Korean text pronunciation prediction, speech synthesis method, related equipment and program products

Abstract

The present application discloses a Korean text pronunciation prediction method, speech synthesis method, apparatus, related equipment and computer program product. For each word in the target Korean text, it is matched with a pronunciation dictionary. If both the word face and the pronunciation rules are matched successfully, the pronunciation of the dictionary that is successfully matched is used as the pronunciation of the current word. If the word face is matched successfully but the pronunciation rule is matched unsuccessfully, the target pronunciation information under the word face that is successfully matched in the pronunciation dictionary is obtained. The pronunciation prediction model is called to predict the candidate pronunciation information of the current word, and the initial phoneme and / or final phoneme in the candidate pronunciation information is used to replace the initial phoneme and / or final phoneme in the target pronunciation information to obtain the pronunciation of the current word. By integrating different pronunciation prediction methods, the pronunciation prediction accuracy of Korean words is improved, especially the prediction accuracy of Korean for the sound change part is improved, the speech synthesis effect is improved, and the difficulty of constructing a pronunciation dictionary is reduced.

Description

Technical Field

[0001] The present application relates to the technical field of pronunciation prediction, and more specifically, to a Korean text pronunciation prediction method, speech synthesis method, apparatus, related equipment, and computer program product. Background Art

[0002] Text-to-speech (TTS) is a human-computer interaction technology that automatically converts written text into natural speech output. The implementation of TTS technology typically involves the following steps: 1. Text analysis; 2. Phoneme conversion; 3. Speech synthesis; and 4. Acoustic post-processing. The second step, phoneme conversion, also known as G2P (Grapheme-to-Phoneme), refers to the process of converting a textual representation (grapheme) into its corresponding phonetic representation (phoneme). It plays a crucial role in the quality of speech synthesis.

[0003] The development of G2P technology is limited to some extent by the complex relationship between speech and text in different languages. A traditional G2P approach relies on a pronunciation dictionary, leveraging linguistic knowledge and rules to construct a pronunciation dictionary. G2P conversion is then performed through rule matching or table lookup. This approach is suitable for languages ​​with relatively standardized phonetic spelling rules, such as English and French. However, Korean has a large number of words, resulting in a large number of out-of-vocabulary (OOV) words (words that do not appear in dictionaries or corpora). Manually collecting all of these words would be prohibitively expensive. Summary of the Invention

[0004] In view of the above problems, this application is proposed to provide a Korean text pronunciation prediction method, speech synthesis method, apparatus, related equipment, and computer program product. These methods integrate multiple different pronunciation prediction methods to improve the pronunciation prediction accuracy of Korean words, while reducing the difficulty of constructing pronunciation dictionaries, lowering costs, and improving speech synthesis performance. The specific solutions are as follows:

[0005] In a first aspect, a Korean text pronunciation prediction method is provided, comprising:

[0006] Perform word segmentation on the target Korean text to obtain several words;

[0007] For each word, the current word is matched with a pre-configured pronunciation dictionary. If both the word face and the pronunciation rules of the current word can be successfully matched, the pronunciation of the dictionary that successfully matches is used as the pronunciation of the current word. The pronunciation dictionary contains the word faces of several words and the pronunciation information corresponding to each word under at least one pronunciation rule.

[0008] If the word-face of the current word is successfully matched but the pronunciation rule is not successfully matched, the target pronunciation information corresponding to any pronunciation rule under the word-face that successfully matched in the pronunciation dictionary is obtained, and the candidate pronunciation information of the current word is predicted by a configured pronunciation prediction model using the current word and its contextual information in the target Korean text, wherein the pronunciation prediction model is trained using sample words and the contextual information of the sample words in the training Korean text as training samples and the pronunciation of the sample words as sample labels;

[0009] The first phoneme and / or the last phoneme in the candidate pronunciation information of the current word are used to replace the corresponding first phoneme and / or the last phoneme in the target pronunciation information, and the replaced pronunciation information is obtained as the pronunciation of the current word.

[0010] In one possible design, in another implementation of the first aspect of the embodiments of the present application, before matching the current word with a preconfigured pronunciation dictionary, the method further includes:

[0011] Convert each word in the target Korean text into Korean letters jamo;

[0012] The pronunciation rules in the pronunciation dictionary include pronunciation rules set according to the jamo of the context words of the word, then, the process of matching the current word with the preconfigured pronunciation dictionary includes:

[0013] The current word is matched word-wise in the pronunciation dictionary. If the word-wise match is successful, the jamo of the context word of the current word is further matched with the pronunciation rules under the word-wise match in the pronunciation dictionary to determine whether the pronunciation rules are matched successfully.

[0014] In one possible design, in another implementation of the first aspect of the embodiments of the present application, the pronunciation rules in the pronunciation dictionary further include: according to the pronunciation rules set according to the part of speech and / or prosodic information of the context words of the word, the process of matching the current word with the preconfigured pronunciation dictionary also includes:

[0015] According to the part of speech and / or prosody information of the context words of the current word, the pronunciation rules under the words that have been successfully matched in the pronunciation dictionary are matched to determine whether the pronunciation rules are successfully matched.

[0016] In one possible design, in another implementation of the first aspect of the embodiments of the present application, a process of predicting candidate pronunciation information of the current word using a configured pronunciation prediction model using the current word and its context information in the target Korean text includes:

[0017] The jamo of the current word and the jamo of the context words of the current word in the target Korean text are fed into the configured pronunciation prediction model to obtain candidate pronunciation information of the current word output by the model.

[0018] In one possible design, in another implementation of the first aspect of the embodiments of the present application, a process of predicting candidate pronunciation information of the current word using a configured pronunciation prediction model using the current word and its context information in the target Korean text includes:

[0019] The jamo of the current word, the jamo of the context words of the current word in the target Korean text, and the prosodic information of the current word and the context words are fed into the configured pronunciation prediction model to obtain the candidate pronunciation information of the current word output by the model.

[0020] In one possible design, in another implementation of the first aspect of the embodiments of the present application, the method further includes:

[0021] If the word face of the current word fails to be matched successfully, the pronunciation information of the current word is predicted by the pronunciation prediction model using the current word and its context information in the target Korean text.

[0022] In one possible design, in another implementation of the first aspect of the embodiments of the present application, the process of predicting the pronunciation information of the current word by the pronunciation prediction model using the current word and its context information in the target Korean text includes:

[0023] The jamo of the current word and the jamo of the context words of the current word in the target Korean text are fed into the configured pronunciation prediction model to obtain the pronunciation information of the current word output by the model;

[0024] or,

[0025] The jamo of the current word, the jamo of the context words of the current word in the target Korean text, and the rhythm information of the current word and the context words are fed into the configured pronunciation prediction model to obtain the pronunciation information of the current word output by the model.

