Data output method and system based on timbre and emotion simulation

Abstract

The invention discloses a data output method and system based on timbre and emotion simulation, and relates to the technical field of data processing, and the method comprises the steps: receiving role playing data and timbre data, carrying out the preprocessing of the role playing data and the timbre data, and obtaining a role playing data set and a timbre data set; inputting the role playing data set into a pre-established lama3 model for training, outputting to obtain a role playing model, inputting the tone data set into a pre-established GPT-SoVITS model for training, and outputting to obtain a tone model; real-time dialogue voice data of the user are obtained, noise removal processing is carried out on the real-time dialogue voice data of the user, phoneme and Chinese character conversion is carried out on the processed real-time dialogue voice data of the user based on a CTC greedy search algorithm, and a recognition text is obtained; the recognition text is input into a pre-established word segmentation model, a parameter file is obtained through a sub-word marking strategy and a byte pair coding method, the recognition text is sliced, the sliced text is input into a lama3 model, prediction is conducted in combination with the parameter file, and a prediction text is obtained through output; and inputting the predicted text into the timbre model to obtain a voice output result, and returning the voice output result to the user.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a data output method and system based on timbre and emotion simulation. Background Art

[0002] Existing emotional companionship and voice simulation technologies are mostly limited to general conversations or simple interactions, and cannot fully adapt to the needs in specific situations. Therefore, they have certain shortcomings in meeting diverse needs. Summary of the invention

[0003] In order to solve the deficiencies mentioned in the above background technology, the purpose of the present invention is to provide a data output method and system based on timbre and emotion simulation.

[0004] In a first aspect, the purpose of the present invention can be achieved by the following technical solution: a data output method based on timbre and emotion simulation, the method comprising the following steps:

[0005] Receiving role-playing data and timbre data, pre-processing the role-playing data and timbre data, and obtaining a role-playing data set and a timbre data set;

[0006] The role-playing dataset is input into the pre-established llama3 model for training, and the role-playing model is output; the timbre dataset is input into the pre-established GPT-SoVITS model for training, and the timbre model is output;

[0007] Acquire the user's real-time conversation voice data, perform noise removal on the user's real-time conversation voice data, and convert the processed user's real-time conversation voice data into phonemes and Chinese characters based on the CTC greedy search algorithm to obtain the recognized text;

[0008] Input the recognized text into the pre-established word segmentation model, obtain the parameter file through the subword tagging strategy and the byte pair encoding method, slice the recognized text, input the sliced ​​text into the llama3 model, predict it in combination with the parameter file, and output the predicted text;

[0009] The predicted text is input into the timbre model to obtain the speech output result, and the speech output result is returned to the user.

[0010] In combination with the first aspect, in some implementations of the first aspect, the method further includes: a preprocessing process of the role-playing data is as follows:

[0011] Select the personality image, character, and way of speaking that you need to simulate, and customize the AI ​​character and speech according to the template;

[0012] Perform simulation based on the user's initial settings, obtain the role-playing identity, and communicate with the customer based on the personality and conversation style.

[0013] In combination with the first aspect, in some implementations of the first aspect, the method further includes: the pre-established llama3 model is as follows:

[0014] y=σ(w2·ReLu(w1·x+b1)+b2)

[0015] Where σ(z) is the activation function of the output layer, x is the input feature vector; ReLU(z) is the nonlinear activation function; w1, w2 are weight matrices; b1, b2 are bias vectors;

[0016] Character data input preprocessing:

[0017] X′=Normalize(X)→Tokenize(X′)→Embedding(x′)

[0018] Normalize(X): normalization processing, mapping feature values ​​to [0,1], Tokenize(X'): split text input into subword units; Embedding(X'): use the embedding layer to convert discrete subword units into continuous vector representations;

[0019] Training and character generation:

[0020] The model is trained through supervision of the character feature label Y to minimize the loss function:

[0021]

[0022] L is the cross entropy loss function; y i is the true label; Predict output for the model;

[0023] The final output of the model is a customized vector for the character's characteristics: R = LLM(X')

[0024] R is the character feature output vector.

