CN111816168A - A model training method, voice playback method, device and storage medium - Google Patents

Abstract

The present application discloses a model training method, a voice playback method, a device and a storage medium, which are applied to the field of artificial intelligence cloud services. The method of the present application includes: obtaining a voice data set to be trained; when N meets a threshold of the number of voice training, obtaining a voice model training instruction; in response to the voice model training instruction, based on the voice data set to be trained, obtain a predicted voice data set through the voice model to be trained ; Based on the predicted speech data set and the to-be-trained speech data set, the to-be-trained speech model is trained to obtain the target speech model; the target speech model is sent, so that the terminal device stores the target speech model. The present application enhances the flexibility of voice model generation, and satisfies the user's voice customization requirements. Secondly, the possibility of selecting a voice model is improved when the user performs voice playback, thereby improving the flexibility of voice playback, thereby improving the user's voice playback experience and flexibility. sex.

Description

A model training method, voice playback method, device and storage medium

technical field

The present application relates to the field of artificial intelligence cloud services, and in particular, to a method for model training, a method, device and storage medium for voice playback.

Background technique

With the rapid development of Internet information technology and the continuous improvement of the quality of life, intelligent terminal equipment is widely used in people's lives, and users' demand for intelligent terminal equipment is also increasing day by day. In order to meet the personalized experience of users on different terminal devices, a way to complete voice playback based on artificial intelligence has been developed.

At present, the way of voice playback is usually by integrating the voice model data of the speaker (such as a public star) into the terminal device, and the user selects the desired speaker, and the terminal device combines the voice of the speaker selected by the user with the desired voice. The played text information is synthesized to obtain target voice data, so that the terminal device plays the target voice data, thereby completing voice playback.

However, since the voice model is integrated in the terminal device, only the voice model integrated in the terminal device can be selected, thereby reducing the selectivity and flexibility of the voice model, thereby reducing the flexibility of voice playback.

SUMMARY OF THE INVENTION

Embodiments of the present application provide a model training method, a voice playback method, a device, and a storage medium, which are used to improve the flexibility of voice model generation, and increase the possibility of voice model selection when a user performs voice playback, thereby improving voice playback. flexibility.

In view of this, on the one hand, the present application provides a method for model training, including:

Acquiring a voice data set to be trained, wherein the voice data set to be trained includes N voice data to be trained, and N is a positive integer greater than or equal to 1;

When N meets the threshold of the number of voice training, obtain the voice model training instruction;

In response to the voice model training instruction, based on the set of voice data to be trained, a set of predicted voice data is obtained through the voice model to be trained, wherein the set of predicted voice data includes N pieces of predicted voice data, and the predicted voice data and the voice data to be trained have a corresponding relationship ;

Based on the predicted speech data set and the to-be-trained speech data set, the to-be-trained speech model is trained to obtain the target speech model;

The target speech model is sent, so that the terminal device stores the target speech model.

Another aspect of the present application provides a method for voice playback, comprising:

Detecting an operation for voice playback, and sending a voice playback request, so that the server obtains service content text data based on the voice data to be parsed, and obtains voiceless voice data according to the service content text data, wherein the voice playback request carries the voice data to be parsed, and the service The content text data is obtained according to the voice data to be parsed, and the timbreless voice data is obtained after voice synthesis is performed on the service content text data;

Receive voiceless voice data;

The target speech data is generated according to the voiceless speech data and the target speech model, wherein the target speech model is obtained by training the speech model to be trained based on the predicted speech data set and the to-be-trained speech data set, and the predicted speech data set is the response of the server to the speech The model training instruction, based on the voice data set to be trained, is obtained through the voice model to be trained;

Play the target voice data.

Another aspect of the present application provides a model training device, comprising:

an acquisition module, configured to acquire a voice data set to be trained, wherein the voice data set to be trained includes N voice data to be trained, and N is a positive integer greater than or equal to 1;

The obtaining module is also used to obtain the training instruction of the voice model when N meets the threshold of the number of voice training;

The acquisition module is further configured to, in response to the voice model training instruction, obtain a predicted voice data set based on the to-be-trained voice data set through the to-be-trained voice model, wherein the predicted voice data set includes N predicted voice data, and the predicted voice data is the same as the to-be-trained voice data set. The training speech data has a corresponding relationship;

The training module is used for training the to-be-trained speech model based on the predicted speech data set and the to-be-trained speech data set to obtain the target speech model;

The sending module is used for sending the target voice model, so that the terminal device stores the target voice model.

In a possible design, in an implementation manner of another aspect of the embodiments of the present application,

An acquisition module, which is specifically used to acquire a voice service request;

In response to the voice service request, a voice data collection instruction is sent to the terminal device, so that the terminal device, in response to the voice data collection instruction, obtains the voice data to be detected, wherein the voice data to be detected is obtained based on the voice training text, and the voice training text is Obtained according to voice data collection instructions;

When the voice data to be detected satisfies the qualified conditions for voice training, determine the voice data to be trained;

Based on the to-be-trained speech data, a to-be-trained speech data set is generated.

In a possible design, in another implementation manner of another aspect of the embodiments of the present application,

A training module, which is specifically used to update the model parameters of the speech model to be trained according to the target loss function based on the predicted speech data set and the to-be-trained speech data set;

If the target loss function reaches convergence, the target speech model is generated according to the model parameters.

In a possible design, in another implementation manner of another aspect of the embodiments of the present application,

an obtaining module, further configured to obtain a voice playback request, wherein the voice playback request carries the voice data to be parsed;

The obtaining module is further configured to obtain service content text data according to the to-be-parsed voice data, wherein the service content text data corresponds to the to-be-parsed voice data;

The acquiring module is further configured to acquire timbreless voice data according to the service content text data, wherein the timbreless voice data is obtained by performing speech synthesis on the service content text data;

The sending module is further configured to send the voiceless voice data to the terminal device, so that the terminal device generates the target voice data according to the voiceless voice data and the target voice model, and plays the target voice data.

Another aspect of the present application provides a voice playback device, comprising:

The sending module is used to detect an operation for voice playback, and send a voice playback request, so that the server obtains service content text data based on the voice data to be parsed, and obtains voiceless voice data according to the service content text data, wherein the voice playback request carries the text data to be parsed. Parse the voice data, the service content text data is obtained according to the voice data to be parsed, and the timbreless voice data is obtained after the service content text data is synthesized by speech;

The acquisition module is used to acquire the voiceless voice data;

The generation module is used to generate target voice data according to the voiceless voice data and the target voice model, wherein the target voice model is obtained by training the voice model to be trained based on the predicted voice data set and the to-be-trained voice data set, and the predicted voice data set It is obtained by the server in response to the voice model training instruction, based on the voice data set to be trained, and obtained through the voice model to be trained;

The playback module is used to play the target voice data.

In a possible design, in an implementation manner of another aspect of the embodiments of the present application,

The sending module is also used to detect the operation for the voice service and send the voice service request;

The obtaining module is also used to obtain the voice data collection instruction when the server responds to the voice service request;

The acquisition module is also used to acquire the voice data to be detected in response to the voice data collection instruction and when the voice data to be detected meets the noise qualified condition, wherein the voice data to be detected is obtained based on the voice training text, and the voice training text is obtained according to the voice Obtained by data collection instructions;

The sending module is also used to send the voice data to be detected, so that the server generates a voice data set to be trained based on the voice data to be detected, the voice data set to be trained includes N voice data to be trained, and N is a positive value greater than or equal to 1. Integer.

In a possible design, in another implementation manner of another aspect of the embodiments of the present application, the voice playback device further includes a storage module,

The sending module is also used to send a voice model training instruction when N meets the voice training quantity threshold, so that the server obtains the target voice model;

The acquisition module is also used to acquire the target speech model;

The storage module is used to store the target speech model.

In a possible design, in another implementation manner of another aspect of the embodiments of the present application,

an acquisition module, further configured to acquire voice data collection status information, wherein the voice data collection status information is used to indicate that voice data to be detected is being acquired;

The acquisition module is further configured to acquire voice data collection completion status information, wherein the voice data collection completion status information is used to indicate that the voice data has been acquired, and the target voice model can be trained;

The acquisition module is further configured to acquire target voice model training status information, wherein the target voice model training status information is used to indicate that the voice model to be trained is being trained;

The obtaining module is further configured to obtain the completion status information of the target voice model, wherein the target voice model completion status information is used to indicate that the target voice model has completed training.

Another aspect of the present application provides a computer-readable storage medium having instructions stored in the computer-readable storage medium, which, when executed on a computer, cause the computer to perform the methods described in the above aspects.

Another aspect of the present application provides a computer program product or computer program comprising computer instructions stored in a computer readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions to cause the computer device to perform the methods provided in various alternative implementations of the above aspects.

As can be seen from the above technical solutions, the embodiments of the present application have the following advantages:

In the embodiment of the present application, a method for model training is provided. The server first obtains a set of to-be-trained speech data including N pieces of speech data to be trained. When N meets the threshold of the number of speech trainings, it obtains a speech model training instruction, and then responds to The voice model training instruction, based on the set of voice data to be trained, obtains the set of predicted voice data through the voice model to be trained, and further, based on the set of predicted voice data and the set of voice data to be trained, trains the voice model to be trained to obtain the target voice model, The target speech model is sent to the terminal device, so that the terminal device stores the target speech model. In the above manner, since the target speech model is obtained by training the to-be-trained speech model based on the predicted speech data set and the to-be-trained speech data set, the terminal device can obtain the target speech model required by the user from the to-be-trained speech data provided by the user , improve the flexibility of voice model generation, and meet the user's voice customization needs. Secondly, the possibility of voice model selection is improved when users play voice playback, thereby improving the flexibility of voice playback, thereby improving the user's voice playback experience and flexibility. .

