first task are provided, including specific practice content, practice time, practice frequency, etc.

[0311] The first task of S5.1-S5.3 of this embodiment focuses on improving the user's eloquence and expression. By establishing the first task based on the eloquence dimension set including clarity, speech speed, posture and language, expression and body language, emotional expression and grammatical accuracy, it can help the user to effectively improve the oral expression ability, expression expression ability and body expression ability. Moreover, through the form of document editing and marking, the first task generated can have strong personal characteristics according to the user's eloquence dimension indicators and the characteristics of the target document, which helps the user to focus on improving the specific problems in his eloquence expression, improve the training efficiency and strengthen the pertinence of weak points.

[0312] In addition, by sorting each of the initial tasks according to the priority level, the focus of the generated first task is clear, allowing users to access the content that needs training and improvement at the first time, which can improve training efficiency.

[0313] S5.4. Verify the training results of the first task through the improvement of eloquence dimension indicators, user satisfaction survey, comparative research and automated evaluation. Specifically:

[0314] 1. Improvement of eloquence dimension indicators: Regularly track the user's eloquence dimension indicators and compare the changes before and after using the first task to verify the improvement effect of the first task on the user's eloquence skills.

[0315] 2. User satisfaction survey: Regularly collect user feedback and suggestions to understand the user's training experience and satisfaction.

[0316] 3. Comparative study: Conduct a comparative study with users who did not use the first task, and analyze the impact of using the first task on the improvement of eloquence skills to verify the effectiveness of the first task.

[0317] 4. Automated evaluation: Automatically evaluate the user's speech or eloquence, and provide feedback and scores to the user based on eloquence dimension indicators and sentiment analysis to directly quantify the training effect.

[0318] S6. Interact with the user in real time through a preset instant messaging function, and generate a second task based on the interaction performance and eloquence dimension indicator set.

[0319] In step S6 of the embodiment of the present invention, S6 includes S6.1 to S6.6; wherein S6.1 is the process of obtaining real-time interactive data and converting it into text, and S6.2 is the process of obtaining User Performance data. i System Feedback i The process of S6.3 is the process of obtaining the EloquenceImprovementIndex, S6.4 is the process of displaying the EloquenceImprovementIndex, S6.5 is the process of generating the second task, and S6.6 is the process of verifying the effect, which is specifically as follows:

[0320] S6.1. Interact with users in real time through a preset instant messaging function to obtain real-time interaction data, and use speech recognition technology to convert the real-time interaction data into text;

[0321] Among them, the instant messaging function is the instant messaging and feedback module, which integrates advanced speech recognition and sentiment analysis technologies; real-time interaction includes asking questions, speaking and requesting feedback to provide real-time interactive functions including text chat, voice calls and video calls, allowing users to ask questions, share progress or request feedback to coaches or other learners at any time.

[0322] S6.2. Based on the mapping dimension indicator set, the text of the real-time interactive data is scored according to the performance level to obtain the eloquence performance data User Performance i ; Among them, the mapping dimension indicator set includes speaking speed, intonation, emotional expression and clarity;

[0323] Use deep learning and machine learning technologies (including acoustic models and sentiment analysis models) to analyze the user's eloquence in real time, and evaluate the eloquence performance data according to the preset score system to obtain eloquence feedback and suggestion data System Feedback i ; Among them, the eloquence feedback suggestion data includes the speed of speech, the monotony of intonation and the lack of emotional expression;

[0324] Furthermore, during the analysis process, natural language processing (NLP) technology is used to understand the text of users’ real-time interactive data in order to better meet users’ needs and provide personalized suggestions, and sentiment analysis technology is used to analyze users’ emotional expressions, including identifying users’ emotional tendencies and emotional intensity, in order to provide more comprehensive feedback;

[0325] And get the eloquence feedback suggestion data System Feedback i After that, relevant eloquence expression feedback is immediately provided to the user based on the eloquence feedback suggestion data. For example, if the user speaks faster than the preset speed, it is recommended to slow down the speaking speed and provide specific examples.

[0326] Among them, the mapping dimension indicator set is established on the basis of the eloquence dimension indicator set, including:

[0327] ① Pace: It is used to reflect the speaking speed of eloquence. The scoring criteria are as follows:

[0328] 1.0: Perfect speaking speed, moderate speed, clear and easy to understand;

[0329] 0.7-0.9: Good speaking speed, slightly faster or slower than moderate, but not affecting understanding;

[0330] 0.4-0.6: medium speaking speed, the speaking speed is obviously too fast or too slow, and may need improvement;

[0331] 0.0-0.3: Poor speaking speed, extremely fast or slow, difficult to understand.

[0332] ② Tone: It is used to reflect the eloquence and the ups and downs of the tone and voice. The scoring criteria are as follows:

[0333] 1.0: Perfect intonation, with appropriate pitch changes and rich expression;

[0334] 0.7-0.9: Good intonation, moderate pitch variation, but may be slightly monotonous;

[0335] 0.4-0.6: Medium intonation, the pitch changes are not obvious enough, need to be improved;

[0336] 0.0-0.3: Poor intonation, with a single tone or large changes in pitch, which affects understanding.

[0337] ③ Emotion: used to reflect the communication and consistency of emotions in eloquence. The scoring criteria are as follows:

[0338] 1.0: Perfect emotional expression, emotional expression is natural, consistent and appropriate;

[0339] 0.7-0.9: Good emotional expression, emotional expression is consistent and appropriate, but may be slightly unnatural;

[0340] 0.4-0.6: Moderate emotional expression, emotional expression is sometimes inconsistent or unnatural, needs improvement;

[0341] 0.0-0.3: Poor emotional expression, inconsistent or inappropriate emotional expression, difficulty perceiving emotions.

[0342] ④ Clarity: It is used to reflect the clarity and comprehensibility of eloquence. The scoring criteria are as follows:

[0343] 1.0: Perfect clarity, clear and understandable expression without ambiguity;

[0344] 0.7-0.9: Good clarity, with most of the expression clear, but there may be some blurriness or lack of clarity;

[0345] 0.4-0.6: medium clarity, the expression is often vague or unclear, and needs improvement;

[0346] 0.0-0.3: Poor clarity, unclear and difficult to understand, needs significant improvement.

[0347] The scoring system includes evaluation criteria and classified feedback information, specifically:

[0348] 1. Establish evaluation criteria: Based on the range and scores of the eloquence dimension indicators, establish evaluation criteria. For example, you can define a speaking speed of 0.8 or above as good, a tone of voice of 0.6 or above as good, etc.

[0349] 2. Classify feedback information: Classify the feedback information provided by the system to ensure that the feedback information is associated with the eloquence dimension indicators. For example, feedback information includes too fast speaking speed, monotonous tone, and insufficient emotional expression.

[0350] In this embodiment S6.2, the definition and scoring rules of the mapping dimension indicator set map the eloquence dimension indicators to a score range of 0 to 1, so that the eloquence performance can be quantitatively evaluated. The user's specific eloquence performance will be scored according to these rules and standards and used to calculate the eloquence improvement index;

[0351] Moreover, through the score mapping method, the eloquence performance data can be used as an exact score to measure the user's eloquence performance; since the eloquence feedback suggestion data includes the speed of speech, the monotony of tone and the lack of emotional expression, the eloquence feedback suggestion data is not a single speech speed feedback data, allowing users to understand the current eloquence training effect from different angles.

[0352] S6.3. Generate an eloquence improvement index EloquenceImprovementIndex based on the number of real-time interactions, eloquence performance data and eloquence feedback suggestion data; the eloquence improvement index can be a positive number or a negative number, a positive number indicates an improvement in eloquence, and a negative number indicates a decline in eloquence, and is used to evaluate the user's overall progress in eloquence; and the eloquence improvement index can be accumulated with multiple interactions of the user to better understand the user's overall eloquence progress;

[0353] Among them, the eloquence improvement index is:

[0354]

[0355] Among them, α and β are weight parameters, which are used to balance the importance of user performance and system feedback, and can be adjusted according to the importance of eloquence performance and feedback to meet user requirements; User Performance i System Feedback for eloquence performance data i Suggest data for eloquence feedback;

[0356] S6.4, displaying the Eloquence Improvement Index on the user interface of the interactive system where the eloquence training is performed, so that the user can understand his or her own progress in real time;

[0357] Users can also adjust their learning strategies based on changes in the progress index, ask more questions or request more feedback to acquire better eloquence skills.

[0358] S6.5, generating a second task according to the eloquence improvement index, the eloquence dimension indicator set and the user's real-time performance; wherein the second task includes specific practice content, practice goals and schedule;

[0359] In addition, users can interact with the instant messaging function at any time to ask questions, share their progress, and request more feedback; and dynamically generate new training tasks based on user needs and performance to update the second task.

