CN114579799A - A method, device, device and medium for generating a recorded manuscript - Google Patents

Abstract

The invention provides a recording manuscript generating method, a device, equipment and a medium, wherein the method comprises the following steps: obtaining a self-defined pinyin combination sequence; converting each pinyin in the pinyin combination sequence into a corresponding Chinese character based on a preset mapping relation between the pinyin sequence and the Chinese character sequence to obtain a Chinese character combination sequence; inputting the Chinese character combination sequence into a trained text generation model to obtain an initial recording manuscript; and correcting the error of the initial recording manuscript to obtain a target recording manuscript. By adopting the embodiment of the invention, the input pinyin sequence can be customized according to actual requirements to generate the recording manuscript which has less characters and contains all pinyin types, thereby further reducing the recording corpus required for constructing the voice synthesis model and lowering the threshold of customizing the sound library.

Description

A method, device, device and medium for generating a recorded manuscript

technical field

The present invention relates to the technical field of audio recording, and in particular, to a method, device, equipment and medium for generating audio recordings.

Background technique

Constructing a speech synthesis model requires the alignment corpus of text and speech, and the alignment corpus of text and speech is generally obtained by dubbing personnel according to the recorded manuscript. Therefore, the content quality of the recorded manuscript directly determines the content quality of the alignment corpus of text and speech. The recording manuscript is a text file used to instruct the dubbing personnel to perform recording, and contains the text content that needs to be recorded. However, the inventor found in the research on the prior art that the method for constructing the recording manuscript in the prior art, or the construction of the recording The number of manuscripts is large, or professional writers are required to write recording manuscripts, and the cost is high, which does not meet the needs of customizing the sound library with a small amount of corpus.

SUMMARY OF THE INVENTION

The present invention provides a method, device, equipment and medium for generating a recording manuscript, which can reduce the recording corpus required for building a speech synthesis model and lower the threshold for customizing a sound library.

To achieve the above purpose, an embodiment of the present invention provides a method for generating a recorded manuscript, comprising the following steps:

Obtain a custom pinyin combination sequence; wherein, the pinyin combination sequence includes a plurality of pinyin and the number of times each of the pinyin appears;

Based on the preset mapping relationship between the pinyin sequence and the Chinese character sequence, each of the pinyin in the pinyin combination sequence is converted into a corresponding Chinese character to obtain a Chinese character combination sequence; wherein, the Chinese character combination sequence includes a plurality of Chinese characters and the number of occurrences of each said Chinese character;

The Chinese character combination sequence is input into the text generation model that the training is completed, and the initial recording manuscript is obtained;

Error correction is performed on the initial recording manuscript to obtain a target audio recording manuscript.

As an optional embodiment, according to the preset mapping relationship between the pinyin sequence and the Chinese character sequence, each of the pinyin in the pinyin combination sequence is converted into a corresponding Chinese character to obtain a Chinese character combination sequence, include:

Based on the pinyin arrangement order in the preset mapping table, sorting each pinyin in the pinyin combination sequence according to the preset order to obtain a sorted pinyin sequence;

Based on the preset mapping table, the sorted pinyin sequence is converted into a corresponding Chinese character sequence, and after each Chinese character in the Chinese character sequence, the number of occurrences of the corresponding pinyin of the Chinese character is increased to obtain a Chinese character combination sequence.

As an optional embodiment, the text generation model is trained in the following manner:

Obtaining a question-and-answer pair corpus; wherein, the question-answer pair corpus includes an original sentence and a question sentence;

The question-and-answer pair corpus is input into a preset text generation model, and the text generation model is trained to obtain a trained text generation model.

As an optional embodiment, the question-and-answer pair corpus is obtained in the following manner:

Obtaining a text corpus, and dividing the text corpus into a plurality of original sentences according to a preset sentence segmentation method;

Based on the preset mapping table, each of the original sentences in the text corpus is converted into a corresponding pinyin sequence;

Count the number of occurrences of each pinyin in each described pinyin sequence, and arrange each pinyin in each described pinyin sequence according to the preset order according to the number of occurrences to obtain a sorted pinyin sequence;

Based on the preset mapping table, the arranged pinyin sequence is converted into a corresponding Chinese character sequence, and the number of occurrences of the corresponding pinyin of the Chinese character is added after each Chinese character in the Chinese character sequence, so that each original sentence corresponds to question sentence.

