Digital human broadcasting method, device, equipment, storage medium and program product
By using pronunciation prediction model in the digital human broadcasting system, combining the content and source of the corpus, accurately judge the pronunciation of polyphonic characters, the problem that digital humans cannot accurately judge the pronunciation of polyphonic characters when playing voice, and improve the accuracy of voice broadcasting .
Patent Information
- Application Number
- CN202210037241.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-01-13
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2042-01-13
AI Technical Summary
Digital people cannot accurately judge the pronunciation of polyphonic characters when playing voice, especially those polyphonic characters that are not configured in the preset pronunciation library.
The pronunciation prediction model is used, combining the content and source of the corpus to be reported, and the target pronunciation of polyphonic characters is determined through a naive Bayes classifier. The model calculates the probability of the occurrence of polyphonic characters under different conditions based on the business areas and publishing agencies of the corpus, and then determines its most likely pronunciation.
Reduce the dependence on preset pronunciation databases, automatically predict the pronunciation of polyphonic characters, eliminate the process of users' continuous enrichment of pronunciation databases, and improve the accuracy of polyphonic characters recognition.
Smart Images

Figure CN114386401B_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the field of artificial intelligence technology, and particularly to a digital human broadcasting method, apparatus, device, storage medium, and program product. Background Art
[0002] A digital human is a virtual human that is based on artificial intelligence technologies such as image recognition, speech recognition and synthesis, semantic understanding, and portrait modeling, has the ability to perceive, recognize, and express the physical world, and uses devices such as electronic screens and VR to achieve human-computer interaction.
[0003] Currently, when a digital human plays speech, it cannot accurately determine the pronunciation of polyphonic characters when encountering them. For example, the "xing" in the two phrases "bank president" and "drive". Therefore, how to accurately determine the pronunciation of polyphonic characters has become an urgent technical problem to be solved. Summary of the Invention
[0004] In view of the above problems, the present disclosure provides a digital human broadcasting method, apparatus, device, storage medium, and program product.
[0005] According to a first aspect of the present disclosure, there is provided a digital human broadcasting method, which includes:
[0006] Obtain the corpus to be broadcast;
[0007] Determine the pronunciation of at least one character in the corpus to be broadcast according to a preset pronunciation library;
[0008] When the corpus to be broadcast contains at least one target polyphonic character, determine the target pronunciation of the target polyphonic character through a pronunciation prediction model according to the content of the corpus to be broadcast and the source of the corpus to be broadcast;
[0009] Wherein, the target polyphonic character includes a polyphonic character whose pronunciation is not configured in the preset pronunciation library.
[0010] According to an embodiment of the present disclosure, the determining the target pronunciation of the target polyphonic character according to the content of the corpus to be broadcast and the source of the corpus to be broadcast includes:
[0011] Determine the business field of the corpus to be broadcast according to the content of the corpus to be broadcast;
[0012] Determine the publishing institution of the corpus to be broadcast according to the source of the corpus to be broadcast;
[0013] Determine the target pronunciation of the target polyphonic character according to the occurrence probability of each pronunciation of the target polyphonic character in the determined business field and the occurrence probability of each pronunciation of the target polyphonic character in the determined publishing institution.
[0014] According to an embodiment of the present disclosure, determining the business domain of the corpus to be broadcast according to the content of the corpus to be broadcast includes: determining the business domain of the corpus to be broadcast according to the body of the corpus to be broadcast and / or the title of the expected report to be broadcast;
[0015] Determining the publishing agency of the corpus to be broadcast according to the source of the corpus to be broadcast includes: determining the publishing agency of the corpus to be broadcast according to the website of the corpus to be broadcast.
[0016] According to an embodiment of the present disclosure, the pronunciation prediction model includes a naive Bayes classifier, and the naive Bayes classifier includes an attribute vector composed of multiple conditions;
[0017] Determining the target pronunciation of the target polyphonetic character according to the occurrence probability of each pronunciation of the target polyphonetic character in the determined business field and the occurrence probability of each pronunciation of the target polyphonetic character in the determined issuing agency comprises:
[0018] Determining the states of a plurality of conditions in the attribute vector according to the business domain and the issuing organization to obtain a target attribute vector;
[0019] For at least one condition in the target attribute vector, determining the conditional probability of each pronunciation of the target polyphonetic character under the condition according to the state of the condition;
[0020] Determining the posterior probability of each pronunciation of the target polyphonetic character in the target attribute vector according to the determined conditional probability;
[0021] The target pronunciation of the target polyphone is determined based on the determined posterior probability.
[0022] According to an embodiment of the present disclosure, the digital human broadcasting method further includes a step of establishing a naive Bayes classifier, and the step of establishing includes:
[0023] Obtaining at least one corpus sample in a training set and a polyphonetic character sample in each of the corpus samples;
[0024] Determine the prior probability of each pronunciation of each polyphonetic character sample according to the occurrence probability of each pronunciation of each polyphonetic character sample in the training set;
[0025] The naive Bayes classifier is constructed according to the business field and publishing organization of each of the corpus samples and the prior probability of each pronunciation of each of the polyphonetic character samples.
[0026] According to an embodiment of the present disclosure, the preset pronunciation library is configured with at least one first phrase and the pronunciation of each character in each first phrase, and determining the pronunciation of at least one character in the to-be-broadcasted corpus according to the preset pronunciation library includes:
[0027] Determine a word with a to-be-determined pronunciation in the to-be-broadcasted corpus to obtain a target word;
[0028] Acquire a second phrase, wherein the second phrase includes a phrase in which the target word is located in the corpus to be broadcasted;
[0029] Determining the first phrase that matches the second phrase from a preset pronunciation library;
[0030] The pronunciation of the target character is determined according to the pronunciation of each character in the first phrase.
