A method, device and medium for converting dialect speech into text
By performing feature extraction and machine learning model training on dialect speech data, the problem of inaccurate dialect speech conversion is solved, and the accurate conversion of dialect speech to text is achieved, which improves the conversion efficiency and accuracy.
Patent Information
- Application Number
- CN202410846346.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-27
- Publication Date
- 2024-12-24
- Estimated Expiration
- 2044-06-27
AI Technical Summary
The prior art is inaccurate in dialect voice conversion, resulting in customers needing to manually adjust text content, consuming time and affecting the experience.
The Mel frequency cepspectral coefficient algorithm, long-term average short-term energy ratio algorithm and speech endpoint detection algorithm are used to extract the feature of dialect speech data, and the dialect speech conversion model is generated through machine learning model training to improve the conversion accuracy.
Through feature extraction and machine learning model training, accurate conversion from dialect speech to text is achieved, improving conversion efficiency and accuracy, and saving customers time.
Smart Images
Figure CN118609570B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of audio processing technology, and more specifically, to a method, device and medium for converting dialect speech into text. Background Art
[0002] The voice platform is a platform that converts the text and voice of an article. During use, it was found that if customers use dialects to record voices, the text content is not accurate when converted, and manual text adjustment is required, which consumes time and affects the customer experience. Therefore, there is an urgent need for a method to accurately convert dialects into text data to solve the technical problem of inaccurate text conversion between different dialects. Summary of the invention
[0003] In view of the deficiencies in the prior art, the present invention provides a method, device and medium for converting dialect speech into text.
[0004] According to one aspect of the present invention, a method for converting dialect speech into text is provided, comprising:
[0005] Collect dialect speech data from different regions, where dialect speech data includes pronunciation, speaking speed and intonation of the dialect;
[0006] The Mel frequency cepstral coefficient algorithm, the long-time average short-time energy ratio algorithm and the speech endpoint detection algorithm are used to extract speech features from the dialect speech data, and obtain various feature audio data of the dialect speech data;
[0007] Train the machine learning model based on multiple feature audio data and text annotations of dialect speech data to generate a dialect speech conversion model;
[0008] Input multiple feature audio data to be converted of the dialect speech data to be converted into the dialect speech conversion model, and output the dialect conversion text of the dialect speech data to be converted, wherein:
[0009] The windowing formula of the Mel frequency cepstral coefficient algorithm is:
[0010] ;
[0011] Where a is the coefficient of the Hamming window function, n is the length of the frame, N Window length.
[0012] Optionally, the method further includes: performing noise reduction processing on the dialect speech data.
[0013] Optionally, a Mel frequency cepstral coefficient algorithm, a long-time average short-time energy ratio algorithm, and a speech endpoint detection algorithm are used to extract speech features from the dialect speech data to obtain a variety of feature audio data of the dialect speech data, including:
[0014] Adjust the number of filters and the filter frequency range of the Mel frequency cepstral coefficient algorithm to extract speech features from dialect speech data of different dialects, and obtain the Mel frequency cepstral coefficients in multiple feature audio data of dialect speech data of different dialects;
[0015] Using a long-term average short-term energy ratio algorithm to extract speech features from dialect speech data of different dialects, and obtaining long-term average short-term energy ratios in multiple feature audio data of dialect speech data of different dialects;
[0016] A speech endpoint detection algorithm is used to extract speech features from dialect speech data of different dialects, and time feature data from multiple feature audio data of dialect speech data of different dialects is obtained.
[0017] Optionally, the formula of the long-time average short-time energy ratio algorithm is:
[0018] ;
[0019] In the formula, LTSE is the long-term standard error; N represents the number of samples; E ( n ) represents the error of the nth sample, that is, the difference between the actual value and the predicted value.
[0020] Optionally, the loss function of the machine learning model is:
[0021] ;
[0022] Where L is the loss function, y is the probability distribution of the actual category label, and y' is the predicted probability distribution;
[0023] The learning function is:
[0024] ;
[0025] Where θ is the model parameter, t is the number of iterations, α=0.001 is the learning rate, and ∇θL(y, y') is the gradient of the loss function with respect to the model parameters.