[0025] In combination with the first aspect, in some implementations of the first aspect, the method further includes: a process of preprocessing the timbre data:

[0026] The timbre data is used as the material. The material is first separated from the human voice by UVR5 and dedeverberation. Then, the speech segmentation tool is used to perform preprocessing operations such as data noise reduction, screening, and slicing to obtain the data set. The pre-established GPT-SoVITS model is used to resample the data set to generate a feature file, and then training is performed.

[0027] The progress is controlled according to the obtained loss rate curve image to prevent the model from overfitting.

[0028] In combination with the first aspect, in some implementations of the first aspect, the method further includes: the pre-established GPT-SoVITS model expression is as follows:

[0029]

[0030] x is the input timbre data feature vector; w1, b1 are the weights and biases of the encoder; g(z) is the nonlinear transformation layer; h(z) is the convolution layer and self-attention mechanism; w2, b2 are the weights and biases of the decoder; z is a random noise vector, For the sound file.

[0031] In combination with the first aspect, in certain implementations of the first aspect, the method also includes: when processing the user's real-time conversation voice data, first performing a certain noise removal process, importing the trained model file into the real-time processed voice, using offline automatic speech recognition ASR, and using end-to-end speech processing technology, combined with the characteristics of convolutional neural network CNN and recurrent neural network RNN ​​for processing.

[0032] In combination with the first aspect, in some implementations of the first aspect, the method further includes: an algorithm process of the CTC greedy search algorithm is as follows:

[0033] Select the output label y with the highest probability from the prediction distribution for each time frame t :

[0034]

[0035] The labels are integrated into a whole sequence, the duplicate labels in the sequence are deleted, the CTC blank placeholders in the sequence are removed, and the sequence is converted into text.

[0036] In a second aspect, in order to achieve the above-mentioned purpose, the present invention discloses a data output system based on timbre and emotion simulation, comprising:

[0037] A data processing module, used for receiving role-playing data and timbre data, pre-processing the role-playing data and timbre data, and obtaining a role-playing data set and a timbre data set;

[0038] A model training module is used to input the role-playing data set into the pre-established llama3 model for training, and output a role-playing model, and input the timbre data set into the pre-established GPT-SoVITS model for training, and output a timbre model;

[0039] The text generation module is used to obtain the user's real-time conversation voice data, perform noise removal on the user's real-time conversation voice data, and convert the processed user's real-time conversation voice data into phonemes and Chinese characters based on the CTC greedy search algorithm to obtain the recognized text;

[0040] The text processing module is used to input the recognized text into the pre-established word segmentation model, obtain the parameter file through the subword tagging strategy and the byte pair encoding method, slice the recognized text, input the sliced ​​text into the llama3 model, perform prediction in combination with the parameter file, and output the predicted text;

[0041] The speech output module is used to input the predicted text into the timbre model, obtain the speech output result, and return the speech output result to the user.

[0042] In another aspect of the present invention, in order to achieve the above-mentioned purpose, a terminal device is disclosed, including a memory, a processor, and a computer program stored in the memory and capable of running on the processor, wherein the memory stores a computer program capable of running on the processor, and when the processor loads and executes the computer program, a data output method based on timbre and emotion simulation as described above is adopted.

[0043] In another aspect of the present invention, in order to achieve the above-mentioned purpose, a computer-readable storage medium is disclosed, in which a computer program is stored. When the computer program is loaded and executed by a processor, a data output method based on timbre and emotion simulation as described above is adopted.