[0031] A second aspect of the present disclosure provides a digital human broadcasting device, comprising:
[0032] The acquisition module is used to obtain the corpus to be broadcast;
[0033] A first pronunciation determination module, used to determine the pronunciation of at least one word in the to-be-broadcasted corpus according to a preset pronunciation library;
[0034] A second pronunciation determination module is used to determine the target pronunciation of the target polyphonetic character by using a pronunciation prediction model according to the content of the corpus to be broadcast and the source of the corpus to be broadcast when the corpus to be broadcast contains at least one target polyphonetic character;
[0035] The target polyphonetic characters include polyphonetic characters for which no pronunciation is configured in the preset pronunciation library.
[0036] The third aspect of the present disclosure provides an electronic device, comprising: one or more processors; a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors execute the above-mentioned digital human broadcasting method.
[0037] The fourth aspect of the present disclosure further provides a computer-readable storage medium on which executable instructions are stored. When the instructions are executed by a processor, the processor executes the above-mentioned digital human broadcasting method.
[0038] The fifth aspect of the present disclosure also provides a computer program product, including a computer program, which implements the above-mentioned digital human broadcasting method when executed by a processor.
[0039] One or more of the above embodiments have the following advantages or beneficial effects:
[0040] The present disclosure can combine the preset pronunciation library and the pronunciation prediction model to perform polyphonic character recognition, thereby realizing voice broadcast. Among them, when there are polyphonic characters that are not configured in the preset pronunciation library in the corpus to be broadcast, the embodiment of the present disclosure can automatically predict the pronunciation of the polyphonic characters according to the content and source of the corpus to be broadcast through the pronunciation prediction model, thereby reducing the dependence on the preset pronunciation library in the polyphonic character recognition process, and eliminating the need for users to continuously enrich the preset pronunciation library. BRIEF DESCRIPTION OF THE DRAWINGS
[0041] The above contents and other purposes, features and advantages of the present disclosure will become more apparent through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, in which:
[0042] Figure 1 The application scenario diagram schematically shows the digital human broadcasting method, device, equipment, storage medium and program product according to the embodiment of the present disclosure;
[0043] Figure 2 A flowchart of a digital human broadcasting method according to an embodiment of the present disclosure is schematically shown;
[0044] Figure 3 A flowchart of determining a pronunciation through a preset pronunciation library according to an embodiment of the present disclosure is schematically shown;
[0045] Figure 4 A flowchart of predicting pronunciation by a pronunciation prediction model according to an embodiment of the present disclosure is schematically shown;
[0046] Figure 5 A flowchart of establishing a naive Bayes classifier according to an embodiment of the present disclosure is schematically shown;
[0047] Figure 6 The structural block diagram of the digital human broadcasting device according to the embodiment of the present disclosure is schematically shown;
[0048] Figure 7 A block diagram of an electronic device suitable for implementing the digital human broadcasting method according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION
[0049] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.
[0050] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.
[0051] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification, and should not be interpreted in an idealized or overly rigid manner.
[0052] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).
[0053] The present disclosure relates to the field of artificial intelligence technology. In the technical solutions of the present disclosure, the collection, storage, use, processing, transmission, provision, disclosure and application of user personal information involved all comply with the provisions of relevant laws and regulations, take necessary confidentiality measures, and do not violate public order and good morals.
[0054] The embodiment of the present disclosure provides a digital human broadcasting method, which relates to the field of artificial intelligence technology. The digital human broadcasting method includes: obtaining a corpus to be broadcast; determining the pronunciation of at least one character in the corpus to be broadcast according to a preset pronunciation library; when the corpus to be broadcast contains at least one target polyphonic character, determining the target pronunciation of the target polyphonic character according to the content in the corpus to be broadcast and the source of the corpus to be broadcast through a pronunciation prediction model; wherein the target polyphonic character includes a polyphonic character for which a pronunciation is not configured in the preset pronunciation library.
[0055] The digital human broadcasting method of the embodiment of the present disclosure can be combined with the preset pronunciation library and the pronunciation prediction model to perform polyphonic character recognition, thereby realizing voice broadcasting. Among them, when there are polyphonic characters that are not configured in the preset pronunciation library in the corpus to be broadcast, the embodiment of the present disclosure can automatically predict the pronunciation of the polyphonic characters according to the content and source of the corpus to be broadcast, through the pronunciation prediction model, thereby reducing the dependence on the preset pronunciation library in the polyphonic character recognition process, and eliminating the need for users to continuously enrich the preset pronunciation library.
[0056] Figure 1 The application scenario diagram of the digital human broadcasting method, device, equipment, storage medium and program product according to the embodiment of the present disclosure is schematically shown. Figure 1As shown, the application scenario 100 according to this embodiment may include terminal devices 101, 102, 103, a network 104 and a server 105. The network 104 is used to provide a medium for a communication link between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.
[0057] Users can use terminal devices 101, 102, 103 to interact with server 105 through network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only examples).
[0058] The terminal devices 101 , 102 , and 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, and desktop computers.
[0059] The server 105 may be a server that provides various services, such as a background management server (only an example) that provides support for websites browsed by users using the terminal devices 101, 102, and 103. The background management server may analyze and process the received data such as user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.
[0060] It should be noted that the digital human broadcasting method provided in the embodiment of the present disclosure can generally be executed by the server 105. Accordingly, the digital human broadcasting device provided in the embodiment of the present disclosure can generally be set in the server 105. The digital human broadcasting method provided in the embodiment of the present disclosure can also be executed by a server or server cluster that is different from the server 105 and can communicate with the terminal devices 101, 102, 103 and / or the server 105. Accordingly, the digital human broadcasting device provided in the embodiment of the present disclosure can also be set in a server or server cluster that is different from the server 105 and can communicate with the terminal devices 101, 102, 103 and / or the server 105.