Description of drawings

Fig. 1 is a schematic diagram of the architecture of model training in the embodiment of the present application;

FIG. 2 is another schematic diagram of the architecture of model training in the embodiment of the present application;

3 is a schematic flowchart of a method for model training in the embodiment of the present application;

4 is a schematic diagram of an embodiment of a method for model training in an embodiment of the present application;

5 is another schematic flowchart of a method for model training in an embodiment of the present application;

6 is a schematic flowchart of a method for voice playback in an embodiment of the present application;

7 is a schematic diagram of an embodiment of a method for voice playback in an embodiment of the present application;

8 is a schematic diagram of an embodiment of an operation for voice playback in an embodiment of the present application;

FIG. 9 is a schematic diagram of another embodiment of a method for voice playback in an embodiment of the present application;

10 is another schematic flowchart of the method for voice playback in the embodiment of the application;

FIG. 11 is a schematic diagram of another embodiment of a method for voice playback in an embodiment of the present application;

FIG. 12 is a schematic diagram of an embodiment of a model training apparatus in an embodiment of the present application;

13 is a schematic diagram of an embodiment of a voice playback device in an embodiment of the application;

FIG. 14 is a schematic diagram of an embodiment of a server in an embodiment of the present application;

FIG. 15 is a schematic diagram of an embodiment of a terminal device in an embodiment of the present application.

Detailed ways

Embodiments of the present application provide a model training method, a voice playback method, a device, and a storage medium, which are used to improve the flexibility of voice model generation and meet the user's voice customization requirements. The possibility of model selection is improved, thereby improving the flexibility of voice playback, thereby improving the user's voice playback experience and flexibility.

The terms "first", "second", "third", "fourth", etc. (if any) in the description and claims of this application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It is to be understood that the data so used may be interchanged under appropriate circumstances such that the embodiments of the application described herein can, for example, be practiced in sequences other than those illustrated or described herein. Furthermore, the terms "comprising" and "corresponding to", and any variations thereof, are intended to cover non-exclusive inclusion, eg, a process, method, system, product or device comprising a series of steps or units not necessarily limited to those expressly listed but may include other steps or units not expressly listed or inherent to these processes, methods, products or devices.

With the rapid development of Internet information technology and the continuous improvement of the quality of life, intelligent terminal equipment is widely used in people's lives, and users' demands for intelligent terminal equipment are also increasing day by day. In order to meet the personalized experience of users on different terminal devices, a way to complete voice playback based on artificial intelligence has been developed. The method provided by this application can be applied to various scenarios in the field of artificial intelligence cloud services, for example, the driver requests voice navigation from the terminal device while driving, or reads a bedtime story by voice on the infant terminal device, or performs During other schedule operations, the terminal device will play news or novels by voice. It should be understood that in practical applications, the artificial intelligence cloud service can also be applied to more other scenarios, so the application scenarios are not exhaustive here.

In order to improve the flexibility of voice playback in the aforementioned scenarios, the present application proposes a model training method and a voice playback method. The method is applied to the model training system shown in FIG. 1 , please refer to FIG. A schematic diagram of the architecture of the model training system in the application embodiment, as shown in the figure, the model training system includes a server and a terminal device, and the client is deployed on the terminal device. The server involved in this application may be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or may provide cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, Cloud servers for basic cloud computing services such as cloud communications, middleware services, domain name services, security services, Content Delivery Network (CDN), and big data and artificial intelligence platforms. The terminal may be a smart phone, a tablet computer, a notebook computer, a desktop computer, a smart speaker, a smart watch, etc., but is not limited thereto. The terminal and the server can be directly or indirectly connected through wired or wireless communication. Secondly, although only six terminal devices, one network and one server are shown in FIG. 1, it should be understood that the example in FIG. 1 is only used for To understand this solution, the number of specific terminal devices, networks and servers should be flexibly determined according to the actual situation.

Specifically, before the user plays the target voice data, the target voice model needs to be acquired first, the terminal device detects the operation for the voice service, and sends the voice service request to the server. The server responds to the voice service request and sends the voice data collection to the terminal device. instruction, the terminal device acquires the voice data to be detected in response to the voice data collection instruction, and sends the voice data to be detected to the server when the voice data to be detected satisfies the noise qualified condition, and the server generates an N voice data to be trained based on the voice data to be trained The voice data set to be trained of the voice data, when N meets the threshold of the number of voice training, the voice model training instruction sent by the terminal device is obtained, and the server responds to the voice model training instruction, based on the voice data set to be trained, and obtains the predicted voice through the voice model to be trained The data set, and then based on the predicted voice data set and the voice data set to be trained, the voice model to be trained is trained to obtain the target voice model, and finally the server sends the target voice model to the terminal device, so that the terminal device stores the target voice model, and when needed When playing the target voice data, the operation for voice playback is detected, and a voice playback request is sent to the server, so that the server obtains the voiceless voice data and sends it to the terminal device, and the terminal device can generate the voiceless voice data and the target voice model according to the voiceless voice data. target voice data, and play the target voice data. Since the target speech model is obtained by training the to-be-trained speech model based on the predicted speech data set and the to-be-trained speech data set, the terminal device can obtain the corresponding target speech model from the to-be-trained speech data provided by the user, improving the performance of the speech model generated. Flexibility, and the possibility of voice model selection is improved when the user performs voice playback, thereby improving the flexibility of voice playback.

Further, please refer to FIG. 2, which is another schematic diagram of the architecture of the model training system in the embodiment of the application. As shown in the figure, the terminal device is integrated with Tencent voice service (tencent voice service, TVS)-software development kit ( software development kit, SDK) and speech synthesis technology (text to speech, TTS)-SDK, and TVS-SDK integrates model management module and message module, and TTS-SDK integrates model management module and voice synthesis. Specifically, the TVS-SDK and TTS-SDK can interact with the TVS service in the server, and the model management module integrated in the TVS-SDK interacts with the TVS-web service in the server, which is specifically used for voice data and model data. For data transmission, the message module integrated in TVS-SDK interacts with the message push service in the server to realize the push of various messages.

Secondly, the server includes TVS service, TVS-web service, TTS customization service, offline training service, message push service and database, wherein TVS service can provide voice services, such as automatic speech recognition technology (automatic speech recognition, ASR), speech synthesis technology, Semantic parsing and skill services, etc. Terminal devices obtain TVS-specific services through semantic query, such as querying news, novels, playing music, navigation and other services. TVS-web service is a gateway service that provides non-voice interactive interfaces, voice The customized interface is provided through this service. The TTS customized service is responsible for receiving the training request initiated by the terminal device, collecting the voice data of the terminal device, creating and triggering the offline training task, writing the task record into the database, and returning the task submission success status. At the same time, the task status will be continuously inquired. If the task is completed, a message will be sent to the terminal device through the message push service to notify the terminal device that the target model training has been completed. The offline training service uses the AI deep learning method to perform feature extraction and training on the training voice data. Finally, the target voice model is obtained, the database is updated, the target training model data is saved, and the task status is changed to a successful status. The message push service has a channel for pushing messages to the terminal device. The channel capability is used in the embodiment of the present application to make the terminal device Know the training result of the target speech model. Therefore, after the terminal device receives the message pushed by the message push server in the server, it jumps to the model page through the message, selects the target voice model, and then triggers the TTS-SDK to store the trained target voice model locally. When the terminal device When TVS services are triggered, such as when a novel is broadcast, TTS-SDK combines the target voice model and the voiceless voice data synthesized by TTS data in the server to generate target voice data, and play the target voice data to realize voice broadcast.

Since the embodiments of the present application are applied in the field of cloud technology, some basic concepts in the field of cloud technology are first introduced before the specific introduction of the voice playback method provided by the embodiments of the present application. Cloud technology refers to a kind of hosting technology that unifies a series of resources such as hardware, software, and network in a wide area network or a local area network to realize the calculation, storage, processing and sharing of data. Cloud technology is a general term for network technology, information technology, integration technology, management platform technology, application technology, etc. based on the application of cloud computing business models. It can form a resource pool, which can be used on demand, flexible and convenient. Cloud computing technology will become an important support. Background services of technical network systems require a lot of computing and storage resources, such as video websites, picture websites and more portal websites. With the high development and application of the Internet industry, in the future, each item may have its own identification mark, which needs to be transmitted to the back-end system for logical processing. Data of different levels will be processed separately, and all kinds of industry data need to be strong. The system backing support can only be achieved through cloud computing.

With the research and progress of cloud technology, cloud technology has been researched in various directions, and artificial intelligence (Artificial Intelligence, AI) cloud services are generally also called artificial intelligence as a service (AI as a Service, AIaaS). This is the current mainstream service method of artificial intelligence platforms. Specifically, the AIaaS platform will split several types of common AI services and provide independent or packaged services in the cloud. This service model is similar to opening an AI-themed mall. All developers can access one or more artificial intelligence services provided by the platform through Application Programming Interface (API). Some experienced developers can also use the AI framework and AI infrastructure provided by the platform to deploy and operate their own cloud AI services.

