Interactive human voice synthesis system based on Frust webpage end

Abstract

The invention relates to the technical field of human voice synthesis, and discloses an interactive human voice synthesis system based on a Frust webpage end, and the system comprises the steps: obtaining a user input code, generating an audio signal in real time according to the user input code, and playing the audio signal; user input is statistically analyzed according to historical data; executing the historical data prediction input strategy, and predicting codes possibly input by a user; executing an audio difference analysis and mistaken touch detection strategy, analyzing the difference between the audio signals, and starting a mistaken touch detection process; real-time feedback and personalized automatic repair suggestions are provided; executing the audio signal flow imaging strategy to generate a visual audio signal flow graph; codes and graphs are combined, and synthesis and modification of audio signals are completed through graphical interface operation; respectively generating corresponding audio signal flow diagrams and comparing the audio signal flow diagrams; after creation is completed, a final audio file is generated; according to the scheme, the audio creation efficiency is improved through the graphical interface, real-time audio synthesis and intelligent repair.

Description

Technical Field

[0001] The present invention relates to the technical field of human voice synthesis, and in particular to an interactive human voice synthesis system based on a Faust webpage. Background Art

[0002] The Faust-based interactive web-based vocal synthesis system enables users to create vocal synthesis directly on the web, eliminating the need to install complex local software environments. This lowers the barrier to entry and allows users to create anytime, anywhere. Furthermore, the powerful expressive power of the Faust language allows users to finely control every aspect of vocal synthesis through code, creating a rich and diverse range of sound effects.

[0003] However, existing code-based voice synthesis systems have several drawbacks. During the code input process, users may make "accidental touches" due to operational errors or lack of knowledge, resulting in the generated audio not matching expectations. Traditional code editing methods make it difficult to intuitively detect and correct these errors. Pure code editing lacks intuitive audio signal feedback, making it difficult for users to quickly understand the impact of the code on the sound, requiring repeated debugging to achieve the desired effect. Even experienced users may struggle to precisely adjust various audio parameters through code to achieve fine-grained sound control.

[0004] This solution aims to overcome the shortcomings of existing technologies by combining code with a graphical interface, providing real-time input suggestions and false touch detection, as well as convenient graphical editing capabilities. Through real-time feedback, automatic repair suggestions, and graphical editing, users can quickly discover and correct errors, improving creative efficiency. The final audio signal flow graph and complete code are saved to facilitate user learning and secondary creation. Summary of the Invention

[0005] The present invention provides an interactive human voice synthesis system based on the Faust webpage, which is used to promote the solution of the problems mentioned in the above background technology.

[0006] In the first aspect, the present application provides an interactive human voice synthesis system based on the Faust webpage, which adopts the following technical solutions: The interactive human voice synthesis system based on the Faust webpage includes:

[0007] Real-time audio generation module: obtains user input code, generates audio signal in real time according to the user input code and plays it;

[0008] Historical data analysis module: obtains user historical data and analyzes user input based on historical data statistics;

[0009] Input strategy prediction module: Executes historical data to predict input strategies, predicts the code that the user may enter, and provides real-time input suggestions and prompts;

[0010] Audio difference analysis and false touch detection module: Executes audio difference analysis and false touch detection strategies, analyzes the differences between audio signals, determines whether there is a false touch, and starts the false touch detection process;

[0011] Automatic repair suggestion module: prompts users of possible accidental touch issues, provides real-time feedback and personalized automatic repair suggestions;

[0012] The user accepts the automatic repair suggestion, and the system automatically repairs the audio signal that was accidentally touched based on historical data;

[0013] If the user rejects the automatic repair suggestion, they can manually adjust the parameters of the repair process at any time to modify the audio signal of the mistakenly touched part;

[0014] Audio signal flow graphing module: executes the audio signal flow graphing strategy according to the audio signal and generates a visual audio signal flow graph;

[0015] Code and graphics combined with interactive operation module: By combining code and graphics, users can complete the synthesis and modification of audio signals through graphical interface operations without having to view and modify the code directly;

[0016] According to the difference between the audio signals before and after repair, corresponding audio signal flow graphs are generated for comparison;

[0017] Audio file generation module: After the creation is completed, the final audio file is generated and can be viewed and edited at any time.