[0026] According to another aspect of the present invention, there is provided a device for converting dialect speech into text, comprising:
[0027] A collection module, used to collect dialect speech data from different regions, wherein the dialect speech data includes the pronunciation, speaking speed and intonation of the dialect;
[0028] An extraction module is used to extract speech features from dialect speech data using a Mel frequency cepstral coefficient algorithm, a long-time average short-time energy ratio algorithm, and a speech endpoint detection algorithm to obtain multiple feature audio data of the dialect speech data;
[0029] A training module is used to train a machine learning model based on multiple feature audio data and text annotations of dialect speech data to generate a dialect speech conversion model;
[0030] The conversion module is used to input a plurality of feature audio data to be converted of the dialect speech data to be converted into the dialect speech conversion model, and output the dialect conversion text of the dialect speech data to be converted, wherein:
[0031] The windowing formula of the Mel frequency cepstral coefficient algorithm is:
[0032] ;
[0033] Where a is the coefficient of the Hamming window function, n is the length of the frame, N Window length.
[0034] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the storage medium stores a computer program, and the computer program is used to execute the method described in any one of the above aspects of the present invention.
[0035] According to another aspect of the present invention, an electronic device is provided, comprising: a processor; a memory for storing instructions executable by the processor; the processor is configured to read the executable instructions from the memory and execute the instructions to implement the method described in any one of the above aspects of the present invention.
[0036] Therefore, this application marks the collected dialect speech data of multiple dialects, extracts features to form a speech pool, and trains the dialect speech conversion model, thereby converting the dialect speech data to be converted, generating corresponding text, and extracting multiple feature audio data of different dialects through an improved feature extraction algorithm to improve the dialect conversion efficiency. Recognizing dialect speech in multiple regions improves the accuracy of converted text and saves time. BRIEF DESCRIPTION OF THE DRAWINGS
[0037] A more complete understanding of exemplary embodiments of the present invention may be obtained by referring to the following drawings:
[0038] Figure 1 It is a flowchart of a method for converting dialect speech into text provided by an exemplary embodiment of the present invention;
[0039] Figure 2It is a schematic diagram of the structure of a device for converting dialect speech into text provided by an exemplary embodiment of the present invention;
[0040] Figure 3 This is a structure of an electronic device provided by an exemplary embodiment of the present invention. DETAILED DESCRIPTION
[0041] Below, the exemplary embodiments according to the present invention will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments of the present invention, and it should be understood that the present invention is not limited to the exemplary embodiments described here.
[0042] It should be noted that the relative arrangement of components and steps, the numerical expressions and numerical values set forth in these embodiments do not limit the scope of the present invention unless specifically stated otherwise.
[0043] Those skilled in the art can understand that the terms "first" and "second" in the embodiments of the present invention are only used to distinguish different steps, devices or modules, etc., and neither represent any specific technical meaning nor indicate the necessary logical order between them.
[0044] It should also be understood that, in the embodiments of the present invention, “plurality” may refer to two or more than two, and “at least one” may refer to one, two or more than two.
[0045] It should also be understood that any component, data or structure mentioned in the embodiments of the present invention can generally be understood as one or more, unless explicitly limited or otherwise indicated in the context.
[0046] In addition, the term "and / or" in the present invention is only a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in the present invention generally indicates that the associated objects before and after are in an "or" relationship.
[0047] It should also be understood that the description of the various embodiments of the present invention focuses on the differences between the various embodiments, and the same or similar aspects thereof can be referenced to each other, and for the sake of brevity, they will not be described one by one.
[0048] At the same time, it should be understood that for the convenience of description, the sizes of the various parts shown in the drawings are not drawn according to the actual proportional relationship.
[0049] The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the invention, its application, or uses.
[0050] Technologies, methods, and equipment known to ordinary technicians in the relevant art may not be discussed in detail, but where appropriate, the technologies, methods, and equipment should be considered part of the specification.
[0051] It should be noted that like reference numerals and letters refer to similar items in the following figures, and therefore, once an item is defined in one figure, it need not be further discussed in subsequent figures.
[0052] Embodiments of the present invention can be applied to electronic devices such as terminal devices, computer systems, servers, etc., which can operate with many other general or special computing system environments or configurations. Examples of well-known terminal devices, computing systems, environments and / or configurations suitable for use with electronic devices such as terminal devices, computer systems, servers, etc. include, but are not limited to: personal computer systems, server computer systems, thin clients, thick clients, handheld or laptop devices, microprocessor-based systems, set-top boxes, programmable consumer electronics, network personal computers, small computer systems, large computer systems, and distributed cloud computing technology environments including any of the above systems, etc.