[0360] The second task of S6.1-S6.5 of this embodiment focuses on generating tasks for training by feedback adjustment based on the user's real-time interactive data; wherein the second task is generated based on the real-time interactive data with the user, so that the user's training progress can be understood in a timely manner; since the eloquence improvement index can reflect the user's progress in different eloquence dimensions, the limited training time can be used to train all eloquence shortcomings to the greatest extent, and improve the user's eloquence skills in a targeted manner. This method of providing instant support and feedback helps to better understand eloquence problems and accelerate learning progress;

[0361] In addition, through instant messaging and feedback, users can obtain real-time feedback and understand their speech performance, including speech clarity, speaking speed, emotional expression, etc. This helps users quickly identify problems and take corrective measures to rapidly improve their eloquence skills.

[0362] S6.6. Verify the training effect of the second task by improving the eloquence dimension indicators, conducting user satisfaction surveys, and comparing the eloquence expression effects of users. Specifically:

[0363] 1. Improvement of eloquence dimension indicators: Continuously track the improvement of users' eloquence dimension indicators, for example, whether the speaking speed has been slowed down, the clarity has been improved, or the emotional expression has been improved, and these changes will be recorded and displayed as part of the effect verification.

[0364] 2. User satisfaction survey: Conduct user satisfaction surveys regularly to understand users’ experience and satisfaction level with the instant messaging and feedback modules, and provide suggestions for improvements.

[0365] 3. Compare the effectiveness of users’ eloquence: Through comparative analysis of the effectiveness of eloquence with the users’ partners and coaches, the uniqueness and effectiveness of users in improving their eloquence can be demonstrated.

[0366] S7. The eloquence training method consists of a to-do list, a first task and a second task.

[0367] Step S7 of the embodiment of the present invention is specifically:

[0368] The eloquence training method consists of a to-do list, a first task and a second task.

[0369] The eloquence training method of the present embodiment includes a to-do list, a first task and a second task. Since the traditional method lacks the ability of real-time analysis and feedback, the user needs to complete the training task under the guidance of a coach, so it is usually impossible to provide a highly personalized training plan; and due to the lack of an effective self-monitoring and improvement mechanism, the user is usually unable to understand his or her own speech performance in a timely manner, and thus cannot quickly discover problems and correct them in time; the eloquence training method of the present embodiment starts from three different eloquence training dimensions, gives full play to the subjective initiative of the user, and helps the user avoid these limited problems of traditional training through document editing and marking, real-time communication and feedback, and generation of personalized training tasks, which helps the user obtain sufficient and practical speaking opportunities and improve his or her eloquence skills.

[0370] It should be noted that the core innovation of this embodiment is:

[0371] 1. Eloquence dimension evaluation algorithm: By developing a series of mathematical formulas and algorithms to evaluate eloquence dimensions, such as speech clarity, speaking speed, emotional expression, speech diversity, expression skills, etc., in order to accurately evaluate eloquence issues.

[0372] 2. Personalized task generation: Intelligently generate personalized training tasks based on the user's eloquence dimension indicators, learning goals and document content to provide a highly personalized learning path.

[0373] 3. Real-time interaction and feedback: Through instant messaging integration, users can activate the marking and suggestion functions at any time to obtain instant feedback on their eloquence.

[0374] 4. Contextual analysis: Through text analysis technology, it automatically identifies the parts of the document that need special attention, which helps users better understand the content of the document and improve the quality of expression.

[0375] 5. Complex Mathematical Operations: By using complex mathematical formulas and algorithms to accurately measure the dimensions of eloquence, such as weight matrices and correlation functions.

[0376] 6. Task management and priority: By automatically generating a to-do list, it helps users manage eloquence training tasks according to eloquence dimension indicators and the priority of the learning plan, thereby providing more convenient task management and learning progress tracking.

[0377] 7. Instant messaging integration: By integrating the instant messaging function into the training method, users can interact, ask questions, and request suggestions with the preset instant messaging function, which helps to better understand the eloquence problems, learning progress, and get instant support and feedback.

[0378] These key technical points constitute the core innovation of this embodiment, which helps to improve the efficiency, personalization and interactivity of eloquence training.

[0379] In general, the embodiments of the present invention have the following beneficial effects:

[0380] In this embodiment, the to-do list focuses on improving the user's eloquence training shortcomings based on the target document. Since the weight set is obtained through correlation analysis, the relevance of different contents can be measured, reflecting the training content that is most needed to be provided to the user in the target document. In addition, the to-do list is generated based on the marked fragments and the text fragments with the highest weights in the weight set, so the most beneficial training content for the user is extracted based on the target document; the first task focuses on improving the user's eloquence expression, which can help the user to effectively improve their oral expression ability, facial expression ability and body expression ability; the second task focuses on generating tasks for training through feedback adjustment based on the user's real-time interactive data. This method of providing instant support and feedback helps to better understand eloquence problems and accelerate learning progress.

[0381] Moreover, since the eloquence training method in this embodiment eliminates the dependence on coaches, the eloquence training resources are more accessible, so that anyone can obtain high-quality eloquence training; through personalized task generation, detailed eloquence dimension indicator analysis and real-time feedback, it helps users to significantly improve their eloquence skills, and effectively improve their speech, expression of emotions, and communication of clear ideas.

[0382] Embodiment 2:

[0383] See also Figure 2 , an embodiment of the present invention provides an eloquence training device based on document editing and communication, comprising an acquisition module 10, a weight module 20, a marking module 30, a list module 40, a construction module 50, a communication module 60 and a combination module 70;

[0384] Wherein, the acquisition module 10 is used to acquire the target document and the user's voice data;

[0385] A weight module 20, configured to obtain a weight set of a text segment set in a target document by calculating a correlation between the target document and a preset eloquence dimension indicator set;

[0386] A marking module 30 is used to mark each text segment in the target document according to the eloquence dimension score of each text segment to obtain a marked segment;

[0387] A list module 40, for generating a to-do list according to the eloquence dimension indicator set, the marked fragments and the text fragment with the highest weight in the weight set;

[0388] A construction module 50, configured to construct a first task according to the textual representation, emotional information and body information of the speech data;

[0389] The communication module 60 is used to interact with the user in real time through a preset instant messaging function, and generate a second task according to the interaction performance and the eloquence dimension indicator set;

[0390] The combination module 70 is used to form an eloquence training method by using the to-do list, the first task and the second task.

[0391] In one embodiment, the acquisition module 10 is specifically:

[0392] Get the target document D j and the user's voice data; wherein the target document refers to a document containing different eloquence training contents.

[0393] In one embodiment, the weight module 20 includes a definition unit, a matrix unit, a feature set unit, a correlation unit and a weight set unit; wherein the definition unit is a process of defining an initial eloquence dimension indicator set, and the matrix unit, the feature set unit, the correlation unit and the weight set unit are processes of obtaining a weight set of a text segment set in a target document, specifically:

[0394] The definition unit is used to define the initial eloquence dimension indicator set; wherein the initial eloquence dimension indicator set includes speech clarity, speaking speed, intonation, emotional expression and expression skills, specifically:

[0395] ① Clarity of Expression: It is used to measure the clarity of a text and can be defined by calculating the number of complex words and the complexity of sentence structure in the text.

[0396] ②Pace: It is used to measure the speaking speed of a text and can be defined by calculating the average number of words or sentence length.

[0397] ③Tone: It is used to measure the tone of the text, which can be defined by sentiment analysis, for example, the distribution of positive, negative or neutral sentiment.

[0398] ④ Emotional Expression: It is used to measure the emotional expression ability of the text, which can be defined by the frequency and diversity of emotional vocabulary.

[0399] ⑤ Expression Techniques: used to measure the expression techniques of the text, which can be defined by vocabulary diversity, rhetoric and the use of metaphors.

[0400] Through machine learning models, natural language processing techniques (such as vocabulary analysis, grammar analysis, and sentiment analysis) are used to analyze the target document D according to the initial eloquence dimension indicator set. j The document content is analyzed and a set of eloquence dimension indicators is generated; the analysis process requires the use of NLP algorithms and preset libraries, such as NLTK and spaCy in Python;

[0401] Moreover, vocabulary analysis refers to identifying key words in a document to help determine clarity and professionalism; grammatical analysis refers to checking sentence structure and grammatical errors to improve fluency; and sentiment analysis refers to analyzing emotional expressions in text to assess the appropriateness and effectiveness of sentiment.

[0402] It should be noted that the initial set of eloquence dimension indicators is stored in the integrated eloquence dimension indicator database, which contains the definitions and relevant data of indicators such as clarity, fluency and emotional expression, and these indicators will be used to automatically identify and mark areas that need improvement based on the document content.

[0403] This embodiment introduces a variety of eloquence dimension indicators, which can comprehensively and accurately evaluate eloquence problems and help users improve their eloquence and expression skills in an all-round way.