[0018] Preferably, the acquiring of user historical data and statistically analyzing user input based on the historical data includes:

[0019] Get the code input by the user, parse the parameters in the code, and generate the corresponding audio signal;

[0020] Collect the user's historical data C, recorded as C = {C1, C2, ..., C n}, where C i Represents the user's i-th creation history data, which includes historical creation records, input pitch, and rhythm;

[0021] Each historical creation record consists of a set of input feature vectors X i =(X i1 , X i2 ,…,X in ) and the corresponding target output y i Composition, where X ij represents the input feature vector of the jth creation;

[0022] Use regression analysis to model historical data C and establish a prediction model Among them, β0, β1, …, β n is the regression coefficient, and ∈ is the error term.

[0023] Preferably, the execution of the historical data prediction input strategy, predicting the code that the user may enter, and providing real-time input suggestions and prompts includes:

[0024] Analyze historical data to obtain input features and predict the next input feature;

[0025] A new set of input features X based on user input new =(X new1 , X new2 ,…,X newm ), predict the corresponding output

[0026] Using mean square error and R 2 The coefficients assess the performance of the regression model;

[0027] Mean Square Error: Among them, y i is the actual output value;

[0028] R 2 coefficient: in, is the output average value;

[0029] According to the predicted output Predict the code that the user may enter and provide real-time input suggestions and prompts;

[0030] Detect user input and adjust the prediction strategy in real time if the input does not match the prediction.

[0031] Preferably, the execution of audio difference analysis and false touch detection strategy, analyzing the differences between audio signals, determining whether a false touch occurs, and starting a false touch detection process includes:

[0032] Get the audio signal S, S(t) = S1(t), S2(t), ..., S k (t), where S(t) represents the value of the audio line at different times t, S c (t) is the currently generated audio signal;

[0033] According to the input predicted by the user's historical data, the expected audio signal is set to S e (t);

[0034] The Euclidean distance is used to calculate the difference measure D(S) between two audio signals. c (t), S e (t)), D Among them, S c，v (t) and S e，v (t) represents two audio signals S c (t) and S e (t) is the value of the vth dimension, t1 and t2 represent the time range of the audio signal;

[0035] Set the difference threshold δ, and calculate the difference metric D(S c (t), S e (t)), determine whether there is an accidental touch;

[0036] When D(S c (t), S e When (t))≤δ, no false touch occurs;

[0037] When D(S c (t), S e When (t))>δ, a false touch occurs, triggering the false touch detection process.

[0038] Preferably, the method of prompting the user that there may be an accidental touch problem and providing real-time feedback and personalized automatic repair suggestions includes:

[0039] When an accidental touch is detected, real-time feedback is provided to the user to remind the user of the accidental touch;

[0040] Calculate the error of the audio signal E(t) = S c (t)-S e (t), provides automatic repair suggestions to minimize the error E(t), G(S c )=S c -E(t);

[0041] The user chooses to accept the automatic repair suggestion, and according to the minimization of the error E(t), the current audio signal is updated to the repaired audio signal and played, that is, S f =G(S c );

[0042] If the user refuses to accept the automatic repair suggestion, he or she can manually adjust the parameters of the repair process at any time to modify the audio signal of the incorrectly touched part;

[0043] After the audio signal modification is completed, the audio signal is updated and played;

[0044] The updated audio signal is recorded in the historical data, and the user historical data is updated.

[0045] Preferably, executing the audio signal flow graphical strategy according to the audio signal to generate a visual audio signal flow graph includes:

[0046] Decomposing audio signals into different modules including pitch, volume, frequency and duration;

[0047] Each module corresponds to an audio processing module, and the audio processing module includes an oscillator, a filter, a volume control and an effector;

[0048] Set the node as a circle representing each audio processing module in the graph, which contains the module name and function description;

[0049] Use signal flow to connect nodes, and use arrows to indicate the direction of signal flow;

[0050] An audio signal flow graph is generated according to the nodes and the signal flow direction and recorded as an initial audio signal flow graph;

[0051] According to the real-time changes of the audio signal, the audio signal flow graph is automatically updated in real time and displays the changing path of the audio signal.