Each original sentence and its corresponding question sentence are formed into a set of question-answer pairs, and multiple sets of question-answer pairs are obtained.

As an optional embodiment, the pinyin combination sequence further includes the tone of each of the pinyin;

Then, based on the preset mapping relationship between the pinyin sequence and the Chinese character sequence, each of the pinyin in the pinyin combination sequence is converted into a corresponding Chinese character, and the Chinese character combination sequence is obtained, including:

Based on the pinyin arrangement order in the preset mapping table, sorting each pinyin in the pinyin combination sequence according to the preset order to obtain a sorted pinyin sequence;

Based on the preset mapping table, according to the pitch of each pinyin, the sorted pinyin sequence is converted into a corresponding Chinese character sequence, and the number of occurrences of the corresponding pinyin of the Chinese character is increased after each Chinese character in the Chinese character sequence , get the Chinese character combination sequence.

As an optional embodiment, the preset mapping table includes a general standard Chinese character table.

As an optional embodiment, the text generation model includes GPT, GPT2, GPT3, LaserTagger, and LSTM.

An embodiment of the present invention provides an apparatus for generating a recorded document, including:

a pinyin combination sequence acquisition module for acquiring a custom pinyin combination sequence; wherein the pinyin combination sequence includes a plurality of pinyin and the number of times each of the pinyin occurs;

A Chinese character combination sequence acquisition module, configured to convert each of the pinyin in the pinyin combination sequence into a corresponding Chinese character based on the preset mapping relationship between the pinyin sequence and the Chinese character sequence, to obtain a Chinese character combination sequence; wherein, the The Chinese character combination sequence includes a plurality of Chinese characters and the number of occurrences of each said Chinese character;

The initial recording manuscript acquisition module is used to input the Chinese character combination sequence into the text generation model that the training is completed to obtain the initial recording manuscript;

The target recording manuscript acquisition module is used to correct the errors of the initial recording manuscript to obtain the target audio recording manuscript.

An embodiment of the present invention provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, where the above implementation is implemented when the processor executes the computer program The method for generating the recorded manuscript described in the example.

An embodiment of the present invention provides a computer-readable storage medium, where the computer-readable storage medium includes a stored computer program, wherein when the computer program runs, the device where the computer-readable storage medium is located is controlled to execute the foregoing embodiments The described vehicle positioning method.

Compared with the prior art, a method, device, equipment and medium for generating a recorded manuscript provided by the embodiment of the present invention can customize the input pinyin sequence according to actual needs, and generate a recorded manuscript with a small number of characters and all types of pinyin, thereby Further reduce the recording corpus required to build a speech synthesis model, and lower the threshold for customizing the sound library.

Description of drawings

1 is a schematic flowchart of a method for generating a recorded manuscript provided by an embodiment of the present invention;

2 is a schematic structural diagram of a device for generating a recorded manuscript provided by an embodiment of the present invention;

FIG. 3 is a schematic structural diagram of a terminal device according to an embodiment of the present invention.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

An embodiment of the present invention provides a method for generating a recorded manuscript. Referring to FIG. 1 , it is a schematic flowchart of a method for generating a recorded manuscript provided by an embodiment of the present invention, and the method includes steps S11 to S14:

S11. Obtain a custom pinyin combination sequence; wherein the pinyin combination sequence includes a plurality of pinyin and the number of occurrences of each pinyin.

It should be noted that the customized pinyin combination sequence is based on sentences.

S12. Based on the preset mapping relationship between the pinyin sequence and the Chinese character sequence, convert each of the pinyin in the pinyin combination sequence into a corresponding Chinese character to obtain a Chinese character combination sequence; wherein, the Chinese character combination sequence includes a plurality of Chinese characters and the number of occurrences of each said Chinese character.

S13. Input the Chinese character combination sequence into the trained text generation model to obtain an initial recording manuscript.

S14. Perform error correction on the initial recorded manuscript to obtain a target recorded manuscript.