[0061] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.
[0062] The following will be based on Figure 1 The scene described by Figure 2 to Figure 5 The digital human broadcasting method of the disclosed embodiment is described in detail.
[0063] Figure 2 The flowchart of the digital human broadcasting method according to the embodiment of the present disclosure is schematically shown. Figure 2 As shown, the digital human broadcasting method of this embodiment includes steps S210 to S230. It should be noted that although Figure 2 The steps in the figure are shown in sequence as indicated by the arrows, but the steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of the steps, and they can be executed in other orders. Moreover, at least some of the steps in the figure may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.
[0064] In step S210, the corpus to be broadcast is obtained.
[0065] In the embodiments of the present disclosure, the content of the corpus to be broadcast can be determined according to actual needs and is not limited here. For example, the corpus to be broadcast can include news information and the like.
[0066] In step S220, the pronunciation of at least one character in the to-be-announced corpus is determined according to a preset pronunciation library.
[0067] In the embodiments of the present disclosure, a preset pronunciation library can be established through a large number of corpus samples based on text categorization (TC) technology and / or machine learning technology. For example, based on certain features of the corpus samples, the category of the corpus samples can be determined according to their content under a predetermined set of category tags (labels), thereby assigning labels (such as pronunciation) to the corpus samples, and then, the preset pronunciation library is composed of corpus samples with labels.
[0068] In the embodiment of the present disclosure, the corpus to be broadcasted may be preprocessed. For example, operations such as word segmentation may be performed on the corpus to be broadcasted to obtain at least one target word. After that, a reference word matching the target word is searched from the preset pronunciation library, and the pronunciation of the target word is determined according to the pronunciation of the reference word. For example, the corpus to be broadcasted contains "Bank", and the preset pronunciation library is configured with "Bank". <hang2>”, then it can be from "bank" in the preset pronunciation library <hang2>”As a reference word, the pronunciation of the character "hang" in "bank" in the to-be-broadcast corpus is further determined.
[0069] In step S230, when the to-be-broadcast corpus contains at least one target polyphonic character, the target pronunciation of the target polyphonic character is determined according to the content and source of the to-be-broadcast corpus through a pronunciation prediction model. It should be understood that when there is no target polyphonic character in the to-be-broadcast corpus, that is, the polyphonic characters in the to-be-broadcast corpus have corresponding configurations in the preset pronunciation library, at this time, the pronunciation of the polyphonic character can be directly determined through step S220 without executing step S230.
[0070] Exemplarily, in the embodiments of the present disclosure, information such as the business field of the to-be-broadcast corpus can be determined according to the content and source of the to-be-broadcast corpus. Furthermore, according to the common pronunciation of the target polyphonic character in this field, the most likely pronunciation of the target polyphonic character, that is, the target pronunciation, can be determined. The pronunciation prediction model can include a classifier capable of performing pronunciation classification. For example, the pronunciation prediction model can include a classifier based on Naive Bayes, a classifier based on the decision tree method, a classifier based on the K-nearest neighbor method, a classifier based on the center vector method, and a classifier based on the support vector machine, etc.
[0071] Optionally, in the embodiments of the present disclosure, the pronunciation prediction model includes a classifier based on Naive Bayes.
[0072] Optionally, in the embodiments of the present disclosure, when constructing the Naive Bayes classifier, the attribute vector of the Naive Bayes classifier can be constructed according to the content and source of the corpus samples in the training set. In step S230, the posterior probability of each pronunciation of the target polyphonic character in the attribute vector is calculated, and then the target pronunciation of the polyphonic character is determined. The specific details will be introduced in detail below and will not be elaborated here.
[0073] After step 230, the digital human broadcast method of the embodiments of the present disclosure may further include step S240 and step S250.
[0074] In step S240, the target broadcast corpus is formed by at least one character whose pronunciation has been determined.
[0075] In step S250, the broadcast is performed according to the target broadcast corpus.
[0076] The digital human broadcasting method of the embodiment of the present disclosure can be combined with the preset pronunciation library and the pronunciation prediction model to perform polyphonic character recognition, thereby realizing voice broadcasting. Among them, when there are polyphonic characters that are not configured in the preset pronunciation library in the corpus to be broadcast, the embodiment of the present disclosure can automatically predict the pronunciation of the polyphonic characters according to the content and source of the corpus to be broadcast, through the pronunciation prediction model, thereby reducing the dependence on the preset pronunciation library in the polyphonic character recognition process, and eliminating the need for users to continuously enrich the preset pronunciation library.
[0077] Combine the following Figures 1 to 5 The digital human broadcasting method of the embodiment of the present disclosure is further described.
[0078] In some specific embodiments, the preset pronunciation library is configured with at least one first phrase and the pronunciation of each character in each first phrase. Figure 3 A flowchart of determining a pronunciation through a preset pronunciation library according to an embodiment of the present disclosure is schematically shown. Figure 3 As shown, step S220 includes steps S221 to S224.
[0079] In step S221, a character whose pronunciation is to be determined in the corpus to be broadcast is determined to obtain a target character.
[0080] In step S222, a second phrase is obtained, where the second phrase includes the phrase where the target word is located in the corpus to be broadcasted.
[0081] In step S223, a first phrase matching the second phrase is determined from the preset pronunciation library.
[0082] In step S224, the pronunciation of the target character is determined according to the pronunciation of each character in the first phrase.
[0083] As an example, the preset pronunciation library stores a large number of phrases and the pronunciation of each word in the phrase. For example, the "line break" can be formatted as "break". <fen1>OK <hang2>” and then store it in a preset pronunciation library. When the digital human encounters the word "line break", it can determine "fen" from the preset pronunciation library. <fen1>OK <hang2>”, and then read out the correct pronunciation.