[0026] In one possible design, in another implementation of the first aspect of the embodiments of the present application, after obtaining the pronunciation of each word in the target Korean text, the method further includes:

[0027] The pronunciation of each word in the target Korean text is tested and corrected using pre-configured special pronunciation rules to obtain the final pronunciation of the target Korean text.

[0028] In a second aspect, a Korean speech synthesis method is provided, comprising:

[0029] For the target Korean text to be synthesized, using the Korean text pronunciation prediction method described in any one of the first aspects of the present application to obtain pronunciation information of the target Korean text;

[0030] Speech synthesis is performed based on the pronunciation information.

[0031] In a third aspect, a Korean text pronunciation prediction device is provided, comprising:

[0032] A word segmentation processing unit is used to perform word segmentation processing on the target Korean text to obtain a number of words;

[0033] a dictionary matching unit for matching each word with a preconfigured pronunciation dictionary; if both the word face and pronunciation rules of the current word are successfully matched, using the pronunciation of the successfully matched dictionary as the pronunciation of the current word; the pronunciation dictionary contains the word faces of several words and the pronunciation information corresponding to each word under at least one pronunciation rule;

[0034] A first model prediction unit is configured to obtain target pronunciation information corresponding to any pronunciation rule under the word face that successfully matches the current word in the pronunciation dictionary if the word face of the current word successfully matches but the pronunciation rule fails to match, and to predict candidate pronunciation information of the current word using a configured pronunciation prediction model using the current word and its contextual information in the target Korean text, wherein the pronunciation prediction model is trained using sample words and their contextual information in the training Korean text as training samples and using the pronunciation of the sample words as sample labels;

[0035] The first and last phoneme replacement unit is used to replace the corresponding first and last phoneme in the target pronunciation information with the first and / or last phoneme in the candidate pronunciation information of the current word, and obtain the replaced pronunciation information as the pronunciation of the current word.

[0036] In a fourth aspect, an electronic device is provided, comprising: a memory and a processor;

[0037] The memory is used to store programs;

[0038] The processor is configured to execute the program to implement the Korean text pronunciation prediction method described in any one of the first aspects of the present application, or to implement the Korean text pronunciation prediction method described in the second aspect of the present application.

[0039] In a fifth aspect, a readable storage medium is provided, on which a computer program is stored, characterized in that when the computer program is executed by a processor, it implements the Korean text pronunciation prediction method described in any one of the first aspects of the present application, or implements the Korean text pronunciation prediction method described in the second aspect of the present application.

[0040] In a sixth aspect, a computer program product is provided, comprising a computer program. When the computer program is executed by a processor, the computer program implements the Korean text pronunciation prediction method described in any one of the first aspects of the present application, or the Korean text pronunciation prediction method described in the second aspect of the present application.

[0041] By means of the above technical solution, the present application combines a dictionary pronunciation prediction method and a data-driven model pronunciation prediction method. Among them, the pronunciation dictionary contains the word faces of several pre-recorded words, as well as the pronunciation information corresponding to each word under at least one pronunciation rule. The target Korean text is segmented and each word is matched with the pronunciation dictionary. If the word face and pronunciation rule of the current word are successfully matched, it means that the pronunciation of the current word has been included in the pronunciation dictionary, and thus the dictionary pronunciation that has successfully matched can be used as the pronunciation of the current word. If the word face of the current word is successfully matched but the pronunciation rule is not successfully matched, it means that the current word is included in the pronunciation dictionary, but the pronunciation information that completely matches its pronunciation rule is not included. On this basis, the target pronunciation information corresponding to any pronunciation rule under the word face that successfully matches in the pronunciation dictionary can be obtained. Considering that the pronunciation of general Korean words is prone to sound changes at the initial and final phonemes, the pronunciation dictionary fails to include pronunciation information that completely matches its pronunciation rule, but the phonemes in the target pronunciation information other than the initial and final phonemes are generally still correct. That is, the following target of the present application scheme promptly further obtains the first phoneme and / or last phoneme of current word, replaces the corresponding first phoneme and / or last phoneme in the target pronunciation information.For this reason, the present application further calls the pronunciation prediction model to predict the candidate pronunciation information of current word, and this model adopts training data drive training in advance.In order to improve the accuracy of model prediction result, in the present application, not only utilize the current word to predict pronunciation information, but make comprehensive use of the current word and the context information of the current word in the target Korean text, predict the candidate pronunciation information of current word by the pronunciation prediction model, by considering the context information of current word, can improve the accuracy of the first and last phoneme prediction result of current word.On this basis, utilize the first phoneme and / or last phoneme in the candidate pronunciation information of current word, replace the first phoneme and / or last phoneme in the target pronunciation information, obtain the pronunciation information after replacement as the pronunciation of current word, so far can obtain the pronunciation of each word in the target Korean text.

[0042] The method provided in the embodiments of this application improves the accuracy of Korean word pronunciation prediction by integrating multiple different pronunciation prediction methods. This improves the accuracy of prediction results for initial and final phonemes, which are prone to sound changes in Korean. This further enhances the effectiveness of Korean speech synthesis. Furthermore, the pronunciation dictionary in this application solution does not need to include all pronunciations for all words, thereby reducing the difficulty and cost of constructing a pronunciation dictionary. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present application. The same reference symbols are used throughout the drawings to represent the same components. In the drawings:

[0044] Figure 1 A schematic diagram of an implementation system architecture of the Korean text pronunciation prediction method provided in an embodiment of the present application;

[0045] Figure 2 A schematic diagram of a terminal structure provided in an embodiment of the present application;

[0046] Figure 3 A schematic diagram of a server structure provided in an embodiment of the present application;

[0047] Figure 4 A schematic diagram of a flow chart of a Korean text pronunciation prediction method provided in an embodiment of the present application;

[0048] Figure 5 Another schematic diagram of a flow chart of a Korean text pronunciation prediction method provided in an embodiment of the present application;

[0049] Figure 6 A schematic diagram of the structure of a Korean text pronunciation prediction device provided in an embodiment of the present application;

[0050] Figure 7 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

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

[0052] This application provides a Korean text pronunciation prediction method and a Korean speech synthesis method, which can be applied to Figure 1The system architecture shown in FIG. 1 may include a terminal 100 and a server 200. The server 200 may include one or more servers ( Figure 1 (This section includes a server as an example).

[0053] The terminal 100 or the server 200 can be used alone to execute the Korean text pronunciation prediction method and / or Korean speech synthesis method provided in the embodiments of the present application. In addition, the terminal 100 and the server 200 can also be used in conjunction to execute the Korean text pronunciation prediction method and / or Korean speech synthesis method provided in the embodiments of the present application.