It should be noted that the text error correction method includes but is not limited to: a statistical-based method, a dictionary-based method, and a neural network-based method. In the actual application process, the method for correcting the errors of the recorded manuscript can be set according to actual needs or experiments, which is not limited here. Among them, it can be understood that if the text generated by the model itself has no errors, the text after the text error correction is consistent with the original text.

In addition, it should be noted that text generation is also called natural language generation, and a natural language generation system can be defined as accepting information in non-linguistic form as input to generate readable text expressions. In the present invention, the pinyin text is specified by the user, and then the Chinese character text is generated. For example, the user specifies that the pinyin of qi3 appears once, lai2 appears once, etc., by using the embodiment of the present invention, a Chinese character text similar to "March 5th building, come on and laugh" is obtained, and the user's preset conditions are met after error correction. .

Further, in the prior art, the method for constructing a recording manuscript generally includes: (1) Selecting several written works written by humans as the recording manuscript; (2) Select some sentence combinations from sentences written by humans as recording manuscripts; for example, divide documents such as news information, prose, novels, etc. into multiple sentences, and then set rules to select some sentences from the segmented sentences. , construct the recorded manuscript with the screened sentences; (3) Professionals write the recorded manuscript according to the preset requirements; for example, specify the subject matter, number of words, genre and other conditions, and the professional writers will write the articles according to these preset requirements, and the articles written are appropriate Process the document as a recording. However, the above method has the following problems: (1) Select several written works written by humans as the recorded manuscript; the recorded manuscript constructed by this method is of high literary quality, rigorous logic and strong readability, but there are many repetitions in pinyin and redundant information. Large, resulting in a large number of manuscripts required to cover all pinyin, and the large number of such corpora required to build a speech synthesis model, which is not conducive to customizing the sound library with a small amount of resources; (2) Select some sentence combinations from sentences written by humans as recordings manuscript; each sentence of the manuscript constructed by this method is logical and readable, but the repetition rate of high-frequency pinyin is high, resulting in a large number of manuscripts needed to cover all pinyin, which is not conducive to customizing the sound library; (3) Professionals write recorded manuscripts according to preset requirements; the manuscripts constructed by this method are of high literary quality, strong logic, and strong readability, but have high requirements for professional writers and high costs, and if the word count requirements and pinyin coverage requirements are specified, It requires professional writers to have a deep understanding of pinyin, and the requirements for professional writers are higher, which is difficult to implement successfully.

Compared with the prior art, the method for generating a recorded manuscript provided by the embodiment of the present invention can customize the input pinyin sequence according to actual needs, and generate a recorded manuscript with a small number of characters and including all pinyin types, thereby further reducing the construction of speech synthesis. The recording corpus required by the model reduces the threshold for customizing the sound library. In addition, the present invention constructs the recorded manuscript in an automatic manner without consuming human resources, thereby further reducing the writing cost.

It should be noted that, by adopting the embodiment of the present invention, when all types of pinyin are included and the number of occurrences of each pinyin is higher than a preset threshold, the number of words in the entire recorded manuscript can be reduced, rather than the number of words corresponding to a single sentence of pinyin. For example, extracting sentences from news manuscripts to form a recording manuscript may require 100,000 words, but this program only needs 10,000 words.

As an optional embodiment, the step S12 includes:

S121, based on the pinyin arrangement order in the preset mapping table, sort each pinyin in the pinyin combination sequence according to the preset order to obtain the sorted pinyin sequence;

S122. Based on the preset mapping table, convert the sorted pinyin sequence into a corresponding Chinese character sequence, and increase the number of occurrences of the corresponding pinyin of the Chinese character after each Chinese character in the Chinese character sequence to obtain a Chinese character combination sequence.

In the actual application process, the customized pinyin combination sequence can be the pinyin combination sequence that has been sorted according to the preset order, and then directly converted based on the preset mapping table to obtain the Chinese character combination sequence, or it can be the unsorted pinyin combination sequence. The pinyin combination sequence is reordered and converted after input, which can be set according to actual needs or experiments, which is not limited here.

It should be noted that the general composition structure of the pinyin combination sequence of custom input is: pinyin + number corresponding to tone + number of occurrences, but in the actual application process, the composition of pinyin + number of occurrences can also be used. The specific setting It is related to the effect to be achieved by the user, which is not limited here.

