English emotion intonation reading device

Abstract

The invention discloses an English emotion intonation reading device, which relates to the technical field of English emotion intonation reading and comprises a text input module, a semantic analysis module, an emotion recognition module, an intonation adjusting module, a speech synthesis module, an audio output module and a self-learning module. The text input module is used for acquiring an English text by typing, uploading a document and adopting an OCR (Optical Character Recognition) image recognition mode to obtain original text data; the semantic analysis module is used for analyzing original text data by adopting a method for analyzing a syntactic structure, keywords and context information to obtain semantic information of a text; the emotion recognition module is used for processing semantic information of the text in combination with a rule base and a deep neural network, recognizing emotion categories in the text and obtaining emotion tags; and the intonation adjusting module is used for adjusting voice expression parameters in the emotion label by adopting a strategy of controlling the speed, pitch, rhythm, pause and accent to obtain a voice regulation and control scheme.

Description

Technical Field

[0001] The present invention relates to the technical field of English emotional intonation reading, and particularly to an English emotional intonation reading device. Background Art

[0002] The English emotional intonation reading technology is an advanced technology integrating semantic understanding, emotion calculation and speech synthesis, aiming to enable the speech generated by a computer to not only accurately convey the text content, but also automatically adjust parameters such as intonation, speech rate, pitch, etc. of the speech according to the emotional color of the text, so as to achieve emotional speech output. This technology shows its importance in many fields. Especially in English speaking competitions, contestants need to precisely control intonation, rhythm and emotional expression to enhance the expressiveness and appeal of their speeches; in the actual English teaching process, it is found that students generally attach importance to fluency while neglecting intonation, that is, they overly focus on pronunciation speed and speech accuracy, while ignoring the importance of emotion conveyance and context change. This phenomenon is particularly prominent when preparing for English speaking competitions. English speaking competitions require contestants to not only have a good language foundation, but also possess excellent emotional expression abilities to impress the judges and win the competition.

[0003] However, the current text-to-speech (TTS) system can only generate speech with a neutral or single emotional color, and it is difficult to adapt to the diverse emotional needs in different scenarios. Moreover, when the traditional TTS system processes the English language, it is prone to ignoring context changes and semantic emotion recognition, resulting in the generated speech showing obvious "mechanicalness" and "flatness". At the same time, in English classroom teaching, students generally attach importance to fluency while neglecting intonation, that is, they overly focus on pronunciation speed and speech accuracy, while ignoring emotion conveyance and context change. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides an English emotional intonation reading device to solve the problem that the current text-to-speech (TTS) system can only generate speech with a neutral or single emotional color and is difficult to adapt to the diverse emotional needs in different scenarios.

[0006] To solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides an English emotional intonation reading device, which includes:

[0008] a text input module, a semantic analysis module, an emotion recognition module, an intonation adjustment module, a speech synthesis module, an audio output module and a self-learning module;

[0009] The text input module is used to obtain English text by means of typing, uploading documents, and OCR image recognition, and obtain the original text data;

[0010] The semantic analysis module is used to analyze the original text data by means of parsing syntactic structures, keywords, and context information to obtain the semantic information of the text;

[0011] The emotion recognition module is used to process the semantic information of the text by combining a rule base and a deep neural network, identify the emotion category in the text, and obtain an emotion label;

[0012] The intonation adjustment module is used to adjust the speech expression parameters in the emotion label by using strategies of controlling speech rate, pitch, rhythm, pause, and stress to obtain a speech regulation scheme;

[0013] The speech synthesis module is used to implement the speech regulation scheme by using a method of generating natural and anthropomorphic speech, and generate English speech with emotional color;

[0014] The audio output module is used to support playing English speech with emotional color through speakers, headphones, and Bluetooth;

[0015] The self-learning module is used to collect user feedback and optimize the emotion recognition accuracy and speech style matching ability between the emotion recognition module and the speech synthesis module.