Secondly, artificial intelligence cloud services are required in the field of artificial intelligence applications, so some basic concepts of artificial intelligence are introduced. Artificial intelligence is a theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technique of computer science that attempts to understand the essence of intelligence and produce a new kind of intelligent machine that can respond in a similar way to human intelligence. Artificial intelligence is to study the design principles and implementation methods of various intelligent machines, so that the machines have the functions of perception, reasoning and decision-making. Artificial intelligence technology is a comprehensive discipline, involving a wide range of fields, including both hardware-level technology and software-level technology. The basic technologies of artificial intelligence generally include technologies such as sensors, special artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation/interaction systems, and mechatronics. Artificial intelligence software technology mainly includes computer vision technology, speech processing technology, natural language processing technology, and machine learning/deep learning. Machine Learning (ML) is a multi-field interdisciplinary subject involving probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory and other disciplines. It specializes in how computers simulate or realize human learning behaviors to acquire new knowledge or skills, and to reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to make computers intelligent, and its applications are in all fields of artificial intelligence. Machine learning and deep learning usually include artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, teaching learning and other technologies.

With the research and progress of artificial intelligence technology, artificial intelligence technology has been researched in various directions. The key technologies of speech technology (SpeechTechnology) include automatic speech recognition technology (ASR) and speech synthesis technology (TTS) and voiceprint recognition technology. Making computers able to hear, see, speak, and feel is the development direction of human-computer interaction in the future, and voice will become one of the most promising human-computer interaction methods in the future. Natural language processing (NLP) is an important direction in the field of computer science and artificial intelligence. It studies various theories and methods that can realize effective communication between humans and computers using natural language. Natural language processing is a science that integrates linguistics, computer science, and mathematics. Therefore, research in this field will involve natural language, the language that people use on a daily basis, so it is closely related to the study of linguistics. Natural language processing technology usually includes text processing, semantic understanding, machine translation, robot question answering, knowledge graph and other technologies.

Based on this, a method for voice playback will be introduced below, please refer to FIG. 3 , which is a schematic flowchart of the method for model training in the embodiment of the application, as shown in the figure, specifically:

In step S1, the terminal device sends the voice data to be trained to the server, where the voice data to be trained is the voice data obtained by the user through voice input.

In step S2, when the quantity of the voice data to be trained sent by the terminal device to the server meets the voice training quantity threshold, the terminal device may also send a voice model training instruction to the server.

In step S3, the server can obtain the to-be-trained voice data set including N to-be-trained voice data through the aforementioned steps, and then the server responds to the voice model training instruction, takes the to-be-trained voice data set as the input of the to-be-trained voice model, and passes The voice model to be trained acquires a predicted voice data set including N predicted voice data.

In step S4, the server trains the speech model to be trained based on the predicted speech data set and the to-be-trained speech data set to obtain the target speech model.

In step S5, the server sends the target speech model to the terminal device.

In step S6, the terminal device stores the target speech model.

In step S7, when the user needs to perform voice playback, the terminal device detects the operation for voice playback, inputs the voice data to be parsed through voice, generates a voice playback request carrying the voice data to be parsed, and sends the voice playback to the server ask.

In step S8, the server obtains service content text data based on the to-be-parsed voice data carried in the voice playback request, and obtains timbreless voice data according to the service content text data.

In step S9, the server sends the voiceless voice data to the terminal device.

In step S10, after acquiring the voiceless voice data, the terminal device generates target voice data according to the acquired voiceless voice data and the stored target voice model.

In step S11, the target voice data is played.

The solutions provided by the embodiments of the present application relate to the artificial intelligence cloud service technology in the cloud technology. In combination with the above introduction, the following will introduce the voice playback method in the present application. Please refer to FIG. 4 , which is a schematic diagram of model training in the embodiments of the present application. A schematic diagram of an embodiment of the method, as shown in the figure, an embodiment of the method for model training in the embodiment of the present application includes:

101. Obtain a voice data set to be trained, wherein the voice data set to be trained includes N voice data to be trained, and N is a positive integer greater than or equal to 1;

In this embodiment, the server acquires the to-be-trained voice data set including N to-be-trained voice data, that is, it can receive the to-be-trained voice data set sent by the terminal device in real time, or it can obtain the to-be-trained voice data set from the database of the terminal device, Or acquire the voice data set to be trained from the database of the server, and specifically, the manner of acquiring the voice data set to be trained is not limited here.

It should be understood that after the server obtains the voice data set to be trained through step 101, if the user needs to train other voice models, the server can also continue to make the server obtain the voice data sets required by the other voice models, that is, the target voice model does not need to complete the training and then Get started.

102. When N meets the voice training quantity threshold, obtain a voice model training instruction;

In this embodiment, when the quantity of the to-be-trained voice data in the to-be-trained voice data set reaches the threshold of the voice training quantity, the terminal device can obtain the voice data collection completion status information, and can learn the terminal device voice data according to the voice data collection completion status information After the acquisition is completed, the training of the target voice model can be started. Therefore, the voice model training instruction is sent to the server, and the server can obtain the voice model training instruction sent by the terminal device.

Specifically, in this embodiment, the threshold for the number of voice training is 20, that is, after the set of voice data to be trained includes 20 voice data to be trained, the terminal device can send a voice model training instruction to the server, and the server obtains the voice model training instruction, and should It is understood that in practical applications, the threshold of the number of voice training can also be other positive integers such as 15, 25, and 30. The specific threshold of the number of voice training should be flexibly determined according to the actual situation and needs of the voice model to be trained.

103. In response to the voice model training instruction, based on the voice data set to be trained, a predicted voice data set is obtained through the voice model to be trained, wherein the predicted voice data set includes N predicted voice data, and the predicted voice data and the to-be-trained voice data have Correspondence;

In this embodiment, after the server receives the voice model training instruction acquired in step 102, it can know that the terminal device needs to train the voice model to be trained, so it can respond to the voice model training instruction to use the voice model to be trained acquired in step 101. The data set is used as the input of the to-be-trained speech model, and the to-be-trained speech model can output a predicted speech data set including N pieces of predicted speech data, and the predicted speech data has a corresponding relationship with the to-be-trained speech data.

Exemplarily, the to-be-trained speech data set includes the to-be-trained speech data A to the to-be-trained speech data T, a total of 20 to-be-trained speech data, then the to-be-trained speech data set is used as the input of the to-be-trained speech model, to be trained. The predicted voice data set output by the voice model includes predicted voice data A to predicted voice data T, a total of 20 predicted voice data, and the to-be-trained voice data A corresponds to the predicted voice data A, that is, the to-be-trained voice data A is used separately. The input of the to-be-trained speech model, the to-be-trained speech model can output predicted speech data A, and the to-be-trained speech data T also corresponds to the predicted speech data T;

104. Based on the predicted speech data set and the to-be-trained speech data set, train the to-be-trained speech model to obtain a target speech model;

In this embodiment, the server performs feature extraction through the offline training module and the deep learning algorithm in combination with the predicted voice data set obtained in step 103 and the to-be-trained voice data set obtained in step 101, and the to-be-trained voice model is trained. , the target speech model can be obtained, and after the training is completed, the obtained target speech model is stored in the database in the server.

Exemplarily, if the user takes the voice of "Dad" as the voice data to be trained, then the target voice model obtained by training can be the voice model of the corresponding voice of "Dad". If the user takes the voice of "Mom" as the voice data to be trained, Then the target speech model obtained by training can be the speech model corresponding to the voice of "mother", and the specific target speech model needs to be flexibly determined by the actual input of the speech data to be trained.

105. Send the target voice model, so that the terminal device stores the target voice model.

In this embodiment, after the server obtains the target voice model through step 104, it can also send the target voice model to the terminal device. After the terminal device receives the target voice model, it can store the target voice model, that is, the terminal device can store the target voice model through the integrated TTS-SDK. The target voice model is loaded into the data structure of the terminal device, and when the terminal device needs to play the target voice, the target voice data can be generated according to the stored target voice model, thereby completing the voice play.

In the embodiment of the present application, a method for model training is provided. Using the above method, since the target speech model is obtained by training the speech model to be trained based on the predicted speech data set and the to-be-trained speech data set, the terminal device can be trained by the user. The provided voice data to be trained gets the corresponding target voice model, which improves the flexibility of voice model generation and meets the user's voice customization requirements. Secondly, the possibility of voice model selection is improved when the user performs voice playback, thereby improving the voice playback performance. flexibility, thereby enhancing the user's voice playback experience and flexibility.

Optionally, on the basis of the above-mentioned embodiment corresponding to FIG. 4 , in an optional embodiment of the model training method provided by the embodiment of the present application, acquiring a set of speech data to be trained specifically includes the following steps:

Get voice service request;

In response to the voice service request, a voice data collection instruction is sent to the terminal device, so that the terminal device, in response to the voice data collection instruction, obtains the voice data to be detected, wherein the voice data to be detected is obtained based on the voice training text, and the voice training text is Obtained according to voice data collection instructions;

When the voice data to be detected satisfies the qualified conditions for voice training, determine the voice data to be trained;

Based on the to-be-trained speech data, a to-be-trained speech data set is generated.