Illustratively, several examples of customizing pinyin input to generate audio transcripts are as follows:

(1) Several pinyin appear multiple times:

Assuming that the pinyin of the custom input and the number of occurrences are: dong 1 wu 1 jia 1qi 2 yue 2 san1you 1lai 2 xiao 1, sort these pinyin combinations to get qi 2 lai 2 yue 2 jia 1 san1you 1 dong 1 wu 1 xiao 1, and then find the first character in the Chinese character corresponding to each pinyin from the "General Standard Chinese Character Table" as the exclusive Chinese character for the pinyin, and convert it into a combination sequence of Chinese characters: 2, 2, 2, 1, 3, 1 tour in February 1 action 1 martial arts 1 effect 1;

(2) Each pinyin appears only once:

Pinyin and its occurrences are: qi3 1 lai2 1 yue4 1 jia1 1 san1 1 you2 1 dong4 1wu3 1 xiao4 1. The text input to the converted text model is: Enterprise 1 to January 1, 1, 3, 1, 1, 1, 1, 1, 1, 1, 1, and 1, and the text output from the text generation model is: March 5th, come on, laugh, and the recorded manuscript after text error correction The text reads: March 5th, come on and laugh.

(3) A single pinyin multiple times:

The pinyin and its occurrence times are: qi3 4, the text input to the converted text model is: Qi 4, the text output by the text generation model is: Qiqiqiqi, and the text of the recorded manuscript after text error correction is: Qiqiqiqi.

(4) Each pinyin appears multiple times:

The pinyin and the number of occurrences of the custom input are: qi3 4 lai2 4 yue4 4. The text of the converted text input model is: Qi 4 to April 4, the text output by the text generation model is: Yueyue jump up, get up and jump up, and the text of the recorded manuscript after text error correction is: Yueyue jump up, get up Get up and jump up.

It should be noted that, in the actual application process, the preset mapping table can be set according to actual requirements or experiments, and the reference standard of the general standard Chinese character table here is only for illustration and not for limitation.

As an optional embodiment, the text generation model is trained in the following manner:

Obtaining a question-and-answer pair corpus; wherein, the question-answer pair corpus includes an original sentence and a question sentence;

The question-and-answer pair corpus is input into a preset text generation model, and the text generation model is trained to obtain a trained text generation model.

As an optional embodiment, the question-and-answer pair corpus is obtained in the following manner:

Obtaining a text corpus, and dividing the text corpus into a plurality of original sentences according to a preset sentence segmentation method;

Based on the preset mapping table, each of the original sentences in the text corpus is converted into a corresponding pinyin sequence;

Count the number of occurrences of each pinyin in each described pinyin sequence, and arrange each pinyin in each described pinyin sequence according to the preset order according to the number of occurrences to obtain a sorted pinyin sequence;

Based on the preset mapping table, the arranged pinyin sequence is converted into a corresponding Chinese character sequence, and the number of occurrences of the corresponding pinyin of the Chinese character is added after each Chinese character in the Chinese character sequence, so that each original sentence corresponds to question sentence.

Each original sentence and its corresponding question sentence are formed into a set of question-answer pairs, and multiple sets of question-answer pairs are obtained.

It should be noted that the sentence segmentation methods include but are not limited to: sentence segmentation based on punctuation marks and sentence segmentation using an NLP pre-trained model.

Exemplarily, the main steps to obtain or construct a question-and-answer pair corpus include:

1. Collect a large number of text corpora with multiple subjects; among them, the channels for collecting text corpus include but are not limited to: news information, online literature, textbooks, prose novels, ancient literature; methods for collecting text corpus include but are not limited to: web crawlers, Open source download, OCR recognition, speech recognition;

2. Convert the text corpus into a question-and-answer corpus;

(1) Divide the corpus into sentence-based texts by the method of clauses to form a list of sentences;

(2) Each pinyin is represented by an exclusive Chinese character to form a mapping table of pinyin and Chinese characters;

3, each sentence in the sentence list is converted into Pinyin from Chinese characters to generate a pinyin sequence; wherein, the methods for Chinese characters to Pinyin include, but are not limited to: Chinese characters based on Pinyin Dictionary, Chinese characters based on statistical learning, based on natural Chinese characters to Pinyin for language processing pre-training models;

4. Generate question sentences for question and answer pairs;

(1) Count the number of occurrences of each pinyin in the pinyin sequence transferred from each sentence.