[0044] Beneficial effects of the present invention:

[0045] The present invention realizes the process of outputting the total data of timbre and emotion simulation through innovative role simulation and timbre generation technology, thereby being able to meet the diverse application needs of users. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0047] Figure 1 It is a schematic flow chart of the method of the present invention;

[0048] Figure 2 It is a schematic diagram of the flow of the timbre simulation and time-space accompaniment model of the present invention;

[0049] Figure 3It is a schematic diagram of the role model pre-training process of the present invention;

[0050] Figure 4 It is a schematic diagram of the system structure of the present invention. DETAILED DESCRIPTION

[0051] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] Embodiment 1:

[0053] like Figure 1 As shown, a data output method based on timbre and emotion simulation comprises the following steps:

[0054] S101: receiving role-playing data and timbre data, and pre-processing the role-playing data and timbre data to obtain a role-playing data set and a timbre data set;

[0055] The preprocessing process of role-playing data is as follows:

[0056] Specifically, first provide customers with a personality setting template (fixed options - including common personality traits and image features, and can also add specific content), and users select the personality image, personality, and speaking style to be simulated. Users first customize the AI ​​personality and speaking style according to the template. More information can be entered to enrich the AI ​​settings.

[0057] The model will first simulate according to the user's initial settings, first understand the role it will play, and then communicate with the customer according to the personality and conversation style. The user can add corresponding events at any time to improve the model's understanding of the person who will play the role and increase the content that can be returned as output.

[0058] This step is based on the llama3 model and innovatively adopts a step-by-step approach to building roles. This step will use the five attributes in psychology to preliminarily measure the character's personality. By scoring the five attributes of emotionality, agreeableness, creativity, extroversion, and responsibility, the model is trained through initial attributes and data, and then the model's role is built by evaluating and updating the role attributes through interview questions. The model's role-playing ability is improved by adjusting and improving the model's prompt words. The model will first simulate according to the user's initial settings, first understand the identity it will play, and communicate with customers according to the personality and dialogue method. Users can add corresponding events at any time to improve the model's understanding of the person who will play the role and increase the content that can be returned and output.

[0059] The pre-built llama3 models are as follows:

[0060] 1. Model formula

[0061] y=σ(w2·ReLu(w1·x+b1)+b2)

[0062] σ(z): activation function of the output layer. The present invention uses sigmoid to normalize the output value for classification.

[0063] x: Input feature vector, representing preprocessed role-playing data (personality labels, dialogue style) or timbre data (spectral features).

[0064] ReLU(z): nonlinear activation function, used to introduce nonlinear features. In the present invention, negative values ​​are set to 0 and positive values ​​are set to z to avoid gradient vanishing and improve the fitting effect of complex emotional data.

[0065] w1,w2: weight matrix, used for linear transformation of features.

[0066] w1 is responsible for projecting the high-dimensional input x (a text file in this invention) to the hidden layer dimension and adjusting the capacity of the hidden layer by controlling its size;

[0067] w2 is used to map the features of the hidden layer to the target dimension of the output layer. In the present invention, the features are the emotion label and the character language style vector.

[0068] b1, b2: bias vectors, used to improve the expressiveness of the model and avoid over-reliance on the origin. In the present invention, the slight differences in the emotion simulation are better captured, such as different emotional expressions for the same timbre.

[0069] 2. Preprocessing of character data input

[0070] The input role-playing data X is a data vector containing multi-dimensional features, which undergoes the following steps:

[0071] X′=Normalize(X)→Tokenize(X′)→Embedding(x′)

[0072] Normalize(X): Normalization processing, mapping the eigenvalues ​​to [0,1].

[0073] Tokenize(X'): Split the text input into subword units to make it easier for the model to process context.

[0074] Embedding(X'): Use an embedding layer to convert discrete subword units into continuous vector representations.

[0075] 3. Training and character generation

[0076] The model is trained through supervision of the character feature label Y to minimize the loss function:

[0077]

[0078] L: cross entropy loss function; y i :True label; Model prediction output.

[0079] The final output of the model is a customized vector for the character's characteristics: R = LLM(X')

[0080] R: Character feature output vector, used as input for subsequent dialogue generation.

[0081] 4.Personalized customization process

[0082] The model generates a character identity through user-customized initial settings, dynamically adjusts the weight distribution based on the input, and finally generates a character model that adapts to user needs. Specifically,

[0083] Application examples:

[0084] 1. Emotional support files for bereaved families

[0085] For families who have lost a loved one, long-term psychological support is provided by building an emotional companionship file.