[0052] Preferably, the step of combining the code with the graphics to generate a graphical interface for display to the user includes:

[0053] Binding the parameters in the module to the controls in the graphical interface, including sliders, input boxes, and knobs;

[0054] The user interacts with the audio signal flow graph through a graphical interface;

[0055] According to the user's operation on the node on the graphical page, the underlying code is modified synchronously in real time, and the audio signal flow graph is updated in real time;

[0056] According to the initial audio signal flow graph, recording graph data in the initial audio signal flow graph, wherein the graph data includes modules, connection paths, and parameters;

[0057] According to the modified audio signal flow graph, recording the graphic data in the modified audio signal flow graph;

[0058] Compare the graphic data of the original audio signal flow graph and the modified audio signal flow graph, use green lines to represent the original graphic data, and red lines to represent the difference between the modified graphic data;

[0059] Provide a description of the differences according to the Differences section.

[0060] Preferably, after the creation is completed, a final audio file is generated, including:

[0061] After the creation is completed, save the final audio signal flow graph and provide a download path;

[0062] Automatically generate complete code based on the final audio signal flow graph for users to view and save;

[0063] Generate the final audio file for users to view and save;

[0064] Users can view and edit the final audio signal flow graph, complete code, and final audio file at any time.

[0065] The present invention has the following beneficial effects:

[0066] 1. The Faust web-based interactive human voice synthesis system provides a complete and convenient creative environment by integrating modules such as real-time audio generation, historical data analysis, input strategy prediction, false touch detection and repair, graphical operation and audio file generation. Users can generate audio in real time by inputting code, and use historical data analysis and input strategy prediction to obtain intelligent input suggestions, which lowers the threshold for creation. More importantly, the audio difference analysis and false touch detection modules can help users find and fix errors in a timely manner, reducing debugging time. The graphical operation interface allows users to modify audio signals in an intuitive way without directly operating the code, greatly improving the creative experience and efficiency. After the creation is completed, the system can also generate the final audio file, which is convenient for users to view and edit at any time, forming a complete creative closed loop.

[0067] 2. The interactive human voice synthesis system based on the Faust webpage can generate corresponding audio signals by obtaining the code input by the user and parsing the various parameters therein, laying the foundation for subsequent analysis and prediction. More importantly, the system can collect the user's historical creation records, including input pitch, rhythm and other information, to build a huge historical database. Using this historical data, the system can use regression analysis and other methods to establish a prediction model to predict user behavior patterns and preferences. This analysis method based on historical data can help the system provide more accurate input suggestions, thereby improving user creation efficiency and reducing trial and error costs. The introduction of the regression analysis model further improves the accuracy and reliability of the prediction, providing users with more intelligent creation assistance.

[0068] 3. The interactive human voice synthesis system based on the Faust webpage obtains input features through historical data analysis. The system can predict the next input feature and, based on a new set of input features entered by the user, predict the corresponding output, thereby providing users with real-time input suggestions and prompts, greatly simplifying the creative process. The use of mean square error and R coefficient can evaluate the effectiveness of the regression model and ensure the accuracy and reliability of the prediction. More importantly, the system can detect user input and adjust the prediction strategy in real time when the input does not match the prediction, ensuring the timeliness and effectiveness of the suggestions. This intelligent input prediction not only improves the user's creative efficiency, but also reduces the learning cost, allowing even beginners to get started quickly.

[0069] 4. The interactive human voice synthesis system based on the Faust webpage obtains the audio signal and compares it with the expected audio signal predicted by the user's historical data. The system can use the Euclidean distance to calculate the difference measure between the two audio signals, thereby quantifying the degree of difference between the audio signals. This difference measurement method can effectively identify abnormal parts in the audio signal and provide a basis for false touch detection. More importantly, the system sets a difference threshold and can automatically determine whether there is a false touch based on the calculated difference measure, and promptly start the false touch detection process. This automated false touch detection mechanism can help users detect errors in a timely manner, reduce debugging time, and improve creative efficiency.