(2) Pinyin is sorted in descending order of the number of occurrences of pinyin. If the number of occurrences of pinyin is the same, it is sorted in ascending order according to the order of the "General Standard Chinese Character Table".

(3) according to the pinyin sequence after the sorting, according to the pinyin Chinese character mapping table, turn into Chinese characters, increase the number of times that the corresponding pinyin of this Chinese character occurs behind each Chinese character, form the question sentence of the question and answer pair corpus;

For example: The original sentence of the sentence list is: Come on, move in March, smile in May.

The generated question sentence is: Enterprise 2 to February 2, 1, 3, 1, swimming, 1, 1, martial, 1, 1.

5. Generate question-and-answer corpus.

(1) Use the sentence in the sentence list as the answer sentence.

(2) Use the question sentence and the answer sentence generated in the previous step to form a question-and-answer pair.

(3) Operate all sentences in the sentence list to generate a question-and-answer pair corpus.

As an optional embodiment, the pinyin combination sequence further includes the tone of each of the pinyin;

Then, based on the preset mapping relationship between the pinyin sequence and the Chinese character sequence, each of the pinyin in the pinyin combination sequence is converted into a corresponding Chinese character, and the Chinese character combination sequence is obtained, including:

Based on the pinyin arrangement order in the preset mapping table, sorting each pinyin in the pinyin combination sequence according to the preset order to obtain a sorted pinyin sequence;

Based on the preset mapping table, according to the pitch of each pinyin, the sorted pinyin sequence is converted into a corresponding Chinese character sequence, and the number of occurrences of the corresponding pinyin of the Chinese character is increased after each Chinese character in the Chinese character sequence , get the Chinese character combination sequence.

Exemplarily, assuming that the pinyin of the custom input and the number of occurrences thereof are: dong4 1 wu3 1 jia1 1 qi32 yue4 2 san1 1 you2 1lai2 2 xiao4 1, sort these pinyin combinations to get qi3 2lai2 2 yue4 2jia1 1 san1 1 you2 1 dong4 1 wu3 1 xiao4 1, and then find the first character in the Chinese character corresponding to each pinyin from the "General Standard Chinese Character Table" as the exclusive Chinese character for the pinyin, and convert it into a Chinese character combination sequence: Qi2 to February 2 households, 1 three, 1 swimming, 1 action, 1 martial arts, 1 effect and 1.

It should be noted that the general composition structure of the pinyin combination sequence of custom input is: pinyin + number corresponding to tone + number of occurrences, but in the actual application process, the composition of pinyin + number of occurrences can also be used. The specific setting It is related to the effect to be achieved by the user, which is not limited here.

As an optional embodiment, the preset mapping table includes a general standard Chinese character table.

As an optional embodiment, the text generation model includes GPT, GPT2, GPT3, LaserTagger, and LSTM.

It should be noted that the text generation model can use an open-source pre-trained model as the basic model, and the model parameters can also be freely customized, including but not limited to: setting parameters according to the amount of training corpus data and setting parameters according to experience, which are not limited here.

An embodiment of the present invention provides an apparatus for generating a recorded document. Referring to FIG. 2 , it is a schematic structural diagram of an apparatus for generating a recorded document 20 provided by an embodiment of the present invention, including:

The pinyin combination sequence acquisition module 21 is used to acquire a custom pinyin combination sequence; wherein, the pinyin combination sequence includes a plurality of pinyin and the number of times each of the pinyin occurs;

The Chinese character combination sequence acquisition module 22 is used to convert each of the pinyin in the pinyin combination sequence into a corresponding Chinese character based on the preset mapping relationship between the pinyin sequence and the Chinese character sequence to obtain the Chinese character combination sequence; Described Chinese character combination sequence comprises a plurality of Chinese characters and the number of times that each described Chinese character appears;

The initial recording manuscript acquisition module 23 is used to input the Chinese character combination sequence into the text generation model that the training is completed to obtain the initial recording manuscript;

The target recording manuscript obtaining module 24 is used for error correction of the initial recording manuscript to obtain the target audio recording manuscript.

