CN115273804A

CN115273804A - Voice conversion method and device based on coding model, electronic equipment and medium

Info

Publication number: CN115273804A
Application number: CN202210908836.6A
Authority: CN
Inventors: 郭洋; 王健宗
Original assignee: Ping An Technology Shenzhen Co Ltd
Current assignee: Ping An Technology Shenzhen Co Ltd
Priority date: 2022-07-29
Filing date: 2022-07-29
Publication date: 2022-11-01

Abstract

The embodiment of the application provides a voice conversion method, a voice conversion device, electronic equipment and a medium based on a coding model, and belongs to the technical field of voice conversion. The method comprises the following steps: acquiring a sample voice set, wherein the sample voice set comprises source speaker voice and reference voice; performing voice synthesis on the source speaker voice and the reference voice in the sample voice set based on the coding model to obtain a target voice; training the coding model according to the target voice, and determining a loss function corresponding to the coding model; updating the coding model according to the loss function to obtain a target coding model; receiving a voice request of a user, wherein the voice request carries tone information; the voice request is input into the target coding model, and the voice in the voice request is converted into the corresponding synthesized voice according to the tone information.

Description

Voice conversion method and device based on coding model, electronic equipment and medium

Technical Field

The present application relates to the field of speech conversion technologies, and in particular, to a speech conversion method and apparatus based on a coding model, an electronic device, and a medium.

Background

With the development of speech signal processing technology, speech conversion gradually becomes an important research branch in the field of speech signal processing, the task of speech conversion is to generate speech with the speech content of a source speaker and the individual characteristics of a target speaker on the premise of giving the speech of the source speaker and the speech of the target speaker to be converted, and the current technical categories of speech conversion mainly include the following three major categories: auto-encoder based, TTS (Text To Speech, from Text To Speech) based, and generative confrontation network based. These technical studies have focused on the speech conversion between speakers in the training data set, and have resulted in higher speech naturalness and tone similarity. But is rarely concerned with the problem of speaker voice conversion outside the training data set. In the prior art, a speaker encoder is pre-trained by using a mass data set containing a large number of speakers, and the speaker encoder is used for decoupling the voice tone and the voice content of the speakers so as to realize voice conversion of any speaker. However, under conditions of limited data and computational resources, it is difficult for a speaker coder to generalize to any speaker, thereby reducing the accuracy of speech conversion.

Disclosure of Invention

The embodiment of the present application mainly aims to provide a speech conversion method, an apparatus, an electronic device and a medium based on a coding model, which can enhance the generalization of the coding model to any speaker and improve the accuracy and the authenticity of speech conversion.

To achieve the above object, a first aspect of an embodiment of the present application proposes a speech conversion method based on a coding model, the method including:

obtaining a sample voice set, wherein the sample voice set comprises source speaker voice and reference voice;

performing speech synthesis on the source speaker speech and the reference speech in the sample speech set based on the coding model to obtain target speech;

training the coding model according to the target voice, and determining a loss function corresponding to the coding model;

updating the coding model according to the loss function to obtain a target coding model;

receiving a voice request of a user, wherein the voice request carries tone information;

and inputting the voice request into the target coding model, and converting the voice in the voice request into corresponding synthesized voice according to the tone information.

In some embodiments, the coding model comprises a content encoder and a vector encoder; the performing speech synthesis on the source speaker speech and the reference speech in the sample speech set based on the coding model to obtain a target speech, including:

inputting the source speaker voice into the content encoder in the encoding model for mapping to obtain the voice content of the source speaker voice;

inputting the reference voice into the vector encoder in the coding model to perform vector extraction, so as to obtain a reference voice vector of the reference voice;

and carrying out voice synthesis according to the reference voice vector and the voice content to obtain the target voice.

In some embodiments, the performing speech synthesis according to the reference speech vector and the speech content to obtain the target speech includes:

inputting the source speaker voice into the vector encoder in the encoding model to perform vector extraction, so as to obtain a source speaker vector of the source speaker voice;

replacing the source speaker vector in the source speaker voice according to the reference voice vector to obtain a replacement result;

and inputting the replacement result and the voice content into a generator in the coding model for synthesis to generate the target voice.

In some embodiments, the loss function comprises an encoding loss function, a generating loss function, and a discriminating loss function; the training the coding model according to the target speech and determining a loss function corresponding to the coding model includes:

training an encoder and a generator in the coding model according to the target voice to generate training voice;

calculating the training speech according to a discriminator in the coding model to obtain a discrimination prediction value;

updating a discrimination loss function of the discriminator and a generation loss function of the generator according to the discrimination prediction value;

updating the encoding loss function of the content encoder according to the speech content.

In some embodiments, the computing the training speech according to a discriminator in the coding model to obtain a discrimination prediction value includes:

acquiring a real voice set, wherein the real voice set consists of real voices of multiple speakers;

inputting the training speech into a discriminator in the coding model so as to compare the training speech with real speech in the real speech set to obtain a comparison result;

and obtaining the discrimination prediction value according to the comparison result.