[0084] The following takes a news report from the official website of a certain bank as the speech corpus to be broadcasted as an example to illustrate the digital human speech method of the present disclosure embodiment.
[0085] The content of the news report is as follows:
[0086] Title: Our Bank Successfully Launched the Notional Fund Pool Business in Macau
[0087] Text: To fully serve the construction of the Guangdong-Hong Kong-Macao Greater Bay Area and promote the improvement of regional competitiveness, recently, the Head Office's Settlement and Cash Management Department, with the strong support of the Financial Technology Department, successfully launched the notional fund pool business in Macau and completed the first business transaction, meeting the cross-currency, multi-account combined interest calculation and payment needs of customers. Since the launch of this business, it has brought approximately HK$1 billion in new deposits and HK$70 billion in new settlement volume to ICBC Macau. At the beginning of this year, our bank learned that a certain customer in Macau faced problems such as multi-currency management and frequent foreign exchange purchases and settlements. The Settlement and Cash Management Department, in cooperation with ICBC Macau, quickly took action and had multiple rounds of communication with the customer. Based on the existing notional fund pool framework of ICBC Asia and the Singapore Branch, and according to the customer's personalized needs and regional characteristics, a notional fund pool solution for Macau was formulated. Through the cross-border fund pool, funds from other countries were automatically aggregated into the offshore account of ICBC Macau, and then, relying on the notional fund pool, the overall operation of multi-currency funds was realized, the multi-account multi-currency funds were netted for interest calculation, the foreign exchange purchase cost was reduced, and the fund sharing efficiency was improved. After the plan was recognized by the customer, our bank immediately launched system R & D. With the strong support of the Financial Technology Department, the first business transaction for the customer has recently been successfully completed, covering six major currencies: RMB, US dollar, euro, British pound, Japanese yen, and Hong Kong dollar. Subsequently, the Settlement and Cash Management Department will closely follow market changes, strive to improve the market competitiveness and customer applicability of global cash management products, give play to the advantages of the interconnected settlement of domestic and overseas finance, and help our bank transform and develop in the new dual-circulation development pattern.
[0088] In this example, taking the character "行 (háng)" as an example, there are multiple first phrases in the preset pronunciation library, and the multiple first phrases include "银 (yín)" <yin2>OK <hang2>,point <fen1>OK <hang2>,total <zong3>OK <hang2>,OK <hang2>long <zhang3>,OK <hang2>Industry <ye4>,OK <xing2>Drive <shi3>,one <yi4>OK <xing2>people <ren2>,Enter <jin4>OK <xing2>,OK <xing2>move <dong4>,OK <xing2>? <ma1>,Can <ke3>OK <xing2>”。
[0089] In the above-mentioned speech corpus to be broadcasted, the second phrases of "行" include "我行", "行动", "分行", "进行", and "总行". In the preset pronunciation library, there are first phrases corresponding to "行动", "分行", "进行", and "总行" in the above-mentioned second phrases. Therefore, the pronunciation of the character "行" in "行动", "分行", "进行", and "总行" can be determined through the preset speech library.
[0090] For "我行" in the above-mentioned speech corpus to be broadcasted, the preset pronunciation library does not have this phrase configured. At this time, the pronunciation of the character "行" in the phrase "我行" can be predicted through step S230.
[0091] Figure 4 Schematically shows a flowchart of predicting pronunciation through a pronunciation prediction model according to an embodiment of the present disclosure, as Figure 4 shown. In some specific embodiments, step S230 includes steps S231 to S233.
[0092] In step S231, according to the content of the speech corpus to be broadcasted, determine the business field of the speech corpus to be broadcasted.
[0093] In step S232, according to the source of the speech corpus to be broadcasted, determine the publishing agency of the speech corpus to be broadcasted.
[0094] In step S233, according to the occurrence probability of each pronunciation of the target polyphonic character in the determined business field, and the occurrence probability of each pronunciation of the target polyphonic character in the determined publishing agency, determine the target pronunciation of the target polyphonic character.
[0095] The content of the speech corpus to be broadcasted may include some or all of the text information in the speech corpus to be broadcasted. Optionally, when the content of the speech corpus to be broadcasted includes some text information in the speech corpus to be broadcasted, the content of the speech corpus to be broadcasted may include keywords in the title of the speech corpus to be broadcasted or keywords in the text, etc. That is, in some specific embodiments, step S231 includes step S2311.
[0096] In step S2311, according to the text of the speech corpus to be broadcasted and / or the title of the speech corpus to be predicted, determine the business field of the speech corpus to be broadcasted.
[0097] Optionally, the speech corpus to be broadcasted may be information in a certain section of a news report. In some specific embodiments, the business field of the speech corpus to be broadcasted may be determined according to the section of the speech corpus to be broadcasted in the news report.
[0098] In some specific embodiments, the source of the speech corpus to be broadcasted may include the address information of the speech corpus to be broadcasted, such as a website address, etc. That is, in some specific embodiments, step S232 includes step S2321.
[0099] In step S2321, according to the website of the corpus to be broadcasted, determine the publishing agency of the corpus to be broadcasted.
[0100] The business field and the publishing agency can be specifically configured according to actual needs. For example, the business field can include the financial field, the technology field, the life field, and so on; the publishing agency can include financial institutions, news agencies, government agencies, and so on.
[0101] In the embodiments of the present disclosure, the most likely pronunciation of the target polyphonic character, that is, the target pronunciation, can be determined according to the common pronunciation of the target polyphonic character in the information previously published by the business field and the publishing agency in the corpus to be broadcasted.