Compared with the prior art, an apparatus for generating a recorded manuscript provided by the embodiment of the present invention can customize the input pinyin sequence according to actual needs, and generate a recorded manuscript with a small number of characters and including all types of pinyin, thereby further reducing the construction of a speech synthesis model. The required recording corpus reduces the threshold for sound library customization.

As one of the optional embodiments, the Chinese character combination sequence acquisition module 22 is specifically used for:

Based on the pinyin arrangement order in the preset mapping table, sorting each pinyin in the pinyin combination sequence according to the preset order to obtain a sorted pinyin sequence;

Based on the preset mapping table, the sorted pinyin sequence is converted into a corresponding Chinese character sequence, and the number of occurrences of the corresponding pinyin of the Chinese character is added after each Chinese character in the Chinese character sequence to obtain a Chinese character combination sequence.

As an optional embodiment, the text generation model is trained in the following manner:

Obtain a question-and-answer pair corpus; wherein, the question-and-answer pair corpus includes original sentences and question sentences;

The question-and-answer pair corpus is input into a preset text generation model, and the text generation model is trained to obtain a trained text generation model.

As an optional embodiment, the question-and-answer pair corpus is obtained in the following manner:

Obtaining a text corpus, and dividing the text corpus into a plurality of original sentences according to a preset sentence segmentation method;

Based on a preset mapping table, converting each of the original sentences in the text corpus into a corresponding pinyin sequence;

Count the number of occurrences of each pinyin in each described pinyin sequence, and arrange each pinyin in each described pinyin sequence according to the preset order according to the number of occurrences to obtain a sorted pinyin sequence;

Based on the preset mapping table, the arranged pinyin sequence is converted into a corresponding Chinese character sequence, and the number of occurrences of the corresponding pinyin of the Chinese character is added after each Chinese character in the Chinese character sequence, so that each original sentence corresponds to question sentence.

Each original sentence and its corresponding question sentence are formed into a set of question-answer pairs, and multiple sets of question-answer pairs are obtained.

As an optional embodiment, the pinyin combination sequence further includes the tone of each of the pinyin;

Then, based on the preset mapping relationship between the pinyin sequence and the Chinese character sequence, each of the pinyin in the pinyin combination sequence is converted into a corresponding Chinese character, and the Chinese character combination sequence is obtained, including:

Based on the pinyin arrangement order in the preset mapping table, sorting each pinyin in the pinyin combination sequence according to the preset order to obtain a sorted pinyin sequence;

Based on the preset mapping table, according to the pitch of each pinyin, the sorted pinyin sequence is converted into a corresponding Chinese character sequence, and the number of occurrences of the corresponding pinyin of the Chinese character is increased after each Chinese character in the Chinese character sequence , get the Chinese character combination sequence.

As an optional embodiment, the preset mapping table includes a general standard Chinese character table.

As an optional embodiment, the text generation model includes GPT, GPT2, GPT3, LaserTagger, and LSTM.

In addition, it should be noted that the specific implementation schemes and beneficial effects of each embodiment of the device for generating a recorded document provided by the embodiment of the present invention are different from the specific implementation schemes and beneficial effects of each embodiment of the method for generating a recorded document provided by the embodiment of the present invention. The implementation scheme and the beneficial effects are correspondingly the same, and are not repeated here.

An embodiment of the present invention provides a terminal device. Referring to FIG. 3 , it is a schematic structural diagram of a terminal device provided by an embodiment of the present invention. The terminal device 3 of this embodiment includes: a processor 30 , a memory 31 , and a computer program stored in the memory 31 and executable on the processor 30 . When the processor 30 executes the computer program, the method for generating a recorded document described in any of the foregoing embodiments is implemented. Alternatively, when the processor 30 executes the computer program, the functions of the modules in the foregoing device embodiments are implemented.

Exemplarily, the computer program can be divided into one or more modules, and the one or more modules are stored in the memory 31 and executed by the processor 30 to complete the present invention. The one or more modules may be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of the computer program in the terminal device 3 .

The terminal device 3 may be a computing device such as a desktop computer, a notebook, a palmtop computer, and a cloud server. The terminal device 3 may include, but is not limited to, a processor 30 and a memory 31 . Those skilled in the art can understand that the schematic diagram is only an example of a terminal device, and does not constitute a limitation to the terminal device, and may include more or less components than the one shown in the figure, or combine some components, or different components, For example, the terminal device 3 may further include an input and output device, a network access device, a bus, and the like.