In some embodiments, the target coding model comprises a target encoder and a target generator, the speech request comprises a request speech vector; the inputting the voice request into the target coding model and generating the synthetic voice corresponding to the voice request according to the tone information includes:

inputting the voice request into a target encoder in the target encoding model for vector extraction to obtain the request voice vector of the voice request;

inputting the request voice vector into the target coding model to carry out relative entropy calculation to obtain a calculation result;

and inputting the calculation result and the tone information into a target generator of the target coding model, and generating the synthetic voice corresponding to the voice request.

In some embodiments, the target encoder comprises a residual layer; the inputting the voice request into a target encoder in the target encoding model for vector extraction to obtain the request voice vector of the voice request includes:

inputting the voice request into a target encoder in the target encoding model for extraction to obtain a Mel cepstrum of the voice request;

and calculating the Mel cepstrum according to the residual error layer of the target encoder to obtain the request speech vector.

To achieve the above object, a second aspect of the embodiments of the present application provides a speech conversion apparatus based on a coding model, the apparatus including:

the system comprises a sample acquisition module, a voice recognition module and a voice recognition module, wherein the sample acquisition module is used for acquiring a sample voice set, and the sample voice set comprises source speaker voice and reference voice;

a speech synthesis module, configured to perform speech synthesis on the source speaker speech and the reference speech in the sample speech set based on the coding model to obtain a target speech;

the model training module is used for training the coding model according to the target voice and determining a loss function corresponding to the coding model;

the model updating module is used for updating the coding model according to the loss function to obtain a target coding model;

the receiving request module is used for receiving a voice request of a user, wherein the voice request carries tone information;

and the voice conversion module is used for inputting the voice request into the target coding model and converting the voice request into corresponding synthesized voice according to the tone information.

In order to achieve the above object, a third aspect of the embodiments of the present application provides an electronic device, which includes a memory, a processor, a program stored in the memory and executable on the processor, and a data bus for implementing connection communication between the processor and the memory, wherein the program, when executed by the processor, implements the speech conversion method based on the coding model according to the first aspect.

In order to achieve the above object, a fourth aspect of the embodiments of the present application proposes a storage medium, which is a computer-readable storage medium for computer-readable storage, and the storage medium stores one or more programs, which are executable by one or more processors to implement the method for converting speech based on coding models according to the first aspect.

The method comprises the steps of firstly, obtaining a sample voice set, synthesizing source speaker voice and reference voice in the sample voice set based on a coding model to obtain target voice, facilitating training of the coding model, then training the coding model through the target voice to obtain a loss function of the coding model, updating the coding model according to the loss function to obtain the target coding model, so that generalization ability of the coding model is enhanced, efficiency and accuracy of voice conversion are improved, finally, receiving a voice request of a user, inputting the voice request into the target coding model, enabling the target coding model to convert the voice according to tone color information in the voice request, and obtaining final synthesized voice.

Drawings

FIG. 1 is a flow chart of a speech conversion method based on coding models according to an embodiment of the present application;

FIG. 2 is a flowchart of step S102 in FIG. 1;

fig. 3 is a flowchart of step S203 in fig. 2;

fig. 4 is a flowchart of step S103 in fig. 1;

FIG. 5 is a flowchart of step S402 in FIG. 4;

FIG. 6 is a flowchart of step S106 in FIG. 1;

fig. 7 is a flowchart of step S601 in fig. 6;

FIG. 8 is a schematic structural diagram of a speech conversion apparatus based on a coding model according to an embodiment of the present application;

fig. 9 is a schematic hardware structure diagram of an electronic device according to an embodiment of the present application.

Detailed Description

In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.

It should be noted that although functional blocks are partitioned in a schematic diagram of an apparatus and a logical order is shown in a flowchart, in some cases, the steps shown or described may be performed in a different order than the partitioning of blocks in the apparatus or the order in the flowchart. The terms first, second and the like in the description and in the claims, as well as in the drawings described above, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order.

Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of the present application only and is not intended to be limiting of the application.

First, several terms referred to in the present application are resolved:

artificial Intelligence (AI): is a new technical science for researching and developing theories, methods, technologies and application systems for simulating, extending and expanding human intelligence; artificial intelligence is a branch of computer science that attempts to understand the essence of intelligence and produces a new intelligent machine that can react in a manner similar to human intelligence, and research in this field includes robotics, language recognition, image recognition, natural language processing, and expert systems, among others. The artificial intelligence can simulate the information process of human consciousness and thinking. Artificial intelligence is also a theory, method, technique and application system that uses a digital computer or a machine controlled by a digital computer to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use the knowledge to obtain the best results.

Natural Language Processing (NLP): NLP uses computer to process, understand and use human language (such as chinese, english, etc.), and belongs to a branch of artificial intelligence, which is a cross discipline between computer science and linguistics, also commonly called computational linguistics. Natural language processing includes parsing, semantic analysis, discourse understanding, and the like. Natural language processing is commonly used in the technical fields of machine translation, character recognition of handwriting and print, speech recognition and text-to-speech conversion, information intention recognition, information extraction and filtering, text classification and clustering, public opinion analysis and viewpoint mining, and relates to data mining, machine learning, knowledge acquisition, knowledge engineering, artificial intelligence research, linguistic research related to language calculation and the like related to language processing.