[0016] As a preferred solution of the English emotion intonation reading device described in the present invention, wherein: the steps of obtaining English text by means of typing, uploading documents, and OCR image recognition to obtain the original text data are as follows:

[0017] Use a high-resolution camera to capture an image containing English text;

[0018] Convert the color image to a grayscale image to reduce computational complexity;

[0019] Convert the grayscale image into a black-and-white image by means of a global threshold method to highlight the text area;

[0020] Apply median filtering technology to remove noise points in the image, and use connected component analysis technology to separate individual characters;

[0021] Use the Tesseract OCR engine for text recognition, and after multiple iterations of optimization, obtain the original text data.

[0022] As a preferred solution of the English emotion intonation reading device described in the present invention, wherein: the steps of analyzing the original text data by means of parsing syntactic structures, keywords, and context information to obtain the semantic information of the text are as follows:

[0023] Obtain the English text data obtained through OCR processing from the text input module;

[0024] Use the dependency parsing algorithm in natural language processing technology NLP to parse the syntactic structure of the text;

[0025] The parsing process constructs a syntax tree by identifying the dependency relationships between various components in the sentence, and introduces an evaluation method to evaluate the accuracy of the syntactic structure. The expression is:

[0026]

[0027] where A is the accuracy factor, P j represents the j-th parsed relationship, R j represents the j-th reference relationship, and C represents the sentence complexity;

[0028] Apply the term frequency-inverse document frequency algorithm TF-IDF to extract keywords from the text;

[0029] Calculate the TF-IDF score of each word, and select the top N words with the highest scores as keywords;

[0030] Combine the context information, and use the recurrent neural network RNN or the Transformer architecture to capture the long-term dependencies between words in the text, so as to better understand the deep meaning of the text;

[0031] Predict the sentiment tendency and semantic role labeling SRL of each sentence through the trained model to assist subsequent sentiment recognition and intonation adjustment.

[0032] As a preferred solution of the English sentiment intonation reading device described in the present invention, wherein: the combined rule base and the deep neural network process the semantic information of the text, identify the emotion categories in the text, and obtain emotion labels. The specific steps are:

[0033] Receive semantic data containing syntactic structure, keywords and context information from the semantic analysis module;

[0034] Use the pre-constructed emotion dictionary and grammar pattern rule base to conduct a preliminary classification of the text;

[0035] The rule base includes words related to joy, sadness, surprise and anger;

[0036] For each sentence, determine the possible emotion categories by searching for and matching the keyword vocabulary and phrases in the rule base, and assign an initial weight;

[0037] Input the results of the preliminary classification and the complete semantic information into the trained deep neural network model;

[0038] Inside the deep neural network model, the accuracy of the sentiment recognition result is evaluated, and the expression is:

[0039]

[0040] where E is the sentiment recognition accuracy factor, F k represents the k-th recognized emotion category, G k represents the k-th actual emotion category, and V represents the text variability;

[0041] Combining the results of rule base matching and deep neural network processing, the final sentiment score is calculated using the weighted average method;

[0042] For each potential emotion category, according to the initial weight of rule base matching and the probability output by the deep learning model, the comprehensive score is calculated, and the expression is:

[0043] S i = αW i + βP i ;

[0044] where S i is the comprehensive score of the i-th emotion category, α is the rule base matching weight factor, β is the deep neural network prediction weight factor, W i is the initial weight of rule base matching, and P i is the probability output by the deep learning model;

[0045] According to the comprehensive score S i of each emotion category, the emotion with the highest score is selected as the sentiment label.

[0046] As a preferred solution of the English emotional intonation reading device described in the present invention, wherein: the speech expression parameters in the sentiment label are adjusted by adopting strategies of controlling speech rate, pitch, rhythm, pause and stress to obtain a speech regulation scheme, and the specific steps are:

[0047] Determine a set of basic speech parameters according to the sentiment label;

[0048] Make personalized adjustments in combination with user preferences and context information;

[0049] According to the basic speech parameters and personalized adjustment parameters, and comprehensively considering the grammatical structure, keywords and context information of the sentence, formulate a specific speech regulation scheme;

[0050] The speech regulation scheme includes speech rate, pitch, rhythm, pause position and stress strategy;

[0051] For each speech element, set a specific numerical range according to the sentiment label and the personalized adjustment result;

[0052] Transfer the finally determined voice control scheme to the speech synthesis module.