In this embodiment, a method for acquiring a voice data set to be trained is provided. When a user needs to perform voice model training, he can perform operations on the terminal device for voice services, including but not limited to clicking on the voice training interface. , double-click the voice training interface, and slide the voice training interface. It should be understood that, in practical applications, this operation can also be controlled by voice and gesture. For example, for voice control, the operation may be the user triggering the voice service through the voice input device (eg, the user says "customize my voice"), and for gesture control, the operation may be triggering the gesture through the user's corresponding gesture instruction, the operation is not limited here.

Therefore, the server can obtain the voice service request, know that the terminal device needs to perform a voice model, trigger the TVS service in the server, and in response to the voice service request, generate a voice data collection instruction, thereby sending voice data collection to the terminal device. Command, the terminal device can know that the server needs to obtain the voice data for training according to the voice data collection command. At this time, the terminal device needs to determine whether the voice data to be detected meets the noise qualification conditions. If the noise qualification conditions are met, the voice data collection command is performed. Obtain the corresponding voice training text, and then the user can perform voice input for the voice training text, and the terminal device obtains the voice data to be detected input by the user's voice, and then sends the voice data to be detected to the server. The TVS-web interface transmits the voice data to be detected to the TTS customization module, and then judges whether the voice data to be detected satisfies the eligibility conditions for voice training. If so, it is determined that the voice data to be detected is the voice data to be trained. The voice input obtains the corresponding voice data to be detected, and sends the voice data to be detected to the server, so that the server judges the voice data to be detected according to the above-mentioned similar steps, and thus obtains N voice data to be trained through the above-mentioned similar steps, that is, A set of speech data to be trained can be generated.

Specifically, the qualified condition for voice training is that the similarity between the voice data to be detected and the voice training text is greater than a preset threshold. In this embodiment, the similarity between the voice data to be detected and the voice training text should be greater than 95%. In the application, the qualified condition for voice training can also be that the similarity between the voice data to be detected and the voice training text should be greater than 98%, or the similarity between the voice data to be detected and the voice training text should be greater than 90%, etc. The training eligibility conditions need to be determined flexibly according to the actual situation and demand. In order to facilitate understanding, the eligibility condition for voice training is that the similarity between the voice data to be detected and the voice training text is greater than 95% as an example, and the voice training text is "jump with me". The text data obtained by semantic recognition of the detected speech data is "jump with me", then it can be said that the similarity between the speech data to be detected and the speech training text is 100%, that is, the qualified condition of speech training is satisfied. Secondly, the voice training text is "Eight hundred pacesetters run to the North Slope, North Slope artillerymen run side by side", and the text data obtained after semantic recognition of the speech data to be detected through the user's voice input is "Eight hundred pacesetters run to the North Slope, North Slope The pacesetter does not need to run”, then the similarity between the speech data to be detected and the speech training text is less than 95%, which does not meet the qualification requirements for speech training.

For ease of understanding, please refer to FIG. 5 , which is another schematic flowchart of the method for model training in the embodiment of the present application. As shown in the figure, in step A1, the terminal device sends a voice service request to the server, and in step A2, After the server receives the voice service request, in response to the voice service request, it sends a voice data collection instruction to the terminal device. In step A3, after the terminal device receives the voice data collection instruction, in response to the voice data collection instruction, it first determines the voice data to be detected. Whether the data meets the noise qualification conditions, if so, obtain the voice training text according to the voice data collection instruction, and then the user performs voice input based on the voice training text to obtain the voice data to be detected, so that the terminal device obtains the voice data to be detected. In step A4 , send the acquired voice data to be detected to the server, in step A5, the server determines that the voice data to be detected is the voice data to be trained when the server determines that the voice data to be detected satisfies the voice training eligibility conditions, in step A6, then repeat From steps A1 to A5, based on the determined speech data to be trained, a set of speech data to be trained is generated.

It should be understood that the foregoing examples are only used to understand this solution. In practical applications, the voice training text needs to be determined by the actual needs and design of the voice model, and the voice data to be detected is determined based on the user's voice input, so whether it satisfies the voice training. Eligibility conditions need to be flexibly determined through the speech training text and the actual situation of the speech data to be detected.

In the embodiment of the present application, a method for obtaining a voice data set to be trained is provided. In the above manner, the voice data to be detected is obtained through a voice data collection instruction, and the voice data to be detected that meets the eligibility conditions for voice training is determined as the voice data to be trained. voice data, thereby generating a set of voice data to be trained including at least one voice data to be trained, thereby providing input data for the voice model to be trained, thereby improving the feasibility of this solution.

Optionally, on the basis of the embodiment corresponding to FIG. 4 above, in another optional embodiment of the model training method provided by the embodiment of the present application, based on the predicted speech data set and the to-be-trained speech data set, the to-be-trained speech model is Carry out training to obtain the target speech model, which specifically includes the following steps:

Based on the predicted speech data set and the to-be-trained speech data set, the model parameters of the to-be-trained speech model are updated according to the target loss function;

If the target loss function reaches convergence, the target speech model is generated according to the model parameters.

In this embodiment, a method for training a target speech model is provided. Since the server needs to target the speech data set to be trained, and train the model parameters of the speech model to be trained, the server obtains one of the predicted speech data sets when After the voice data is predicted, a voice data to be trained corresponding to the aforementioned one predicted voice data can be obtained from the set of voice data to be trained, and then a target loss function can be generated according to the predicted voice data and the corresponding voice data to be trained. value, according to the value of the target loss function to determine whether the target loss function reaches the convergence condition, if the convergence condition is not reached, use the value of the target loss function to update the model parameters of the speech model to be trained. It can be seen from this that the server can perform the aforementioned operation once every time the speech model to be trained generates a predicted speech data, until the target loss function reaches the convergence condition, and then generates the target speech model according to the model parameters obtained after updating the model parameters for the last time.

The objective loss function may be the sum of the differences between the predicted speech data and the corresponding to-be-trained speech data, or may be the sum of the absolute values of the differences between the predicted speech data and the corresponding to-be-trained speech data , it can also be the square of the sum of the difference between the predicted speech data and the corresponding to-be-trained speech data, or it can be an objective loss function of other forms, etc. The choice of the specific objective loss function can be determined according to the actual situation. There are no restrictions. The convergence condition of the objective loss function may be that the value of the objective loss function is less than or equal to the first preset threshold, as an example, for example, the value of the first preset threshold may be 0.005, 0.01, 0.02 or other values approaching 0; It can also be that the difference between two adjacent values of the objective loss function is less than or equal to a second preset threshold, and the value of the second threshold can be the same as or different from the value of the first threshold. The value of the threshold may be 0.005, 0.01, 0.02, or other values close to 0, etc. The server may also adopt other convergence conditions, etc., which are not limited here.

In the embodiment of the present application, a method for training a target speech model is provided. Based on the predicted speech data set and the to-be-trained speech data set, the model parameters of the to-be-trained speech model are updated according to the target loss function. The model parameters generate the target speech model. The above method provides a specific implementation method for the server to update the parameters of the speech model to be trained, and thereby generates the target speech model, thus improving the practicability of the solution.

Optionally, on the basis of the embodiment corresponding to FIG. 4 above, in another optional embodiment of the model training method provided by the embodiment of the present application, the model training method further includes the following steps:

obtaining a voice playback request, wherein the voice playback request carries the voice data to be parsed;

Obtaining service content text data according to the voice data to be parsed, wherein the service content text data corresponds to the voice data to be parsed;

Acquiring timbreless voice data according to the service content text data, wherein the timbreless voice data is obtained by performing speech synthesis on the service content text data;

The voiceless voice data is sent to the terminal device, so that the terminal device generates the target voice data according to the voiceless voice data and the target voice model, and plays the target voice data.

In this embodiment, when the user needs to perform voice playback, the terminal device can detect the operation for voice playback, and input the voice data to be parsed through voice, thereby generating a voice play request carrying the voice data to be parsed, and sending it to the server A voice playback request, from which the server obtains the voice playback request, and performs voice recognition on the to-be-parsed voice data carried in the voice playback request. The specific voice recognition can be semantic-to-text and semantic understanding, etc., and then determines the user's intention through voice recognition. and demand, so as to determine the service content text data, and then perform speech synthesis on the service content text data to complete the conversion of the text data to the voice data, thereby obtaining the voiceless voice data, the server sends the TTS command to the terminal, and the TTS command carries whether or not timbre voice data, so after the terminal obtains the TTS instruction carrying the timbre-free voice data, it can generate target voice data according to the timbre-free voice data and the stored target voice model, and play the target voice data, thereby completing voice playback.

Exemplarily, take the voice data to be parsed carried in the voice playback request as "I want to listen to the ugly duckling" as an example to introduce, please refer to Figure 6, Figure 6 is a schematic flowchart of the method for voice playback in the embodiment of the application, as shown in Figure 6. As shown in the figure, in step F1, the user initiates a voice playback request to the terminal device according to the requirements, the voice data to be parsed carried in the voice playback request is "I want to listen to the ugly duckling", and in step F2, the terminal device sends the voice playback to the server Request, in step F3, the TVS service in the server can determine that the user's intention is to listen to the story after the process of speech recognition such as semantic-to-text and semantic understanding, and the content that he wants to listen to is "the ugly duckling story". The skill service in the TVS service initiates a request, specifically for the children's story content service, and then the skill service can feed back the service content text data of "The Ugly Duckling Story" in the children's story content. In step F4, the server sends the service content text data "The Ugly Duckling Story" "Conduct text to voice conversion, thereby obtaining voiceless voice data, the voiceless voice data is voiceless voice data whose voice content is the fairy tale "The Ugly Duckling" voiceless voice data, and then in step F5, the server sends no voice to the terminal device. Tone voice data, in step F6, the terminal device generates target voice data according to the voiceless voice data and the target voice model, and plays the target voice data in step F7, that is, the story content of "The Ugly Duckling" is played through the target voice, for example, " Once upon a time, there was an ugly duckling...". It should be understood that the foregoing examples are only used to understand this solution, and in practical applications, the specific voiceless voice data needs to be flexibly determined according to the actual situation of the voice data to be parsed.