Information Extraction (Information Extraction): and extracting the fact information of entities, relations, events and the like of specified types from the natural language text, and forming a text processing technology for outputting structured data. Information extraction is a technique for extracting specific information from text data. The text data is composed of specific units, such as sentences, paragraphs and chapters, and the text information is composed of small specific units, such as words, phrases, sentences and paragraphs or combinations of these specific units. The extraction of noun phrases, names of people, names of places, etc. in the text data is text information extraction, and of course, the information extracted by the text information extraction technology can be various types of information.

Mapping (mapping): data Mapping (Data Mapping) gives two Data models, and a correspondence relationship of Data elements is established between the models, and the process is called Data Mapping. Data mapping is the first step of many data integration tasks, such as: data migration (data migration), data cleansing (data cleansing), data integration, semantic web construction, and p2p information systems. The data mapping method has two modes: manual coding (Hand-coded) and visual manipulation (Graphical manual). The manual coding is to directly define the data corresponding relation by using a programming language like XSLT, JAVA and C + +. Visualization operations typically support a user drawing a line between data items to define a correspondence between the data items. Some tools that support visualization operations may automatically establish such correspondence. This automatically established correspondence typically requires that the data items have the same name. Whether the relationship is established manually or automatically, the tool is finally required to automatically convert the corresponding relationship of the graphic representation into executable programs such as XSLT, JAVA and C + +.

Robustness (Robust): robust is the transliteration of Robust, i.e. in the sense of being Robust and strong. It is also the ability of the system to survive abnormal and dangerous conditions. For example, whether computer software is halted or crashed in the case of input error, disk failure, network overload, or intentional attack is the robustness of the software. By "robustness", it is also meant that the control system maintains some other characteristic under certain (structural, size) parameter perturbation. According to different definitions of performance, stable robustness and performance robustness can be divided. A fixed controller designed with the robustness of a closed loop system as a target is called a robust controller.

Loss Function (Loss Function): a loss function or cost function (cost function) is a function that maps the value of a random event or its related random variable to a non-negative real number to represent the "risk" or "loss" of the random event. In application, the loss function is usually associated with the optimization problem as a learning criterion, i.e. the model is solved and evaluated by minimizing the loss function. For example, it is used in statistical and machine learning for parameter estimation (parameter estimation) of models, in macro-economics for risk management (risk management) and decision making, and in control theory for optimal control theory (optimal control theory).

Relative Entropy (Relative Entropy): also known as Kullback-Leibler divergence or information divergence is a measure of asymmetry in the difference between two probability distributions. In information theory, the relative entropy is equivalent to the difference between the information entropies (Shannon entropy) of two probability distributions.

Mel-Frequency Spectrum (MFC): a spectrum for representing short-term audio, which is based on a logarithmic spectrum (spectrum) represented by a nonlinear Mel scale (mel scale) and a linear cosine transform (linear cosine transform) thereof; and Mel-Frequency cepstral Coefficients (MFCCs) are a set of key Coefficients used to create the Mel-Frequency cepstrum. From the segments of the music signal, we can obtain a set of cepstrum that sufficiently represents the music signal, and the mel cepstral coefficients are the cepstrum (i.e. the spectrum of the spectrum) derived from the cepstrum. Unlike the general cepstrum, the mel cepstrum is most characterized in that the bands on the mel cepstrum are uniformly distributed on the mel scale, that is, such bands are closer to the linear cepstrum representation method seen by the general people and the human nonlinear auditory system (audio system). For example: in the audio compression technique, the mel frequency cepstrum is often used for processing.

Based on this, the embodiment of the present application provides a speech conversion method and apparatus based on a coding model, an electronic device, and a storage medium, and aims to enhance generalization of the coding model to any speaker and improve accuracy of speech conversion.

Specifically, the following embodiments are provided to explain a speech conversion method and apparatus based on a coding model, an electronic device, and a storage medium, and first describe the speech conversion method based on a coding model in the embodiments of the present application.

The embodiment of the application can acquire and process related data based on an artificial intelligence technology. Among them, artificial Intelligence (AI) is a theory, method, technique and application system that simulates, extends and expands human Intelligence using a digital computer or a machine controlled by a digital computer, senses the environment, acquires knowledge and uses the knowledge to obtain the best result.

The artificial intelligence infrastructure generally includes technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technologies, operation/interaction systems, mechatronics, and the like. The artificial intelligence software technology mainly comprises a computer vision technology, a robot technology, a biological recognition technology, a voice processing technology, a natural language processing technology, machine learning/deep learning and the like.

The embodiment of the application provides a speech conversion method based on a coding model, and relates to the technical field of artificial intelligence. The speech conversion method based on the coding model provided by the embodiment of the application can be applied to a terminal, a server side and software running in the terminal or the server side. In some embodiments, the terminal may be a smartphone, tablet, laptop, desktop computer, or the like; the server side can be configured into an independent physical server, a server cluster or a distributed system formed by a plurality of physical servers, and cloud servers for providing basic cloud computing services such as cloud service, a cloud database, cloud computing, cloud functions, cloud storage, network service, cloud communication, middleware service, domain name service, security service, CDN (content delivery network) and big data and artificial intelligence platforms; the software may be an application or the like that implements a speech conversion method based on a coding model, but is not limited to the above form.