[0086] Case 1: After losing a loved one, users can upload past voice, photos and other materials to generate a "replacement personality for the loved one". Through voice interaction, users can recall common life fragments, gradually establish positive emotional connections and reduce the risk of depression.

[0087] Case 2: Elderly people feel lonely because their children live in different places. They can customize family-style care through archival records and regularly remind users of important family dates, shared memories, etc.

[0088] S102: inputting the role-playing data set into a pre-established llama3 model for training, and outputting a role-playing model; inputting the timbre data set into a pre-established GPT-SoVITS model for training, and outputting a timbre model;

[0089] The process of preprocessing timbre data:

[0090] The timbre data is used as the material. The material is first separated from the human voice by UVR5 and dedeverberation. Then, the speech segmentation tool is used to perform preprocessing operations such as data noise reduction, screening, and slicing to obtain the data set. The pre-established GPT-SoVITS model is used to resample the data set to generate a feature file, and then training is performed.

[0091] The progress is controlled according to the obtained loss rate curve image to prevent the model from overfitting.

[0092] The pre-built GPT-SoVITS model expression is as follows:

[0093] 1. Tone data preprocessing

[0094] Use the UVR5 separation model and speech segmentation tool for combined processing to achieve noise reduction, segmentation, and resampling.

[0095] 1. Model formula

[0096]

[0097] x: Input timbre data feature vector, representing the audio segment after preprocessing (such as noise reduction, slicing, etc.).

[0098] w1, b1: Encoder weight and bias, used to extract the characteristic representation of timbre.

[0099] g(z): nonlinear transformation layer (the present invention adopts LeakyReLU to avoid negative values ​​and convergence performance), used to map timbre features to latent space.

[0100] h(z): Convolutional layer (RNN+BiLSTM) with self-attention mechanism. The convolutional layer is used to model the temporal relationship between audio clips and capture short-term audio characteristics. The self-attention mechanism is used to capture long-distance dependencies.

[0101] w2, b2: Decoder weights and biases, which convert latent space features into feature representations of the target audio segment.

[0102] z: The present invention adds a random noise vector to enhance the diversity of generated timbre.

[0103] Sound file.

[0104] S103: obtaining user real-time conversation voice data, performing noise removal processing on the user real-time conversation voice data, and converting the processed user real-time conversation voice data into phonemes and Chinese characters based on a CTC greedy search algorithm to obtain a recognized text;

[0105] When processing user real-time conversation voice data, a certain amount of noise removal processing is first performed, and the real-time processed voice is imported into the trained model file, offline automatic speech recognition ASR is used, and end-to-end voice processing technology is used, combined with the characteristics of convolutional neural network CNN and recurrent neural network RNN ​​for processing.

[0106] The algorithm process of the CTC greedy search algorithm is as follows:

[0107] 1. Select the output label yt with the highest probability from the predicted distribution at each time frame.

[0108]

[0109] Combine tags into a sequence.

[0110] 2. Delete the duplicate tags in the sequence and remove the CTC blank placeholder in the sequence.

[0111] 3. Convert the sequence to text.

[0112] S104: input the recognized text into a pre-established word segmentation model, obtain a parameter file through a subword tagging strategy and a byte pair encoding method, slice the recognized text, input the sliced ​​text into the llama3 model, perform prediction in combination with the parameter file, and output the predicted text;

[0113] S105: Input the predicted text into the timbre model to obtain a speech output result, and return the speech output result to the user.

[0114] The client is divided into software, API, and hardware forms, and is the medium for users to interact with this method.

[0115] Specifically, the present invention is further described below by way of embodiments:

[0116] This method can be applied to:

[0117] 1. Emotional Marketing in Customer Service

[0118] In the service industry, emotion simulation technology can be used to improve service quality and provide customers with a highly personalized interactive experience. This method can be applied to virtual customer service to imitate the brand image or the style familiar to users for conversation.