In the embodiment of the present application, a method for playing voice data based on a target voice model is provided. The server obtains service content text data according to the voice data to be parsed carried in the voice play request, and sends the service content text data obtained according to the service content text data to the terminal device. For voiceless voice data, in the above manner, the terminal device can generate target voice data according to the voiceless voice data and the target voice model, and play the target voice data, thus improving the feasibility of this solution.

The method for voice playback in the present application will be introduced below. Please refer to FIG. 7 . FIG. 7 is a schematic diagram of an embodiment of the method for voice playback in the embodiment of the present application. As shown in the figure, the method for voice playback in the embodiment of the present application is implemented. Examples include:

201. Detect an operation for voice playback, and send a voice playback request, so that the server obtains service content text data based on the voice data to be parsed, and obtains voiceless voice data according to the service content text data, wherein the voice playback request carries the voice data to be parsed. , the service content text data is obtained according to the voice data to be parsed, and the timbreless voice data is obtained after voice synthesis is performed on the service content text data;

In this embodiment, when the user needs to perform voice playback, the terminal device can detect the operation for voice playback, and perform voice input through the voice input interface to obtain the voice data to be parsed, and generate a voice playback request carrying the voice data to be parsed, Then the terminal device can send a voice playback request to the server through the TVS-SDK, so that the general TVS voice service of the server performs voice recognition on the voice data to be parsed carried in the voice playback request to obtain the service content text data, and then performs the service content text data on the service content text data. Speech synthesis results in voiceless speech data.

Specifically, the operations for voice playback include, but are not limited to, single-clicking the voice training interface, double-clicking the voice training interface, and sliding the voice training interface. For ease of understanding, the terminal device is a mobile phone, and the operation for voice playback is to click the voice The training interface is introduced as an example, please refer to FIG. 8 . FIG. 8 is a schematic diagram of an embodiment of the operation for voice playback in the embodiment of the application. As shown in the figure, B1 is used to indicate the voice playback interface, and B2 is used to indicate the voice playback. Input interface, Figure (A) in Figure 8 is used to instruct the user to click on the voice playback interface B1, and then jump to the interface shown in Figure (B) in Figure 8, Figure (B) in Figure 8 An interface used to instruct the user to perform voice input through the voice input interface B2. For example, after the user clicks on the voice input interface B2 on the interface shown in (B) in Figure 8, and says "I want to listen to the ugly duckling" to the terminal device, the terminal device obtains To the speech data to be parsed "I want to listen to the ugly duckling". It should be understood that the foregoing examples are only used to understand this solution, and in practical applications, the operation may also be control by voice and control by gesture. For example, for voice control, the operation may be that the user triggers a voice service through a voice input device (for example, the user speaks "voice play"), and for gesture control, the operation may be to trigger a gesture instruction through the user's corresponding gesture, Specifically, this operation is not limited here. Secondly, in practical applications, the shape, size, position and corresponding text information of the voice playback interface and the voice input interface need to be flexibly determined according to the actual situation.

202. Obtain voiceless voice data;

In this embodiment, after obtaining the voiceless voice data in step 201, the server sends the voiceless voice data to the terminal device, and the TVS-SDK integrated in the terminal device obtains the voiceless voice data thereby. For example, if the voice data to be parsed is "I want to listen to the ugly duckling", it can be seen from the foregoing embodiment that the voiceless voice data is the voice data of the fairy tale "The Ugly Duckling" with no tonal timbre. If the voice data to be parsed is "I want to listen to it" "Ugly", similar to the aforementioned embodiment, the voiceless voice data may be voice data with the voice content of the song "Ugly" and no timbre. It should be understood that the foregoing examples are only used to understand this solution. In practical applications, the specific voice data to be parsed needs to be flexibly determined according to the actual situation of the user's needs, and the specific voiceless voice data needs to be flexibly determined according to the actual situation of the voice data to be parsed. .

203. Generate target voice data according to the voiceless voice data and the target voice model, wherein the target voice model is obtained by training the voice model to be trained based on the predicted voice data set and the voice data set to be trained, and the predicted voice data set is the server response. Based on the voice model training instruction, based on the voice data set to be trained, obtained through the voice model to be trained;

In this embodiment, after the TVS-SDK integrated in the terminal device obtains the voiceless voice data, it transmits the voiceless voice data to the TTS-SDK, and the user can select the target voice model at this time. It can be seen from the foregoing embodiments that the target voice model is Based on the predicted voice data set and the to-be-trained voice data set, the voice model to be trained is obtained by training the to-be-trained voice model. The predicted voice data set is obtained by the server in response to the voice model training instruction and based on the to-be-trained voice data set through the to-be-trained voice model. Therefore, It is not repeated here.

Exemplarily, if the user takes the voice of "Dad" as the voice data to be trained, then the target voice model can be the voice model corresponding to the voice of "Dad". If the user takes the voice of "Mom" as the voice data to be trained, then the target voice model A speech model that can correspond to a voice for "mom". If the target voice model selected by the user is the voice model corresponding to the voice of "Mom", then TTS-SDK will use the voiceless voice data as the input of the target voice model, and the target voice model will output the target voice data, that is, output the voice tone of "Mom" corresponding voice data. It should be understood that the foregoing examples are only used to understand this solution. In practical applications, the specific target speech data needs to be flexibly determined according to the timbreless speech data and the actual situation of the target speech model selected by the user.

204. Play the target voice data.

In this embodiment, after acquiring the target voice data, the terminal device can play the target voice data. For ease of understanding, the target voice model is the voice model corresponding to the voice of "mother" as an example for description, please refer to FIG. 9 . FIG. 9 is a schematic diagram of another embodiment of the voice playback method in the embodiment of the application, as shown in the figure , C1 is used to indicate the voice input interface. After the user clicks on the voice input interface C1, he says "I want to listen to the ugly duckling" to the terminal device, then the terminal device obtains the voice data to be parsed "I want to listen to the ugly duckling". The step can obtain the target voice data whose text content is the fairy tale "The Ugly Duckling", so the terminal device plays the target voice data "Once upon a time, there was an ugly duckling..." in the timbre of "Mama". It should be understood that the foregoing examples are only used to understand this solution. In practical applications, the specific target speech data needs to be flexibly determined according to the timbreless speech data and the actual situation of the target speech model selected by the user.

In the embodiment of the present application, a method for playing voice is provided. By adopting the above method, the terminal device can generate target voice data according to the voiceless voice data and the target voice model. Since the target voice model is the voice data to be trained provided by the user Therefore, the flexibility of generating the voice model is improved, thereby improving the flexibility of generating the target voice data. Secondly, the generated target voice data is played. Due to the improvement of the flexibility of generating the target voice data, the voice playback can be improved. flexibility.

Optionally, on the basis of the above-mentioned embodiment corresponding to FIG. 7 , in an optional embodiment of the voice playback method provided by the embodiment of the present application, the voice playback method further includes the following steps:

Detect operations for voice services, and send voice service requests;

When the server responds to the voice service request, obtain the voice data collection instruction;

In response to the voice data collection instruction, and when the to-be-detected voice data meets the noise qualified condition, acquire the to-be-detected voice data, wherein the to-be-detected voice data is obtained based on the voice training text, and the voice training text is obtained according to the voice data collection instruction;

The voice data to be detected is sent, so that the server generates a voice data set to be trained based on the voice data to be detected. The voice data set to be trained includes N voice data to be trained, where N is a positive integer greater than or equal to 1.

In this embodiment, a method for sending voice data to be detected is provided. When the user needs to perform voice model training, the terminal equipment can be operated for the voice service. The terminal equipment detects the operation for the voice service. After sending a voice service request, the server can obtain the voice service request, learn the needs of the terminal device to perform a voice model, trigger the TVS service in the server, and in response to the voice service request, generate a voice data collection instruction, thereby sending it to the terminal device. Voice data collection instruction, the terminal device can know that the server needs to obtain the voice data for training according to the voice data collection instruction, and then first judges whether the voice data to be detected satisfies the noise qualified condition, and if so, obtains the corresponding voice data according to the voice data collection instruction Then the user can perform voice input for the voice training text, and the terminal device obtains the to-be-detected voice data input by the user's voice.

Further, the terminal device sends the voice data to be detected to the server, and after the server obtains the voice data to be detected, it transmits the voice data to be detected to the TTS customization module through the TVS-web interface, and judges whether the voice data to be detected satisfies the voice training eligibility conditions, If it is satisfied, it is determined that the to-be-detected voice data is the to-be-trained voice data. If not, the terminal device continues to send the to-be-detected voice data that meets the noise qualified conditions to the server, and the server judges the to-be-detected voice data according to the above similar steps. After obtaining N pieces of speech data to be trained in the above similar steps, a set of speech data to be trained can be generated.