[0053] As a preferred solution of the English emotional intonation reading device described in the present invention, wherein: the method of generating natural and anthropomorphic speech is adopted to implement the voice control scheme, and English speech with emotional color is generated. The specific steps are as follows:

[0054] Receive the voice control scheme including speech rate, pitch, rhythm, pause and stress strategy from the intonation adjustment module;

[0055] Initialize the relevant parameters of the speech synthesis engine according to the received voice control scheme;

[0056] The parameters include fundamental frequency, pronunciation speed, pause time, and stress processing of specific words or phrases;

[0057] Based on the text content and context information, adjust the parameters in the speech synthesis process in real time, and introduce an evaluation formula to adjust and optimize the effect in the speech synthesis process. The expression is:

[0058]

[0059] where D a is the dynamic adjustment factor, T l represents the l-th target adjustment value, R l represents the l-th actually implemented adjustment value, and S represents the speech fluency stability;

[0060] Use the WaveNe speech synthesis technology to generate a high-quality speech waveform according to the adjusted parameters;

[0061] Perform post-processing on the generated speech to remove unnecessary noise, smooth the transition area, and enhance clarity;

[0062] Transfer the optimized speech data to the audio output module and play it through speakers, headphones or Bluetooth.

[0063] As a preferred solution of the English emotional intonation reading device described in the present invention, wherein: the steps for supporting the playback of English speech with emotional color through speakers, headphones and Bluetooth are as follows:

[0064] Convert and optimize the speech data according to different output devices;

[0065] The user selects the output device to be used through the interface, and the system automatically detects and connects the corresponding device according to the user's selection;

[0066] Configure the corresponding audio parameters according to the selected output device type;

[0067] Send the optimized voice data to the selected output device for playback;

[0068] After the playback ends, collect the user's feedback information and pass this data to the self-learning module.

[0069] As a preferred solution of the English emotional intonation reading device described in the present invention, wherein: collect the user feedback, and optimize the emotional recognition accuracy and speech style matching ability between the emotional recognition module and the speech synthesis module. The specific steps are as follows:

[0070] After each voice playback ends, collect the user's feedback information through interface prompts or in the form of a questionnaire;

[0071] The feedback information includes ratings on voice naturalness and emotional expression accuracy, as well as specific improvement suggestions;

[0072] Classify and analyze the collected user feedback to identify the key factors affecting the user experience;

[0073] When most users report that a certain type of emotion recognition is inaccurate, focus on the performance of this emotion category in the emotional recognition module;

[0074] According to the results of the user feedback analysis, make targeted adjustments to the rule base and deep neural network model in the emotional recognition module;

[0075] Combine the opinions on voice naturalness and emotional expression in the user feedback to optimize the parameters of the speech synthesis module;

[0076] Apply the adjusted emotional recognition model and the optimized speech synthesis strategy to the actual scenario and observe the changes in the user experience.

[0077] In a second aspect, the present invention provides a computer device, including a memory and a processor. The memory stores a computer program, wherein: when the computer program is executed by the processor, it implements any step of the English emotional intonation reading device described in the first aspect of the present invention.

[0078] In a third aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, wherein: when the computer program is executed by the processor, it implements any step of the English emotional intonation reading device described in the first aspect of the present invention.