[0070] 5. The interactive human voice synthesis system based on the Faust webpage can provide real-time feedback to users when it detects that the user has made an accidental touch, prompting the user to make an accidental touch and correct the error in time. More importantly, the system can calculate the error of the audio signal and provide automatic repair suggestions to minimize the error, helping users to quickly return to the expected creative state. If the user accepts the automatic repair suggestion, the system can automatically update the current audio signal to the repaired audio signal and play it, which is simple and efficient to operate. Even if the user refuses to accept the automatic repair suggestion, the parameters of the repair process can be manually adjusted at any time to modify the audio signal of the accidentally touched part, providing a flexible repair method. This automatic repair suggestion function not only improves the user's creative efficiency, but also reduces the learning cost, allowing users to focus more on the creation itself.

[0071] 6. This interactive vocal synthesis system based on the Faust webpage can build a clear audio signal processing framework by decomposing audio signals into different modules such as pitch, volume, frequency and duration, and corresponding to audio processing modules such as oscillators, filters, volume controls and effects. More importantly, the system can represent each audio processing module as a node in a graph and connect the nodes using signal flow to form an intuitive audio signal flow graph. By observing the audio signal flow graph, users can clearly understand the generation and processing process of audio signals, thereby better understanding the principles of audio synthesis. The automatic real-time update of the audio signal flow graph allows users to observe the changing path of the audio signal in real time and deepen their understanding of the audio signal.

[0072] 7. This interactive vocal synthesis system based on the Faust webpage enables users to interact with the audio signal flow graph through the graphical interface by binding the parameters in the module to the controls in the graphical interface, such as sliders, input boxes, and knobs. Users can modify the underlying code in real time and update the audio signal flow graph in real time by operating the nodes on the graphical page. The operation is simple and intuitive. More importantly, the system can record the graphical data in the initial audio signal flow graph and the modified audio signal flow graph, and by comparing the differences between the two, use color lines to display the differences and provide explanations of the differences. This visual difference comparison can help users quickly understand the impact of modifications on the audio signal and improve creative efficiency.

[0073] 8. This interactive vocal synthesis system based on the Faust webpage can save the final audio signal flow graph after the creation is completed and provide a download path for users to view and edit at any time. More importantly, the system can automatically generate complete code based on the final audio signal flow graph for users to view and save, making it easier for users to learn and create secondary works. The system can also generate the final audio file for users to view and save, ensuring that users can review their creative achievements at any time. Users can view and edit the final audio signal flow graph, complete code and final audio file at any time, forming a complete creative process, making it easier for users to create vocal synthesis. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 Schematic diagram of the system structure of the present invention. DETAILED DESCRIPTION

[0075] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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 making creative efforts are within the scope of protection of the present invention.

[0076] Example 1, refer to Figure 1 , based on the Faust web-based interactive voice synthesis system, including:

[0077] Real-time audio generation module: obtains user input code, generates audio signal in real time according to the user input code and plays it;

[0078] Historical data analysis module: obtains user historical data and analyzes user input based on historical data statistics;

[0079] Input strategy prediction module: Executes historical data to predict input strategies, predicts the code that the user may enter, and provides real-time input suggestions and prompts;

[0080] Audio difference analysis and false touch detection module: Executes audio difference analysis and false touch detection strategies, analyzes the differences between audio signals, determines whether there is a false touch, and starts the false touch detection process;

[0081] Automatic repair suggestion module: prompts users of possible accidental touch issues, provides real-time feedback and personalized automatic repair suggestions;

[0082] The user accepts the automatic repair suggestion, and the system automatically repairs the audio signal that was accidentally touched based on historical data;

[0083] If the user rejects the automatic repair suggestion, they can manually adjust the parameters of the repair process at any time to modify the audio signal of the mistakenly touched part;

[0084] Audio signal flow graphing module: executes the audio signal flow graphing strategy according to the audio signal and generates a visual audio signal flow graph;

[0085] Code and graphics combined with interactive operation module: By combining code and graphics, users can complete the synthesis and modification of audio signals through graphical interface operations without having to view and modify the code directly;

[0086] According to the difference between the audio signals before and after repair, corresponding audio signal flow graphs are generated for comparison;

[0087] Audio file generation module: After the creation is completed, the final audio file is generated and can be viewed and edited at any time.