[0053] Electronic devices such as terminal devices, computer systems, servers, etc. can be described in the general context of computer system executable instructions (such as program modules) executed by computer systems. Generally, program modules can include routines, programs, object programs, components, logic, data structures, etc., which perform specific tasks or implement specific abstract data types. Computer systems / servers can be implemented in a distributed cloud computing environment, where tasks are performed by remote processing devices linked through a communication network. In a distributed cloud computing environment, program modules can be located on local or remote computing system storage media including storage devices.
[0054] Exemplary Methods
[0055] Figure 1 FIG. 1 is a flow chart of a method for converting dialect speech into text provided by an exemplary embodiment of the present invention. This embodiment can be applied to electronic devices, such as Figure 1 As shown, the method 100 for converting dialect speech into text includes the following steps:
[0056] Step 101, collecting dialect speech data from different regions, wherein the dialect speech data includes the pronunciation, speaking speed and intonation of the dialect;
[0057] Step 1021, using a Mel frequency cepstral coefficient algorithm, a long-time average short-time energy ratio algorithm, and a speech endpoint detection algorithm to extract speech features from the dialect speech data, and obtain multiple feature audio data of the dialect speech data;
[0058] Step 103, training a machine learning model based on the text annotations of the multiple feature audio data and the dialect speech data to generate a dialect speech conversion model;
[0059] Step 104, inputting a plurality of feature audio data to be converted of the dialect speech data to be converted into a dialect speech conversion model, and outputting a dialect conversion text of the dialect speech data to be converted, wherein:
[0060] The windowing formula of the Mel frequency cepstral coefficient algorithm is:
[0061] ;
[0062] Where a is the coefficient of the Hamming window function, n is the length of the frame, N Window length.
[0063] Specifically, the specific steps of the dialect speech text conversion method proposed in this application are as follows:
[0064] By collecting a large amount of dialect data and performing feature extraction to generate a big data pool, a speech recognition model is trained through a machine learning algorithm, and finally dialect speech recognition is performed through the model.
[0065] Solution steps:
[0066] 1. Data collection and preprocessing:
[0067] Collect a large amount of dialect speech data, including changes in pronunciation, speaking speed, intonation, etc. of different dialects.
[0068] The collected voice data is pre-processed, including noise reduction, removal of unnecessary sounds and interference, etc.
[0069] 2. Feature extraction:
[0070] Extract the characteristics of dialect speech, such as the spectral characteristics of audio signals, Mel-frequency cepstral coefficients (MFCC), etc.
[0071] These features can be extracted by using digital signal processing (DSP) techniques or other audio processing algorithms.
[0072] Among them, the number of filters in MFCC, the filter frequency range and other parameters are adjusted to adapt to the speech characteristics of different dialects. By improving the windowing formula of the Mel frequency cepstral coefficient algorithm, after the signal is divided into frames, a window function is multiplied for each frame to increase the continuity of the left and right ends of the frame, thereby reducing the leakage of the spectrum.
[0073] By combining the Mel frequency cepstral coefficients, the long-term average short-term energy ratio LTSE, time feature data and other features, such as the duration of the speech segment, the spectral characteristics of the speech segment, etc., for more accurate dialect recognition.
[0074] The long-term average short-term energy ratio LTSE The calculation formula is:
[0075] ;
[0076] In the formula, LTSE is the long-term standard error; N represents the number of samples; E ( n ) represents the error of the nth sample, that is, the difference between the actual value and the predicted value.
[0077] 3. Training of dialect speech recognition model:
[0078] Use machine learning (such as deep learning) algorithms to train a dialect speech recognition model based on the collected dialect speech data and its corresponding text annotations.
[0079] This model can be based on recurrent neural networks, convolutional neural networks, or probability-based hidden Markov models (HMM).
[0080] a Loss function: For speech recognition tasks, the commonly used loss function is the cross-entropy loss function. The cross-entropy loss function can measure the difference between the predicted probability distribution and the probability distribution of the actual category label.
[0081] The formula is as follows: Where L is the loss function, y is the probability distribution of the actual class label, and y' is the predicted probability distribution.
[0082] bLearning function: The learning function is used to update the model parameters. The commonly used learning function is the gradient descent method. The gradient descent method updates the model parameters by calculating the gradient of the loss function to minimize the loss function.