The application is operational with numerous general purpose or special purpose computing system environments or configurations. For example: personal computers, server computers, hand-held or portable devices, tablet-type devices, multiprocessor systems, microprocessor-based systems, set top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments that include any of the above systems or devices, and the like. The application may be described in the general context of computer-executable instructions, such as program modules, being executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The application may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media including memory storage devices.

Fig. 1 is an alternative flowchart of a speech conversion method based on a coding model according to an embodiment of the present application, where the method in fig. 1 may include, but is not limited to, steps S101 to S106.

Step S101, a sample voice set is obtained;

it should be noted that the sample speech set includes the source speaker speech and the reference speech.

It is understood that the reference speech may be the speech of other speakers, the speech of a cartoon character, or a synthesized speech, and the reference speech may be manually input or directly called from a speech library, which is not limited in this embodiment.

Step S102, performing voice synthesis on the source speaker voice and the reference voice in the sample voice set based on the coding model to obtain target voice;

step S103, training the coding model according to the target voice, and determining a loss function corresponding to the coding model;

step S104, updating the coding model according to the loss function to obtain a target coding model;

step S105, receiving a voice request of a user;

it should be noted that the voice request carries the tone information.

And step S106, inputting the voice request into the target coding model, and converting the voice in the voice request into corresponding synthesized voice according to the tone information.

In steps S101 to S106 illustrated in this embodiment of the present application, first, a sample speech set is obtained, and a source speaker speech and a reference speech in the sample speech set are synthesized based on a coding model to obtain a target speech, which is convenient for training the coding model, then the coding model is trained by the target speech to obtain a loss function of the coding model, and the coding model is updated according to the loss function to obtain a target coding model, so as to enhance generalization ability of the coding model and improve efficiency and accuracy of speech conversion, and finally, a speech request of a user is received and input into the target coding model, so that the target coding model converts speech according to timbre information in the speech request to obtain a final synthesized speech, thereby enhancing generalization ability of the coding model to any speaker, improving accuracy of speech conversion, and enhancing authenticity of the synthesized speech by improving the timbre information.

In step S102 of some embodiments, speech synthesis is performed on the source speaker speech and the reference speech in the sample speech set based on the coding model to obtain a target speech, which is convenient for subsequent training of the coding model by the target speech.

It should be noted that the coding model includes an encoder, a generator, and a discriminator, where the encoder includes two parts, a content encoder and a vector encoder.

In step S103 of some embodiments, the coding model is trained according to the target speech, and a loss function corresponding to the coding model is determined, so as to improve the efficiency of speech conversion and speech synthesis of the coding model.

In step S104 of some embodiments, the coding model is updated according to the loss function to obtain the target coding model, so as to enhance the generalization capability of the coding model and facilitate the conversion of the input speech.

In step S105 of some embodiments, a voice request carrying tone color information by a user is received, where the voice request is voice information that the user wants to convert.

In step S106 of some embodiments, the voice request is input into the target coding model, and the voice in the voice request is converted into the corresponding synthesized voice according to the tone information, so as to implement the conversion of the voice and improve the authenticity of the voice conversion.

It should be noted that, inputting the voice request into the target coding model can make the target coding model convert the voice in the voice request according to the tone color information, thereby improving the authenticity of voice conversion and avoiding the situations of too large tone color difference, inaccurate voice conversion, etc.

Referring to fig. 2, in some embodiments, step S102 may include, but is not limited to, step S201 to step S203:

it should be noted that the coding model includes a content encoder and a vector encoder.

Step S201, mapping a content encoder in a source speaker voice input encoding model to obtain the voice content of the source speaker voice;

step S202, vector extraction is carried out on a vector encoder in a reference voice input encoding model to obtain a reference voice vector of a reference voice;

and step S203, performing voice synthesis according to the reference voice vector and the voice content to obtain the target voice.

In step S201 of some embodiments, the content encoder in the source speaker speech input encoding model is mapped to obtain the speech content of the source speaker speech, so as to avoid situations such as speech content error and speech content loss.

The speech content includes the content of the speaker, the phoneme of the speech, the prosody of the speech, and the like.

In step S202 of some embodiments, a vector encoder in the reference speech input coding model performs vector extraction to obtain a reference speech vector of the reference speech, so as to facilitate subsequent speech synthesis.

In step S203 of some embodiments, speech synthesis is performed according to the reference speech vector and the speech content to obtain a target speech, where the target speech includes the speech content of the source speaker speech and the reference speech vector of the reference speech, so as to retain the content of the source speaker speech, and in a case where the speech content is not changed, the target speech is synthesized with the reference speech vector to obtain the target speech, thereby avoiding missing of the speech content in the target speech.

Referring to fig. 3, in some embodiments, step S203 may include, but is not limited to, step S301 to step S303:

step S301, vector extraction is carried out on a vector encoder in a source speaker voice input encoding model to obtain a source speaker vector of the source speaker voice;

step S302, a source speaker vector in the voice of a source speaker is replaced according to a reference voice vector to obtain a replacement result;

step S303, the substitution result and the speech content are input to the generator in the coding model and synthesized, and the target speech is generated.