[0404] Matrix unit, used to calculate the target document D j and an eloquence dimension indicator set, establishing an eloquence dimension weight matrix by means of a training data set and using a machine learning algorithm; wherein the eloquence dimension weight matrix includes the correlation weights between the eloquence dimension indicator set and each text segment in the target document;

[0405] Feature set unit, used to j Each text segment T in i Perform natural language processing on the target document D j Extract each text segment T i The relevant features with the eloquence dimension indicator set are used to obtain the relevant feature set;

[0406] The relevance unit is used to calculate the relevant feature set for each text segment T i The correlation function F is obtained by comparing the correlation of each eloquence dimension index k in the eloquence dimension index set. k (T i );

[0407] The weight set unit is used to calculate the weight and correlation function F of the eloquence dimension weight matrix. k (T i ) and the number of text fragments in the target document, and calculate the number of each tag M in the text fragment set i In the target document D j The weight M in ij, get the weight set;

[0408] Among them, the calculation method of each weight in the weight set is:

[0409]

[0410] Among them, M ij For the target document D j The mark M in i Weights, SpeechDimensionWeights ik is the weight of the i-th row and k-th column in the eloquence dimension weight matrix; F k (T i ) is the text segment T i The correlation function between the eloquence dimension index k in the eloquence dimension index set, for example, TF-IDF can be used to represent the correlation; D j1 For the target document D j The number of text snippets in .

[0411] In the matrix unit, feature set unit, correlation unit and weight set unit of this embodiment, since the correlation function can represent the degree of correlation between the vocabulary frequency, emotional expression and complexity between the text fragment and the eloquence dimension indicator set, the weight set established on this basis can reflect the training content that is most needed to be provided to the user in the target document; and the weight set generated by combining the weights of the eloquence dimension weight matrix and the number of text fragments can provide the user with the most valuable practice content while measuring all the text fragments in the target document, which is helpful for the generation of the subsequent to-do list;

[0412] In addition, by learning the eloquence dimension weight matrix through training data sets and machine learning algorithms, more eloquence dimension indicators can be taken into account to ensure the comprehensiveness of the matrix; and the weights of document tags are calculated through the eloquence dimension weight matrix and correlation function, which can improve the accuracy and credibility of the calculation.

[0413] In one embodiment, the marking module 30 includes a score unit, a speech rate unit, an emotion unit, a speech unit, a marking unit, a feature unit, an evaluation unit, a display unit and a first verification unit; wherein the score unit, the speech rate unit, the emotion unit and the speech unit are processes for obtaining different scores, the marking unit is a process for marking text segments, the feature unit and the evaluation unit are processes for obtaining different display scores, the display unit is a process for starting the marking display, and the first verification unit is a process for effect verification;

[0414] Moreover, the score unit, the speech rate unit, the emotion unit, the speech unit and the marking unit are the complete process of marking the text segment, and the feature unit, the evaluation unit and the display unit are the complete process of displaying the marking;

[0415] Specifically:

[0416] Among them, the score unit is used to measure the target document D j The speech clarity of each text segment in T is obtained to obtain the speech clarity score; and the speech clarity score is used to measure the text segment T i the clarity of the rhetoric to identify areas for improvement;

[0417] The speech rate unit is used to obtain a speech rate control score by evaluating the speech rate of each text segment; and the speech rate control score is used to evaluate the speech rate of the text segment T i to identify whether the speaking rate needs to be changed;

[0418] The sentiment unit is used to obtain a sentiment expression score by evaluating the sentiment expression of each text segment; and the sentiment expression score is used to evaluate the text segment T i emotional expression to determine whether emotional expression needs to be enhanced;

[0419] The speech unit is used to obtain a speech diversity score by measuring the speech diversity of each text segment; and the speech diversity score is used to measure the text segment T i diversity of rhetoric to determine whether there is a need for greater diversity;

[0420] Specifically:

[0421] 1. Speech clarity scores are:

[0422]

[0423] Where N is the text segment T i The number of sentences in Conf j (T i ) is the clarity confidence score of the jth sentence.

[0424] 2. The speed control score is:

[0425]

[0426] Among them, Speed j (T i ) is the speaking speed score of the jth sentence.

[0427] 3. Emotional expression scores are:

[0428]

[0429] Among them, Sentiment j (T i) is the sentiment score of the jth sentence, which can be obtained through sentiment vocabulary analysis and sentiment intensity analysis.

[0430] 4. The verbal diversity score is:

[0431]

[0432] Among them, Vocabulary (T i ) is the text segment T i The number of different words in TotalWords(T i ) is the text segment T i The total vocabulary in .

[0433] The marked segments in the score unit, speech rate unit, emotion unit and speech unit of this embodiment take into account a variety of practice measurement criteria such as speech clarity score, speech rate control score, emotion expression score and speech diversity score, while adding personalized needs, which can truly provide users with the most suitable and most needed training benchmarks, and provide strong data support for subsequent calculations.

[0434] A marking unit, used to mark each text segment according to a speech clarity score, a speech speed control score, an emotional expression score, a speech diversity score and preset personalized needs, to obtain a marked segment;

[0435] Among them, the marking method is adaptive marking triggering, which can adaptively adjust the weight distribution and the threshold of triggering marking according to the user's eloquence level and personalized needs to ensure that the user obtains the training suggestions that best meet their needs; and the weight distribution refers to adjusting the weight of the eloquence dimension indicators according to the user's preferences, and the threshold refers to the condition for determining when to trigger the eloquence marking. The threshold can be set by the user, or the threshold of triggering marking can be adjusted by the user according to the user's eloquence performance and progress;

[0436] And, based on the user's choice, by combining the eloquence dimension indicator set, it provides users with instant editing and tagging options;

[0437] Additionally, users can choose to trigger the tagging feature manually or activate it by clicking or selecting a specific section in the document.

[0438] Feature unit, used for natural language processing, the target document D j The text fragment T i Decompose into sentences, token segments and text segments T i Perform feature extraction to obtain diverse features;

[0439] An evaluation unit is used to evaluate the eloquence dimension of the diversity features through a preset formula set to obtain a display score of speech clarity, a display score of speech speed, a display score of emotional expression, and a display score of speech diversity; and in the process of calculating each score, by combining the range mapping of the indicator and the Sigmoid function, to ensure that the label range is between 0 and 1;

[0440] Among them, the speech clarity display score is used to present the text segment T i A mark of clarity of speech, ranging from 0 to 1, with 1 indicating the highest level of need for improvement;

[0441] The speaking rate display score is used to present the text segment T i The speech rate marker ranges from 0 to 1, with 1 indicating the highest level of improvement required;

[0442] The emotional expression display score is used to present the text segment T i The emotional expression mark ranges from 0 to 1, with 1 indicating the highest degree of improvement;

[0443] The speech diversity display score is used to present the text segment T i The rhetorical diversity mark ranges from 0 to 1, with 1 indicating the highest level of need for improvement;

[0444] Specifically:

[0445] ① Speech clarity score:

[0446]

[0447] Among them, μ C Clarity(T i ) is the text segment T i An indicator of speech clarity, ranging from 0 to 1, with 1 indicating complete clarity.

[0448] ②Speech speed display score is:

[0449]

[0450] Among them, μ P is the average value of the speaking speed index; Pace(T i ) is the text segment T i The speaking speed index ranges from 0 to 1, where 1 indicates an appropriate speaking speed. In addition, an appropriate speaking speed needs to meet the following requirements: ensuring the clarity and fluency of the text, clearly presenting the context and purpose of the speech, ensuring that the audience can understand the content of the speech, and accurately emphasizing information and expressing emotions.

[0451] ③The emotional expression score is:

[0452]

[0453] Among them, μ E is the average value of the emotion expression index; Emotion(T i ) is the text segment T i The emotional expression index ranges from 0 to 1, with 1 indicating the highest emotional expression.

[0454] ④ The scores of verbal diversity are:

[0455]

[0456] Among them, μ D is the average value of the speech diversity index; Diversity(T i ) is the text segment T i An indicator of speech diversity, ranging from 0 to 1, with 1 indicating the highest diversity.

[0457] It should be noted that μ C , μ P , μ E and μ D is through the target document D j The calculation is performed on a large number of text fragments in the target document D j The overall clarity, speaking speed, emotional expression and verbal diversity of Chinese text segments.

[0458] Display unit, used to display the score ClarityDisplay (T i )、Speech speed display score PaceDisplay(T i ), Emotion Display score EmotionDisplay(T i ) and DiversityDisplay(T i ), is the target document D j Visually mark each text segment in the text set, and highlight the visually marked results and the text segment with the highest weight in the weight set by presetting specific colors, highlighting and marking icons;

[0459] For example, clarity can be marked in blue, speech rate can be marked in red, and emotional expression can be marked in yellow to help users more intuitively understand where improvements are needed.