[0054] Next describe Figure 1 The product form of the mid-terminal 100;

[0055] The terminal 100 in the embodiment of the present application can be a mobile phone, a tablet computer, a learning machine, a translation machine, a teaching screen, a wearable device, a vehicle-mounted device, a conference terminal, an augmented reality (AR) / virtual reality (VR) device, a laptop computer, an ultra-mobile personal computer (UMPC), a netbook, a personal digital assistant (PDA), etc., and the embodiment of the present application does not impose any restrictions on this.

[0056] Figure 2 A schematic diagram of an optional hardware structure of the terminal 100 is shown.

[0057] refer to Figure 2 As shown, the terminal 100 may include a radio frequency unit 110, a memory 120, an input unit 130, a display unit 140, a camera 150 (optional), an audio circuit 160, a speaker 161, a microphone 162, an earphone jack 163 (optional), a processor 170, an external interface 180, a power supply 190 and other components. Those skilled in the art will understand that Figure 2 These are merely examples of terminals or multi-function devices and do not limit the terminal or multi-function device. The terminal or multi-function device may include more or fewer components than shown in the figure, or may combine certain components or different components.

[0058] The input unit 130 can be used to receive input digital or character information and generate key signal input related to user settings and function control of the portable multifunction device. Specifically, the input unit 130 may include a touch screen 131 and / or other input devices 132. The touch screen 131 can detect user touch operations on or near it (for example, operations performed on or near the touch screen using a finger, joint, stylus, or any other suitable object) and drive corresponding connected devices according to pre-set programs. The touch screen can detect user touch actions on the touch screen, convert the touch actions into touch signals and transmit them to the processor 170. It can also receive and execute commands sent by the processor 170; the touch signals include at least touch point coordinate information. The touch screen 131 provides an input interface and an output interface between the terminal 100 and the user. Touch screens can be implemented using various types, including resistive, capacitive, infrared, and surface acoustic wave. In addition to the touch screen 131, the input unit 130 may also include other input devices. Specifically, the other input devices 132 may include, but are not limited to, one or more of a physical keyboard, function keys (such as a volume control key, a switch key, etc.), a trackball, a mouse, a joystick, and the like.

[0059] Among them, other input devices 132 can receive input data and so on.

[0060] The display unit 140 may be used to display information input by or provided to the user, various menus of the terminal 100, interactive interfaces, file display, and / or playback of any multimedia file. In the embodiment of the present application, the display unit 140 may be used to display various interactive interfaces and processing results of the Korean text pronunciation prediction method and / or the Korean speech synthesis method.

[0061] Memory 120 can be used to store instructions and data. It primarily includes an instruction storage area and a data storage area. The data storage area can store various data, such as multimedia files and text. The instruction storage area can store software units such as the operating system, applications, and instructions required for at least one function, or subsets or extensions thereof. It may also include non-volatile random access memory (RAM). It provides processor 170 with management functions for the hardware, software, and data resources within the computing and processing device, supporting control software and applications. It is also used to store multimedia files and running programs and applications.

[0062] The processor 170 is the control center of the terminal 100. It connects all components of the terminal 100 using various interfaces and circuits. By executing instructions stored in the memory 120 and accessing data stored therein, it executes various functions of the terminal 100 and processes data, thereby providing overall control of the terminal device. Optionally, the processor 170 may include one or more processing units. Preferably, the processor 170 may integrate an application processor and a modem processor, with the application processor primarily processing the operating system, user interface, and application programs, while the modem processor primarily handles wireless communications. It is understood that the modem processor may not be integrated into the processor 170. In some embodiments, the processor and memory may be implemented on a single chip; in other embodiments, they may be implemented on separate chips. The processor 170 may also generate corresponding operational control signals and send them to the corresponding components of the computing and processing device. It may also read and process data in the software, particularly the data and programs in the memory 120, to enable the various functional modules therein to perform their corresponding functions, thereby controlling the corresponding components to operate as instructed.

[0063] Among them, the memory 120 can be used to store software codes related to the Korean text pronunciation prediction method and / or the Korean speech synthesis method, the processor 170 can execute the steps of the Korean text pronunciation prediction method and / or the Korean speech synthesis method, and can also schedule other units (such as the above-mentioned input unit 130 and the display unit 140) to implement corresponding functions.

[0064] The RF unit 110 (optional) can be used to send and receive information or receive and send signals during a call. For example, after receiving downlink information from the base station, it is passed to the processor 170 for processing; in addition, the designed uplink data is sent to the base station. Generally, the RF circuit includes but is not limited to an antenna, at least one amplifier, a transceiver, a coupler, a low noise amplifier (LNA), a duplexer, etc. In addition, the RF unit 110 can also communicate with network devices and other devices via wireless communication. This wireless communication can use any communication standard or protocol, including but not limited to Global System of Mobile communication (GSM), General Packet Radio Service (GPRS), Code Division Multiple Access (CDMA), Wideband Code Division Multiple Access (WCDMA), Long Term Evolution (LTE), email, Short Messaging Service (SMS), etc.

[0065] In this embodiment of the present application, the RF unit 110 can send data to the server 200 and receive processing results sent by the server 200. For example, the RF unit 110 sends the received target Korean text to the server 200, and the server 200 obtains pronunciation information of the target Korean text and returns the pronunciation information to the terminal 100 for speech synthesis. Alternatively, after obtaining the pronunciation information of the target Korean text, the server 200 can further perform speech synthesis and return the synthesized speech to the terminal 100.

[0066] It should be understood that the radio frequency unit 110 is optional and can be replaced by other communication interfaces, such as a network port.

[0067] The terminal 100 also includes a power supply 190 (such as a battery) for supplying power to various components. Preferably, the power supply can be logically connected to the processor 170 through a power management system, thereby managing functions such as charging, discharging, and power consumption through the power management system.

[0068] The terminal 100 further includes an external interface 180 , which may be a standard Micro USB interface or a multi-pin connector, and may be used to connect the terminal 100 to other devices for communication, or to connect a charger to charge the terminal 100 .

[0069] Although not shown, the terminal 100 may also include a flashlight, a wireless fidelity (WiFi) module, a Bluetooth module, sensors with different functions, etc., which will not be described in detail here. Some or all of the methods described below can be applied to Figure 2 In the terminal 100 shown.

[0070] Next describe Figure 1 The product form of the server 200;

[0071] Figure 3 A structural diagram of a server 200 is provided, such as Figure 3 As shown, the server 200 includes a bus 201, a processor 202, a communication interface 203, and a memory 204. The processor 202, the memory 204, and the communication interface 203 communicate with each other via the bus 201.