[0079] The beneficial effects of the present invention are as follows: By analyzing the syntactic structure, keywords, and context information of the original text data, comprehensive text semantic information is obtained. The method can not only accurately parse the sentence structure but also capture the complex emotions and semantic roles contained in the text, providing more accurate data support for the emotion recognition module. In the educational scenario, it helps to more accurately guide students to master the correct intonation and emotional expression methods, improving the effect of language learning. By combining the rule base with the deep neural network to process the semantic information, the emotion category in the text is identified and an emotion label is obtained. The dual recognition method greatly improves the accuracy and robustness of emotion recognition. For diverse emotional expressions, especially for non-native language learners in English speeches or drama performances, it can help them convey emotions more naturally, enhancing the resonance and participation of the audience. By controlling speech rate, pitch, rhythm, pause, and stress strategies to adjust the speech expression parameters in the emotion label, a personalized speech regulation scheme is generated. The personalized adjustment not only improves the naturalness and authenticity of the speech but also makes the reading more vivid and interesting. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0081] Figure 1 It is a schematic diagram of the module connection of the English emotion intonation reading device in Embodiment 1. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0082] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will provide a detailed description of the specific embodiments of the present invention with reference to the drawings in the specification.

[0083] In the following description, many specific details are set forth to facilitate a thorough understanding of the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar generalizations without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0084] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that can be included in at least one implementation manner of the present invention. The "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment that is mutually exclusive with other embodiments.

[0085] Embodiment 1, refer to Figure 1, which is the first embodiment of the present invention. This embodiment provides an English emotional intonation reading device, including the following steps:

[0086] A text input module, a semantic analysis module, an emotion recognition module, an intonation adjustment module, a speech synthesis module, an audio output module, and a self-learning module;

[0087] The text input module is used to obtain English text by means of typing, uploading documents, and OCR image recognition, and obtain the original text data;

[0088] Furthermore, a high-resolution camera is used to capture images containing English text;

[0089] Convert the color image to a grayscale image to reduce the computational complexity;

[0090] Convert the grayscale image into a black-and-white image through the global threshold method to highlight the text area;

[0091] Apply median filtering technology to remove noise points in the image, and use connected component analysis technology to separate individual characters;

[0092] Use the Tesseract OCR engine for text recognition. After multiple iterations of optimization, the original text data is obtained;

[0093] It should be noted that by obtaining English text in multiple ways, not only the flexibility and application scope of the system are improved, but also the accuracy and integrity of the original text data are ensured. In particular, using a high-resolution camera to capture images and combining with the OCR technology of multiple iterations of optimization greatly improves the accuracy of extracting text from pictures, providing a solid foundation for subsequent processing.

[0094] The semantic analysis module is used to analyze the original text data by using methods such as parsing syntactic structures, keywords, and context information to obtain the semantic information of the text;

[0095] Furthermore, obtain the English text data obtained through OCR processing from the text input module;

[0096] Use the dependency parsing algorithm in the natural language processing technology NLP to parse the syntactic structure of the text;

[0097] In the parsing process, a syntax tree is constructed by identifying the dependency relationships between the various components in the sentence, and an evaluation method is introduced to evaluate the accuracy of the syntactic structure. The expression is:

[0098]

[0099] Among them, A is the accuracy factor, P j represents the jth parsed relationship, R jDenote the j-th reference relationship, and C represents the sentence complexity;

[0100] Apply the term frequency-inverse document frequency algorithm TF-IDF to extract keywords in the text;

[0101] Calculate the TF-IDF scores of each word, and select the top N words with the highest scores as keywords;

[0102] Combine the context information, and use the recursive neural network RNN or the Transformer architecture to capture the long-term dependencies between words in the text, so as to better understand the deep meaning of the text;

[0103] Predict the sentiment tendency of each sentence and the semantic role labeling SRL through the trained model to assist subsequent sentiment recognition and intonation adjustment;

[0104] It should be noted that during the process of parsing the syntactic structure, introducing an evaluation method to measure the accuracy of the syntactic structure helps to dynamically adjust the parsing parameters, thereby improving the quality of the parsing results. In addition, combining the TF-IDF algorithm to extract keywords and using the RNN or Transformer architecture to understand the deep meaning can more comprehensively grasp the sentiment tendency and semantic role of the text, providing more accurate data support for sentiment recognition.