[0460] The display unit of this embodiment can present the exercise content that needs attention to the user in the most intuitive way according to different scores through highlighting, helping the user to easily identify and distinguish the direction of improvement. This visual marking method can enable the user to escape from the boring and large amount of training data in the target document, grasp the key points of the exercise at the first time, and achieve good training results.

[0461] The first verification unit is used to verify the practicality of the eloquence dimension mark and the effect of helping users improve their eloquence by obtaining user feedback and regularly monitoring the effect of eloquence training;

[0462] Also, conduct A / B testing to compare text with Eloquent Dimensions tags with text without tags to evaluate the tagging effect and user experience.

[0463] In one embodiment, the list module 40 includes a synthesis unit, a calculation unit, a time unit, a ranking unit, an indicator unit, a score unit, a result unit, an impact value unit, an expansion unit, a processing unit, an evaluation unit, a context unit, a weighting unit, an optimization unit, a performance unit, a synthesis unit, an improvement unit, an interface unit, and a second verification unit;

[0464] Among them, the comprehensive unit is the process of obtaining the comprehensive eloquence dimension indicator set, and the calculation unit is the process of obtaining the time factor set TimeToEvent (T i ), the time unit is the process of obtaining a to-do priority set, the sorting unit is the process of generating a to-do list, and the synthesis unit, the calculation unit, the time unit and the sorting unit are the complete process of generating a to-do list;

[0465] The indicator unit is the process of defining the editing indicator set of the eloquence dimension, the score unit is the process of obtaining the current eloquence dimension score (CS), the result unit is the process of obtaining the eloquence dimension score (ES), and the influence value unit is the process of generating user editing optimization suggestions. In addition, the indicator unit, the score unit, the result unit and the influence value unit are the complete process of generating user editing optimization suggestions.

[0466] The expansion unit is used for the process of obtaining the extended set of eloquence dimension indicators, the processing unit is the process of obtaining the text analysis result, the evaluation unit is the process of obtaining the eloquence dimension evaluation result, the context unit is the process of obtaining the context analysis result, the weighting unit is the process of obtaining the comprehensive eloquence dimension score (CS), and the optimization unit is the process of generating eloquence optimization suggestions, and the expansion unit, the processing unit, the evaluation unit, the context unit, the weighting unit and the optimization unit are the complete process of generating eloquence optimization suggestions;

[0467] The performance unit is the process of defining the performance indicator set of the eloquence dimension, and the comprehensive unit is the process of obtaining the comprehensive eloquence score (Stotal ), the improvement unit is the process of generating interactive tag suggestions, the interface unit is the process of using the interactive interface to display information, the second verification unit is the process of performing effect verification, and the performance unit, the comprehensive unit, the improvement unit and the interface unit are the complete process of performing interface interaction with the user;

[0468] Specifically:

[0469] The comprehensive unit is used to perform comprehensive calculation on the text segments in the marked segments and the text segments with the highest weight in the weight set according to the eloquence dimension indicator set to obtain a comprehensive eloquence dimension indicator set; wherein the comprehensive eloquence dimension indicator P in the comprehensive eloquence dimension indicator set is i , used to represent the text segment T i The comprehensive eloquence dimension index ranges from 0 to 1, with 1 representing the best eloquence performance and 0 representing the worst;

[0470] Among them, the comprehensive eloquence dimension index P in the comprehensive eloquence dimension index set i for:

[0471] P i =α1·Clarity(T i )+α2·Pace(T i )+α3·Emotion(T i )+α4·Diversity(T i )+α5

[0472] ·Confidence(T i )

[0473] Among them, α1, α2, α3, α4 and α5 are weight coefficients used to balance the importance of different eloquence dimensions, and each coefficient corresponds to an eloquence dimension; the values ​​of these weight coefficients can be adjusted according to the relative importance of the eloquence dimensions, the user's eloquence training needs and preferences, so as to better reflect the user's eloquence training needs;

[0474] Clarity(T i ) represents the text segment T i Symbol for the Clarity Eloquence Dimension Index; The Clarity Eloquence Dimension Index is used to measure the clarity of the verbal expression in a text. The clarity of a text can be determined by vocabulary choice and sentence structure.

[0475] Pace(T i ) represents the text segment T i The symbol of the speed and eloquence dimension index; the speed and eloquence dimension index is used to measure the speed of speech expression in a text, which can be calculated based on the speed changes, pauses and stress in the text;

[0476] Emotion(T i ) represents the text segment T i The symbol of the emotional expression eloquence dimension index; the emotional expression eloquence dimension index is used to measure the degree of emotional communication in the text, which can be calculated by the emotional vocabulary, tone and emotional intensity in the text;

[0477] Diversity(T i ) represents the text segment T i The symbol of the diversity eloquence dimension index; the diversity eloquence dimension index is used to measure the degree of diversity of words and expressions in a text, which can be calculated through vocabulary diversity, changes in sentence structure, etc.

[0478] Confidence(T i ) represents the text segment T i The symbol of the confidence eloquence dimension index; the confidence eloquence dimension index is used to measure the confidence level of verbal expression in a text, which can be calculated through the confident vocabulary, tone and confidence level in the text; and the confidence level can be specifically evaluated through the stability of speech speed and tone, clarity of speech, body language and facial expressions, word choice and user feedback.

[0479] A calculation unit, used to obtain the user's speech target completion time and speech date;

[0480] The calculation unit is further used to determine the practice time that the user needs to spend on the eloquence dimension indicator set when performing eloquence training according to the speech target completion time, the speech date and the voice data, and obtain the time factor set;

[0481] Among them, the time factor set is used to measure the time that users need to spend to complete relevant exercises before the formal speech or eloquence expression event; each time factor in the time factor set is expressed as TimeToEvent (T i ), and TimeToEvent(T i ) and the text fragment T i Corresponding.

[0482] The time unit is used to combine the comprehensive eloquence dimension indicator set with the preset time factor set to generate a to-do priority set; wherein the to-do priority Priority (T i ) is the comprehensive eloquence dimension index P i The function takes into account factors such as time urgency and is used to represent the text segment T i The priority of the to-do items, ranging from 0 to 1, with 1 being the highest priority;

[0483] Among them, the priority of the to-do item in the to-do priority set is Priority (T i )for:

[0484] Priority(T i )=f(P i ,TimeToEvent(T i ))

[0485] Among them, f(P i ,TimeToEvent(T i )) is a function used to calculate the priority of a to-do item, which accepts the eloquence dimension indicator P i and each time factor in the time factor set TimeToEvent(T i ) as input, and the specific form of the function can be adjusted according to the system design and user needs. The function is used to comprehensively consider the eloquence performance and time urgency to determine the priority; TimeToEvent(T i ) represents each time factor in the time factor set.

[0486] Since the time factor set of the time unit in this embodiment is calculated based on the speech target completion time and voice data, it can objectively measure the gap between the user's current speech control time and the target time, so that the user can better consider the user's time constraints and speech dates in different dimensions of the eloquence dimension indicator set, formulate the most suitable training content for the user, and ensure that the user's training progress is always in an appropriate state.

[0487] A sorting unit, used for sorting the text segments in the marked segments and the text segments with the highest weight in the weight set according to the to-do priority set, to obtain a to-do list;

[0488] Each to-do item in the to-do list includes a related text segment T i , and suggested directions for improvement in the eloquence dimension; for example, asking users to practice improving clarity, controlling speech speed, or enhancing emotional expression;

[0489] Furthermore, the to-do list is used to present the generated to-do items to the user for viewing and management.

[0490] The to-do list in the comprehensive unit, calculation unit, time unit and sorting unit of this embodiment focuses on improving the user's shortcomings in eloquence training according to the target document; wherein, the to-do list is obtained by sorting the text fragments according to the priority of the weight, so that the user can be exposed to the content that needs to be practiced most first; and, by combining the time factor set, the practice content in the to-do list can be reasonably planned according to the user's personal preset time, so as to achieve the best training effect within a limited time.

[0491] The indicator unit is used to define the editing indicator set of the eloquence dimension; the editing indicator set of the eloquence dimension includes speech clarity, speaking speed, intonation, emotional expression, body language and word choice, specifically:

[0492] ① Speech clarity (C): used to indicate the clarity of the text, ranging from 0 to 100.

[0493] ②Speed ​​(S): used to indicate the speaking speed of the text, ranging from 0 to 100.

[0494] ③ Tone (T): used to indicate the tone of the text; the range is -100 to 100, negative numbers indicate low tones, and positive numbers indicate high tones.

[0495] ④ Emotional expression (E): used to indicate the emotional expression of the text; the range is -100 to 100, negative numbers represent negative emotions, and positive numbers represent positive emotions.