Specifically, the qualified condition for noise is that the ambient noise is lower than the preset threshold of decibels. In this embodiment, the qualified condition for noise is that the ambient noise is lower than 45 decibels. It should be understood that in practical applications, if more accurate speech data to be detected is required, The qualified noise condition can also be that the ambient noise is lower than 40 decibels, or the voice data to be detected needs to be obtained more quickly. The qualified noise condition can also be that the ambient noise is lower than 50 decibels. The specific noise qualified condition can be determined flexibly according to the actual situation and needs.

For ease of understanding, please refer to FIG. 10. FIG. 10 is another schematic flowchart of the method for voice playback in the embodiment of the application. As shown in the figure, in step D1, the user performs an operation for voice service, thereby generating a voice service request, In step D2, the terminal device detects the operation for the voice service, and sends a voice service request to the server. In step D3, the server responds to the voice service request and sends a voice data collection instruction to the terminal device. In step D4, the terminal device In response to the voice data collection instruction, it is judged whether the voice data to be detected satisfies the noise qualified condition, and if so, step D5 is performed. In step D5, the terminal device obtains the voice training text according to the voice data collection instruction, and the user enters the voice input to be detected based on the voice training text. Voice data, in step D6, send the voice data to be detected to the server, in step D7, the server determines that the voice data to be detected is the voice data to be trained when the server determines that the voice data to be detected satisfies the voice training eligibility conditions, and based on the determined voice data The voice data to be trained is generated, and a voice data set to be trained is generated. In step D8, when the number of voice data to be trained in the voice data set to be trained reaches the threshold of the voice training quantity, the terminal device can obtain the voice data collection completion status information, according to The voice data collection completion status information shows that the terminal device voice data has been acquired, and the target voice model can be trained, so the server sends a voice model training instruction to the server. In step D9, after the server receives the voice model training instruction, it can know that the terminal device needs to train the voice model to be trained, so it can perform model training in response to the voice model training instruction to obtain the target voice model. In step D10, the server The target voice model is sent to the terminal device, and the terminal device stores the target voice model in step D11. In step D12, when the user needs to customize voice playback, the terminal device can send a voice playback request to the server. In step D13, the server obtains service content text data according to the voice data to be parsed. In step D14, the server obtains service content text data according to the service content text. The data obtains voiceless voice data, then in step D15, the server sends voiceless voice data to the terminal device, in step D16, the terminal device generates target voice data according to the voiceless voice data and the target voice model, and in step D17, the terminal device Play the target voice data. The specific manner in the foregoing steps has been introduced in the foregoing embodiments, and thus will not be repeated here. It should be understood that the foregoing examples are only used to understand this solution. In practical applications, whether the specific voice data to be detected is to be sent by the server needs to be flexibly determined according to the user's voice input.

In the embodiment of the present application, a method for sending voice data to be detected is provided. By adopting the above method, when the qualified noise condition is satisfied, the voice data to be detected is obtained according to user requirements, and the voice data to be detected is sent to the server, thereby improving the The accuracy of the voice data to be detected is used to facilitate the feature extraction of the voice data to be detected by the voice model to be trained, thereby improving the accuracy and efficiency of generating the target voice model, thereby improving the accuracy and efficiency of voice playback.

Optionally, on the basis of the embodiment corresponding to FIG. 7 above, in another optional embodiment of the voice playback method provided by the embodiment of the present application, the voice playback method further includes the following steps:

When N meets the voice training quantity threshold, send a voice model training instruction so that the server obtains the target voice model;

Get the target speech model;

Store the target speech model.

In this embodiment, a method for storing a target voice model is provided. After the terminal device sends the voice data to be detected to the server through the foregoing steps, when the server determines that the quantity of the voice data to be trained in the set of voice data to be trained reaches the threshold of the voice training quantity The terminal device can obtain the voice data collection completion status information. According to the voice data collection completion status information, it can be known that the terminal device voice data has been acquired, and the model training can be started. Therefore, the voice model training instruction is sent to the server, and the server responds to the voice The model training instruction, based on the set of voice data to be trained, obtains the set of predicted voice data through the voice model to be trained, and trains the voice model to be trained based on the set of predicted voice data and the set of voice data to be trained, obtains the target voice model, and then sends it to the terminal The device sends the target voice model, and the terminal device stores the acquired target voice model. When the terminal device needs to play the target voice, it can generate target voice data according to the stored target voice model, so as to complete the voice play.

In the embodiment of the present application, a method for storing a target voice model is provided. Using the above method, the training of the target voice model can be started according to user requirements, and the obtained target voice model can be stored in this way. The voice model generates target voice data to complete voice playback and improve the feasibility of this solution.

Optionally, on the basis of the embodiment corresponding to FIG. 7 above, in another optional embodiment of the voice playback method provided by the embodiment of the present application, the voice playback method further includes the following steps:

acquiring voice data collection status information, wherein the voice data collection status information is used to indicate that voice data to be detected is being acquired;

Obtaining voice data collection completion status information, wherein the voice data collection completion status information is used to indicate that the voice data has been acquired, and the target voice model can be trained;

Acquiring training status information of the target voice model, wherein the training status information of the target voice model is used to indicate that the voice model to be trained is being trained;

Acquiring the completion status information of the target voice model, where the target voice model completion status information is used to indicate that the target voice model has completed training.

In this embodiment, the terminal device can also acquire different status information, and learn the status of the model training based on the different status information. Specifically, the terminal device can acquire voice data collection status information, voice data collection completion status information, and target voice data. Model training status information and target speech model completion status information. Among them, the terminal device can know that the voice data to be detected is being acquired through the voice data collection status information, and the terminal device can know that the voice data has been obtained through the voice data collection completion status information, that is, N meets the voice training quantity threshold, and can start training. The target voice model, so the terminal device sends voice model training instructions to the server based on the voice data collection completion status information. Further, the terminal device can know that the server is training the voice model to be trained through the target voice model training status information, and the terminal device can know that the server target voice model has completed training through the target voice model completion status information, that is, the server has obtained. The target voice model, and the target voice model is stored in the data of the server, and then the server sends the target voice model to the terminal device, and the terminal device stores the target voice model that has been trained. When the user requests the TTV service through the operation for voice playback , the terminal device can use the stored target voice model and the voiceless voice data to synthesize target voice data.

For ease of understanding, please refer to FIG. 11. FIG. 11 is a schematic diagram of another embodiment of the voice playback method in this embodiment of the application. As shown in the figure, in step E1, the terminal device continuously requests the server for detection model training status information. The voice data to be detected is being acquired. In step E2, the terminal device may acquire voice data collection status information. If the server generates a to-be-trained voice data set including N to-be-trained voice data, and N satisfies the voice training quantity threshold, in step E2, the terminal device may obtain voice data collection completion status information. At this point, the terminal device sends a voice model training instruction to the server, and the server starts model training. In step E3, the TTS customization service requests model status information from the database. If the model training is completed to obtain the target voice model, step E4 is performed. If not completed, step E2 is performed, and in step E2, the training state information of the target speech model is sent to the terminal device. In step E4, the offline training service sends the target voice model to the database. At this time, the TTS customization service can know that the model has been trained. Therefore, in step E5, the TTS customization service sends the target voice model completion status information to the message push service. , in step E2, the message push service in the server sends the target voice model completion status information to the terminal device. Secondly, after the TVS-SDK of the terminal device receives the completion status information of the target voice model, it can also trigger the interface to call the TTS-SDK to check the model status. After determining that the target voice model has been generated through the completion status information of the target voice model, the model management module obtains the target The voice model is saved to the local model library. When the user chooses to use the target voice model, TTS-SDK loads the marked voice model into the data structure for generating the target voice data with the voiceless voice data.

In the embodiment of the present application, another method for voice playback is provided. By using the above method, the terminal device can determine the model training process through different status information, so the data required by the target voice model can be provided in time, and the data required by the target voice model can be provided in time after the training is completed. The target speech model is acquired and stored in time, thereby improving the efficiency of subsequent generation of target speech data.

The model training device in the present application will be described in detail below. Please refer to FIG. 12. FIG. 12 is a schematic diagram of an embodiment of the model training device in the embodiment of the present application. As shown in the figure, the model training device 30 includes:

The acquisition module 301 is configured to acquire a voice data set to be trained, wherein the voice data set to be trained includes N voice data to be trained, and N is a positive integer greater than or equal to 1;

The obtaining module 301 is further configured to obtain a voice model training instruction when N meets the voice training quantity threshold;

The obtaining module 301 is further configured to, in response to the voice model training instruction, obtain a predicted voice data set based on the to-be-trained voice data set through the to-be-trained voice model, wherein the predicted voice data set includes N predicted voice data, and the predicted voice data is the same as the predicted voice data set. The speech data to be trained has a corresponding relationship;

The training module 302 is used for training the to-be-trained speech model based on the predicted speech data set and the to-be-trained speech data set to obtain the target speech model;

The sending module 303 is configured to send the target voice model, so that the terminal device stores the target voice model.

Optionally, on the basis of the embodiment corresponding to FIG. 12 above, in another embodiment of the model training apparatus 30 provided by the embodiment of the present application,

an obtaining module 301, specifically configured to obtain a voice service request;

In response to the voice service request, a voice data collection instruction is sent to the terminal device, so that the terminal device, in response to the voice data collection instruction, obtains the voice data to be detected, wherein the voice data to be detected is obtained based on the voice training text, and the voice training text is Obtained according to voice data collection instructions;

When the voice data to be detected satisfies the qualified conditions for voice training, determine the voice data to be trained;

Based on the to-be-trained speech data, a to-be-trained speech data set is generated.