[0105] The sentiment recognition module is used to process the semantic information of the text by combining the rule base and the deep neural network, identify the emotion categories in the text, and obtain the sentiment labels;

[0106] Furthermore, receive the semantic data including syntactic structure, keywords, and context information from the semantic analysis module;

[0107] Use the pre-constructed emotion dictionary and grammar pattern rule base to preliminarily classify the text;

[0108] The rule base includes words related to joy, sadness, surprise, and anger;

[0109] For each sentence, determine the possible emotion categories and assign initial weights by searching for and matching the keyword vocabulary and phrases in the rule base;

[0110] Input the results of the preliminary classification and the complete semantic information into the trained deep neural network model;

[0111] Inside the deep neural network model, evaluate the accuracy of the sentiment recognition results. The expression is:

[0112]

[0113] Among them, E is the sentiment recognition accuracy factor, F kRepresents the k-th recognized emotion category, G k Represents the k-th actual emotion category, V represents text variability;

[0114] Combining the results of rule base matching and deep neural network processing, use the weighted average method to calculate the final sentiment score;

[0115] For each potential emotion category, calculate the comprehensive score according to the initial weight of rule base matching and the probability output by the deep learning model. The expression is:

[0116] S i = αW i + βP i ;

[0117] Among them, S i is the comprehensive score of the i-th emotion category, α is the rule base matching weight factor, β is the deep neural network prediction weight factor, W i is the initial weight of rule base matching, P i is the probability output by the deep learning model;

[0118] According to the comprehensive score S i of each emotion category, select the emotion with the highest score as the sentiment label;

[0119] It should be noted that by combining the rule base and the deep neural network for emotion recognition, the predefined emotion dictionary and the results of complex model training can be fully utilized, which not only ensures the speed of preliminary classification but also improves the accuracy of the final sentiment label. The method of calculating the comprehensive score by weighted average further enhances the system's ability to understand complex emotional expressions, making emotion recognition closer to the human cognitive mode.

[0120] The intonation adjustment module is used to adjust the speech expression parameters in the sentiment label by adopting strategies of controlling speech speed, pitch, rhythm, pause and stress to obtain a speech regulation plan;

[0121] Furthermore, according to the sentiment label, determine a set of basic speech parameters;

[0122] Make personalized adjustments by combining user preferences and context information;

[0123] According to the basic speech parameters and personalized adjustment parameters, and comprehensively considering the syntactic structure, keywords and context information of the sentence, formulate a specific speech regulation plan;

[0124] The speech regulation plan includes speech speed, pitch, rhythm, pause position and stress strategy;

[0125] For each speech element, set a specific numerical range according to the sentiment label and personalized adjustment results;

[0126] Transfer the finally determined voice control scheme to the speech synthesis module;

[0127] It should be noted that customizing the voice control scheme based on emotional tags and personal preferences not only takes into account the basic voice elements but also makes personalized adjustments in combination with context information, which greatly improves the naturalness and expressiveness of the voice output. Especially for application scenarios such as education and entertainment, personalized voice control can significantly enhance the user's immersion and interactive experience.

[0128] The speech synthesis module is used to implement the voice control scheme by adopting a method of generating natural and anthropomorphic voices and generate English voices with emotional colors;

[0129] Furthermore, receive the voice control scheme including speech rate, pitch, rhythm, pause, and stress strategies from the intonation adjustment module;

[0130] Initialize the relevant parameters of the speech synthesis engine according to the received voice control scheme;

[0131] The parameters include fundamental frequency, pronunciation speed, pause time, and stress processing of specific words or phrases;

[0132] Based on the text content and context information, adjust the parameters in the speech synthesis process in real time and introduce an evaluation formula to adjust and optimize the effect in the speech synthesis process. The expression is:

[0133]

[0134] where D a is the dynamic adjustment factor, T l represents the l-th target adjustment value, R l represents the l-th actually implemented adjustment value, and S represents the speech fluency stability;

[0135] Utilize the WaveNe speech synthesis technology to generate a high-quality speech waveform according to the adjusted parameters;

[0136] Perform post-processing on the generated speech to remove unnecessary noise, smooth the transition regions, and enhance the clarity;

[0137] Transfer the optimized speech data to the audio output module and play it through a speaker, headphones, or Bluetooth;

[0138] It should be noted that during the process of generating natural and anthropomorphic voices, various parameters of speech synthesis are adjusted in real time, and the synthesis effect is optimized through dynamic adjustment factors, ensuring the quality and consistency of voice output. The application of advanced speech synthesis technologies such as WaveNet not only improves the realism and clarity of the voice, but also can flexibly adjust the pronunciation style according to different needs to meet diverse usage scenarios.