[0072] The bus 201 may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 3 Only one thick line is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0073] The processor 202 may be any one or more of a central processing unit (CPU), a graphics processing unit (GPU), a microprocessor (MP), or a digital signal processor (DSP).

[0074] The memory 204 may include volatile memory, such as random access memory (RAM). The memory 204 may also include non-volatile memory, such as read-only memory (ROM), flash memory, hard disk drive (HDD), or solid state drive (SSD).

[0075] Among them, the memory 204 can be used to store software codes related to the Korean text pronunciation prediction method and / or the Korean speech synthesis method, the processor 202 can execute the steps of the Korean text pronunciation prediction method and / or the Korean speech synthesis method, and can also schedule other units to implement corresponding functions.

[0076] It should be understood that the above-mentioned terminal 100 and server 200 can be centralized or distributed devices, and the processors in the above-mentioned terminal 100 and server 200 (such as processor 170 and processor 202) can be hardware circuits (such as application specific integrated circuit (ASIC), field-programmable gate array (FPGA), general-purpose processor, digital signal processor (DSP), microprocessor or microcontroller, etc.), or a combination of these hardware circuits. For example, the processor can be a hardware system with an instruction execution function, such as a CPU, DSP, etc., or a hardware system without an instruction execution function, such as an ASIC, FPGA, etc., or a combination of the above-mentioned hardware systems without an instruction execution function and hardware systems with an instruction execution function.

[0077] The development of G2P technology is limited to a certain extent by the complex relationship between speech and text in different languages. G2P algorithms targeting different languages ​​have been studied, and the G2P algorithms used in different languages ​​may be different.

[0078] Current G2P technology for Korean generally uses a pronunciation dictionary approach. This involves matching Korean words against a pronunciation dictionary. If a match is found, the dictionary pronunciation is used as the word's pronunciation. However, given the large number of Korean words, there are a large number of oov words, and manually collecting all of them would be prohibitively expensive. The applicant in this case sought to predict pronunciation by combining a pronunciation dictionary with a data-driven pronunciation prediction model.

[0079] One possible implementation is that when a word fails to match a dictionary, the current word can be directly fed into a pre-trained pronunciation prediction model to predict the pronunciation of the current word. This can reduce the difficulty of building a pronunciation dictionary.

[0080] However, research has found that there are still the following defects:

[0081] First, the pronunciation prediction model based on data is inherently a "black box" model, and its prediction results are subject to certain errors. Second, Korean has many phonetic variations, and the same word can have different pronunciations in different sentences. The above method only predicts pronunciation based on the current word, without considering context. This can easily lead to inaccurate predictions when the initial or final phonemes undergo sound changes, further reducing the accuracy of the pronunciation prediction results.

[0082] After trying and thinking about the above solutions, this application further provides an improved solution to improve the defects of the above solutions.

[0083] The present invention provides a method for predicting the pronunciation of Korean text. Figure 1 Taking the computer device in the example as an example, the computer device can be Figure 1 The terminal 100 or the system consisting of the terminal 100 and the server 200. Figure 4 , the Korean text pronunciation prediction method specifically includes the following steps:

[0084] Step S100: performing word segmentation processing on the target Korean text to obtain a number of words.

[0085] Specifically, the target Korean text may be segmented using spaces as demarcations to obtain segmented words. Furthermore, a segmentation tool may be used to segment the target Korean text.

[0086] Optionally, before or after word segmentation of the target Korean text, the target Korean text may be preprocessed, such as removing stop words, removing obviously erroneous characters, etc., which will not be described in detail here.

[0087] Step S110: For each word, the current word is matched with a pre-configured pronunciation dictionary. If both the word face and the pronunciation rules of the current word are successfully matched, step S120 is executed. If the word face of the current word is successfully matched but the pronunciation rules are not successfully matched, step S130 is executed.

[0088] The pronunciation dictionary contains the word faces of several pre-recorded words and the pronunciation information corresponding to each word under at least one pronunciation rule.

[0089] The following example shows an alternative dictionary format:

[0090] Word: (pronunciation information) [Rule 1, Rule 2, …]

[0091] The word before the colon ":" is the word's literal meaning, the part within the parentheses "()" is the word's pronunciation information, and the part within the square brackets "[]" is the pronunciation rule corresponding to the pronunciation information. Different words can have one or more pronunciation rules corresponding to their pronunciation information, and multiple pronunciation rules can be separated by commas. As long as any pronunciation rule within the square brackets is met, the pronunciation of the word can be determined according to the corresponding pronunciation information.

[0092] If a word has multiple pronunciations, you can use a slash " / " to separate the different pronunciation information and corresponding pronunciation rules, as shown in the following example:

[0093] Word: (pronunciation information 1) [rule 1, rule 2] / (pronunciation information 2) [rule 3, rule 4]…

[0094] Of course, the above is only an example of an optional way of dictionary format. In addition, other formats can also be used to construct a pronunciation dictionary.

[0095] The pronunciation rules can be pre-arranged based on the pronunciation patterns of Korean words. The pronunciation rules may include pronunciation rules set based on the Korean letters "jamo" of the context words of the word, or pronunciation rules set based on the part of speech and prosodic information of the context words of the word, etc. In other words, the pronunciation rules can be configured based on any one or more combinations of part of speech, "jamo", and prosodic information.

[0096] In this embodiment, several optional examples of pronunciation rules are given:

[0097] 1. When the word before the current word is a verb and the word after it is any word or no word, the pronunciation of the current word is X1;

[0098] 2. When the previous word of the current word is any word or no word, and the part of speech of the next word is an adverb, the pronunciation of the current word is X2;

[0099] 3. The last jamo after the previous word of the current word is split into jamo is , the first jamo after the next word is split into jamo is When , the pronunciation of the current word is X3;

[0100] 4. The rhythm of the word before the current word is a long pause, and the first jamo after the next word is split into jamos is , the pronunciation of the current word is X4.

[0101] Among them, X1-X4 represent 4 different pronunciation information.

[0102] Step S120: Use the successfully matched dictionary pronunciation as the pronunciation of the current word.

[0103] Specifically, when matching the current word with the pronunciation dictionary, if both the word face and the pronunciation rules of the current word can be successfully matched, the dictionary pronunciation of the successful match can be used as the pronunciation of the current word. Among them, the word face and the pronunciation rules of the current word can be successfully matched can be understood as that the word face of the current word exists in the pronunciation dictionary, and the current word meets a certain pronunciation rule under the word face, and the pronunciation information corresponding to the pronunciation rule that meets the match can be used as the dictionary pronunciation that is successfully matched, and then as the pronunciation of the current word.