In step S301 of some embodiments, a vector encoder in the source speaker speech input coding model is used to perform vector extraction, so as to obtain a source speaker vector of the source speaker speech, which is convenient for subsequent speech synthesis.

In step S302 of some embodiments, the source speaker vector in the source speaker speech is replaced according to the reference speech vector to obtain a replacement result, so as to implement conversion of speech information and avoid the problems of discontinuous conversion speech due to discrete speech vectors.

In step S303 of some embodiments, the replacement result and the speech content are input to the generator in the coding model for synthesis, and the target speech is generated, thereby completing the conversion and synthesis of the speech.

Referring to fig. 4, in some embodiments, step S103 may include, but is not limited to, step S401 to step S404:

note that the loss function includes an encoding loss function, a generating loss function, and a discriminating loss function.

Step S401, training an encoder and a generator in the coding model according to the target voice to generate training voice;

step S402, calculating the training speech according to the discriminator in the coding model to obtain a discrimination predicted value;

step S403, updating the discrimination loss function of the discriminator and the generation loss function of the generator according to the discrimination prediction value;

step S404, updating the coding loss function of the content encoder according to the speech content.

In step S401 of some embodiments, an encoder and a generator in the coding model are trained according to the target speech to generate training speech, where the training speech is synthesized speech generated by the encoder and the generator, and facilitates subsequent training of the discriminator.

In step S402 of some embodiments, a discriminant prediction value is obtained by calculating the training speech according to a discriminant in the coding model, so as to facilitate subsequent updating of the coding loss function.

It should be noted that, first, the encoder and the generator are trained through the target speech to obtain the training speech, so that the discriminator can be trained according to the training speech, the discriminator can judge whether the training speech is a synthesized speech, the difference between the training speech and the real speech is obtained, and the discrimination prediction value is obtained according to the difference between the training speech and the real speech.

In step S403 of some embodiments, the discrimination loss function of the discriminator and the generation loss function of the generator are updated according to the discrimination prediction value, so that the discriminator and the generator are updated through the discrimination loss function and the generation loss function, and the efficiency of model conversion is improved.

It should be noted that the discrimination loss function and the generation loss function are both antagonistic loss functions.

In some embodiments, formula (1) identifying the loss function and generating the loss function (2) are represented as follows:

it is understood that in formula (1) and formula (2), D is the discriminator, G is the generator, c is the speech content of the source speaker,

to train the speech vectors of speech, E_cIn the form of a content encoder, the content encoder,

the discriminant prediction value calculated for the discriminant on the training speech.

In step S404 of some embodiments, the encoding loss function of the content encoder is updated according to the speech content, so that the converted synthesized speech maintains the speech content of the source speaker and avoids alteration or loss of the speech content.

Equation (3) for the coding loss function in some embodiments is expressed as follows: :

it can be understood that E in the formula (3)_c(x) Is the result of inputting the source speaker's speech into the content encoder.

Referring to fig. 5, in some embodiments, step S402 may further include, but is not limited to, step S501 to step S502:

step S501, acquiring a real voice set;

it should be noted that the real voice collection is composed of real voices of multiple speakers.

Step S502, inputting the training speech into a discriminator in the coding model so as to compare the training speech with the real speech in the real speech set to obtain a comparison result;

in step S503, a discrimination prediction value is obtained according to the comparison result.

In step S501 of some embodiments, a real speech set is obtained, and the real speech set is composed of real speeches of multiple speakers, where the real speech includes real speech content, real speech vectors, and the like, so that a subsequent discriminator can compare the real speech with a training speech.

In step S502 of some embodiments, the training speech is input into the discriminator in the coding model, so that the training speech is compared with the real speech in the real speech set to obtain a comparison result, thereby accurately obtaining the difference between the training speech and the real speech, and facilitating to accurately determine the discrimination prediction value.

It should be noted that, comparing the training speech with the real speech includes comparing timbre information, audio information, phoneme information, and the like of the training speech and the real speech, so as to obtain a plurality of comparison results.

In step S503 of some embodiments, the multiple comparison results obtained in step S502 are integrated to obtain a discrimination prediction value, so as to update the coding model.

Referring to fig. 6, in some embodiments, step S106 includes, but is not limited to, steps S601 to S602:

it should be noted that the target coding model includes a target encoder and a target generator.

Step S601, inputting a voice request into a target encoder in a target encoding model to perform vector extraction, so as to obtain a request voice vector of the voice request;

step S602, inputting the request voice vector into a target coding model to perform relative entropy calculation to obtain a calculation result;

step S603 inputs the calculation result and the tone information to the target generator of the target coding model, and generates a synthesized speech corresponding to the speech request.

In step S601 in some embodiments, a target encoder that inputs a speech request into a target coding model performs vector extraction to obtain a requested speech vector of the speech request, so as to facilitate subsequent speech conversion according to the requested speech vector.