[0088] By integrating core modules such as real-time audio generation, historical data analysis, input strategy prediction, false touch detection and repair, graphical manipulation, and audio file generation, the system creates a highly collaborative and user-friendly Faust web-based interactive voice synthesis environment. Users can not only generate audio in real time but also receive personalized creative suggestions through historical data analysis and intelligent prediction, significantly reducing the complexity of creation.

[0089] Obtain user historical data and analyze user input based on historical data statistics, including:

[0090] Get the code input by the user, parse the parameters in the code, and generate the corresponding audio signal;

[0091] Collect the user's historical data C, recorded as C = {C1, C2, ..., C n}, where C i Represents the user's i-th creation history data, which includes historical creation records, input pitch, and rhythm;

[0092] Each historical creation record consists of a set of input feature vectors X i =(Xi1 , X i2 ,…,X in ) and the corresponding target output y i Composition, where X ij represents the input feature vector of the jth creation;

[0093] Use regression analysis to model historical data C and establish a prediction model Among them, β0, β1, …, β n is the regression coefficient, and ∈ is the error term.

[0094] By deeply mining the codes input by the user, the system can parse various parameters and generate key audio signals, laying a solid foundation for subsequent intelligent analysis. More importantly, the system can continuously accumulate users' historical creative data, including key information such as pitch and rhythm, and build a powerful historical database. With the help of this data, the system uses advanced methods such as regression analysis to establish accurate prediction models and gain insights into users' behavioral patterns and creative preferences. This intelligent analysis based on historical data can provide users with highly personalized creative suggestions, significantly improve creative efficiency, and effectively reduce trial and error costs. The application of regression analysis models further enhances the accuracy and reliability of predictions, ensuring that users receive more intelligent creative assistance.

[0095] Execute historical data prediction input strategy to predict the code that the user may enter and provide real-time input suggestions and prompts, including:

[0096] Analyze historical data to obtain input features and predict the next input feature;

[0097] A new set of input features X based on user input new =(X new1 , X new2 ,…,X newm ), predict the corresponding output

[0098] Using mean square error and R 2 The coefficients assess the performance of the regression model;

[0099] Mean Square Error: Among them, y i is the actual output value;

[0100] R 2 coefficient: in, is the output average value;

[0101] According to the predicted output Predict the code that the user may enter and provide real-time input suggestions and prompts;

[0102] Detect user input and adjust the prediction strategy in real time if the input does not match the prediction.

[0103] By conducting in-depth analysis of historical data, the system can accurately predict input features and infer the corresponding output results based on the new features currently input by the user, thereby providing users with real-time intelligent input suggestions and prompts, greatly simplifying the creative process. Using indicators such as mean square error and R coefficient, the system can continuously evaluate the performance of the regression model to ensure the accuracy and reliability of the predictions. More importantly, the system can monitor user input in real time and dynamically adjust the prediction strategy when it finds deviations between input and prediction to ensure the timeliness and effectiveness of the suggestions. This intelligent input prediction not only significantly improves the user's creative efficiency, but also lowers the learning threshold, allowing beginners to quickly get started and experience the joy of creation.

[0104] Perform audio difference analysis and false touch detection strategies, analyze the differences between audio signals, determine whether there is a false touch, and start the false touch detection process, including:

[0105] Get the audio signal S, S(t) = S1(t), S2(t), ..., S k (t), where S(t) represents the value of the audio line at different times t, S c (t) is the currently generated audio signal;

[0106] According to the input predicted by the user's historical data, the expected audio signal is set to S e (t);

[0107] The Euclidean distance is used to calculate the difference measure D(S) between two audio signals. c (t), S e (t)), D Among them, S c，v (t) and S e，v (t) represents two audio signals S c (t) and S e (t) is the value of the vth dimension, t1 and t2 represent the time range of the audio signal;

[0108] Set the difference threshold δ, and calculate the difference metric D(S c (t), S e (t)), determine whether there is an accidental touch;

[0109] When D(S c (t), S e When (t))≤δ, no false touch occurs;

[0110] When D(S c (t), Se When (t))>δ, a false touch occurs, triggering the false touch detection process.