[0496] ⑤Body language (B): used to indicate the accuracy of body language and expressions in the text, ranging from 0 to 100.

[0497] ⑥Word Choice (W): used to indicate whether the word choice of the text is appropriate, ranging from 0 to 100.

[0498] A score unit, used to calculate the current eloquence dimension score (CS) of each to-do item in the to-do list according to the text content of the to-do list and the eloquence dimension editing indicator set;

[0499] Among them, the current eloquence dimension score is:

[0500] CS=(C1+S1+T1+E1+B1+W1) / 6

[0501] Among them, C1, S1, T1, E1, B1 and W1 are respectively the speech clarity index, speaking speed index, intonation index, emotional expression index, body language index and word selection index of each to-do list before editing; and the above indicators are calculated based on the indicators in the eloquence dimension editing indicator set for the text content of the to-do list.

[0502] A result unit, which is used to obtain the results of the user's edits to the to-do list, including speech clarity, speaking speed, tone, emotional expression, body language, and word choice;

[0503] According to the text content in the editing result, the eloquence dimension score (ES) of the user after editing is calculated;

[0504] Among them, the scores of eloquence dimension are:

[0505] ES=(C2+S2+T2+E2+B2+W2) / 6

[0506] Among them, C2, S2, T2, E2, B2 and W2 respectively represent the speech clarity index, speaking speed index, intonation index, emotional expression index, body language index and word choice index of each to-do item in the to-do list after editing.

[0507] The influence value unit is used to calculate the difference between the eloquence dimension score after the user edits and the eloquence dimension score before the edits to obtain the influence value (IE);

[0508] The influence value unit is also used to generate user editing optimization suggestions based on the influence value (IE) and the eloquence dimension editing indicator set; specifically: if the influence value is positive, the user is encouraged to continue similar editing; if the influence value is negative, improvement suggestions are provided, such as improving the accuracy of body language or improving vocabulary selection;

[0509] The impact value is:

[0510] IE=ES-CS

[0511] Among them, ES is the eloquence dimension score, and CS is the current eloquence dimension score.

[0512] The expansion unit is used to expand and redefine the initial eloquence dimension indicator set to obtain an extended set of eloquence dimension indicators; wherein the extended set of eloquence dimension indicators includes speech clarity, speaking speed, intonation, emotional expression, discourse logic and expression confidence, specifically:

[0513] ① Speech clarity (C): used to indicate the clarity of the text, ranging from 0 to 100.

[0514] ②Speed ​​(S): used to indicate the speaking speed of the text, ranging from 0 to 100.

[0515] ③ Tone (T): used to indicate the tone of the text; the range is -100 to 100, negative numbers indicate low tones, and positive numbers indicate high tones.

[0516] ④ Emotional expression (E): used to indicate the emotional expression of the text; the range is -100 to 100, negative numbers represent negative emotions, and positive numbers represent positive emotions.

[0517] ⑤Discourse logic (L): used to indicate the logic of the text, ranging from 0 to 100.

[0518] ⑥ Expression confidence (CF): used to indicate the expression confidence of the text, ranging from 0 to 100.

[0519] Processing unit, used for processing target document D j Perform natural language processing (NLP) analysis to identify key elements in the text, such as topics, sentiment, keywords, and logical structure, and obtain text analysis results.

[0520] The evaluation unit is used to analyze the target document D according to the text analysis result for each eloquence dimension indicator in the eloquence dimension indicator extension set. j The elements related to the indicator are used to obtain the evaluation results of the eloquence dimension;

[0521] For example, speech clarity may be related to sentence structure and vocabulary, speaking rate may be related to sentence length and pauses, intonation may be related to modal particles and pitch changes, emotional expression may be related to emotional vocabulary and emotional emotions, discourse logic may be related to sentence connection and argument expression, and expression confidence may be related to assertive vocabulary and tone.

[0522] Context unit, used to j Conduct context analysis on the content of the preceding and following texts to obtain the context analysis results;

[0523] Specifically: based on the relevance and logical relationship of text elements, context weighting is applied to each eloquence dimension indicator in the extended set of eloquence dimension indicators, and the context analysis result is obtained according to the context weighted result analysis; wherein, the weight of the context weighting is determined according to the importance of the context elements.

[0524] A weighting unit is used to obtain a comprehensive eloquence dimension score (CS) by contextual weighted scoring of the comprehensive eloquence dimension indicators;

[0525] Among them, the comprehensive eloquence dimension score is:

[0526] CS=(C*WC+S*WS+T*WT+E*WE+L*WL+CF*WCF) / (WC+WS+WT+WE

[0527] +WL+WCF)

[0528] Among them, C, S, T, E, L and CF respectively represent the speech clarity index, speaking speed index, intonation index, emotional expression index, discourse logic index and expression confidence index of the current text in the eloquence dimension evaluation results, and the above indicators correspond one-to-one to the indicators in the eloquence dimension indicator extension set; WC, WS, WT, WE, WL and WCF respectively represent the context weighting of the corresponding indicators.

[0529] The optimization unit is used to generate eloquence optimization suggestions based on the comprehensive eloquence dimension score (CS) and context analysis results; for example, if the discourse logic score is lower than the preset value, it is recommended that the user improve the argument connection method and expression logic;

[0530] Furthermore, users can view the eloquence optimization suggestions and choose whether to accept them.

[0531] The processing unit, evaluation unit, context unit, weighting unit and optimization unit of this embodiment identify the parts of the document that need special attention through context analysis, and provide interactive eloquence optimization suggestions, which help users to train their eloquence more focused and efficiently, so that users can better understand the areas that need improvement, improve the quality of expression, and provide more targeted improvement suggestions.

[0532] The performance unit is used to define the performance indicator set of the eloquence dimension; the eloquence dimension performance indicator set includes language fluency, emotional coherence, tone adjustment and topic consistency, specifically:

[0533] ① Language Fluency (F): used to measure the fluency and coherence of the text; the range is 0 to 100, where 100 means the most fluent.

[0534] ② Emotional coherence (CE): used to consider the consistency of emotional expression in the text; the range is 0 to 100.

[0535] Where 100 means the most consistent.

[0536] ③ Tone Adjustment (MT): used to detect the tone in the text; the range is -100 to 100, negative numbers represent negative tone, and positive numbers represent positive tone.

[0537] ④ Thematic consistency (CT): used to consider the consistency of each part of the text with the theme; the range is 0 to

[0538] 100, with 100 being the most consistent.

[0539] ⑤ Comprehensive unit, used to calculate the comprehensive eloquence score (S total );

[0540] Among them, the comprehensive eloquence score is:

[0541] S total =w1C+w2S+w3T+w4E+w5L+w6CF+w7F+w8CE+w9MT+w 10 CT

[0542] Among them, w1 to w 10 They are the weights of each eloquence dimension in the eloquence dimension performance index set, and can be adjusted according to user needs.

[0543] Improvement unit for comprehensive eloquence score (S total ) determines the importance level of the suggestions; texts with lower scores will receive more suggestions for improvement;

[0544] The improvement unit is also used to generate relevant interactive tag suggestions for each indicator according to the eloquence dimension performance indicator set. For example, if the emotional coherence (CE) score is low, it is recommended that the user maintain consistent emotional expression in the text; and the user can view the interactive tag suggestions and choose whether to accept them;

[0545] Improvement unit, also used to score according to the comprehensive eloquence score (S total ), optionally generating goal settings to optimize interactive tagging suggestions; for example, if a user's rating is low, setting achievable goals to help the user improve over time.

[0546] an interface unit for developing a user interactive interface using front-end development technology and back-end development technology, and integrating the suggestions with the user interactive interface to display the suggestions, goals, and the user's progress;

[0547] In addition, users can activate the tagging function and edit to-do items, as well as view tags, suggestions and their own improvement progress, and choose which suggestions to implement, set goals, and then track the corresponding improvements;

[0548] Among them, front-end development refers to the use of front-end technologies such as HTML, CSS, and JavaScript to create interactive user interfaces; back-end development refers to the use of server-side scripting languages ​​(such as Python and Node.js, etc.) to process user requests and data exchanges;

[0549] Among them, suggestions are made through a preset core algorithm and are associated with eloquence dimension indicators; for example, after the user marks a piece of text, similar practice content will be recommended to help the user concentrate on practicing specific eloquence skills. Users can view these suggestions on the interactive interface and perform corresponding exercises.

[0550] It should be noted that, in the above according to the target document D jIn the process of generating tags and suggestions for the eloquence dimension indicator set, the eloquence dimension indicator extension set, and the eloquence dimension performance indicator set, machine learning algorithms are used. Specifically, the training models in the machine learning algorithms are used to predict the eloquence dimension indicators, and these prediction models are optimized based on the existing training data. These algorithms will generate tags and suggestions based on different document types and eloquence dimension indicators. The specific technical means include:

[0551] ① Feature engineering: Convert text data into features that can be used by machine learning algorithms, such as word frequency and grammatical structure.