In step S602 of some embodiments, the request speech vector is input into the target coding model to perform relative entropy calculation, so as to obtain a calculation result, thereby implementing constraint on the request speech vector, improving robustness of the target coding model to the input speech request, and making the request speech vector smoother and continuous.

In step S603 of some embodiments, the calculation result and the timbre information are input to the target generator of the target coding model, and a synthesized speech corresponding to the speech request is generated, so that the synthesized speech is consistent with the timbre information in the requested speech, and the timbre similarity of the speech conversion is improved.

In some embodiments, the target generator may also be trained by the timbre information, so as to add a reconstruction loss function corresponding to the target generator, and the reconstruction loss function (4) is as follows:

L_fm(G，E_c)＝||D(x)-D(G(c，z))|| (4)

it is understood that in formula (4), D is the discriminator, G is the generator, c is the phonetic content of the source speaker, z is the request phonetic vector, E_cIs a content encoder.

Referring to fig. 7, in some embodiments, step S601 may include, but is not limited to include, step S701 to step S702:

note that the target encoder includes a residual layer.

In some embodiments, the target encoder of the target encoder includes four layers of upsampled layers and four layers of residual layers, and the target generator includes four layers of residual layers and four layers of downsampled layers.

Step S701, inputting a voice request into a target encoder in a target encoding model for extraction to obtain a Mel cepstrum of the voice request;

step S702, the Mel cepstrum is calculated according to the residual error layer of the target encoder, and the request voice vector is obtained.

In step S701 of some embodiments, the target encoder inputting the voice request into the target encoding model is extracted to obtain the mel cepstrum of the voice request, so as to reduce the requirement of the target encoding model for the corpus.

In step S702 of some embodiments, the mel cepstrum is calculated according to the residual layer of the target encoder to obtain the requested speech vector, which is convenient for performing speech conversion according to the requested speech vector.

In some embodiments, the mel cepstrum in the target speech request is extracted first, then the mel cepstrum is input into a residual layer of a target encoder to be calculated, so that a request speech vector is obtained, and finally the request speech vector is input into a target encoding model to be calculated according to the relative entropy, so that a calculation result is obtained.

It should be noted that the relative entropy function (5) for inputting the requested speech vector into the target coding model to perform relative entropy calculation is as follows:

L_kl(E_s)＝log(p(z；μ，σ))-log(N(z；0，1)) (5)

it is understood that N (z; 0, 1) represents a standard normal distribution. p (z; μ, σ) represents the normal distribution of the speaker vector z following a mean μ and a standard deviation σ. Wherein the mean μ and standard deviation σ are encoder E_sThe mean μ and standard deviation σ are the results of the computation of the residual layer that inputs the mel-frequency cepstrum to the target encoder.

In some embodiments, the loss function (6) of the target coding network is obtained by equations (1) - (5), which are as follows:

L(G，E_c，E_s)＝L_adv(G，E_c)+L_fm(G，E_c)+L_con(G，E_c)+L_kl(E_s) (6)

referring to fig. 8, an embodiment of the present application further provides a speech conversion apparatus based on a coding model, which can implement the speech conversion method based on the coding model, and the apparatus includes:

a sample obtaining module 801, configured to obtain a sample voice set, where the sample voice set includes a source speaker voice and a reference voice;

a speech synthesis module 802, configured to perform speech synthesis on a source speaker speech and a reference speech in a sample speech set based on a coding model to obtain a target speech;

a model training module 803, configured to train a coding model according to the target speech, and determine a loss function corresponding to the coding model;

the model updating module 804 is used for updating the coding model according to the loss function to obtain a target coding model;

a receiving request module 805, configured to receive a voice request of a user, where the voice request carries tone information;

and a voice conversion module 806, configured to input the voice request into the target coding model, and convert the voice request into corresponding synthesized voice according to the tone information.

The specific implementation of the speech conversion apparatus based on coding model is substantially the same as the specific implementation of the speech conversion method based on coding model, and is not described herein again.

An embodiment of the present application further provides an electronic device, where the electronic device includes: : the speech conversion system comprises a memory, a processor, a program stored in the memory and capable of running on the processor, and a data bus for realizing connection communication between the processor and the memory, wherein the program realizes the speech conversion method based on the coding model when being executed by the processor. The electronic equipment can be any intelligent terminal including a tablet computer, a vehicle-mounted computer and the like.

Referring to fig. 9, fig. 9 illustrates a hardware structure of an electronic device according to another embodiment, where the electronic device includes:

the processor 901 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an Application Specific Integrated Circuit (ASIC), or one or more integrated circuits, and is configured to execute a relevant program to implement the technical solution provided in the embodiment of the present Application;

the Memory 902 may be implemented in the form of a Read Only Memory (ROM), a static storage device, a dynamic storage device, or a Random Access Memory (RAM). The memory 902 may store an operating system and other application programs, and when the technical solution provided by the embodiments of the present specification is implemented by software or firmware, the relevant program codes are stored in the memory 902 and called by the processor 901 to execute the speech conversion method based on the coding model according to the embodiments of the present application;

an input/output interface 903 for implementing information input and output;

a communication interface 904, configured to implement communication interaction between the device and another device, where communication may be implemented in a wired manner (e.g., USB, network cable, etc.), or in a wireless manner (e.g., mobile network, WIFI, bluetooth, etc.);

a bus 905 that transfers information between various components of the device (e.g., the processor 901, memory 902, input/output interface 903, and communication interface 904);

wherein the processor 901, the memory 902, the input/output interface 903 and the communication interface 904 are communicatively connected to each other within the device via a bus 905.