The so-called processor 30 may be a central processing unit (Central Processing Unit, CPU), and may also be other general-purpose processors, digital signal processors (Digital Signal Processors, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), Field-Programmable Gate Array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor. part.

The memory 31 can be used to store the computer programs and/or modules, and the processor 30 can run or execute the computer programs and/or modules stored in the memory 31 and call the data stored in the memory 31, Various functions of the terminal device 3 are realized. The memory 31 may mainly include a stored program area and a stored data area, wherein the stored program area may store an operating system, an application program required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the storage data area may Stores data (such as audio data, phonebook, etc.) created according to the use of the mobile phone, and the like. In addition, the memory 31 may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, internal memory, plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card, at least one magnetic disk storage device, flash memory device, or other volatile solid state storage device.

Wherein, if the modules integrated in the terminal device 3 are implemented in the form of software functional units and sold or used as independent products, they may be stored in a computer-readable storage medium. Based on this understanding, the present invention can implement all or part of the processes in the methods of the above embodiments, and can also be completed by instructing relevant hardware through a computer program, and the computer program can be stored in a computer-readable storage medium. When executed by the processor 30, the steps of the various method embodiments described above may be implemented. Wherein, the computer program includes computer program code, and the computer program code may be in the form of source code, object code, executable file or some intermediate form, and the like. The computer-readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a U disk, a removable hard disk, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM, Read-Only Memory) , Random Access Memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content contained in the computer-readable media may be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction, for example, in some jurisdictions, according to legislation and patent practice, the computer-readable media Electric carrier signals and telecommunication signals are not included.

It should be noted that the device embodiments described above are only schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical unit, that is, it can be located in one place, or it can be distributed over multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution in this embodiment. In addition, in the drawings of the apparatus embodiments provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. Those of ordinary skill in the art can understand and implement it without creative effort.

An embodiment of the present invention provides a computer-readable storage medium, where the computer-readable storage medium includes a stored computer program, wherein when the computer program runs, the device where the computer-readable storage medium is located is controlled to execute the above-mentioned The method for generating the recorded transcript described.

Those skilled in the art can understand that the modules in the apparatus in the embodiment can be adaptively changed and arranged in one or more apparatuses different from the embodiment. The modules or units in the embodiments may be combined into one module or unit, and further they may be divided into multiple sub-modules or sub-units. All features disclosed in this specification (including accompanying claims, abstract and drawings) and any method so disclosed may be employed in any combination, unless at least some of such features and/or procedures or elements are mutually exclusive. All processes or units of equipment are combined. Each feature disclosed in this specification (including accompanying claims, abstract and drawings) may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise.

It should be noted that the above-described embodiments illustrate rather than limit the invention, and that alternative embodiments may be devised by those skilled in the art without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in a claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements.

Furthermore, those skilled in the art will appreciate that although some of the embodiments described herein include certain features, but not others, included in other embodiments, that combinations of features of different embodiments are intended to be within the scope of the invention within and form different embodiments. For example, in the claims of this invention, any of the claimed embodiments may be used in any combination.

The above are the preferred embodiments of the present invention. It should be pointed out that for those skilled in the art, without departing from the principles of the present invention, several improvements and modifications can be made, and these improvements and modifications may also be regarded as It is the protection scope of the present invention.

Claims (10)