[0111] By acquiring the audio signal in real time and comparing it with the expected signal predicted based on the user's historical data, the system can use methods such as Euclidean distance to accurately calculate the difference between the two, thereby accurately quantifying the abnormal part of the audio signal. This difference measurement method can effectively identify potential problems in the audio signal and provide strong support for false touch detection. More importantly, the system has preset intelligent difference thresholds, which can automatically determine whether there is a false touch operation based on the calculated difference measurement and immediately initiate the corresponding detection process. This automated false touch detection mechanism can help users discover potential errors in a timely manner, reduce debugging time, and thus significantly improve creative efficiency and audio quality.

[0112] Prompt users of possible accidental touch issues, provide real-time feedback and personalized automatic repair suggestions, including:

[0113] When an accidental touch is detected, real-time feedback is provided to the user to remind the user of the accidental touch;

[0114] Calculate the error of the audio signal E(t) = S c (t)-S e (t), provides automatic repair suggestions to minimize the error E(t), G(S c )=S c -E(t);

[0115] The user chooses to accept the automatic repair suggestion, and according to the minimization of the error E(t), the current audio signal is updated to the repaired audio signal and played, that is, S f =G(S c );

[0116] If the user refuses to accept the automatic repair suggestion, he or she can manually adjust the parameters of the repair process at any time to modify the audio signal of the incorrectly touched part;

[0117] After the audio signal modification is completed, the audio signal is updated and played;

[0118] The updated audio signal is recorded in the historical data, and the user historical data is updated.

[0119] By intelligently detecting user operations, the system can provide real-time feedback to users when it detects accidental touches, prompting errors and guiding users to make timely corrections. More importantly, the system can proactively provide automatic repair suggestions based on error calculations, guiding users to minimize errors and quickly return to the ideal creative state. Users can choose to accept the automatic repair suggestions, and the system will automatically update the audio signal and play it immediately, making the operation simple and efficient. At the same time, the system also allows users to reject automatic repair suggestions, manually adjust parameters, and flexibly make precise modifications to the accidentally touched parts. This automatic repair function not only significantly improves creative efficiency, but also lowers the learning threshold, allowing users to focus on the creation itself and unleash their creativity.

[0120] According to the audio signal, the audio signal flow graphical strategy is executed to generate a visual audio signal flow graph, including:

[0121] Decomposing audio signals into different modules including pitch, volume, frequency and duration;

[0122] Each module corresponds to an audio processing module, and the audio processing module includes an oscillator, a filter, a volume control and an effector;

[0123] Set the node as a circle representing each audio processing module in the graph, which contains the module name and function description;

[0124] Use signal flow to connect nodes, and use arrows to indicate the direction of signal flow;

[0125] An audio signal flow graph is generated according to the nodes and the signal flow direction and recorded as an initial audio signal flow graph;

[0126] According to the real-time changes of the audio signal, the audio signal flow graph is automatically updated in real time and displays the changing path of the audio signal.

[0127] By decomposing complex audio signals into easy-to-understand modules such as pitch, volume, frequency, and duration, and establishing connections with corresponding audio processing modules such as oscillators, filters, volume controls, and effects, the system builds a clear and structured audio signal processing framework. More importantly, the system presents each audio processing module as a graphical node and connects them with arrows to form an intuitive audio signal flow graph. By observing the flow graph, users can clearly understand the generation and processing process of audio signals, thereby gaining a deep understanding of the principles of audio synthesis. The real-time dynamic update of the audio signal flow graph allows users to intuitively observe the changing path of the audio signal, further deepening their understanding of the audio effects.

[0128] Combine code with graphics to generate a graphical interface and display it to users, including:

[0129] Binding the parameters in the module to the controls in the graphical interface, including sliders, input boxes, and knobs;

[0130] The user interacts with the audio signal flow graph through a graphical interface;

[0131] According to the user's operation on the node on the graphical page, the underlying code is modified synchronously in real time, and the audio signal flow graph is updated in real time;

[0132] According to the initial audio signal flow graph, recording graph data in the initial audio signal flow graph, wherein the graph data includes modules, connection paths, and parameters;

[0133] According to the modified audio signal flow graph, recording the graphic data in the modified audio signal flow graph;

[0134] Compare the graphic data of the original audio signal flow graph and the modified audio signal flow graph, use green lines to represent the original graphic data, and red lines to represent the difference between the modified graphic data;

[0135] Provide a description of the differences according to the Differences section.