[0102] For example, the target polyphonic character is "我行 (wǒ xíng)". Through step S231, according to the content in the corpus to be broadcasted, it is determined that the business field of the corpus to be broadcasted is the financial field; through step S232, according to the source in the corpus to be broadcasted, it is determined that the publishing agency of the corpus to be broadcasted is a banking institution; through step S233, it is determined that in the financial field and the information previously published by the banking institution, the pronunciation probability of "行 (xíng)" in "我行 (wǒ xíng)" is relatively high. Furthermore, it can be determined that the pronunciation of "行 (xíng)" in "我行 (wǒ xíng)" in the corpus to be broadcasted is "hang2".
[0103] To better explain the prediction process of the target polyphonic character in the embodiments of the present disclosure, the following first introduces the establishment process of the pronunciation prediction model in the embodiments of the present disclosure.
[0104] In some specific embodiments, the pronunciation prediction model includes a Naive Bayes classifier, and the digital human broadcast method further includes the establishment steps of the Naive Bayes classifier. Figure 5 Schematically shows a flowchart of establishing a Naive Bayes classifier according to an embodiment of the present disclosure, as Figure 5 shown, the establishment steps include step S310 to step S330.
[0105] In step S310, obtain at least one corpus sample in the training set and the polyphonic character samples in each corpus sample.
[0106] In step S320, according to the occurrence probability of each pronunciation of each polyphonic character sample in the training set, determine the prior probability of each pronunciation of each polyphonic character sample.
[0107] In the embodiments of the present disclosure, the training set may be from previously published news reports. The prior probability refers to the prior knowledge of each pronunciation of each polyphonic character sample obtained from the training set. For example, for the polyphonic character sample "行", there are two possible hypotheses: that is, the pronunciation of "行" is "xing2", or the pronunciation of "行" is "hang2". Suppose in the training set, "行" appears 14 times, among which, the pronunciation is "hang2" 9 times and the pronunciation is "xing2" 5 times. Then, the prior probability Pa(hang2) of the pronunciation of "行" being "hang2" = 9 / 14, and the prior probability Pa(xing2) of the pronunciation of "行" being "xing2" = 5 / 14.
[0108] It should be noted that in the embodiments of the present disclosure, the source of the training set is only for illustrative purposes. Depending on different actual needs, the training set may also come from other channels, and no specific limitation is made here.
[0109] In step S330, a Naive Bayes classifier is constructed according to the business field and publishing agency of each corpus sample and the prior probability of each pronunciation of each polyphonic character sample.
[0110] In some specific embodiments, step S330 includes step S331.
[0111] In step S331, according to the business field and publishing agency of each corpus sample, the attribute vector in the Naive Bayes classifier is determined.
[0112] In some specific embodiments, the attribute vector of the Naive Bayes classifier includes multiple conditions, and the business field (or publishing agency) in the corpus sample can be used as one of the conditions.
[0113] Optionally, in the embodiments of the present disclosure, the business field (or publishing agency) can be further segmented, and the segmented business field (or publishing agency) is used as one of the conditions in the attribute vector. For example, the publishing agency includes banking institutions and government agencies, and the banking institutions and government agencies can be used as one of the conditions in the attribute vector respectively. The business field includes the financial field and the life field, and the financial field and the life field can be used as one of the conditions in the attribute vector respectively.
[0114] Optionally, in the embodiments of the present disclosure, the process of determining the business domain (or the publishing agency) can be subdivided, and the subdivided process can be used as a condition in the attribute vector. For example, the business domain of the corpus sample can be determined according to the section of the news report where the corpus sample is located. Therefore, whether the section of the news report where the corpus sample is located is a certain section can be used as a condition in the attribute vector; for another example, the business domain of the corpus sample can be determined according to the keywords in the title of the corpus sample. Therefore, whether the title of the corpus sample includes the keywords in a certain business domain can be used as a condition in the attribute vector; for another example, the publishing agency of the corpus sample can be determined according to the website address of the corpus sample. Therefore, whether the website address of the corpus sample is the website address of a certain publishing agency can be used as a condition in the attribute vector
[0115] In some specific embodiments, step S330 further includes the conditional probability of each pronunciation of each polyphonic character sample under each condition in the attribute vector.
[0116] For example, the attribute vector includes 4 conditions, which are respectively: (1) whether the news section to which the corpus to be broadcast belongs in the news report is the financial section; (2) whether the title of the corpus to be broadcast includes the keywords in the financial business domain; (3) whether the website address of the corpus to be broadcast is the website address of a banking institution; (4) whether the website address of the corpus to be broadcast is the website address of a government agency.
[0117] As an example, in the above training set, for condition (1), when the news section to which the corpus to be broadcast belongs in the news report is the financial section, the sample with the pronunciation "hang2" of the character "行" appears 5 times. According to the prior probability Pa(hang2), the conditional probability Pb(condition 1 = yes|hang2) = 5 / 9 can be obtained; the conditional probability Pb(condition 1 = no|hang2) = 4 / 9. Among them, P(Y|X) represents the probability that Y occurs on the premise that X occurs.
[0118] When the news section to which the corpus to be broadcast belongs in the news report is the financial section, the sample with the pronunciation "xing2" of the character "行" appears 2 times. According to the prior probability Pa(xing2), the conditional probability Pb(condition 1 = yes|xing2) = 2 / 5 can be obtained, and the conditional probability Pb(condition 1 = no|hang2) = 3 / 5.
[0119] In some specific embodiments, the situation where the polyphonic characters in the corpus to be broadcast can find corresponding reference words in the preset pronunciation library can also be taken into consideration. Specifically, conditions can be added to the attribute vector. For example, add condition (5): whether the phrase to which the polyphonic character belongs (i.e., the second phrase mentioned above) is any one of "bank, branch, head office, president, industry". Each of these phrases has a corresponding first phrase matched in the preset pronunciation library, and the pronunciation of "行" is "hang2". Therefore, when condition (5) is satisfied, that is, the situation where the pronunciation of "行" can be directly determined as "hang2" according to the preset pronunciation library through step S220. For another example, add condition (6): whether the phrase to which the polyphonic character belongs is any one of "drive, a group of people, conduct, act, okay, feasible". The pronunciation of "行" is "xing2". Therefore, when condition (6) is satisfied, that is, the situation where the pronunciation of "行" can be directly determined as "xing2" according to the preset pronunciation library through step S220.