[0139] An audio output module, which is used to support the playback of emotionally rich English voices through speakers, headphones, and Bluetooth.

[0140] Furthermore, the speech data is format-converted and optimized according to different output devices.

[0141] The user selects the output device to be used through the interface, and the system automatically detects and connects the corresponding device according to the user's selection.

[0142] Configure the corresponding audio parameters according to the selected output device type.

[0143] Send the optimized speech data to the selected output device for playback.

[0144] After the playback is completed, collect the user's feedback information and transfer this data to the self-learning module.

[0145] It should be noted that format-converting and optimizing the speech data according to the characteristics of different output devices ensures high-quality playback effects on speakers, headphones, or Bluetooth devices. The user can select the most suitable output method through the interface. The user-friendly design not only improves the user experience but also reflects the high adaptability of the system to different application scenarios.

[0146] A self-learning module, which is used to collect user feedback and optimize the emotional recognition accuracy and speech style matching ability between the emotional recognition module and the speech synthesis module.

[0147] Furthermore, after each voice playback is completed, collect the user's feedback information through interface prompts or questionnaires.

[0148] The feedback information includes ratings on the naturalness of the voice, the accuracy of emotional expression, and specific improvement suggestions.

[0149] Classify and analyze the collected user feedback to identify the key factors affecting the user experience.

[0150] When most users report that a certain type of emotion recognition is inaccurate, focus on the performance of this emotion category in the emotional recognition module.

[0151] According to the results of user feedback analysis, the rule base and deep neural network model in the emotion recognition module are adjusted in a targeted manner;

[0152] Based on user feedback on speech naturalness and emotional expression, the parameters of the speech synthesis module were optimized;

[0153] Apply the adjusted emotion recognition model and optimized speech synthesis strategy to actual scenarios and observe the changes in user experience;

[0154] It should be noted that by collecting user feedback and continuously optimizing the system, a closed-loop management from user use to system improvement is achieved. The mechanism can not only quickly respond to changes in user needs, but also continuously learn and accumulate experience, gradually improving the accuracy of emotion recognition and the naturalness of speech synthesis. In the long run, this will greatly enhance the intelligence level and service quality of the system, forming a virtuous cycle of development.

[0155] This embodiment also provides a computer device, which is applicable to the case of an English emotional intonation reading device, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions to implement the English emotional intonation reading device proposed in the above embodiment.

[0156] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through Wi-Fi, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0157] This embodiment also provides a storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the English emotional intonation reading device proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, abbreviated as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, abbreviated as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, abbreviated as EPROM), programmable read-only memory (Programmable Red-Only Memory, abbreviated as PROM), read-only memory (Read-Only Memory, abbreviated as ROM), magnetic memory, flash memory, magnetic disk or optical disc.

[0158] In summary, the present invention deeply analyzes the original text data by parsing syntactic structures, keywords and context information, and obtains comprehensive text semantic information. The method can not only accurately parse sentence structures, but also capture complex emotions and semantic roles contained in the text, providing more accurate data support for the emotion recognition module. In an educational scenario, it helps to more accurately guide students to master correct intonation and emotional expression methods, improving the effect of language learning. By combining a rule base with a deep neural network to process semantic information, the emotion category in the text is recognized and an emotion label is obtained. The dual recognition method greatly improves the accuracy and robustness of emotion recognition. For diverse emotional expressions, especially for non-native language learners in English speeches or drama performances, it can help them convey emotions more naturally, enhancing the resonance and participation of the audience. By controlling speech rate, pitch, rhythm, pause and stress strategies to adjust the speech expression parameters in the emotion label, a personalized speech regulation scheme is generated. The personalized adjustment not only improves the naturalness and authenticity of the speech, but also makes the reading more vivid and interesting.