[0552] ② Training model: Use existing data sets to train machine learning models, such as recurrent neural networks (RNNs) or deep learning models, to learn the complex relationship between text fragments and eloquence dimensions to predict eloquence dimension indicators.

[0553] ③Model evaluation and optimization: Evaluate the performance of machine learning models through methods such as cross-validation, and adjust and optimize them according to user requirements.

[0554] The interface unit of this embodiment can activate the marking and suggestion functions at any time to obtain instant feedback on eloquence expression. This real-time interactive method helps users correct problems, improve performance, and provide a more efficient eloquence training experience.

[0555] The second verification unit is used to verify the effectiveness of the above training program through user satisfaction surveys, eloquence skill assessments, data analysis and comparative studies, specifically:

[0556] 1. User satisfaction survey: Conduct user satisfaction surveys regularly to collect user feedback and suggestions to understand user satisfaction and experience with the training program. User feedback will be used to improve the training program.

[0557] 2. Oratory skills assessment: The degree of improvement is assessed by regularly evaluating the user's speech or eloquence performance before and after using the training program. This can be achieved by comparing the user's eloquence skill score or actual performance.

[0558] 3. Data analysis: Analyze the user's markings, suggestions, and practice data in the training program to evaluate the user's learning progress and improvement; for example, you can track the number of exercises completed by the user and the improvement in the eloquence dimension.

[0559] 4. Comparative study: compare with traditional eloquence training methods to prove the originality and effectiveness of the eloquence training program in this embodiment; and the objects of comparison may include indicators such as user improvement speed, learning effect and satisfaction.

[0560] In one embodiment, the construction module 50 includes an initial unit, a weight unit, a first subunit, a second subunit, a third subunit, and a third verification unit;

[0561] Among them, the initial unit is the process of obtaining the initial weight set and the score of the eloquence dimension, the weight unit is the process of obtaining the task weight set, the first subunit, the second subunit and the third subunit are the process of generating the first task, and the third verification unit is the process of verifying the effect, specifically:

[0562] The initial unit is used to convert the voice data recorded in real time by the microphone into text form by using natural language processing and speech recognition engines, and use natural language processing technology to identify the emotional polarity and emotional intensity of the speech text in the voice data to obtain a text conversion result; wherein the text conversion result includes the emotional analysis result and the speech content; the natural language processing and speech recognition engines can select Google's Speech-to-Text or Microsoft's Azure SpeechService; the natural language processing technology can select an emotional vocabulary library and an emotional analysis algorithm, such as VADER emotional analyzer;

[0563] The initial unit is also used to determine the initial weights of each eloquence dimension in the eloquence dimension set according to the transformation result, and obtain the initial weight set; wherein the eloquence dimension set includes clarity (C), speech speed (P), posture and language (B), expression and body language (F), emotional expression (E) and grammatical accuracy (G); the weights in the initial weight set include W C , W P , W B , W F , W E and W G , and, as indicated by the letter subscripts of each weight, each weight in the initial weight set and each eloquence dimension in the eloquence dimension set are in one-to-one correspondence;

[0564] The initial unit is also used to calculate the text conversion results according to the eloquence dimension set using a preset mathematical model based on a hybrid method of acoustic analysis, text analysis and sentiment analysis to obtain the score of each eloquence dimension, including the clarity score C Score , speaking speed score P Score , Posture and Language Score B Score , Expression and body language score F Score , Emotional expression score E Score and grammatical accuracy score G Score ;

[0565] The weight unit is used to calculate the task weight set based on the initial weight set and the user's target score on the eloquence dimension set; the task weight set includes the clarity weight TaskWeightC , Speech rate weight Task WeightP , Posture and Language Weight Task WeightB , facial expression and body language weight Task WeightF , emotional expression weight Tas k WeightE and grammatical accuracy weight Task WeightG , specifically:

[0566] Task WeightC =W C ΔC;

[0567] Task WeightP =W P ΔP;

[0568] Task WeightB =W B ΔB;

[0569] Task WeightF =W F ΔF;

[0570] Task WeightE =W E ΔE;

[0571] Task WeightG =W G ΔG

[0572] And, for the parameters ΔC, ΔP, ΔB, ΔF, ΔE and ΔG:

[0573] ΔC=C Tanget ―C Score ;

[0574] ΔP=P Tanget ―P Score ;

[0575] ΔB=B Target ―B Score ;

[0576] ΔF=F Target ―F Score ;

[0577] ΔE=E Target ―E Score ;

[0578] ΔG=G Tanget ―G Score

[0579] Among them, C Tanget , P Tanget , B Target 、F Target 、ETarget and G Tanget is the user's target score in each eloquence dimension in each eloquence dimension set. These target scores can be set by the user or according to the suggestions of the eloquence training program; W C , W P , W B , W F , W E and W G are all weights in the initial weight set; parameters ΔC, ΔP, ΔB, ΔF, ΔE, and ΔG represent the score differences of each eloquence dimension in the eloquence dimension set, C Score , P Score , B Score 、F Score 、E Score and G Score They are clarity score, speaking speed score, posture and language score, expression and body language score, emotional expression score and grammatical accuracy score.

[0580] The first subunit is used to generate a number of corresponding initial tasks on different dimensions of the eloquence dimension set according to the eloquence dimension set and the voice data; specifically, using supervised learning, reinforcement learning or generative adversarial networks (GANs) and other machine learning technologies, according to the user's eloquence dimension set, text conversion results (the results include sentiment analysis results and speech content) and the user's relevant training data, to generate a number of corresponding initial tasks on different dimensions of the eloquence dimension set; and the number of initial tasks includes specific suggestions, examples and practice methods, and the user can choose whether to accept these suggestions later and then start practicing the speech or eloquence expression task;

[0581] The second subunit is used to determine the priority of each task in the initial tasks according to the weight of each task in the task weight set;

[0582] A third subunit is used to sort each task in the plurality of initial tasks according to the priority level, and form a first task from the sorted plurality of initial tasks;

[0583] In addition, detailed suggestions for each training task in the first task are provided, including specific practice content, practice time, practice frequency, etc.

[0584] The first tasks of the first subunit, the second subunit and the third subunit of this embodiment focus on improving the user's eloquence and expression. By establishing the first tasks according to the eloquence dimension set including clarity, speech speed, posture and language, expression and body language, emotional expression and grammatical accuracy, the user can be effectively improved in terms of oral expression ability, expression expression ability and body expression ability. Moreover, through document editing and marking, the first tasks generated can have strong personal characteristics according to the user's eloquence dimension indicators and the characteristics of the target document, which helps the user to focus on improving the specific problems in his eloquence expression, improve the training efficiency and strengthen the pertinence of weak points.

[0585] In addition, by sorting each of the initial tasks according to the priority level, the focus of the generated first task is clear, allowing users to access the content that needs training and improvement at the first time, which can improve training efficiency.

[0586] The third verification unit is used to verify the effect of the training results of the first task by improving the eloquence dimension indicators, user satisfaction survey, comparative research and automated evaluation, specifically:

[0587] 1. Improvement of eloquence dimension indicators: Regularly track the user's eloquence dimension indicators and compare the changes before and after using the first task to verify the improvement effect of the first task on the user's eloquence skills.

[0588] 2. User satisfaction survey: Regularly collect user feedback and suggestions to understand the user's training experience and satisfaction.

[0589] 3. Comparative study: Conduct a comparative study with users who did not use the first task, and analyze the impact of using the first task on the improvement of eloquence skills to verify the effectiveness of the first task.

[0590] 4. Automated evaluation: Automatically evaluate the user's speech or eloquence, and provide feedback and scores to the user based on eloquence dimension indicators and sentiment analysis to directly quantify the training effect.

[0591] In one embodiment, the communication module 60 includes an interaction unit, a data unit, a progress unit, a display unit, a construction unit, and a fourth verification unit;

[0592] The interactive unit is the process of obtaining real-time interactive data and converting it into text, and the data unit is the process of obtaining eloquence performance data User Performance i System Feedback iThe improvement unit is the process of obtaining the eloquence improvement index EloquenceImprovementIndex, the display unit is the process of displaying the eloquence improvement index EloquenceImprovementIndex, the construction unit is the process of generating the second task, and the fourth verification unit is the process of verifying the effect, which is specifically:

[0593] The interactive unit is used to interact with the user in real time through a preset instant messaging function, obtain real-time interactive data, and convert the real-time interactive data into text using speech recognition technology;

[0594] Among them, the instant messaging function is the instant messaging and feedback module, which integrates advanced speech recognition and sentiment analysis technologies; real-time interaction includes asking questions, speaking and requesting feedback to provide real-time interactive functions including text chat, voice calls and video calls, allowing users to ask questions, share progress or request feedback to coaches or other learners at any time.