[0120] After introducing condition (5) and condition (6), assuming that in the above training set, there are "bank" and "drive", at this time, the conditional probability of each pronunciation of the polyphonic character sample will change. Specifically, after introducing condition (5) and condition (6), the conditional probability Pb(condition 1 = no|hang2) = 3 / 9, and the conditional probability Pb(condition 1 = no|hang2) = 3 / 5.
[0121] Table 1
[0122]
[0123] Assume that the training set includes 14 corpus samples, and the 14 corpus samples are shown in Table 1. Then, by analogy in the above manner, it can be obtained that:
[0124] Pb(condition 2 = yes|hang2) = 6 / 9, Pb(condition 2 = no|xing2) = 2 / 5, Pb(condition 3 = yes|hang2) = 6 / 9, Pb(condition 3 = no|xing2) = 2 / 5, Pb(condition 4 = yes|hang2) = 5 / 9, Pb(condition 4 = no|xing2) = 2 / 5, Pb(condition 2 = no|hang2) = 2 / 9, Pb(condition 2 = yes|xing2) = 2 / 5, Pb(condition 3 = no|hang2) = 2 / 9, Pb(condition 3 = yes|xing2) = 2 / 5, Pb(condition 4 = no|hang2) = 3 / 9, Pb(condition 4 = yes|xing2) = 2 / 5.
[0125] Next, return to step S233 to illustrate the prediction process of the "行" in the target polyphonic character "我行" in the embodiment of the present disclosure.
[0126] In some specific embodiments, step S233 includes steps S2331 to S2334.
[0127] In step S2331, according to the business field and the publishing agency, determine the states of multiple conditions in the attribute vector to obtain the target attribute vector.
[0128] Exemplarily, assume that the corpus to be broadcast belongs to the financial sector, its theme includes keywords corresponding to the financial field, the corpus to be broadcast comes from a bank website rather than a government website. Then, the attribute vector T = (condition 1 = yes, condition 2 = yes, condition 3 = yes, condition 4 = no).
[0129] In step S2332, for at least one condition in the target attribute vector, according to the state of this condition, determine the conditional probability of each pronunciation of the target polyphone under this condition.
[0130] Since the attribute vector T = (condition 1 = yes, condition 2 = yes, condition 3 = yes, condition 4 = no), four conditional probabilities for the pronunciation "hang2" of "行" can be determined, which are respectively the conditional probability Pb(condition 1 = yes|hang2) = 5 / 9, the conditional probability Pb(condition 2 = yes|hang2) = 6 / 9, the conditional probability Pb(condition 3 = yes|hang2) = 6 / 9, and the conditional probability Pb(condition 4 = no|hang2) = 3 / 9. Similarly, four conditional probabilities for the pronunciation "xing2" of "行" can also be determined, which are respectively the conditional probability Pb(condition 1 = yes|xing2) = 2 / 5, Pb(condition 2 = yes|xing2) = 2 / 5, Pb(condition 3 = yes|xing2) = 2 / 5, and Pb(condition 4 = no|xing2) = 2 / 5.
[0131] In step S2333, according to the determined conditional probabilities, determine the posterior probability of each pronunciation of the target polyphone in the target attribute vector.
[0132] In the embodiments of the present disclosure, the posterior probability Pc(hang2|T) of the pronunciation "hang2" of "行" = Pb(condition 1 = yes|hang2) * Pb(condition 2 = yes|hang2) * Pb(condition 3 = yes|hang2) * Pb(condition 4 = no|hang2) = 5 / 9 * 6 / 9 * 6 / 9 * 3 / 9 = 0.0823, and the posterior probability Pc(xing2|T) of the pronunciation "xing2" of "行" = Pb(condition 1 = yes|xing2) * Pb(condition 2 = yes|xing2) * Pb(condition 3 = yes|xing2) * Pb(condition 4 = no|xing2) = 2 / 5 * 2 / 5 * 2 / 5 * 2 / 5 = 0.0256.
[0133] In step S2334, based on the determined posterior probability, determine the target pronunciation of the target polyphonic character.
[0134] In an embodiment of the present disclosure, Pc(hang2|T) > Pc(xing2|T). Therefore, in the corpus to be broadcasted, the probability that the pronunciation of "行" in "我行" is "hang2" is relatively high. From this, it can be inferred that the pronunciation of "行" in "我行" is "hang 2", that is, the target pronunciation is "hang".
[0135] Based on the above digital human broadcasting method, the present disclosure also provides a digital human broadcasting device. The following will be combined with Figure 6 Describe this device in detail.
[0136] Figure 6 Schematically shows a structural block diagram of a digital human broadcasting device according to an embodiment of the present disclosure. As Figure 6 shown, the digital human broadcasting device 600 of this embodiment includes an acquisition module 610, a first pronunciation determination module 620, and a second pronunciation determination module 630.
[0137] The acquisition module 610 is used to acquire the corpus to be broadcasted. In one embodiment, the acquisition module 610 can be used to execute step S210 described above, which will not be elaborated here.
[0138] The first pronunciation determination module 620 is used to determine the pronunciation of at least one character in the corpus to be broadcasted according to a preset pronunciation library. In one embodiment, the first pronunciation determination module 620 can be used to execute step S220 described above, which will not be elaborated here.