[0083] The formula is as follows: Where θ is the model parameter, t is the number of iterations, α is the learning rate, and ∇θL(y, y') is the gradient of the loss function with respect to the model parameter.
[0084] c Parameter setting: The learning rate (α) is an important parameter in the gradient descent method, which is used to control the step size of each iteration. The choice of learning rate has an important impact on the convergence speed and stability of the model and needs to be adjusted according to the specific situation. Usually, the learning rate can be set to a smaller value, such as 0.01, 0.001, etc.
[0085] 4. Voice Recognition:
[0086] Use the trained dialect speech recognition model to convert the input dialect speech into text.
[0087] This can be done by extracting the input dialect speech features and inputting them into the model, and then using the model to output the corresponding text.
[0088] 5. Post-processing and error correction:
[0089] Post-process the recognition results, such as removing repeated words, correcting grammatical errors, etc. Use natural language processing technology to further improve and correct the recognition results.
[0090] Algorithms: post-processing algorithms (such as deduplication and grammatical error correction), NLP technologies (such as grammatical analysis and semantic analysis). Among them, NLP technologies can be used for grammatical analysis and semantic understanding of speech recognition results, thereby improving the accuracy and readability of recognition results.
[0091] Therefore, this application marks the collected dialect speech data of multiple dialects, extracts features to form a speech pool, and trains the dialect speech conversion model, thereby converting the dialect speech data to be converted, generating corresponding text, and extracting multiple feature audio data of different dialects through an improved feature extraction algorithm to improve the dialect conversion efficiency. Recognizing dialect speech in multiple regions improves the accuracy of converted text and saves time.
[0092] Exemplary Devices
[0093] Figure 2 FIG. 1 is a schematic diagram of a structure of a device for converting dialect speech into text provided by an exemplary embodiment of the present invention. Figure 2 As shown, the device 200 includes:
[0094] A collection module 210 is used to collect dialect speech data from different regions, wherein the dialect speech data includes the pronunciation, speaking speed and intonation of the dialect;
[0095] An extraction module 220 is used to extract speech features from the dialect speech data using a Mel frequency cepstral coefficient algorithm, a long-time average short-time energy ratio algorithm, and a speech endpoint detection algorithm to obtain multiple feature audio data of the dialect speech data;
[0096] A training module 230 is used to train a machine learning model based on multiple feature audio data and text annotations of dialect speech data to generate a dialect speech conversion model;
[0097] The conversion module 240 is used to input a plurality of feature audio data to be converted of the dialect speech data to be converted into the dialect speech conversion model, and output a dialect conversion text of the dialect speech data to be converted, wherein:
[0098] The windowing formula of the Mel frequency cepstral coefficient algorithm is:
[0099] ;
[0100] Where a is the coefficient of the Hamming window function, n is the length of the frame, N Window length.
[0101] Optionally, the device 200 further includes: a processing module, configured to perform noise reduction processing on the dialect speech data.
[0102] Optionally, the extraction module 220 includes:
[0103] The first extraction submodule is used to adjust the number of filters and the filter frequency range of the Mel frequency cepstral coefficient algorithm to extract speech features from dialect speech data of different dialects, and obtain the Mel frequency cepstral coefficients in multiple feature audio data of dialect speech data of different dialects;
[0104] The second extraction submodule is used to extract speech features from dialect speech data of different dialects using a long-term average short-term energy ratio algorithm to obtain long-term average short-term energy ratios in multiple feature audio data of dialect speech data of different dialects;
[0105] The third extraction submodule is used to extract speech features from dialect speech data of different dialects using a speech endpoint detection algorithm, and obtain time feature data from a variety of feature audio data of dialect speech data of different dialects.
[0106] Optionally, the formula of the long-time average short-time energy ratio algorithm is:
[0107] ;
[0108] In the formula, LTSE is the long-term standard error; N represents the number of samples; E ( n ) represents the error of the nth sample, that is, the difference between the actual value and the predicted value.
[0109] Optionally, the loss function of the machine learning model is:
[0110] ;
[0111] Where L is the loss function, y is the probability distribution of the actual category label, and y' is the predicted probability distribution;
[0112] The learning function is:
[0113] ;
[0114] Where θ is the model parameter, t is the number of iterations, α=0.001 is the learning rate, and ∇θL(y, y') is the gradient of the loss function with respect to the model parameters.