[0104] Step S130: Obtain target pronunciation information corresponding to any pronunciation rule under the successfully matched word in the pronunciation dictionary, and predict candidate pronunciation information of the current word through the configured pronunciation prediction model using the current word and its context information in the target Korean text.

[0105] Here, the word "candidate" in "candidate pronunciation information" is intended to indicate that the pronunciation information of the current word predicted by the pronunciation prediction model is not directly used as the final pronunciation result.

[0106] The pronunciation prediction model is a data-driven model that learns the mapping relationship between words and their contextual information and the pronunciation of words through machine learning methods. The pronunciation prediction model can be based on traditional machine learning methods, such as hidden Markov model HMM, maximum entropy model MaxEnt, conditional random field CRF, etc., or it can be based on a neural network method, such as using recurrent neural network RNN, long short-term memory network LSTM, attention mechanism-based Attention model, Transformer model, etc. The pronunciation prediction model can be obtained by using sample words and the contextual information of the sample words in the training Korean text as training samples, and the pronunciation of the sample words as sample labels for training.

[0107] When matching the current word with the pronunciation dictionary, if the lexical match of the current word is successful but the pronunciation rule is not successfully matched, it means that the lexical match of the current word is included in the pronunciation dictionary, but the current word in the target Korean text does not meet any pronunciation rules under the lexical match of the current word in the pronunciation dictionary. Therefore, the pronunciation information in the pronunciation dictionary cannot be directly used as the pronunciation of the current word.

[0108] However, considering the pronunciation habits of Korean, it is very likely that the current word has a different context in the target Korean text, resulting in a sound change in the initial and / or final phonemes of the current word, and this situation is not included in the pronunciation dictionary, which in turn causes the pronunciation rule to fail to match successfully. In this case, the pronunciation information of the current word in the pronunciation dictionary (which can be the pronunciation information corresponding to any pronunciation rule) is generally correct except for the initial and / or final phonemes. Therefore, in this step, the target pronunciation information corresponding to any pronunciation rule under the word that successfully matched in the pronunciation dictionary can be obtained.

[0109] Furthermore, in this step, the model can be used to predict the initial phoneme and / or final phoneme of the current word. To improve the accuracy of the pronunciation prediction results of the initial phoneme and / or final phoneme, this step considers both the current word in the target Korean text and the context of the current word. Based on this, the pronunciation prediction model is used to predict candidate pronunciation information for the current word.

[0110] In this embodiment, since the pronunciation prediction model can refer to the context information of the current word, it can more accurately predict the correct pronunciation of the initial phoneme and the final phoneme of the current word. In this embodiment, it can be determined that the initial phoneme and the final phoneme in the candidate pronunciation information of the current word predicted by the pronunciation prediction model are correct.

[0111] Step S140: Using the initial phoneme and / or final phoneme in the candidate pronunciation information of the current word, the corresponding initial phoneme and / or final phoneme in the target pronunciation information is replaced to obtain the pronunciation of the current word.

[0112] Specifically, in this step, you can choose to replace only the initial phoneme, or only the final phoneme, or both the initial phoneme and the final phoneme, and obtain the replaced pronunciation information as the pronunciation of the current word.

[0113] The target pronunciation information obtained in the previous step may have inaccurate initial and / or final phonemes. However, the remaining phonemes in the target pronunciation information, except for the initial and final phonemes, can still be determined to be correct. The initial and final phonemes obtained by the pronunciation prediction model have a relatively higher accuracy rate (and can be determined to be correct in this embodiment). Therefore, in this step, the initial and / or final phonemes obtained by the pronunciation prediction model can be used to replace the corresponding initial and / or final phonemes in the target pronunciation information to obtain the pronunciation of the current word.

[0114] The Korean text pronunciation prediction method provided in the embodiment of the present application improves the pronunciation accuracy of Korean words and reduces the difficulty of constructing a pronunciation dictionary by combining a pronunciation dictionary and a pronunciation prediction model.

[0115] This application makes full use of manually proofread pronunciation dictionary resources and contextual information in the text to improve the accuracy of Korean prediction for sound changes.

[0116] Combine Figure 5 In the embodiment of the present application, another optional embodiment of the Korean text pronunciation prediction method is provided, comprising the following steps:

[0117] Step S200: performing word segmentation processing on the target Korean text to obtain a plurality of words.

[0118] Specifically, the target Korean text may be segmented using spaces as demarcations to obtain segmented words. Furthermore, a segmentation tool may be used to segment the target Korean text.

[0119] Optionally, before or after word segmentation of the target Korean text, the target Korean text may be preprocessed, such as removing stop words, removing obviously erroneous characters, etc., which will not be described in detail here.

[0120] Step S210: For each word, the current word is matched against a pre-configured pronunciation dictionary. If both the word face and the pronunciation rules of the current word are successfully matched, step S220 is executed. If the word face of the current word is successfully matched but the pronunciation rules are not successfully matched, step S230 is executed. If the word face of the current word is not successfully matched, step S250 is executed.

[0121] Step S220: Use the successfully matched dictionary pronunciation as the pronunciation of the current word.

[0122] Step S230: Obtain target pronunciation information corresponding to any pronunciation rule under the successfully matched word in the pronunciation dictionary, and predict candidate pronunciation information of the current word through the configured pronunciation prediction model using the current word and its context information in the target Korean text.

[0123] Step S240: Use the initial phoneme and / or final phoneme in the candidate pronunciation information of the current word to replace the corresponding initial phoneme and / or final phoneme in the target pronunciation information to obtain the pronunciation of the current word.

[0124] The above steps S200-S240 correspond to the aforementioned steps S100-S140 one by one. Please refer to the above description for details and will not be repeated here.

[0125] Step S250: using the current word and its context information in the target Korean text, predict the pronunciation information of the current word through a pronunciation prediction model.

[0126] In step S210, when matching the current word with the pronunciation dictionary, it is possible that the word face of the current word is not successfully matched, indicating that the pronunciation dictionary does not include the current word, and thus no pronunciation information of the current word can be obtained through the pronunciation dictionary. In this case, a pronunciation prediction model can be used in this step to predict the pronunciation information of the current word.

[0127] The input of the pronunciation prediction model in this step may be the same as the input of the pronunciation prediction model in the aforementioned step S230 , that is, the pronunciation prediction model uses the current word and its context information in the target Korean text to predict the pronunciation information of the current word.

[0128] Different from step S230, in this step, the pronunciation information output by the pronunciation prediction model can be directly used as the pronunciation information of the current word.

[0129] In this embodiment, a processing method for the third matching situation when the current word is matched with the pronunciation dictionary is further provided, that is, when the word face of the current word is not included in the pronunciation dictionary, the pronunciation information of the current word can be directly predicted through the pronunciation prediction model to ensure that the complete pronunciation information of the target Korean text can be obtained.