[0159] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. An English emotional intonation reading device, characterized in that: Including: A text input module, a semantic analysis module, an emotion recognition module, an intonation adjustment module, a speech synthesis module, an audio output module, and a self-learning module; The text input module is used to obtain English text by means of typing, uploading documents, and OCR image recognition, and obtain the original text data; The semantic analysis module is used to analyze the original text data by using methods of parsing syntactic structures, keywords, and context information to obtain the semantic information of the text; The emotion recognition module is used to process the semantic information of the text by combining a rule base and a deep neural network, identify the emotion category in the text, and obtain an emotion label; The intonation adjustment module is used to adjust the speech expression parameters in the emotion label by using strategies of controlling speech rate, pitch, rhythm, pause, and stress to obtain a speech regulation scheme; The speech synthesis module is used to implement the speech regulation scheme by using a method of generating natural and anthropomorphic speech and generate English speech with emotional color; The audio output module is used to support playing English speech with emotional color in ways of speaker, earphone, and Bluetooth; The self-learning module is used to collect user feedback and optimize the emotion recognition accuracy and speech style matching ability between the emotion recognition module and the speech synthesis module.

2. The English emotional intonation reading device according to claim 1, characterized in that: The steps of obtaining English text by means of typing, uploading documents, and OCR image recognition to obtain the original text data are as follows: Use a high-resolution camera to capture an image containing English text; Convert the color image to a grayscale image to reduce computational complexity; Convert the grayscale image to a black-and-white image through a global threshold method to highlight the text area; Apply median filtering technology to remove noise points in the image, and use connected component analysis technology to separate individual characters; Use the Tesseract OCR engine for text recognition, and after multiple iterations of optimization, obtain the original text data.

3. The English emotional intonation reading device according to claim 2, characterized in that: The steps of analyzing the original text data by using methods of parsing syntactic structures, keywords, and context information to obtain the semantic information of the text are as follows: Obtain the English text data processed by OCR from the text input module; Use the dependency parsing algorithm in natural language processing technology NLP to parse the syntactic structure of the text; In the parsing process, a syntax tree is constructed by identifying the dependency relationships between various components in the sentence, and an evaluation method is introduced to evaluate the accuracy of the syntactic structure. The expression is: where A is the accuracy factor, P j represents the j-th parsed relationship, R j represents the j-th reference relationship, and C represents the sentence complexity; Apply the term frequency-inverse document frequency algorithm TF-IDF to extract keywords in the text; Calculate the TF-IDF scores of each word, and select the top N words with the highest scores as keywords; Combined with context information, use a recurrent neural network RNN or a Transformer architecture to capture the long-term dependencies between words in the text, so as to better understand the deep meaning of the text; Predict the emotion tendency of each sentence and semantic role labeling SRL through a trained model to assist subsequent emotion recognition and intonation adjustment.

4. The English emotional intonation reading device according to claim 3, characterized in that: The steps of processing the semantic information of the text by combining a rule base and a deep neural network, identifying the emotion category in the text, and obtaining an emotion label are as follows: Receive semantic data containing syntactic structure, keywords, and context information from the semantic analysis module; Use a pre - built emotion lexicon and grammar pattern rule base to perform a preliminary classification of the text; The rule base includes words related to joy, sadness, surprise, and anger; For each sentence, determine the possible emotion categories by looking up and matching the keyword vocabulary and phrases in the rule base, and assign an initial weight; Input the results of the preliminary classification and the complete semantic information into a trained deep neural network model; Inside the deep neural network model, evaluate the accuracy of the emotion recognition results, with the expression: Among them, E is the emotional recognition accuracy factor, F k represents the k-th recognized emotion category, G k represents the k-th actual emotion category, and V represents the text variability; Combine the results of rule - base matching and deep neural network processing, and use the weighted average method to calculate the final emotion score; For each potential emotion category, calculate the comprehensive score according to the initial weight from rule - base matching and the probability output by the deep learning model, with the expression: S i = αW i + βP i ; Among them, S i is the comprehensive score of the i-th emotion category, α is the rule base matching weight factor, β is the deep neural network prediction weight factor, W i is the initial weight of the rule base matching, P i is the probability output by the deep learning model; Based on the comprehensive score S for each emotion category i , select the emotion with the highest score as the emotion label.