[0119] 2. Educational assistance through knowledge interaction and personalized companionship

[0120] This method can simulate familiar role images (such as teachers and study partners) to provide customized learning companion services for students. Users input students' subject needs and learning goals, and set students' learning styles (visual / auditory, etc.) as role data input. This method is used to generate emotional voice interaction content, including answering questions, simulating classroom teaching, or providing emotional encouragement. The model dynamically adjusts the teaching progress and feedback tone by identifying the problems or emotional states in the students' voice input, forming a learning companion file to record the students' growth trajectory and learning status.

[0121] Embodiment 2: In the second aspect, as Figure 4 As shown, in order to achieve the above-mentioned purpose, the present invention discloses a data output system based on timbre and emotion simulation, comprising:

[0122] The data processing module 11 is used to receive the role-playing data and the timbre data, and pre-process the role-playing data and the timbre data to obtain the role-playing data set and the timbre data set;

[0123] A model training module 12 is used to input the role-playing data set into a pre-established llama3 model for training, and output a role-playing model, and input the timbre data set into a pre-established GPT-SoVITS model for training, and output a timbre model;

[0124] The text generation module 13 is used to obtain the user's real-time conversation voice data, perform noise removal processing on the user's real-time conversation voice data, and convert the processed user's real-time conversation voice data into phonemes and Chinese characters based on the CTC greedy search algorithm to obtain the recognized text;

[0125] The text processing module 14 is used to input the recognized text into a pre-established word segmentation model, obtain a parameter file through a subword tagging strategy and a byte pair encoding method, slice the recognized text, input the sliced ​​text into the llama3 model, perform prediction in combination with the parameter file, and output the predicted text;

[0126] The speech output module 15 is used to input the predicted text into the timbre model, obtain a speech output result, and return the speech output result to the user.

[0127] Based on the same inventive concept, the present invention also provides a computer device, which includes: one or more processors, and a memory for storing one or more computer programs; the program includes program instructions, and the processor is used to execute the program instructions stored in the memory. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is used to implement one or more instructions, specifically for loading and executing one or more instructions in a computer storage medium to implement the above method.

[0128] It needs to be further explained that, based on the same inventive concept, the present invention also provides a computer storage medium, on which a computer program is stored, and the computer program is executed by a processor to execute the above method. The storage medium can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electrical, magnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium can be any tangible medium containing or storing a program, which can be used by an instruction execution system, device or device or used in combination with it.

[0129] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present disclosure. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0130] The above shows and describes the basic principles, main features and advantages of the present disclosure. Those skilled in the art should understand that the present disclosure is not limited by the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present disclosure. Without departing from the spirit and scope of the present disclosure, the present disclosure may have various changes and improvements, and these changes and improvements fall within the scope of the present disclosure to be protected.

Claims

1. A data output method based on timbre and emotion simulation, characterized in that: The method comprises the following steps: Receiving role-playing data and timbre data, pre-processing the role-playing data and timbre data, and obtaining a role-playing data set and a timbre data set; The role-playing dataset is input into the pre-established llama3 model for training and fine-tuning, and the role-playing model is output; the timbre dataset is input into the pre-established GPT-SoVITS model for training, and the timbre model is output; Acquire the user's real-time conversation voice data, process the user's real-time conversation voice data, convert the processed user's real-time conversation voice data into phonemes and Chinese characters based on the CTC greedy search algorithm, and obtain the recognized text; Input the recognized text into the pre-established word segmentation model, obtain the parameter file through the subword tagging strategy and the byte pair encoding method, slice the recognized text, input the sliced ​​text into the llama3 model, predict it in combination with the parameter file, and output the predicted text; The predicted text is input into the timbre model to obtain the speech output result, and the speech output result is returned to the user.

2. The data output method based on timbre and emotion simulation according to claim 1, characterized in that: The preprocessing process of the role-playing data is as follows: Select the personality image, character, and way of speaking that you want to simulate, and customize the AI's character and speech according to the template; Perform simulation based on the user's initial settings, obtain the role-playing identity, and communicate with the customer based on the personality and conversation style.