Optionally, on the basis of the embodiment corresponding to FIG. 12 above, in another embodiment of the model training apparatus 30 provided by the embodiment of the present application,

The training module 302 is specifically configured to update the model parameters of the speech model to be trained according to the target loss function based on the predicted speech data set and the to-be-trained speech data set;

If the target loss function reaches convergence, the target speech model is generated according to the model parameters.

Optionally, on the basis of the embodiment corresponding to FIG. 12 above, in another embodiment of the model training apparatus 30 provided by the embodiment of the present application,

The obtaining module 301 is further configured to obtain a voice play request, wherein the voice play request carries voice data to be parsed;

The obtaining module 301 is further configured to obtain service content text data according to the to-be-parsed voice data, wherein the service content text data corresponds to the to-be-parsed voice data;

The obtaining module 301 is further configured to obtain voiceless voice data according to the service content text data, wherein the voiceless voice data is obtained by performing speech synthesis on the service content text data;

The sending module 303 is further configured to send the voiceless voice data to the terminal device, so that the terminal device generates the target voice data according to the voiceless voice data and the target voice model, and plays the target voice data.

The following is a detailed description of the voice playback device in the present application. Please refer to FIG. 13. FIG. 13 is a schematic diagram of an embodiment of the voice playback device in the embodiment of the present application. As shown in the figure, the voice playback device 40 includes:

The sending module 401 is used to detect an operation for voice playback, and send a voice playback request, so that the server obtains service content text data based on the voice data to be parsed, and obtains voiceless voice data according to the service content text data, wherein the voice playback request carries The voice data to be parsed, the service content text data is obtained according to the voice data to be parsed, and the timbreless voice data is obtained after voice synthesis is performed on the service content text data;

Obtaining module 402, for obtaining voiceless voice data;

The generation module 403 is used to generate target voice data according to the voiceless voice data and the target voice model, wherein the target voice model is based on the predicted voice data set and the voice data set to be trained, the voice model to be trained is trained and obtained, the predicted voice data is obtained. The set is obtained by the server through the voice model to be trained based on the voice data set to be trained in response to the voice model training instruction;

The playing module 404 is used for playing the target voice data.

Optionally, on the basis of the embodiment corresponding to FIG. 13 above, in another embodiment of the voice playback device 40 provided in the embodiment of the present application,

The sending module 401 is further configured to detect the operation for the voice service and send the voice service request;

The obtaining module 402 is further configured to obtain the voice data collection instruction when the server responds to the voice service request;

The acquisition module 402 is further configured to acquire the voice data to be detected in response to the voice data collection instruction and when the voice data to be detected meets the noise qualification conditions, wherein the voice data to be detected is obtained based on the voice training text, and the voice training text is obtained according to the voice data collection instruction;

The sending module 401 is further configured to send the voice data to be detected, so that the server generates a voice data set to be trained based on the voice data to be detected, the voice data set to be trained includes N voice data to be trained, and N is greater than or equal to 1 positive integer.

Optionally, on the basis of the embodiment corresponding to FIG. 13 above, in another embodiment of the voice playback device 40 provided by the embodiment of the present application, the voice playback device 40 further includes a storage module 405,

The sending module 401 is also used to send a voice model training instruction when N meets the voice training quantity threshold, so that the server obtains the target voice model;

an obtaining module 402, further configured to obtain a target speech model;

The storage module 405 is used to store the target speech model.

Optionally, on the basis of the embodiment corresponding to FIG. 13 above, in another embodiment of the voice playback device 40 provided in the embodiment of the present application,

The acquiring module 402 is further configured to acquire voice data collection status information, wherein the voice data collection status information is used to indicate that voice data to be detected is being acquired;

The acquisition module 402 is further configured to acquire voice data collection completion status information, wherein the voice data collection completion status information is used to indicate that the voice data has been acquired, and the target voice model can be trained;

The obtaining module 402 is further configured to obtain target voice model training status information, wherein the target voice model training status information is used to indicate that the voice model to be trained is being trained;

The obtaining module 402 is further configured to obtain the completion status information of the target voice model, wherein the target voice model completion status information is used to indicate that the target voice model has completed training.

An embodiment of the present application also provides a server. Please refer to FIG. 14 . FIG. 14 is a schematic structural diagram of a server in an embodiment of the present application. As shown in the figure, the server 500 may vary greatly due to different configurations or performances. , may include one or more central processing units (CPU) 522 (eg, one or more processors) and memory 532, one or more storage media 530 (eg, a or more than one mass storage device). Among them, the memory 532 and the storage medium 530 may be short-term storage or persistent storage. The program stored in the storage medium 530 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations on the server. Furthermore, the central processing unit 522 may be configured to communicate with the storage medium 530 to execute a series of instruction operations in the storage medium 530 on the server 500 .

Server 500 may also include one or more power supplies 526, one or more wired or wireless network interfaces 550, one or more input and output interfaces 558, and/or, one or more operating systems 541, such as Windows Server ^™ , Mac OS X ^TM , Unix ^TM , Linux ^TM , FreeBSD ^TM and many more.

The steps performed by the server in the above embodiment may be based on the server structure shown in FIG. 14 .

An embodiment of the present application further provides a terminal device, where a client is deployed on the terminal device, as shown in FIG. 15 , for the convenience of description, only the part related to the embodiment of the present application is shown, and the specific technical details are not disclosed. Please refer to the method section of the embodiment of the present application. Terminal devices can include but are not limited to any terminal device such as smart phones, tablet computers, notebook computers, PDAs, personal computers, smart watches, TV sets, etc. Taking the terminal device as a mobile phone as an example:

FIG. 15 is a block diagram showing a partial structure of a mobile phone related to a terminal device provided by an embodiment of the present application. Referring to FIG. 15 , as shown in the figure, the mobile phone includes: a radio frequency (RF) circuit 610, a memory 620, an input unit 630, a display unit 640, a sensor 630, a voice circuit 660, and a wireless fidelity (WiFi) module 670, the processor 680, and the power supply 690 and other components. Those skilled in the art can understand that the structure of the mobile phone shown in FIG. 15 does not constitute a limitation on the mobile phone, and may include more or less components than shown, or combine some components, or arrange different components.

The following is a detailed introduction to each component of the mobile phone with reference to Figure 15:

The RF circuit 610 can be used for receiving and sending signals during sending and receiving information or during a call. In particular, after receiving the downlink information of the base station, it is processed by the processor 680; in addition, the designed uplink data is sent to the base station. Typically, the RF circuit 610 includes, but is not limited to, an antenna, at least one amplifier, a transceiver, a coupler, a Low Noise Amplifier (LNA), a duplexer, and the like. In addition, RF circuitry 610 may communicate with networks and other devices via wireless communications. The above-mentioned wireless communication can use any communication standard or protocol, including but not limited to Global System of Mobile communication (GSM), General Packet Radio Service (General Packet Radio Service, GPRS), Code Division Multiple Access (Code Division Multiple Access) , CDMA), Wideband Code Division Multiple Access (Wideband Code Division Multiple Access, WCDMA), Long Term Evolution (Long Term Evolution, LTE), email, Short Messaging Service (Short Messaging Service, SMS) and the like.

The memory 620 can be used to store software programs and modules, and the processor 680 executes various functional applications and data processing of the mobile phone by running the software programs and modules stored in the memory 620 . The memory 620 may mainly include a stored program area and a stored data area, wherein the stored program area may store an operating system, an application program (such as a sound playback function, an image playback function, etc.) required for at least one function, and the like; The data created by the use of the mobile phone (such as voice data, phone book, etc.), etc. Additionally, memory 620 may include high-speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other volatile solid state storage device.

The input unit 630 may be used for receiving inputted numerical or character information, and generating key signal input related to user setting and function control of the mobile phone. Specifically, the input unit 630 may include a touch panel 631 and other input devices 632 . The touch panel 631, also referred to as a touch screen, can collect the user's touch operations on or near it (such as the user's finger, stylus, etc., any suitable object or attachment on or near the touch panel 631). operation), and drive the corresponding connection device according to the preset program. Optionally, the touch panel 631 may include two parts, a touch detection device and a touch controller. Among them, the touch detection device detects the user's touch orientation, detects the signal brought by the touch operation, and transmits the signal to the touch controller; the touch controller receives the touch information from the touch detection device, converts it into contact coordinates, and then sends it to the touch controller. To the processor 680, and can receive the commands sent by the processor 680 and execute them. In addition, the touch panel 631 can be implemented in various types such as resistive, capacitive, infrared, and surface acoustic waves. Besides the touch panel 631 , the input unit 630 may further include other input devices 632 . Specifically, other input devices 632 may include, but are not limited to, one or more of physical keyboards, function keys (such as volume control keys, switch keys, etc.), trackballs, mice, joysticks, and the like.

The display unit 640 may be used to display information input by the user or information provided to the user and various menus of the mobile phone. The display unit 640 may include a display panel 641, and optionally, the display panel 641 may be configured in the form of a liquid crystal display (Liquid Crystal Display, LCD), an organic light-emitting diode (Organic Light-Emitting Diode, OLED), or the like. Further, the touch panel 631 can cover the display panel 641, and when the touch panel 631 detects a touch operation on or near it, it transmits it to the processor 680 to determine the type of the touch event, and then the processor 680 determines the type of the touch event according to the touch event. Type provides corresponding visual output on display panel 641 . Although in FIG. 15, the touch panel 631 and the display panel 641 are used as two independent components to realize the input and input functions of the mobile phone, in some embodiments, the touch panel 631 and the display panel 641 can be integrated to form Realize the input and output functions of the mobile phone.