1. a recording manuscript generation method, is characterized in that, comprises: Obtain a custom pinyin combination sequence; wherein, the pinyin combination sequence includes a plurality of pinyin and the number of times each of the pinyin appears; Based on the preset mapping relationship between the pinyin sequence and the Chinese character sequence, each of the pinyin in the pinyin combination sequence is converted into a corresponding Chinese character to obtain a Chinese character combination sequence; wherein, the Chinese character combination sequence includes a plurality of Chinese characters and the number of occurrences of each said Chinese character; The Chinese character combination sequence is input into the text generation model that the training is completed, and the initial recording manuscript is obtained; Error correction is performed on the initial recording manuscript to obtain a target audio recording manuscript. 2. recording manuscript generation method according to claim 1, is characterized in that, described based on the preset mapping relation between pinyin sequence and Chinese character sequence, each described pinyin in described pinyin combination sequence is converted into corresponding Chinese characters, get the Chinese character combination sequence, including: Based on the pinyin arrangement order in the preset mapping table, sorting each pinyin in the pinyin combination sequence according to the preset order to obtain a sorted pinyin sequence; Based on the preset mapping table, the sorted pinyin sequence is converted into a corresponding Chinese character sequence, and after each Chinese character in the Chinese character sequence, the number of occurrences of the corresponding pinyin of the Chinese character is increased to obtain a Chinese character combination sequence. 3. recording manuscript generation method according to claim 1, is characterized in that, described text generation model is trained in the following manner: Obtaining a question-and-answer pair corpus; wherein, the question-answer pair corpus includes an original sentence and a question sentence; The question-and-answer pair corpus is input into a preset text generation model, and the text generation model is trained to obtain a trained text generation model. 4. recording manuscript generation method according to claim 3, is characterized in that, obtains question-and-answer pair language material by following way: Obtaining a text corpus, and dividing the text corpus into a plurality of original sentences according to a preset sentence segmentation method; Based on the preset mapping table, each of the original sentences in the text corpus is converted into a corresponding pinyin sequence; Count the number of occurrences of each pinyin in each described pinyin sequence, and arrange each pinyin in each described pinyin sequence according to the preset order according to the number of occurrences to obtain a sorted pinyin sequence; Based on the preset mapping table, the arranged pinyin sequence is converted into a corresponding Chinese character sequence, and the number of occurrences of the corresponding pinyin of the Chinese character is added after each Chinese character in the Chinese character sequence, so that each original sentence corresponds to question sentence. Each original sentence and its corresponding question sentence are formed into a set of question-answer pairs, and multiple sets of question-answer pairs are obtained. 5. recording manuscript generation method according to claim 1, is characterized in that, described pinyin combination sequence also comprises the pitch of each described pinyin; Then, based on the preset mapping relationship between the pinyin sequence and the Chinese character sequence, each of the pinyin in the pinyin combination sequence is converted into a corresponding Chinese character, and the Chinese character combination sequence is obtained, including: Based on the pinyin arrangement order in the preset mapping table, sorting each pinyin in the pinyin combination sequence according to the preset order to obtain a sorted pinyin sequence; Based on the preset mapping table, according to the pitch of each pinyin, the sorted pinyin sequence is converted into a corresponding Chinese character sequence, and the number of occurrences of the corresponding pinyin of the Chinese character is increased after each Chinese character in the Chinese character sequence , get the Chinese character combination sequence. 6 . The method for generating a recorded manuscript according to claim 1 , wherein the preset mapping table comprises a general standard Chinese character table. 7 . 7. The method for generating a recorded document according to claim 1, wherein the text generation model comprises GPT, GPT2, GPT3, LaserTagger, and LSTM. 8. a recording document generation device, is characterized in that, comprises: a pinyin combination sequence acquisition module for acquiring a custom pinyin combination sequence; wherein the pinyin combination sequence includes a plurality of pinyin and the number of times each of the pinyin occurs; A Chinese character combination sequence acquisition module, configured to convert each of the pinyin in the pinyin combination sequence into a corresponding Chinese character based on the preset mapping relationship between the pinyin sequence and the Chinese character sequence, to obtain a Chinese character combination sequence; wherein, the The Chinese character combination sequence includes a plurality of Chinese characters and the number of occurrences of each said Chinese character; The initial recording manuscript acquisition module is used to input the Chinese character combination sequence into the text generation model that the training is completed to obtain the initial recording manuscript; The target recording manuscript acquisition module is used to correct the errors of the initial recording manuscript to obtain the target audio recording manuscript. 9. A terminal device, characterized by comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, when the processor executes the computer program, the computer program as claimed in the claim is implemented The method for generating a recorded manuscript described in any one of requirements 1 to 7. 10. A computer-readable storage medium, characterized in that the computer-readable storage medium comprises a stored computer program, wherein, when the computer program is run, the device where the computer-readable storage medium is located is controlled to perform as claimed in the claims The recording manuscript generation method described in any one of 1 to 7.

Images