[0139] The second pronunciation determination module 630 is used to, when the corpus to be broadcasted contains at least one target polyphonic character, determine the target pronunciation of the target polyphonic character through a pronunciation prediction model according to the content and source of the corpus to be broadcasted. In one embodiment, the second pronunciation determination module 630 can be used to execute step S230 described above, which will not be elaborated here.
[0140] Among them, the target polyphonic characters include polyphonic characters whose pronunciations are not configured in the preset pronunciation library.
[0141] Using the digital human broadcasting device of the embodiment of the present disclosure, it can combine a preset pronunciation library and a pronunciation prediction model for polyphonic character recognition, and then realize voice broadcasting. Among them, when there are polyphonic characters in the corpus to be broadcasted that are not configured in the preset pronunciation library, the embodiment of the present disclosure can automatically predict the pronunciation of the polyphonic characters according to the content and source of the corpus to be broadcasted through the pronunciation prediction model, thereby reducing the dependence on the preset pronunciation library in the process of polyphonic character recognition and eliminating the process that users need to continuously enrich the preset pronunciation library.
[0142] According to an embodiment of the present disclosure, any multiple modules in the acquisition module 610, the first pronunciation determination module 620, and the second pronunciation determination module 630 can be combined in one module for implementation, or any one of the modules can be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules can be combined with at least part of the functions of other modules and implemented in one module. According to an embodiment of the present disclosure, at least one of the acquisition module 610, the first pronunciation determination module 620, and the second pronunciation determination module 630 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application-specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or implemented in any one of the three implementation modes of software, hardware, and firmware, or in any appropriate combination of any of them. Alternatively, at least one of the acquisition module 610, the first pronunciation determination module 620, and the second pronunciation determination module 630 can be at least partially implemented as a computer program module, and when the computer program module is run, the corresponding function can be performed.
[0143] Figure 7 The block diagram of an electronic device suitable for implementing the digital human broadcasting method according to an embodiment of the present disclosure is schematically shown. Figure 7 As shown, the electronic device 700 according to an embodiment of the present disclosure includes a processor 701, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 702 or a program loaded from a storage part 708 into a random access memory (RAM) 703. The processor 701 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or a related chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 701 may also include an onboard memory for caching purposes. The processor 701 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present disclosure.
[0144] In RAM 703, various programs and data required for the operation of electronic device 700 are stored. Processor 701, ROM 702 and RAM 703 are connected to each other via bus 704. Processor 701 performs various operations of the method flow according to the embodiment of the present disclosure by executing the program in ROM 702 and / or RAM 703. It should be noted that the program can also be stored in one or more memories other than ROM 702 and RAM 703. Processor 701 can also perform various operations of the method flow according to the embodiment of the present disclosure by executing the program stored in the one or more memories.
[0145] According to an embodiment of the present disclosure, the electronic device 700 may further include an input / output (I / O) interface 705, which is also connected to the bus 704. The electronic device 700 may further include one or more of the following components connected to the I / O interface 705: an input portion 706 including a keyboard, a mouse, etc.; an output portion 707 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage portion 708 including a hard disk, etc.; and a communication portion 709 including a network interface card such as a LAN card, a modem, etc. The communication portion 709 performs communication processing via a network such as the Internet. A drive 710 is also connected to the I / O interface 705 as needed. A removable medium 711, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 710 as needed, so that a computer program read therefrom is installed into the storage portion 708 as needed.
[0146] The present disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or may exist independently without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed, the digital human broadcasting method according to the embodiment of the present disclosure is implemented.
[0147] According to an embodiment of the present disclosure, the computer-readable storage medium may be a non-volatile computer-readable storage medium, for example, may include but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, an apparatus or a device. For example, according to an embodiment of the present disclosure, a computer-readable storage medium may include the ROM 702 and / or RAM 703 described above and / or one or more memories other than ROM 702 and RAM 703.
[0148] The embodiment of the present disclosure also includes a computer program product, which includes a computer program, and the computer program contains program code for executing the method shown in the flowchart. When the computer program product is run in a computer system, the program code is used to enable the computer system to implement the digital human broadcasting method provided by the embodiment of the present disclosure.
[0149] The above functions defined in the system / device of the embodiment of the present disclosure are performed when the computer program is executed by the processor 701. According to the embodiment of the present disclosure, the system, device, module, unit, etc. described above can be implemented by a computer program module.
[0150] In one embodiment, the computer program may rely on tangible storage media such as optical storage devices, magnetic storage devices, etc. In another embodiment, the computer program may also be transmitted and distributed in the form of signals on a network medium, and downloaded and installed through the communication part 709, and / or installed from the removable medium 711. The program code contained in the computer program may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.
[0151] In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 709, and / or installed from the removable medium 711. When the computer program is executed by the processor 701, the above functions defined in the system of the embodiment of the present disclosure are performed. According to the embodiment of the present disclosure, the system, device, means, module, unit, etc. described above can be implemented by a computer program module.
[0152] According to an embodiment of the present disclosure, the program code for executing the computer program provided by the embodiment of the present disclosure can be written in any combination of one or more programming languages. Specifically, these computing programs can be implemented using high-level process and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages include, but are not limited to, Java, C++, python, "C" language or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on the remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (e.g., using an Internet service provider to connect through the Internet).
[0153] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.
[0154] It will be appreciated by those skilled in the art that the features described in the various embodiments and / or claims of the present disclosure may be combined and / or combined in a variety of ways, even if such combinations and / or combinations are not explicitly described in the present disclosure. In particular, the features described in the various embodiments and / or claims of the present disclosure may be combined and / or combined in a variety of ways without departing from the spirit and teachings of the present disclosure. All of these combinations and / or combinations fall within the scope of the present disclosure.