[0115] Exemplary Electronic Devices
[0116] Figure 3 This is a structure of an electronic device provided by an exemplary embodiment of the present invention. Figure 3 As shown, the electronic device 30 includes one or more processors 31 and a memory 32 .
[0117] The processor 31 may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device to perform desired functions.
[0118] The memory 32 may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, a random access memory (RAM) and / or a cache memory (cache), etc. The non-volatile memory may include, for example, a read-only memory (ROM), a hard disk, a flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 31 may run the program instructions to implement the methods of the software programs of the various embodiments of the present invention described above and / or other desired functions. In one example, the electronic device may also include: an input device 33 and an output device 34, which are interconnected via a bus system and / or other forms of connection mechanisms (not shown).
[0119] In addition, the input device 33 may also include, for example, a keyboard, a mouse, and the like.
[0120] The output device 34 can output various information to the outside. The output device 34 can include, for example, a display, a speaker, a printer, a communication network and a remote output device connected thereto.
[0121] Of course, to simplify, Figure 3 Only some of the components related to the present invention in the electronic device are shown, and components such as a bus, an input / output interface, etc. are omitted. In addition, according to specific application conditions, the electronic device may also include any other appropriate components.
[0122] Exemplary computer program products and computer-readable storage media
[0123] In addition to the above-mentioned methods and devices, an embodiment of the present invention may also be a computer program product, which includes computer program instructions, which, when executed by a processor, enable the processor to execute the steps of the method according to various embodiments of the present invention described in the above-mentioned "Exemplary Method" section of this specification.
[0124] The computer program product may be written in any combination of one or more programming languages to write program code for performing the operations of the embodiments of the present invention, including object-oriented programming languages such as Java, C++, etc., and conventional procedural programming languages such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.
[0125] In addition, an embodiment of the present invention may also be a computer-readable storage medium having computer program instructions stored thereon, which, when executed by a processor, enables the processor to execute the steps of the method for information mining of historical change records according to various embodiments of the present invention described in the above “Exemplary Method” section of this specification.
[0126] The computer readable storage medium can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can include, for example, but is not limited to, a system, system or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable 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.
[0127] The basic principle of the present invention is described above in conjunction with specific embodiments. However, it should be pointed out that the advantages, strengths, effects, etc. mentioned in the present invention are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. must be possessed by each embodiment of the present invention. In addition, the specific details disclosed above are only for the purpose of illustration and facilitation of understanding, rather than limitation, and the above details do not limit the present invention to being implemented by adopting the above specific details.
[0128] Each embodiment in this specification is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system embodiment, since it basically corresponds to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.
[0129] The block diagrams of the devices, systems, equipment, and systems involved in the present invention are only illustrative examples and are not intended to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagram. As will be appreciated by those skilled in the art, these devices, systems, equipment, and systems can be connected, arranged, and configured in any manner. Words such as "including", "comprising", "having", etc. are open words, referring to "including but not limited to", and can be used interchangeably with them. The words "or" and "and" used here refer to the words "and / or" and can be used interchangeably with them, unless the context clearly indicates otherwise. The word "such as" used here refers to the phrase "such as but not limited to", and can be used interchangeably with it.
[0130] The method and system of the present invention may be implemented in many ways. For example, the method and system of the present invention may be implemented by software, hardware, firmware or any combination of software, hardware, firmware. The above order of steps for the method is only for illustration, and the steps of the method of the present invention are not limited to the order specifically described above, unless otherwise specifically stated. In addition, in some embodiments, the present invention may also be implemented as a program recorded in a recording medium, which includes machine-readable instructions for implementing the method according to the present invention. Thus, the present invention also covers a recording medium storing a program for executing the method according to the present invention.
[0131] It should also be noted that in the system, device and method of the present invention, each component or each step can be decomposed and / or recombined. These decompositions and / or recombinations should be regarded as equivalent schemes of the present invention. The above description of the disclosed aspects is provided to enable any technician in the field to make or use the present invention. Various modifications to these aspects are very obvious to those skilled in the art, and the general principles defined here can be applied to other aspects without departing from the scope of the present invention. Therefore, the present invention is not intended to be limited to the aspects shown here, but in accordance with the widest range consistent with the principles and novel features disclosed here.
[0132] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present invention to the forms disclosed herein. Although a number of example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof.