[0595] Data unit, used to map the text of real-time interactive data according to the performance level according to the mapping dimension indicator set, and obtain the eloquence performance data User Performance i ; Among them, the mapping dimension indicator set includes speaking speed, intonation, emotional expression and clarity;

[0596] The data unit is also used to use deep learning and machine learning technologies (including acoustic models and sentiment analysis models) to analyze the user's eloquence in real time, and evaluate the eloquence performance data according to the preset score system to obtain eloquence feedback suggestion data System Feedback i ; Among them, the eloquence feedback suggestion data includes the speed of speech, the monotony of intonation and the lack of emotional expression;

[0597] Furthermore, during the analysis process, natural language processing (NLP) technology is used to understand the text of users’ real-time interactive data in order to better meet users’ needs and provide personalized suggestions, and sentiment analysis technology is used to analyze users’ emotional expressions, including identifying users’ emotional tendencies and emotional intensity, in order to provide more comprehensive feedback;

[0598] And get the eloquence feedback suggestion data System Feedback i After that, relevant eloquence expression feedback is immediately provided to the user based on the eloquence feedback suggestion data. For example, if the user speaks faster than the preset speed, it is recommended to slow down the speaking speed and provide specific examples.

[0599] Among them, the mapping dimension indicator set is established on the basis of the eloquence dimension indicator set, including:

[0600] ① Pace: It is used to reflect the speaking speed of eloquence. The scoring criteria are as follows:

[0601] 1.0: Perfect speaking speed, moderate speed, clear and easy to understand;

[0602] 0.7-0.9: Good speaking speed, slightly faster or slower than moderate, but not affecting understanding;

[0603] 0.4-0.6: medium speaking speed, the speaking speed is obviously too fast or too slow, and may need improvement;

[0604] 0.0-0.3: Poor speaking speed, extremely fast or slow, difficult to understand.

[0605] ② Tone: It is used to reflect the eloquence and the ups and downs of the tone and voice. The scoring criteria are as follows:

[0606] 1.0: Perfect intonation, with appropriate pitch changes and rich expression;

[0607] 0.7-0.9: Good intonation, moderate pitch variation, but may be slightly monotonous;

[0608] 0.4-0.6: Medium intonation, the pitch changes are not obvious enough, need to be improved;

[0609] 0.0-0.3: Poor intonation, with a single tone or large changes in pitch, which affects understanding.

[0610] ③ Emotion: used to reflect the communication and consistency of emotions in eloquence. The scoring criteria are as follows:

[0611] 1.0: Perfect emotional expression, emotional expression is natural, consistent and appropriate;

[0612] 0.7-0.9: Good emotional expression, emotional expression is consistent and appropriate, but may be slightly unnatural;

[0613] 0.4-0.6: Moderate emotional expression, emotional expression is sometimes inconsistent or unnatural, needs improvement;

[0614] 0.0-0.3: Poor emotional expression, inconsistent or inappropriate emotional expression, difficulty perceiving emotions.

[0615] ④ Clarity: It is used to reflect the clarity and comprehensibility of eloquence. The scoring criteria are as follows:

[0616] 1.0: Perfect clarity, clear and understandable expression without ambiguity;

[0617] 0.7-0.9: Good clarity, with most of the expression clear, but there may be some blurriness or lack of clarity;

[0618] 0.4-0.6: medium clarity, the expression is often vague or unclear, and needs improvement;

[0619] 0.0-0.3: Poor clarity, unclear and difficult to understand, needs significant improvement.

[0620] The scoring system includes evaluation criteria and classified feedback information, specifically:

[0621] 1. Establish evaluation criteria: Based on the range and scores of the eloquence dimension indicators, establish evaluation criteria. For example, you can define a speaking speed of 0.8 or above as good, a tone of voice of 0.6 or above as good, etc.

[0622] 2. Classify feedback information: Classify the feedback information provided by the system to ensure that the feedback information is associated with the eloquence dimension indicators. For example, feedback information includes too fast speaking speed, monotonous tone, and insufficient emotional expression.

[0623] The definition and scoring rules of the data unit mapping dimension indicator set in this embodiment map the eloquence dimension indicators to a score range of 0 to 1, so that the eloquence performance can be quantitatively evaluated. The user's specific eloquence performance will be scored according to these rules and standards and used to calculate the eloquence improvement index;

[0624] Moreover, through the score mapping method, the eloquence performance data can be used as an exact score to measure the user's eloquence performance; since the eloquence feedback suggestion data includes the speed of speech, the monotony of tone and the lack of emotional expression, the eloquence feedback suggestion data is not a single speech speed feedback data, allowing users to understand the current eloquence training effect from different angles.

[0625] The improvement unit is used to generate an eloquence improvement index EloquenceImprovementIndex according to the number of real-time interactions, eloquence performance data and eloquence feedback suggestion data; wherein the eloquence improvement index can be a positive number or a negative number, a positive number indicates that the eloquence has improved, and a negative number indicates that the eloquence has deteriorated, and is used to evaluate the user's overall progress in eloquence; and the eloquence improvement index can be accumulated with multiple interactions of the user to better understand the user's overall eloquence progress;

[0626] Among them, the eloquence improvement index is:

[0627]

[0628] Among them, α and β are weight parameters, which are used to balance the importance of user performance and system feedback, and can be adjusted according to the importance of eloquence performance and feedback to meet user requirements; User Performance i System Feedback for eloquence performance data i Suggest data for eloquence feedback;

[0629] A display unit is used to display the eloquence improvement index EloquenceImprovementIndex on the user interface of the interactive system where the eloquence training is located, so that the user can understand his or her own progress in real time;

[0630] Users can also adjust their learning strategies based on changes in the progress index, ask more questions or request more feedback to acquire better eloquence skills.

[0631] A construction unit is used to generate a second task according to the eloquence improvement index, the eloquence dimension indicator set and the real-time performance of the user; wherein the second task includes specific practice content, practice goals and schedule;

[0632] In addition, users can interact with the instant messaging function at any time to ask questions, share their progress, and request more feedback; and dynamically generate new training tasks based on user needs and performance to update the second task.

[0633] The second task of the interaction unit, data unit, progress unit, display unit and construction unit of this embodiment focuses on generating tasks for training through feedback adjustment based on the real-time interaction data of the user; wherein the second task is generated based on the real-time interaction data with the user, and the training progress of the user can be understood in a timely manner; since the eloquence improvement index can reflect the progress of the user in different eloquence dimensions, the limited training time can be used to train all eloquence shortcomings to the greatest extent, and improve the eloquence skills of the user in a targeted manner. This method of providing instant support and feedback helps to better understand eloquence problems and accelerate learning progress;

[0634] In addition, through instant messaging and feedback, users can obtain real-time feedback and understand their speech performance, including speech clarity, speaking speed, emotional expression, etc. This helps users quickly identify problems and take corrective measures to rapidly improve their eloquence skills.

[0635] The fourth verification unit is used to verify the training effect of the second task by improving the eloquence dimension index, conducting a user satisfaction survey, and comparing the user's eloquence expression effect, specifically:

[0636] 1. Improvement of eloquence dimension indicators: Continuously track the improvement of users' eloquence dimension indicators, for example, whether the speaking speed has been slowed down, the clarity has been improved, or the emotional expression has been improved, and these changes will be recorded and displayed as part of the effect verification.

[0637] 2. User satisfaction survey: Conduct user satisfaction surveys regularly to understand users’ experience and satisfaction level with the instant messaging and feedback modules, and provide suggestions for improvements.

[0638] 3. Compare the effectiveness of users’ eloquence: Through comparative analysis of the effectiveness of eloquence with the users’ partners and coaches, the uniqueness and effectiveness of users in improving their eloquence can be demonstrated.

[0639] In one embodiment, the combination module 70 is specifically:

[0640] The eloquence training method consists of a to-do list, a first task and a second task.

[0641] The eloquence training method of the present embodiment includes a to-do list, a first task and a second task. Since the traditional method lacks the ability of real-time analysis and feedback, the user needs to complete the training task under the guidance of a coach, so it is usually impossible to provide a highly personalized training plan; and due to the lack of an effective self-monitoring and improvement mechanism, the user is usually unable to understand his or her own speech performance in a timely manner, and thus cannot quickly discover problems and correct them in time; the eloquence training method of the present embodiment starts from three different eloquence training dimensions, gives full play to the subjective initiative of the user, and helps the user avoid the problem of limited training through document editing and marking, real-time communication and feedback, and the generation of personalized training tasks, which helps the user obtain sufficient and practical speaking opportunities and improves his or her eloquence skills.