[0136] By binding the parameters in the module to the controls in the graphical interface, the system enables convenient interaction, allowing users to directly manipulate the audio signal flow graph through the graphical interface. Users can modify the underlying code in real time by operating the nodes on the graphical interface, and immediately observe the updates to the audio signal flow graph. The operation is simple and intuitive. More importantly, the system can record the initial and modified audio signal flow graph data, intuitively display the differences through color lines, and provide detailed explanations of the differences. This visual difference comparison can help users quickly understand the impact of modifications on the audio signal, effectively improving creative efficiency and control.

[0137] After the creation is completed, the final audio file is generated, including:

[0138] After the creation is completed, save the final audio signal flow graph and provide a download path;

[0139] Automatically generate complete code based on the final audio signal flow graph for users to view and save;

[0140] Generate the final audio file for users to view and save;

[0141] Users can view and edit the final audio signal flow graph, complete code, and final audio file at any time.

[0142] After the creation is completed, the system can automatically save the final audio signal flow graph and provide a convenient download path, allowing users to view and edit it at any time. More importantly, the system can automatically generate complete code based on the final audio signal flow graph for users to learn and save, promoting knowledge sharing and skill improvement. At the same time, the system can also automatically generate the final audio file, allowing users to listen back and share their creations at any time. Users can access and edit the final audio signal flow graph, complete code, and audio files at any time, forming a complete creative closed loop, making vocal synthesis creation easier and more efficient.

[0143] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0144] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. Based on the Faust web-based interactive voice synthesis system, it is characterized by: include, Real-time audio generation module: obtains user input code, generates audio signal in real time according to the user input code and plays it; Historical data analysis module: obtains user historical data and analyzes user input based on historical data statistics; Input strategy prediction module: Executes historical data to predict input strategies, predicts the code that the user may enter, and provides real-time input suggestions and prompts; Audio difference analysis and false touch detection module: Executes audio difference analysis and false touch detection strategies, analyzes the differences between audio signals, determines whether there is a false touch, and starts the false touch detection process; Automatic repair suggestion module: prompts users of possible accidental touch issues, provides real-time feedback and personalized automatic repair suggestions; The user accepts the automatic repair suggestion, and the system automatically repairs the audio signal that was accidentally touched based on historical data; If the user rejects the automatic repair suggestion, they can manually adjust the parameters of the repair process at any time to modify the audio signal of the mistakenly touched part; Audio signal flow graphing module: executes the audio signal flow graphing strategy according to the audio signal and generates a visual audio signal flow graph; Code and graphics combined with interactive operation module: Combine code and graphics to generate a graphical interface and display it to users; According to the difference between the audio signals before and after repair, corresponding audio signal flow graphs are generated for comparison; Audio file generation module: After the creation is completed, the final audio file is generated.

2. The interactive human voice synthesis system based on the Faust webpage according to claim 1 is characterized in that: The obtaining of user historical data and statistical analysis of user input based on the historical data includes: Get the code input by the user, parse the parameters in the code, and generate the corresponding audio signal; Collect the user's historical data C, recorded as C = {C1, C2, ..., C n }, where C i Represents the user's i-th creation history data, which includes historical creation records, input pitch, and rhythm; Each historical creation record consists of a set of input feature vectors X i =(X i1 , X i2 ,…,X in ) and the corresponding target output y i Composition, where X ij represents the input feature vector of the jth creation; Use regression analysis to model historical data C and establish a prediction model Among them, β0, β1, …, β n is the regression coefficient, and ∈ is the error term.

3. The Faust web-based interactive voice synthesis system according to claim 2 is characterized in that: The execution of historical data prediction input strategy, predicting the code that the user may enter, and providing real-time input suggestions and prompts, includes: Analyze historical data to obtain input features and predict the next input feature; A new set of input features X based on user input new =(X new1 , X new2 ,…,X newm ), predict the corresponding output Using mean square error and R 2 The coefficients assess the performance of the regression model; Mean Square Error: Among them, y i is the actual output value; R 2 coefficient: in, is the output average value; According to the predicted output Predict the code that the user may enter and provide real-time input suggestions and prompts; Detect user input and adjust the prediction strategy in real time if the input does not match the prediction.