The embodiment of the present application further provides a storage medium, which is a computer-readable storage medium for a computer-readable storage, and the storage medium stores one or more programs, and the one or more programs are executable by one or more processors to implement the above-mentioned speech conversion method based on the coding model.

The memory, which is a non-transitory computer readable storage medium, may be used to store non-transitory software programs as well as non-transitory computer executable programs. Further, the memory may include high speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid state storage device. In some embodiments, the memory optionally includes memory located remotely from the processor, and these remote memories may be connected to the processor through a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.

According to the voice conversion method based on the coding model, the voice conversion device based on the coding model, the electronic equipment and the storage medium provided by the embodiment of the application, firstly, a sample voice set is obtained, a source speaker voice and a reference voice in the sample voice set are synthesized based on the coding model to obtain a target voice, the training of the coding model is facilitated, then the coding model is trained through the target voice to obtain a loss function of the coding model, the coding model is updated according to the loss function to obtain the target coding model, so that the generalization capability of the coding model is enhanced, the efficiency and the accuracy of voice conversion are improved, finally, a voice request of a user is received, the voice request is input into the target coding model, the voice is converted through the target coding model according to the tone color information in the voice request, and the final synthesized voice is obtained, therefore the generalization performance of the coding model to any speaker is enhanced, the accuracy of the voice conversion is improved, and the authenticity of the synthesized voice is enhanced through the improvement of the tone color information.

The embodiments described in the embodiments of the present application are for more clearly illustrating the technical solutions of the embodiments of the present application, and do not constitute a limitation to the technical solutions provided in the embodiments of the present application, and it is obvious to those skilled in the art that the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems with the evolution of technology and the emergence of new application scenarios.

It will be appreciated by those skilled in the art that the solutions shown in fig. 1-7 are not intended to limit the embodiments of the present application and may include more or fewer steps than those shown, or some of the steps may be combined, or different steps may be included.

While certain embodiments of the present application have been described above, other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily have to be in the particular order shown or in sequential order to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.

The embodiments in the present application are described in a progressive manner, and the same and similar parts among the embodiments can be referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the apparatus, device, and computer-readable storage medium embodiments, since they are substantially similar to the method embodiments, the description is relatively simple, and reference may be made to some descriptions of the method embodiments for relevant points.

The apparatus, the device, the computer-readable storage medium, and the method provided in the embodiments of the present application correspond to each other, and therefore, the apparatus, the device, and the nonvolatile computer storage medium also have advantageous technical effects similar to those of the corresponding method.

In the 90 s of the 20 th century, improvements in a technology could clearly distinguish between improvements in hardware (e.g., improvements in circuit structures such as diodes, transistors, switches, etc.) and improvements in software (improvements in process flow). However, as technology advances, many of today's process flow improvements have been seen as direct improvements in hardware circuit architecture. Designers almost always obtain the corresponding hardware circuit structure by programming an improved method flow into the hardware circuit. Thus, it cannot be said that an improvement in the process flow cannot be realized by hardware physical modules.

For example, a Programmable Logic Device (PLD), such as a Field Programmable Gate Array (FPGA), is an integrated circuit whose Logic functions are determined by programming the Device by a user. A digital system is "integrated" on a PLD by the designer's own programming without requiring the chip manufacturer to design and fabricate application-specific integrated circuit chips. Furthermore, nowadays, instead of manually manufacturing an integrated circuit chip, such programming is often implemented by "logic compiler" software, which is similar to a software compiler used in program development, and the original code before compiling is also written by a specific programming Language, which is called Hardware Description Language (HDL), and the HDL is not only one kind but many kinds, for example:

ABEL (Advanced Boolean Expression Language); AHDL (Altera Hardware Description Language); confluence; CUPL (corner University Programming Language); HDCal; and JHDL (Java Hardware Description Language); lava, lola, myHDL, PALSM, RHDL (Ruby Hardware Description Language), etc.; at present, VHDL (Very-High-Speed Integrated Circuit Hardware Description Language) and the Language Verilog are more commonly used among the technologies in the art by comparison. It will also be apparent to those skilled in the art that hardware circuitry that implements the logical method flows can be readily obtained by merely slightly programming the method flows into an integrated circuit using the hardware description languages described above.

The controller may be implemented in any suitable manner, for example, the controller may take the form of, for example, a microprocessor or processor and a computer readable medium that stores computer readable program code (e.g., software or firmware) executable by the (micro) processor, logic gates, switches, an Application Specific Integrated Circuit (ASIC), a programmable logic controller, and embedded microcontrollers, examples of which include, but are not limited to, the following microcontrollers:

ARC 625D, atmel AT91SAM, microchIP address PIC18F26K20, and Silicone Labs C8051F320, the memory controller may also be implemented as part of the control logic of the memory. Those skilled in the art will also appreciate that, in addition to implementing the controller as pure computer readable program code, the same functionality can be implemented by logically programming method steps such that the controller is in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers, embedded microcontrollers and the like. Such a controller may thus be considered a hardware component, and the means included therein for performing the various functions may also be considered as a structure within the hardware component. Or even means for performing the functions may be regarded as being both a software module for performing the method and a structure within a hardware component.