Claims
1. A method for converting dialect speech into text, characterized in that: include: Collecting dialect speech data from different regions, wherein the dialect speech data includes pronunciation, speaking speed and intonation of the dialect; Using a Mel frequency cepstral coefficient algorithm, a long-time average short-time energy ratio algorithm, and a speech endpoint detection algorithm to extract speech features from the dialect speech data, and obtain multiple feature audio data of the dialect speech data; Training a machine learning model based on the multiple characteristic audio data and text annotations of the dialect speech data to generate a dialect speech conversion model; Input multiple feature audio data to be converted of the dialect speech data to be converted into the dialect speech conversion model, and output the dialect conversion text of the dialect speech data to be converted, wherein: The windowing formula of the Mel frequency cepstral coefficient algorithm is: ； Where a is the coefficient of the Hamming window function, n is the length of the frame, N Window length.
2. The method according to claim 1, characterized in that Also includes: The dialect speech data is subjected to noise reduction processing.
3. The method according to claim 1, characterized in that The Mel frequency cepstral coefficient algorithm, the long-time average short-time energy ratio algorithm and the speech endpoint detection algorithm are used to extract speech features from the dialect speech data to obtain multiple feature audio data of the dialect speech data, including: Adjusting the number of filters and the filter frequency range of the Mel frequency cepstral coefficient algorithm to extract speech features from the dialect speech data of different dialects, and obtaining Mel frequency cepstral coefficients from a plurality of the feature audio data of the dialect speech data of different dialects; Using the long-time average short-time energy ratio algorithm to extract speech features from the dialect speech data of different dialects, and obtaining the long-time average short-time energy ratios in the plurality of feature audio data of the dialect speech data of different dialects; The speech endpoint detection algorithm is used to extract speech features from the dialect speech data of different dialects, and time feature data from the multiple feature audio data of the dialect speech data of different dialects is obtained.
4. The method according to claim 3, characterized in that The formula of the long-time average short-time energy ratio algorithm is: ； In the formula, LTSE is the long-term standard error; N represents the number of samples; E ( n ) represents the error of the nth sample, that is, the difference between the actual value and the predicted value.
5. The method according to claim 1, characterized in that The loss function of the machine learning model is: ； Where L is the loss function, y is the probability distribution of the actual category label, and y' is the predicted probability distribution; The learning function is: ； Where θ is the model parameter, t is the number of iterations, α=0.001 is the learning rate, and ∇θL(y, y') is the gradient of the loss function with respect to the model parameters.
6. A device for converting dialect speech into text, characterized in that: include: A collection module, used to collect dialect speech data from different regions, wherein the dialect speech data includes the pronunciation, speaking speed and intonation of the dialect; An extraction module, used to extract speech features from the dialect speech data using a Mel frequency cepstral coefficient algorithm, a long-time average short-time energy ratio algorithm, and a speech endpoint detection algorithm, to obtain a variety of feature audio data of the dialect speech data; A training module, used to train a machine learning model based on the multiple characteristic audio data and the text annotations of the dialect speech data to generate a dialect speech conversion model; The conversion module is used to input a plurality of feature audio data to be converted of the dialect speech data to be converted into the dialect speech conversion model, and output the dialect conversion text of the dialect speech data to be converted, wherein: The windowing formula of the Mel frequency cepstral coefficient algorithm is: ； Where a is the coefficient of the Hamming window function, n is the length of the frame, N Window length.
7. The device according to claim 6, characterized in that Also includes: A processing module is used to perform noise reduction processing on the dialect speech data.
8. The device according to claim 6, characterized in that Extraction modules, including: A first extraction submodule is used to adjust the number of filters and the filter frequency range of the Mel frequency cepstral coefficient algorithm to extract speech features from the dialect speech data of different dialects, and obtain Mel frequency cepstral coefficients from a plurality of feature audio data of the dialect speech data of different dialects; A second extraction submodule is used to extract speech features from the dialect speech data of different dialects using the long-time average short-time energy ratio algorithm to obtain the long-time average short-time energy ratios in the multiple feature audio data of the dialect speech data of different dialects; The third extraction submodule is used to extract speech features from the dialect speech data of different dialects using the speech endpoint detection algorithm, and obtain time feature data from the multiple feature audio data of the dialect speech data of different dialects.
9. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and the computer program is used to execute the method according to any one of claims 1 to 5.
10. An electronic device, characterized in that: The electronic device comprises: processor; a memory for storing instructions executable by the processor; The processor is used to read the executable instructions from the memory and execute the instructions to implement the method described in any one of claims 1 to 5.