3. The data output method based on timbre and emotion simulation according to claim 1, characterized in that: The pre-built llama3 model is as follows: y=σ(w2·PeLu(w1·x+b1)+b2) Where σ(z) is the activation function of the output layer, x is the input feature vector; ReLU(z) is the nonlinear activation function; w1, w2 are weight matrices; b1, b2 are bias vectors; Character data input preprocessing: X ′ =Normalize(X)→Tokenize(X ′ )→Embedding(x ′ ) Normalize(X): normalization processing, mapping feature values ​​to [0,1], Tokenize(X'): split text input into subword units; Embedding(X'): use the embedding layer to convert discrete subword units into continuous vector representations; Training and character generation: The model is trained through supervision of the character feature label Y to minimize the loss function: L is the cross entropy loss function; y i is the true label; Predict output for the model; The final output of the model is a customized vector for the character's characteristics: R = LLM(X') R is the character feature output vector.

4. The data output method based on timbre and emotion simulation according to claim 1, characterized in that: The process of preprocessing the timbre data: The timbre data is used as the material. The material is first separated from the human voice by UVR5 and dedeverberation. Then, the speech segmentation tool is used to perform preprocessing operations such as data noise reduction, screening, and slicing to obtain the data set. The pre-established GPT-SoVITS model is used to resample the data set to generate a feature file, and then training is performed. The progress is controlled according to the obtained loss rate curve image to prevent the model from overfitting.

5. The data output method based on timbre and emotion simulation according to claim 4, characterized in that: The pre-established GPT-SoVITS model expression is as follows: x is the input timbre data feature vector; w1, b1 are the weights and biases of the encoder; g(z) is the nonlinear transformation layer; h(z) is the convolution layer and self-attention mechanism; w2, b2 are the weights and biases of the decoder; z is a random noise vector, For the sound file.

6. The data output method based on timbre and emotion simulation according to claim 1, characterized in that: The process of processing the user's real-time conversation voice data: First, perform noise removal, import the trained model file into the real-time processed voice, use offline automatic speech recognition ASR, and use end-to-end voice processing technology, combined with the characteristics of convolutional neural network CNN and recurrent neural network RNN ​​for processing.

7. The data output method based on timbre and emotion simulation according to claim 1, characterized in that: The algorithm process of the CTC greedy search algorithm is as follows: Select the output label y with the highest probability from the prediction distribution for each time frame t : The labels are integrated into a whole sequence, the duplicate labels in the sequence are deleted, the CTC blank placeholders in the sequence are removed, and the sequence is converted into text.

8. A data output system based on timbre and emotion simulation, characterized in that: include: A data processing module, used for receiving role playing data and timbre data, pre-processing the role playing data and timbre data, and obtaining a role playing data set and a timbre data set; A model training module is used to input the role-playing data set into the pre-established llama3 model for training, and output a role-playing model, and input the timbre data set into the pre-established GPT-SoVITS model for training, and output a timbre model; The text generation module is used to obtain the user's real-time conversation voice data, perform noise removal on the user's real-time conversation voice data, and convert the processed user's real-time conversation voice data into phonemes and Chinese characters based on the CTC greedy search algorithm to obtain the recognized text; The text processing module is used to input the recognized text into the pre-established word segmentation model, obtain the parameter file through the subword tagging strategy and the byte pair encoding method, slice the recognized text, input the sliced ​​text into the llama3 model, predict it in combination with the parameter file, and output the predicted text; The speech output module is used to input the predicted text into the timbre model, obtain the speech output result, and return the speech output result to the user.

9. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that: The memory stores a computer program that can be run on the processor. When the processor loads and executes the computer program, a data output method based on timbre and emotion simulation as described in any one of claims 1 to 7 is adopted.

10. A computer-readable storage medium having a computer program stored therein, characterized in that: When the computer program is loaded and executed by the processor, a data output method based on timbre and emotion simulation as described in any one of claims 1 to 7 is adopted.

Images