The cell phone may also include at least one sensor 630, such as a light sensor, a motion sensor, and other sensors. Specifically, the light sensor may include an ambient light sensor and a proximity sensor, wherein the ambient light sensor may adjust the brightness of the display panel 641 according to the brightness of the ambient light, and the proximity sensor may turn off the display panel 641 and/or when the mobile phone is moved to the ear. or backlight. As a kind of motion sensor, the accelerometer sensor can detect the magnitude of acceleration in all directions (usually three axes), and can detect the magnitude and direction of gravity when it is stationary. games, magnetometer attitude calibration), vibration recognition related functions (such as pedometer, tapping), etc.; as for other sensors such as gyroscope, barometer, hygrometer, thermometer, infrared sensor, etc. Repeat.

The voice circuit 660, the speaker 661, and the microphone 662 can provide a voice interface between the user and the mobile phone. The voice circuit 660 can transmit the received voice data converted electrical signal to the speaker 661, and the speaker 661 converts it into a sound signal for output; on the other hand, the microphone 662 converts the collected voice signal into an electrical signal, which is converted by the voice circuit 660 After receiving, it is converted into voice data, and then the voice data is output to the processor 680 for processing, and then sent to, for example, another mobile phone through the RF circuit 610, or the voice data is output to the memory 620 for further processing.

WiFi is a short-distance wireless transmission technology. The mobile phone can help users to send and receive emails, browse web pages, and access streaming media through the WiFi module 670. It provides users with wireless broadband Internet access. Although FIG. 15 shows the WiFi module 670, it can be understood that it is not a necessary component of the mobile phone.

The processor 680 is the control center of the mobile phone, using various interfaces and lines to connect various parts of the entire mobile phone, by running or executing the software programs and/or modules stored in the memory 620, and calling the data stored in the memory 620. Various functions of the mobile phone and processing data, so as to monitor the mobile phone as a whole. Optionally, the processor 680 may include one or more processing units; preferably, the processor 680 may integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, user interface, and application programs, etc. , the modem processor mainly deals with wireless communication. It can be understood that, the above-mentioned modulation and demodulation processor may not be integrated into the processor 680 .

The mobile phone also includes a power supply 690 (such as a battery) for supplying power to various components. Preferably, the power supply can be logically connected to the processor 680 through a power management system, so as to manage charging, discharging, and power consumption management functions through the power management system.

Although not shown, the mobile phone may also include a camera, a Bluetooth module, and the like, which will not be repeated here.

In this embodiment of the present application, the processor 680 included in the terminal device may perform the functions in the foregoing embodiments, and details are not described herein again.

Embodiments of the present application further provide a computer-readable storage medium, where a computer program is stored in the computer-readable storage medium, and when it runs on a computer, the computer executes the method as described in the foregoing embodiments that is executed by the terminal device. or, causing the computer to execute the steps executed by the server in the methods described in the foregoing embodiments.

Embodiments of the present application also provide a computer program product including a program, which, when running on a computer, causes the computer to perform the steps performed by the terminal device in the methods described in the foregoing embodiments, or causes the computer to execute the steps performed by the terminal device as described in the foregoing embodiments. The steps performed by the server in the method described in the example.

Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the system, device and unit described above may refer to the corresponding process in the foregoing method embodiments, which will not be repeated here.

In the several embodiments provided in this application, it should be understood that the disclosed system, apparatus and method may be implemented in other manners. For example, the apparatus embodiments described above are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components may be combined or Can be integrated into another system, or some features can be ignored, or not implemented. On the other hand, the shown or discussed mutual coupling or direct coupling or communication connection may be through some interfaces, indirect coupling or communication connection of devices or units, and may be in electrical, mechanical or other forms.

The units described as separate components may or may not be physically separated, and components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution in this embodiment.

In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically alone, or two or more units may be integrated into one unit. The above-mentioned integrated units may be implemented in the form of hardware, or may be implemented in the form of software functional units.

The integrated unit, if implemented in the form of a software functional unit and sold or used as an independent product, may be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the present application can be embodied in the form of software products in essence, or the parts that contribute to the prior art, or all or part of the technical solutions, and the computer software products are stored in a storage medium , including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: U disk, removable hard disk, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk and other media that can store program codes .

As mentioned above, the above embodiments are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand: The technical solutions described in the embodiments are modified, or some technical features thereof are equivalently replaced; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions in the embodiments of the present application.

Claims (10)

1. A method of model training, comprising:

acquiring a voice data set to be trained, wherein the voice data set to be trained comprises N voice data to be trained, and N is a positive integer greater than or equal to 1;

when N meets the threshold of the number of voice training, acquiring a voice model training instruction;

responding to the voice model training instruction, and based on the voice data set to be trained, obtaining a prediction voice data set through a voice model to be trained, wherein the prediction voice data set comprises N prediction voice data, and the prediction voice data and the voice data to be trained have a corresponding relation;

training the voice model to be trained based on the predicted voice data set and the voice data set to be trained to obtain a target voice model;

and sending the target voice model so that the terminal equipment stores the target voice model.

2. The method of claim 1, wherein the obtaining the set of speech data to be trained comprises:

acquiring a voice service request;

responding to the voice service request, sending a voice data collection instruction to the terminal device, so that the terminal device responds to the voice data collection instruction to obtain voice data to be detected, wherein the voice data to be detected is obtained based on a voice training text, and the voice training text is obtained according to the voice data collection instruction;

when the voice data to be tested meets the qualified voice training conditions, determining the voice data to be trained;

and generating the voice data set to be trained based on the voice data to be trained.

3. The method of claim 1, wherein the training the speech model to be trained based on the predicted speech data set and the speech data set to be trained to obtain the target speech model comprises:

updating model parameters of the voice model to be trained according to a target loss function based on the predicted voice data set and the voice data set to be trained;

and if the target loss function reaches convergence, generating the target voice model according to the model parameters.

4. The method according to any one of claims 1 to 3, further comprising:

acquiring a voice playing request, wherein the voice playing request carries voice data to be analyzed;

acquiring service content text data according to the voice data to be analyzed, wherein the service content text data corresponds to the voice data to be analyzed;

obtaining unvoiced sound data according to the service content text data, wherein the unvoiced sound data is obtained by performing voice synthesis on the service content text data;

and sending the unvoiced sound data to a terminal device, so that the terminal device generates target sound data according to the unvoiced sound data and the target sound model, and playing the target sound data.

5. A method for voice playback, comprising:

detecting operation aiming at voice playing, and sending a voice playing request to enable a server to obtain service content text data based on voice data to be analyzed and obtain unvoiced voice data according to the service content text data, wherein the voice playing request carries the voice data to be analyzed, the service content text data is obtained according to the voice data to be analyzed, and the unvoiced voice data is obtained after voice synthesis is carried out on the service content text data;

acquiring the unvoiced sound data;

generating target voice data according to the unvoiced voice data and a target voice model, wherein the target voice model is obtained by training a voice model to be trained based on a prediction voice data set and a voice data set to be trained, and the prediction voice data set is obtained by a server responding to a voice model training instruction, based on the voice data set to be trained, through the voice model to be trained;

and playing the target voice data.

6. The method of claim 5, further comprising:

detecting an operation for a voice service, and sending a voice service request;

when the server responds to the voice service request, acquiring a voice data collection instruction;

responding to the voice data collection instruction, and acquiring the voice data to be detected when the voice data to be detected meets a noise qualified condition, wherein the voice data to be detected is acquired based on a voice training text which is acquired according to the voice data collection instruction;

and sending the voice data to be detected so that the server generates a voice data set to be trained based on the voice data to be detected, wherein the voice data set to be trained comprises N voice data to be trained, and N is a positive integer greater than or equal to 1.

7. The method of claim 6, further comprising:

when N meets the threshold of the number of voice training, sending a voice model training instruction to enable the server to obtain a target voice model;

acquiring a target voice model;

and storing the target voice model.

8. The method according to any one of claims 5 to 7, further comprising:

acquiring voice data collection state information, wherein the voice data collection state information is used for indicating that voice data to be detected is being acquired;

acquiring voice data collection completion state information, wherein the voice data collection completion state information is used for indicating that voice data is completely acquired and training of the target voice model can be started;

acquiring target voice model training state information, wherein the target voice model training state information is used for indicating that the voice model to be trained is being trained;

and acquiring target voice model completion state information, wherein the target voice model completion state information is used for indicating that the target voice model is trained completely.

9. A server, comprising: a memory, a transceiver, a processor, and a bus system;

wherein the memory is used for storing programs;

the processor for executing a program in the memory, the processor for performing the method of any of claims 1 to 4 according to instructions in the program code;

the bus system is used for connecting the memory and the processor so as to enable the memory and the processor to communicate.

10. A terminal device, comprising: a memory, a transceiver, a processor, and a bus system;

wherein the memory is used for storing programs;

the processor for executing a program in the memory, the processor for performing the method of any of claims 5 to 8 according to instructions in the program code;

the bus system is used for connecting the memory and the processor so as to enable the memory and the processor to communicate.

Images