5. The English emotional intonation reading device according to claim 4, wherein: Adopt strategies for controlling speech rate, pitch, rhythm, pause, and stress to adjust the speech expression parameters in the emotion label, and obtain a speech regulation plan. The specific steps are as follows: Determine a set of basic speech parameters according to the emotion label; Make personalized adjustments in combination with user preferences and context information; According to the basic speech parameters and personalized adjustment parameters, and comprehensively considering the syntactic structure, keywords, and context information of the sentence, formulate a specific speech regulation plan; The speech regulation plan includes speech rate, pitch, rhythm, pause position, and stress strategy; For each speech element, set a specific numerical range according to the emotion label and personalized adjustment results; Transmit the finally determined speech regulation plan to the speech synthesis module.

6. The English emotional intonation reading device according to claim 5, characterized in that: Adopt a method of generating natural and anthropomorphic speech to implement the speech regulation plan and generate English speech with emotional color. The specific steps are as follows: Receive the speech regulation plan containing speech rate, pitch, rhythm, pause, and stress strategy from the intonation adjustment module; According to the received speech regulation plan, initialize the relevant parameters of the speech synthesis engine; The parameters include fundamental frequency, pronunciation speed, pause time, and stress processing of specific words or phrases; Based on the text content and context information, adjust the parameters in the speech synthesis process in real - time, and introduce an evaluation formula to adjust and optimize the effect in the speech synthesis process, with the expression: Among them, D a is the dynamic adjustment factor, T l represents the l-th target adjustment value, R l represents the l-th actually implemented adjustment value, and S represents the speech fluency stability; Use WaveNe speech synthesis technology to generate high - quality speech waveforms according to the adjusted parameters; Perform post - processing on the generated speech to remove unnecessary noise, smooth transition regions, and enhance clarity; Transmit the optimized speech data to the audio output module and play it through speakers, headphones, or Bluetooth.

7. The English emotional intonation reading device according to claim 6, characterized in that: The steps for supporting the playback of English speech with emotional color through speakers, headphones, and Bluetooth are as follows: Perform format conversion and optimization on the speech data according to different output devices; The user selects the output device to be used through the interface, and the system automatically detects and connects to the corresponding device according to the user's selection; Configure the corresponding audio parameters according to the selected output device type; Send the optimized speech data to the selected output device for playback; After the playback ends, collect the user's feedback information and transmit this data to the self - learning module.

8. The English emotional intonation reading device according to claim 6, characterized in that: Collect user feedback and optimize the emotion recognition accuracy and speech style matching ability between the emotion recognition module and the speech synthesis module. The specific steps are as follows: After each voice playback ends, collect user feedback information through interface prompts or questionnaires; The feedback information includes ratings on speech naturalness and emotion expression accuracy, as well as specific improvement suggestions; Classify and analyze the collected user feedback to identify key factors affecting the user experience; When most users report that a certain type of emotion recognition is inaccurate, focus on the performance of this emotion category in the emotion recognition module; According to the results of the user feedback analysis, make targeted adjustments to the rule base and deep neural network model in the emotion recognition module; Combine the opinions on speech naturalness and emotion expression in the user feedback to optimize the parameters of the speech synthesis module; Apply the adjusted emotion recognition model and the optimized speech synthesis strategy to the actual scenario and observe the changes in the user experience.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the English emotional intonation reading device according to any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the English emotional intonation reading device according to any one of claims 1 to 8.

Images