[0130] In addition, the input of the pronunciation prediction model includes both the current word and its context information, which can improve the prediction accuracy of the Korean phonetic changes.

[0131] In some possible implementations, for the methods provided in the aforementioned embodiments, the following processing steps may be further added before matching the current word with the pronunciation dictionary:

[0132] Convert each word in the target Korean text into Korean letters jamo.

[0133] By splitting Korean words into jamo, the data information of the words can be fully mined. On this basis, the pronunciation rules in the pronunciation dictionary configured in this application can include: pronunciation rules set according to the jamo of the context word of the word.

[0134] Examples of pronunciation rules may include:

[0135] The last jamo after the word before the current word is split into jamo is , the first jamo after the next word is split into jamo is When , the pronunciation of the current word is X3. Wherein, X3 is a set pronunciation information.

[0136] The staff can set the above pronunciation rules in advance based on Korean linguistics knowledge.

[0137] Based on the pronunciation dictionary provided in this embodiment, the process of matching the current word with the pronunciation dictionary in the aforementioned step may include:

[0138] The current word is matched word-wise in the pronunciation dictionary. If the word-wise match is successful, the jamo of the context word of the current word is further matched with the pronunciation rules under the word-wise match in the pronunciation dictionary to determine whether the pronunciation rules are matched successfully.

[0139] By configuring word-based jamo pronunciation rules in the pronunciation dictionary, you can more accurately summarize the pronunciation rules of Korean words and improve the probability of successful pronunciation dictionary matching.

[0140] Further optionally, in step S230 and step S250 of the aforementioned embodiment, the process of predicting the pronunciation information of the current word (predicting candidate pronunciation information in step S230, collectively referred to herein as pronunciation information) by using the pronunciation prediction model using the current word and its contextual information in the target Korean text may include:

[0141] The jamo of the current word and the jamo of the context words of the current word in the target Korean text are fed into the configured pronunciation prediction model to obtain the pronunciation information of the current word output by the model.

[0142] By splitting words into jamo forms and feeding the jamo of the current word and the jamo of the context words into the pronunciation prediction model, the model can refer to more fine-grained and richer data information, thereby improving the accuracy of pronunciation prediction.

[0143] In some possible implementations, for the methods provided in the aforementioned embodiments, the following processing steps may be further added before matching the current word with the pronunciation dictionary:

[0144] The part of speech and / or prosodic information of each word in the target Korean text is determined.

[0145] Correspondingly, the pronunciation rules in the pronunciation dictionary may also include pronunciation rules set according to the context of the word, the part of speech of the word, and / or prosody information.

[0146] Examples of pronunciation rules may include:

[0147] When the part of speech of the word before the current word is a verb and the word after it is any word or no word, the pronunciation of the current word is X1;

[0148] When the word preceding the current word is any word or no word, and the part of speech of the following word is an adverb, the pronunciation of the current word is X2;

[0149] The rhythm of the word before the current word is a long pause, and the first jamo after the next word is split into jamos is , the pronunciation of the current word is X4.

[0150] Among them, X1, X2, and X4 represent different pronunciation information.

[0151] The staff can set the above pronunciation rules in advance based on Korean linguistics knowledge.

[0152] Based on the pronunciation dictionary provided in this embodiment, the process of matching the current word with the pronunciation dictionary in the aforementioned step may further include:

[0153] According to the part of speech and / or prosody information of the context words of the current word, the pronunciation rules under the words that have been successfully matched in the pronunciation dictionary are matched to determine whether the pronunciation rules are successfully matched.

[0154] By configuring the pronunciation rules in the pronunciation dictionary through any one or more combinations of part of speech, jamo, and prosody information, more diverse pronunciation rules of Korean words can be covered, thereby increasing the probability of successful pronunciation dictionary matching.

[0155] Further optionally, in step S230 and step S250 of the aforementioned embodiment, the process of predicting the pronunciation information of the current word (predicting candidate pronunciation information in step S230, collectively referred to herein as pronunciation information) by using the pronunciation prediction model using the current word and its contextual information in the target Korean text may include:

[0156] The jamo of the current word, the jamo of the context words of the current word in the target Korean text, and the rhythm information of the current word and the context words are fed into the configured pronunciation prediction model to obtain the pronunciation information of the current word output by the model.

[0157] Compared with the above embodiment, this embodiment further feeds the prosodic information of the context words of the current word into the pronunciation prediction model, so that the model can further refer to the prosodic information of the context to help improve the accuracy of pronunciation prediction.

[0158] On the basis of the above embodiments, the Korean text pronunciation prediction method of the present application can further add the following processing steps:

[0159] After obtaining the pronunciation of each word in the target Korean text according to the method of any of the above embodiments, the pronunciation of each word in the target Korean text can also be tested and corrected using pre-configured special pronunciation rules to obtain the final pronunciation of the target Korean text.

[0160] Specifically, some words in Korean have special pronunciation rules. For example, some words have fixed pronunciations when they appear at the beginning or end of a sentence; some words have fixed pronunciations when they appear after a verb; and some words have fixed pronunciations in specific sentence structures. The staff can pre-record these special pronunciation rules. After obtaining the pronunciation of the target Korean text according to the aforementioned embodiments, they can further use the special pronunciation rules to verify and correct the pronunciation of each word in the target Korean text to obtain the final pronunciation result of the target Korean text.

[0161] The Korean text pronunciation prediction method provided in this embodiment adopts a deep combination of pronunciation dictionaries, model predictions and special pronunciation rules. It can simultaneously combine multiple pronunciation prediction methods to predict the pronunciation of Korean text, which can improve the accuracy of pronunciation prediction results.

[0162] In some embodiments of the present application, a Korean speech synthesis method is further provided. For the target Korean text to be synthesized, the Korean text pronunciation prediction method of any of the above embodiments can be used to obtain pronunciation information, and on this basis, speech synthesis is performed based on the pronunciation information.

[0163] In view of the effect brought by the Korean text pronunciation prediction method, the Korean speech synthesis effect in this embodiment can be further optimized.

[0164] The Korean text pronunciation prediction device provided in an embodiment of the present application is described below. The Korean text pronunciation prediction device described below and the Korean text pronunciation prediction method described above can refer to each other.

[0165] See also Figure 6 , Figure 6 This is a schematic diagram of the structure of a Korean text pronunciation prediction device disclosed in an embodiment of the present application.