[0642] It should be noted that the core innovation of this embodiment is:

[0643] 1. Eloquence dimension evaluation algorithm: By developing a series of mathematical formulas and algorithms to evaluate eloquence dimensions, such as speech clarity, speaking speed, emotional expression, speech diversity, expression skills, etc., in order to accurately evaluate eloquence issues.

[0644] 2. Personalized task generation: Intelligently generate personalized training tasks based on the user's eloquence dimension indicators, learning goals and document content to provide a highly personalized learning path.

[0645] 3. Real-time interaction and feedback: Through instant messaging integration, users can activate the marking and suggestion functions at any time to obtain instant feedback on their eloquence.

[0646] 4. Contextual analysis: Through text analysis technology, it automatically identifies the parts of the document that need special attention, which helps users better understand the content of the document and improve the quality of expression.

[0647] 5. Complex Mathematical Operations: By using complex mathematical formulas and algorithms to accurately measure the dimensions of eloquence, such as weight matrices and correlation functions.

[0648] 6. Task management and priority: By automatically generating a to-do list, it helps users manage eloquence training tasks according to eloquence dimension indicators and the priority of the learning plan, thereby providing more convenient task management and learning progress tracking.

[0649] 7. Instant messaging integration: By integrating the instant messaging function into the training method, users can interact, ask questions, and request suggestions with the preset instant messaging function, which helps to better understand the eloquence problems, learning progress, and get instant support and feedback.

[0650] These key technical points constitute the core innovation of this embodiment, which helps to improve the efficiency, personalization and interactivity of eloquence training.

[0651] In general, the embodiments of the present invention have the following beneficial effects:

[0652] In this embodiment, the to-do list focuses on improving the user's eloquence training shortcomings based on the target document. Since the weight set is obtained through correlation analysis, the relevance of different contents can be measured, reflecting the training content that is most needed to be provided to the user in the target document. In addition, the to-do list is generated based on the marked fragments and the text fragments with the highest weights in the weight set, so the most beneficial training content for the user is extracted based on the target document; the first task focuses on improving the user's eloquence expression, which can help the user to effectively improve their oral expression ability, facial expression ability and body expression ability; the second task focuses on generating tasks for training through feedback adjustment based on the user's real-time interactive data. This method of providing instant support and feedback helps to better understand eloquence problems and accelerate learning progress.

[0653] Moreover, since the eloquence training method in this embodiment eliminates the dependence on coaches, the eloquence training resources are more accessible, so that anyone can obtain high-quality eloquence training; through personalized task generation, detailed eloquence dimension indicator analysis and real-time feedback, it helps users to significantly improve their eloquence skills, and effectively improve their speech, expression of emotions, and communication of clear ideas.

[0654] Embodiment three:

[0655] An embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the eloquence training method based on document editing and communication;

[0656] Wherein, the eloquence training method based on document editing and communication, if implemented in the form of a software functional unit and used as an independent product, can be stored in a computer-readable storage medium. Based on such an understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0657] The above are preferred embodiments of the present invention. It should be noted that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for training eloquence based on document editing and communication, characterized in that: include: Obtain target documents and user's voice data; By calculating the correlation between the target document and a preset eloquence dimension indicator set, a weight set of the text segment set in the target document is obtained; Marking each text segment in the target document according to the eloquence dimension score of each text segment to obtain a marked segment; Generate a to-do list according to the eloquence dimension indicator set, the marked segment and the text segment with the highest weight in the weight set; specifically: according to the eloquence dimension indicator set, comprehensively calculate the text segment in the marked segment and the text segment with the highest weight in the weight set to obtain a comprehensive eloquence dimension indicator set; combine the comprehensive eloquence dimension indicator set with a preset time factor set to generate a to-do priority set; according to the to-do priority set, prioritize the text segment in the marked segment and the text segment with the highest weight in the weight set to obtain the to-do list; Constructing a first task based on the textual representation, emotional information and body language information of the speech data; Conducting real-time interaction with the user through a preset instant messaging function, and generating a second task according to the interaction performance and the eloquence dimension indicator set; The to-do list, the first task and the second task constitute an eloquence training method.

2. The method for training eloquence based on document editing and communication as claimed in claim 1, characterized in that: By calculating the correlation between the target document and the preset eloquence dimension indicator set, the weight set of the text segment set in the target document is obtained, specifically: Establishing an eloquence dimension weight matrix according to the target document and the eloquence dimension indicator set; Extracting relevant features of each text segment and the eloquence dimension indicator set in the target document to obtain a relevant feature set; Calculating the correlation between each text segment in the relevant feature set and the eloquence dimension indicator set to obtain a correlation function; According to the weights of the eloquence dimension weight matrix, the correlation function and the number of text segments in the target document, a weight set of the text segment set in the target document is calculated.

3. The method for training eloquence based on document editing and communication as claimed in claim 1, characterized in that: According to the eloquence dimension score of each text segment in the target document, each text segment is marked to obtain a marked segment, specifically: Obtaining a speech clarity score by measuring the speech clarity of each text segment in the target document; By evaluating the speech rate of each text segment, a speech rate control score is obtained; By evaluating the emotional expression of each text segment, an emotional expression score is obtained; By measuring the speech diversity of each of the text segments, a speech diversity score is obtained; According to the speech clarity score, the speech rate control score, the emotional expression score, the speech diversity score and preset personalized needs, each text segment is marked to obtain the marked segment.

4. The method for training eloquence based on document editing and communication as claimed in claim 1, characterized in that: Before generating a to-do list according to the eloquence dimension indicator set, the marked segment and the text segment with the highest weight in the weight set, the method further includes: Extracting features from the marker fragments to obtain diversity features; The diversity features are evaluated in eloquence dimension by using a preset formula set to obtain a speech clarity display score, a speech speed display score, an emotional expression display score, and a speech diversity display score; According to the speech clarity display score, the speaking speed display score, the emotional expression display score and the speech diversity display score, each text segment in the target document is visually marked, and the visual marking results and the text segment with the highest weight in the weight set are highlighted in a preset manner.

5. The method for training eloquence based on document editing and communication as claimed in claim 1, characterized in that: The time factor set is specifically: Obtain the user's speech target completion time; According to the completion time of the speech target and the voice data, the practice time that the user needs to spend on the eloquence dimension indicator set when performing eloquence training is determined to obtain the time factor set.

6. The method for training eloquence based on document editing and communication as claimed in claim 1, characterized in that: The first task is constructed according to the textual expression, emotional information and body information of the speech data, specifically: According to the speech data, determining the initial weight of each eloquence dimension in the eloquence dimension set to obtain an initial weight set; wherein the eloquence dimension set includes clarity, speech speed, posture and language, expression and body language, emotional expression and grammatical accuracy; Calculating a task weight set according to the initial weight set and the user's target score on the eloquence dimension set; The first task is generated according to the eloquence dimension indicator set and in a manner of determining the priority by the task weights in the task weight set.

7. The method for training eloquence based on document editing and communication as claimed in claim 6, characterized in that: According to the eloquence dimension indicator set, the first task is generated in a manner of determining the priority by the task weights in the task weight set, specifically: According to the eloquence dimension indicator set and the speech data, generating a number of corresponding initial tasks on different dimensions of the eloquence dimension set; Determining the priority of each task in the plurality of initial tasks according to the weight of each task in the task weight set; The tasks in the plurality of initial tasks are sorted according to the priority levels, and the sorted plurality of initial tasks constitute the first task.

8. The method for training eloquence based on document editing and communication as claimed in claim 1, characterized in that: Interact with the user in real time through a preset instant messaging function, and generate a second task based on the interaction performance and the eloquence dimension indicator set, specifically: Interacting with the user in real time through a preset instant messaging function to obtain real-time interaction data; According to a preset mapping dimension indicator set, the real-time interactive data is subjected to eloquence performance quantification and eloquence performance feedback evaluation to obtain eloquence performance data and eloquence feedback suggestion data; generating an eloquence improvement index according to the number of real-time interactions, the eloquence performance data and the eloquence feedback suggestion data; The second task is generated according to the eloquence improvement index and the eloquence dimension indicator set.

9. The method for training eloquence based on document editing and communication as claimed in claim 8, characterized in that: According to the preset mapping dimension indicator set, the real-time interactive data is subjected to eloquence performance quantification and eloquence performance feedback evaluation to obtain eloquence performance data and eloquence feedback suggestion data, specifically: According to the mapping dimension indicator set, the real-time interactive data is score-mapped according to the performance degree to obtain the eloquence performance data; wherein the mapping dimension indicator set includes speech speed, intonation, emotional expression and clarity; The eloquence performance data is evaluated according to a preset score system to obtain the eloquence feedback suggestion data; wherein the eloquence feedback suggestion data includes the speed of speech, the monotony of intonation and the lack of emotional expression.

Images