[0155] The embodiments of the present disclosure are described above. However, these embodiments are only for illustrative purposes and are not intended to limit the scope of the present disclosure. Although the embodiments are described above separately, this does not mean that the measures in the various embodiments cannot be used in combination to advantage. The scope of the present disclosure is defined by the attached claims and their equivalents. Without departing from the scope of the present disclosure, those skilled in the art may make a variety of substitutions and modifications, which should all fall within the scope of the present disclosure.
Claims
1. A digital human broadcasting method, characterized in that: include: Get the corpus to be broadcast; Determine the pronunciation of at least one character in the to-be-broadcasted corpus according to a preset pronunciation library; When the corpus to be broadcast contains at least one target polyphone, a target pronunciation of the target polyphone is determined by a pronunciation prediction model according to the content of the corpus to be broadcast and the source of the corpus to be broadcast; Wherein, the target polyphonetic characters include polyphonetic characters for which no pronunciation is configured in the preset pronunciation library; Determining the target pronunciation of the target polyphonetic character according to the content of the to-be-broadcasted corpus and the source of the to-be-broadcasted corpus includes: Determine the business field of the corpus to be broadcast according to the content of the corpus to be broadcast; determine the publishing agency of the corpus to be broadcast according to the source of the corpus to be broadcast; determine the target pronunciation of the target polyphone according to the probability of occurrence of each pronunciation of the target polyphone in the determined business field and the probability of occurrence of each pronunciation of the target polyphone in the determined publishing agency; The pronunciation prediction model includes a naive Bayes classifier, and the naive Bayes classifier includes an attribute vector composed of multiple conditions; Determining the target pronunciation of the target polyphonetic character according to the occurrence probability of each pronunciation of the target polyphonetic character in the determined business field and the occurrence probability of each pronunciation of the target polyphonetic character in the determined issuing agency comprises: According to the business field and the issuing organization, the states of multiple conditions in the attribute vector are determined to obtain a target attribute vector; for at least one condition in the target attribute vector, the conditional probability of each pronunciation of the target polyphonetic character under the condition is determined according to the state of the condition; according to the determined conditional probability, the posterior probability of each pronunciation of the target polyphonetic character in the target attribute vector is determined; according to the determined posterior probability, the target pronunciation of the target polyphonetic character is determined.
2. The digital human broadcasting method according to claim 1, characterized in that: Determining the business domain of the corpus to be broadcast according to the content of the corpus to be broadcast includes: determining the business domain of the corpus to be broadcast according to the text of the corpus to be broadcast and / or the title of the expected report to be broadcast; Determining the publishing agency of the corpus to be broadcast according to the source of the corpus to be broadcast includes: determining the publishing agency of the corpus to be broadcast according to the website of the corpus to be broadcast.
3. The digital human broadcasting method according to claim 1, characterized in that: The digital human broadcasting method further comprises a step of establishing a naive Bayes classifier, and the step of establishing comprises: Obtaining at least one corpus sample in a training set and a polyphonetic character sample in each of the corpus samples; Determine the prior probability of each pronunciation of each polyphonetic character sample according to the occurrence probability of each pronunciation of each polyphonetic character sample in the training set; The naive Bayes classifier is constructed according to the business field and publishing organization of each of the corpus samples and the prior probability of each pronunciation of each of the polyphonetic character samples.
4. The digital human broadcasting method according to any one of claims 1 to 3, characterized in that: The preset pronunciation library is configured with at least one first phrase and the pronunciation of each character in each first phrase, and determining the pronunciation of at least one character in the to-be-broadcasted corpus according to the preset pronunciation library comprises: Determine a word with a to-be-determined pronunciation in the to-be-broadcasted corpus to obtain a target word; Acquire a second phrase, wherein the second phrase includes a phrase in which the target word is located in the corpus to be broadcasted; Determining the first phrase that matches the second phrase from a preset pronunciation library; The pronunciation of the target character is determined according to the pronunciation of each character in the first phrase.
5. A digital human broadcasting device, characterized in that: include: The acquisition module is used to obtain the corpus to be broadcast; A first pronunciation determination module, used to determine the pronunciation of at least one word in the to-be-broadcasted corpus according to a preset pronunciation library; A second pronunciation determination module is used to determine the target pronunciation of the target polyphonetic character by using a pronunciation prediction model according to the content of the corpus to be broadcast and the source of the corpus to be broadcast when the corpus to be broadcast contains at least one target polyphonetic character; Wherein, the target polyphone includes polyphones that are not configured with pronunciation in the preset pronunciation library; the pronunciation prediction model includes a naive Bayes classifier, and the naive Bayes classifier includes an attribute vector composed of multiple conditions; The second pronunciation determination module is further used to determine the business field of the corpus to be broadcast according to the content of the corpus to be broadcast; determine the publishing agency of the corpus to be broadcast according to the source of the corpus to be broadcast; determine the target pronunciation of the target polyphone according to the occurrence probability of each pronunciation of the target polyphone in the determined business field and the occurrence probability of each pronunciation of the target polyphone in the determined publishing agency; The second pronunciation determination module is also used to determine the status of multiple conditions in the attribute vector according to the business field and the issuing organization to obtain a target attribute vector; for at least one condition in the target attribute vector, determine the conditional probability of each pronunciation of the target polyphonetic character under the condition according to the status of the condition; determine the posterior probability of each pronunciation of the target polyphonetic character in the target attribute vector according to the determined conditional probability; and determine the target pronunciation of the target polyphonetic character according to the determined posterior probability.
6. An electronic device, characterized in that: include: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors execute the digital human broadcasting method according to any one of claims 1 to 4.
7. A computer-readable storage medium, characterized in that: Executable instructions are stored thereon, and when the instructions are executed by the processor, the processor executes the digital human broadcasting method according to any one of claims 1 to 4.
8. A computer program product, characterized in that It comprises a computer program, which, when executed by a processor, implements the digital human broadcasting method according to any one of claims 1 to 4.