[0166] like Figure 6 As shown, the device may include:

[0167] A word segmentation processing unit 11 is used to perform word segmentation processing on the target Korean text to obtain a plurality of words;

[0168] A dictionary matching unit 12 is configured to match each word with a preconfigured pronunciation dictionary. If both the word face and pronunciation rules of the current word are successfully matched, the pronunciation of the dictionary that successfully matches is used as the pronunciation of the current word. The pronunciation dictionary contains the word faces of several words and the pronunciation information corresponding to each word under at least one pronunciation rule.

[0169] A first model prediction unit 13 is configured to obtain target pronunciation information corresponding to any pronunciation rule under the word face that successfully matches the current word in the pronunciation dictionary if the word face of the current word successfully matches but the pronunciation rule fails to match, and to predict candidate pronunciation information of the current word using a configured pronunciation prediction model using the current word and its contextual information in the target Korean text, wherein the pronunciation prediction model is trained using sample words and their contextual information in the training Korean text as training samples and using the pronunciation of the sample words as sample labels;

[0170] The first and last phoneme replacing unit 14 is used to replace the corresponding first and last phoneme in the target pronunciation information with the first and / or last phoneme in the candidate pronunciation information of the current word, and obtain the replaced pronunciation information as the pronunciation of the current word.

[0171] In some possible implementations, the apparatus of the present application may further include:

[0172] The jamo conversion unit is configured to convert each word in the target Korean text into the Korean letter jamo before the dictionary matching unit is executed. Accordingly, the pronunciation rules in the pronunciation dictionary include pronunciation rules set according to the jamo of the context words of the word. Then, the dictionary matching unit matches the current word with the preconfigured pronunciation dictionary, including:

[0173] The current word is matched word-wise in the pronunciation dictionary. If the word-wise match is successful, the jamo of the context word of the current word is further matched with the pronunciation rules under the word-wise match in the pronunciation dictionary to determine whether the pronunciation rules are matched successfully.

[0174] In some possible implementations, the pronunciation rules in the pronunciation dictionary further include: pronunciation rules set according to the part of speech and / or prosodic information of the context words of the word, and the process of the dictionary matching unit matching the current word with the preconfigured pronunciation dictionary further includes:

[0175] According to the part of speech and / or prosody information of the context words of the current word, the pronunciation rules under the words that have been successfully matched in the pronunciation dictionary are matched to determine whether the pronunciation rules are successfully matched.

[0176] In some possible implementations, the process of the first model prediction unit using the current word and its context information in the target Korean text to predict candidate pronunciation information of the current word through the configured pronunciation prediction model includes:

[0177] The jamo of the current word and the jamo of the context words of the current word in the target Korean text are fed into the configured pronunciation prediction model to obtain candidate pronunciation information of the current word output by the model.

[0178] In some possible implementations, the process of the first model prediction unit using the current word and its context information in the target Korean text to predict candidate pronunciation information of the current word through the configured pronunciation prediction model includes:

[0179] The jamo of the current word, the jamo of the context words of the current word in the target Korean text, and the prosodic information of the current word and the context words are fed into the configured pronunciation prediction model to obtain the candidate pronunciation information of the current word output by the model.

[0180] In some possible implementations, the apparatus of the present application may further include:

[0181] The second model prediction unit is used to predict the pronunciation information of the current word through the pronunciation prediction model by using the current word and its context information in the target Korean text if the word face of the current word fails to be matched successfully.

[0182] In some possible implementations, the second model prediction unit uses the current word and its context information in the target Korean text to predict the pronunciation information of the current word through the pronunciation prediction model, including:

[0183] The jamo of the current word and the jamo of the context words of the current word in the target Korean text are fed into the configured pronunciation prediction model to obtain the pronunciation information of the current word output by the model;

[0184] or,

[0185] The jamo of the current word, the jamo of the context words of the current word in the target Korean text, and the rhythm information of the current word and the context words are fed into the configured pronunciation prediction model to obtain the pronunciation information of the current word output by the model.

[0186] In some possible implementations, the apparatus of the present application may further include:

[0187] The special pronunciation rule filtering unit is used to, after obtaining the pronunciation of each word in the target Korean text, use pre-configured special pronunciation rules to test and correct the pronunciation of each word in the target Korean text to obtain the final pronunciation of the target Korean.

[0188] An electronic device is also provided in an embodiment of the present application. Figure 7 , which shows a schematic diagram of the structure of an electronic device suitable for implementing the embodiment of the present application. The electronic device in the embodiment of the present application may include but is not limited to fixed terminals such as mobile phones, tablet computers, teaching large screens, wearable devices, etc. Figure 7The electronic device shown is merely an example and should not limit the functions and scope of use of the embodiments of the present application.

[0189] like Figure 7 As shown, the electronic device may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage device 608 into a random access memory (RAM) 603 to implement the Korean text pronunciation prediction method and / or Korean speech synthesis method of the aforementioned embodiments of the present application. When the electronic device is powered on, the RAM 603 also stores various programs and data required for the operation of the electronic device. The processing device 601, ROM 602, and RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0190] Typically, the following devices may be connected to the I / O interface 605: an input device 606 including, for example, a touch screen, a touchpad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 607 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 608 including, for example, a memory card, a hard disk, etc.; and a communication device 609. The communication device 609 may allow the electronic device to communicate with other devices wirelessly or by wire to exchange data. Figure 7 The electronic device is shown with various devices, but it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed instead.

[0191] An embodiment of the present application also provides a computer program product including computer-readable instructions. When the computer-readable instructions are executed on an electronic device, the electronic device implements any one of the Korean text pronunciation prediction methods or Korean speech synthesis methods provided in the embodiments of the present application.

[0192] A computer-readable storage medium is also provided in an embodiment of the present application. The storage medium carries one or more computer programs. When the one or more computer programs are executed by an electronic device, the electronic device can implement any Korean text pronunciation prediction method or Korean speech synthesis method provided in the embodiment of the present application.

[0193] It should also be noted that the device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed across multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment. In addition, in the drawings of the device embodiments provided in this application, the connection relationship between the modules indicates that there is a communication connection between them, which can be specifically implemented as one or more communication buses or signal lines.

[0194] Through the description of the above embodiments, those skilled in the art can clearly understand that the present application can be implemented by means of software plus necessary general hardware, and of course can also be implemented by special hardware including application-specific integrated circuits, special CPUs, special memories, special components, etc. In general, all functions performed by computer programs can be easily implemented with corresponding hardware, and the specific hardware structures used to implement the same function can also be diverse, such as analog circuits, digital circuits or special circuits, etc. However, for the present application, software program implementation is a better implementation method in most cases. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a readable storage medium, such as a computer's floppy disk, USB flash drive, mobile hard disk, ROM, RAM, magnetic disk or optical disk, etc., and includes a number of instructions to enable a computer device (which can be a personal computer, training equipment, or network equipment, etc.) to execute the methods described in each embodiment of the present application.