The systems, devices, modules or units illustrated in the above embodiments may be implemented by a computer chip or an entity, or by a product with certain functions. One typical implementation device is a computer. In particular, the computer may be, for example, a personal computer, a laptop computer, a cellular telephone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

For convenience of description, the above devices are described as being divided into various units by function, respectively. Of course, the functions of the units can be implemented in the same software and/or hardware or in a plurality of software and/or hardware when implementing the embodiment of the present application.

As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, embodiments of the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, embodiments of the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

The present specification has been described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

In a typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.

The memory may include forms of volatile memory in a computer readable medium, random Access Memory (RAM) and/or non-volatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). Memory is an example of a computer-readable medium.

Computer-readable media, including both permanent and non-permanent, removable and non-removable media, may implement the information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer readable medium does not include a transitory computer readable medium such as a modulated data signal and a carrier wave.

It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrases "comprising one of 8230; \8230;" 8230; "does not exclude the presence of additional like elements in a process, method, article, or apparatus that comprises the element.

In the embodiments of the present application, "at least one" means one or more, "a plurality" means two or more. "and/or" describes the association relationship of the associated objects, and means that there may be three relationships, for example, a and/or B, and may mean that a exists alone, a and B exist simultaneously, and B exists alone. Wherein A and B can be singular or plural. The character "/" generally indicates that the former and latter associated objects are in an "or" relationship. "at least one of the following" and the like, refer to any combination of these items, including any combination of singular or plural items. For example, at least one of a, b, and c may represent: a, b, c, a and b, a and c, b and c or a and b and c, wherein a, b and c can be single or multiple.

Embodiments of the application may be described in the general context of computer-executable instructions, such as program modules, being executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. Embodiments of the application may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media including memory storage devices.

The embodiments in the present application are described in a progressive manner, and the same and similar parts among the embodiments can be referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the system embodiment, since it is substantially similar to the method embodiment, the description is simple, and for the relevant points, reference may be made to the partial description of the method embodiment.

The above are merely examples of the present application and are not intended to limit the present application. Various modifications and changes may occur to those skilled in the art to which the present application pertains. Any modification, equivalent replacement, improvement or the like made within the spirit and principle of the present application shall be included in the scope of the claims of the present application.

Claims

1. A method for speech conversion based on a coding model, the method comprising:

acquiring a sample voice set, wherein the sample voice set comprises source speaker voice and reference voice;

2. The coding-model-based speech conversion method of claim 1, wherein the coding model comprises a content coder and a vector coder; the performing speech synthesis on the source speaker speech and the reference speech in the sample speech set based on the coding model to obtain a target speech, including:

3. The method of claim 2, wherein the performing speech synthesis according to the reference speech vector and the speech content to obtain the target speech comprises:

4. The coding-model-based speech conversion method of claim 2, wherein the loss function comprises a coding loss function, a generation loss function, and a discrimination loss function; the training the coding model according to the target speech and determining a loss function corresponding to the coding model includes:

5. The method of claim 4, wherein the computing the training speech according to the discriminator in the coding model to obtain a discrimination prediction value comprises:

inputting the training voice into a discriminator in the coding model so as to compare the training voice with real voice in the real voice set to obtain a comparison result;

and obtaining the identification predicted value according to the comparison result.

6. The coding-model-based speech conversion method of claim 1, wherein the target coding model comprises a target encoder and a target generator; the inputting the voice request into the target coding model and generating the synthetic voice corresponding to the voice request according to the tone information includes:

inputting the voice request into a target encoder in the target encoding model to perform vector extraction, so as to obtain a request voice vector of the voice request;

and inputting the calculation result and the tone information into a target generator of the target coding model to generate the synthetic voice corresponding to the voice request.

7. The coding model-based speech conversion method of claim 6, wherein the target encoder comprises a residual layer; the inputting the voice request into a target encoder in the target encoding model for vector extraction to obtain the request voice vector of the voice request includes:

8. An apparatus for speech conversion based on a coding model, the apparatus comprising:

the system comprises a sample acquisition module, a voice analysis module and a voice analysis module, wherein the sample acquisition module is used for acquiring a sample voice set, and the sample voice set comprises source speaker voice and reference voice;

9. An electronic device, characterized in that the electronic device comprises a memory, a processor, a program stored on the memory and executable on the processor, and a data bus for enabling a connection communication between the processor and the memory, the program, when executed by the processor, implementing the steps of the coding model based speech conversion method according to any of claims 1 to 7.

10. A storage medium, which is a computer-readable storage medium for a computer-readable storage, characterized in that the storage medium stores one or more programs, which are executable by one or more processors to implement the steps of the coding model based speech conversion method according to any one of claims 1 to 7.