4. The interactive human voice synthesis system based on the Faust webpage according to claim 1 is characterized in that: The execution of the audio difference analysis and false touch detection strategy, analyzing the differences between the audio signals, determining whether there is a false touch, and starting the false touch detection process includes: Get the audio signal S, S(t) = S1(t), S2(t), ..., S k (t), where S(t) represents the value of the audio line at different times t, S c (t) is the currently generated audio signal; According to the input predicted by the user's historical data, the expected audio signal is set to S e (t); The Euclidean distance is used to calculate the difference measure D(S) between two audio signals. c (t), S e (t)), Among them, S c，v (t) and S e，v (t) represents two audio signals S c (t) and S e (t) is the value in the vth dimension, where t1 and t2 represent the time range of the audio signal; Set the difference threshold δ, and calculate the difference metric D(S c (t), S e (t)), determine whether there is an accidental touch; When D(S c (t), S e When (t))≤δ, no false touch occurs; When D(S c (t), S e When (t))>δ, a false touch occurs, triggering the false touch detection process.

5. The interactive human voice synthesis system based on the Faust webpage according to claim 4 is characterized in that: The system prompts users that there may be accidental touch issues and provides real-time feedback and personalized automatic repair suggestions, including: When an accidental touch is detected, real-time feedback is provided to the user to remind the user of the accidental touch; Calculate the error of the audio signal E(t) = S c (t)-S e (t), provides automatic repair suggestions to minimize the error E(t), G(S c )=S c -E(t); The user chooses to accept the automatic repair suggestion, and according to the minimization of the error E(t), the current audio signal is updated to the repaired audio signal and played, that is, S f =G(S c ); If the user refuses to accept the automatic repair suggestion, he or she can manually adjust the parameters of the repair process at any time to modify the audio signal of the incorrectly touched part; After the audio signal modification is completed, the audio signal is updated and played; The updated audio signal is recorded in the historical data, and the user historical data is updated.

6. The interactive human voice synthesis system based on the Faust webpage according to claim 1 is characterized in that: The step of executing the audio signal flow graphical strategy according to the audio signal to generate a visual audio signal flow graph includes: Decomposing audio signals into different modules including pitch, volume, frequency and duration; Each module corresponds to an audio processing module, and the audio processing module includes an oscillator, a filter, a volume control and an effector; Set the node as a circle representing each audio processing module in the graph, which contains the module name and function description; Use signal flow to connect nodes, and use arrows to indicate the direction of signal flow; An audio signal flow graph is generated according to the nodes and the signal flow direction and recorded as an initial audio signal flow graph; According to the real-time changes of the audio signal, the audio signal flow graph is automatically updated in real time and displays the changing path of the audio signal.

7. The interactive human voice synthesis system based on the Faust webpage according to claim 6 is characterized in that: The code is combined with the graphics to generate a graphical interface for display to the user, including: Binding the parameters in the module to the controls in the graphical interface, including sliders, input boxes, and knobs; The user interacts with the audio signal flow graph through a graphical interface; According to the user's operation on the node on the graphical page, the underlying code is modified synchronously in real time, and the audio signal flow graph is updated in real time; According to the initial audio signal flow graph, recording graph data in the initial audio signal flow graph, wherein the graph data includes modules, connection paths, and parameters; According to the modified audio signal flow graph, recording the graphic data in the modified audio signal flow graph; Compare the graphic data of the original audio signal flow graph and the modified audio signal flow graph, use green lines to represent the original graphic data, and red lines to represent the difference between the modified graphic data; Provide a description of the differences according to the Differences section.

8. The interactive human voice synthesis system based on the Faust webpage according to claim 1 is characterized in that: After the creation is completed, the final audio file is generated, including: After the creation is completed, save the final audio signal flow graph and provide a download path; Automatically generate complete code based on the final audio signal flow graph for users to view and save; generate the final audio file for users to view and save; Users can view and edit the final audio signal flow graph, complete code, and final audio file at any time.

Images