[0195] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware, or any combination thereof. When implemented by software, all or part of the embodiments may be implemented in the form of a computer program product.

[0196] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website, a computer, a training device or a data center by wired (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, microwave, etc.) mode to another website, computer, training device or data center. The computer-readable storage medium can be any available medium that a computer can store or a data storage device such as a training device, a data center that includes one or more available media integrations. The available medium can be a magnetic medium, (for example, a floppy disk, a hard disk, a tape), an optical medium (for example, a DVD), or a semiconductor medium (for example, a solid-state drive (SSD)).

[0197] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The various embodiments can be combined as needed, and the same or similar parts can be referenced to each other.

Claims

1. A Korean text pronunciation prediction method, characterized in that: include: Perform word segmentation on the target Korean text to obtain several words; For each word, the current word is matched with a pre-configured pronunciation dictionary. If both the word face and the pronunciation rules of the current word can be successfully matched, the pronunciation of the dictionary that successfully matches is used as the pronunciation of the current word. The pronunciation dictionary contains the word faces of several words and the pronunciation information corresponding to each word under at least one pronunciation rule. If the word-face of the current word is successfully matched but the pronunciation rule is not successfully matched, the target pronunciation information corresponding to any pronunciation rule under the word-face that successfully matched in the pronunciation dictionary is obtained, and the candidate pronunciation information of the current word is predicted by a configured pronunciation prediction model using the current word and its contextual information in the target Korean text, wherein the pronunciation prediction model is trained using sample words and the contextual information of the sample words in the training Korean text as training samples and the pronunciation of the sample words as sample labels; The first phoneme and / or the last phoneme in the candidate pronunciation information of the current word are used to replace the corresponding first phoneme and / or the last phoneme in the target pronunciation information, and the replaced pronunciation information is obtained as the pronunciation of the current word.

2. The method according to claim 1, characterized in that Before matching the current word with the preconfigured pronunciation dictionary, it also includes: Convert each word in the target Korean text into Korean letters jamo; The pronunciation rules in the pronunciation dictionary include pronunciation rules set according to the jamo of the context words of the word, then, the process of matching the current word with the preconfigured pronunciation dictionary includes: The current word is matched word-wise in the pronunciation dictionary. If the word-wise match is successful, the jamo of the context word of the current word is further matched with the pronunciation rules under the word-wise match in the pronunciation dictionary to determine whether the pronunciation rules are matched successfully.

3. The method according to claim 2, characterized in that The pronunciation rules in the pronunciation dictionary also include: according to the pronunciation rules set by the word context word part of speech and / or prosodic information, the process of matching the current word with the pre-configured pronunciation dictionary also includes: According to the part of speech and / or prosody information of the context words of the current word, the pronunciation rules under the words that have been successfully matched in the pronunciation dictionary are matched to determine whether the pronunciation rules are successfully matched.

4. The method according to claim 2, characterized in that The process of predicting candidate pronunciation information of the current word using the configured pronunciation prediction model by using the current word and its context information in the target Korean text includes: The jamo of the current word and the jamo of the context words of the current word in the target Korean text are fed into the configured pronunciation prediction model to obtain candidate pronunciation information of the current word output by the model.

5. The method according to claim 2, characterized in that The process of predicting candidate pronunciation information of the current word using the configured pronunciation prediction model by using the current word and its context information in the target Korean text includes: The jamo of the current word, the jamo of the context words of the current word in the target Korean text, and the prosodic information of the current word and the context words are fed into the configured pronunciation prediction model to obtain the candidate pronunciation information of the current word output by the model.

6. The method according to claim 1, characterized in that Also includes: If the word face of the current word fails to be matched successfully, the pronunciation information of the current word is predicted by the pronunciation prediction model using the current word and its context information in the target Korean text.

7. The method according to claim 6, characterized in that The process of predicting the pronunciation information of the current word by the pronunciation prediction model using the current word and its context information in the target Korean text includes: The jamo of the current word and the jamo of the context words of the current word in the target Korean text are fed into the configured pronunciation prediction model to obtain the pronunciation information of the current word output by the model; or, The jamo of the current word, the jamo of the context words of the current word in the target Korean text, and the rhythm information of the current word and the context words are fed into the configured pronunciation prediction model to obtain the pronunciation information of the current word output by the model.

8. The method according to any one of claims 1 to 7, characterized in that After obtaining the pronunciation of each word in the target Korean text, it also includes: The pronunciation of each word in the target Korean text is tested and corrected using pre-configured special pronunciation rules to obtain the final pronunciation of the target Korean text.

9. A Korean speech synthesis method, characterized in that: include: For a target Korean text to be synthesized, obtaining pronunciation information of the target Korean text using the Korean text pronunciation prediction method according to any one of claims 1 to 8; Speech synthesis is performed based on the pronunciation information.

10. A Korean text pronunciation prediction device, characterized in that: include: A word segmentation processing unit is used to perform word segmentation processing on the target Korean text to obtain a number of words; a dictionary matching unit for matching each word with a preconfigured pronunciation dictionary; if both the word face and pronunciation rules of the current word are successfully matched, using the pronunciation of the successfully matched dictionary as the pronunciation of the current word; the pronunciation dictionary contains the word faces of several words and the pronunciation information corresponding to each word under at least one pronunciation rule; A first model prediction unit is configured to obtain target pronunciation information corresponding to any pronunciation rule under the word face that successfully matches the current word in the pronunciation dictionary if the word face of the current word successfully matches but the pronunciation rule fails to match, and to predict candidate pronunciation information of the current word using a configured pronunciation prediction model using the current word and its contextual information in the target Korean text, wherein the pronunciation prediction model is trained using sample words and their contextual information in the training Korean text as training samples and using the pronunciation of the sample words as sample labels; The first and last phoneme replacement unit is used to replace the corresponding first and last phoneme in the target pronunciation information with the first and / or last phoneme in the candidate pronunciation information of the current word, and obtain the replaced pronunciation information as the pronunciation of the current word.

11. An electronic device, characterized in that: include: memory and processor; The memory is used to store programs; The processor is configured to execute the program to implement the steps of the Korean text pronunciation prediction method according to any one of claims 1 to 8, or to implement the steps of the Korean speech synthesis method according to claim 9.

12. A readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the computer program implements the steps of the Korean text pronunciation prediction method according to any one of claims 1 to 8, or the steps of the Korean speech synthesis method according to claim 9.

13. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the computer program implements the steps of the Korean text pronunciation prediction method according to any one of claims 1 to 8, or the steps of the Korean speech synthesis method according to claim 9.

Images