Speech privacy inference method and device based on variational quantum circuit and storage medium

Abstract

The application relates to the technical field of financial technology, and provides a speech privacy inference method and device based on a variational quantum circuit, an electronic device and a computer readable storage medium, the method comprising the following steps: acquiring a speech signal; pre-processing the speech signal to obtain speech data; performing quantum computing processing on the speech data based on a first variational quantum circuit model to obtain speech spectrum information, wherein the first variational quantum circuit model is obtained by performing sparse training processing on a preset second variational quantum circuit model; performing identification processing on the speech spectrum information based on a pre-trained cloud network model to obtain speech recognition information, wherein the speech recognition information comprises semantic information; and performing inference processing on the semantic information to obtain a speech inference result. Through the technical solution, the calculation time of speech inference can be reduced, and the efficiency of speech inference in the financial industry can be improved.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to, but are not limited to, the technical field of financial technology, and in particular to a speech privacy inference method and device based on a variational quantum circuit, an electronic device, and a computer readable storage medium. BACKGROUND

[0002] With the continuous development of social economy and the continuous progress of science and technology, people's living standards are also constantly improving, and people are also paying more and more attention to the protection of personal privacy. Privacy computing refers to a collection of technologies that achieve data analysis and calculation while protecting data from being leaked to the outside, achieving the purpose of "available" but "invisible" to data. In the financial industry, under the premise of fully protecting data and privacy security, it is necessary to realize the transformation and release of data value. However, the current speech privacy inference method in the financial industry has large calculation amount and is time-consuming, resulting in low efficiency of speech inference in the financial industry. SUMMARY

[0003] The following is a summary of the subject matter described in detail herein. This summary is not intended to limit the scope of the claims.

[0004] To solve the problems mentioned in the background art, the embodiments of the present application provide a speech privacy inference method and device based on a variational quantum circuit, an electronic device, and a computer readable storage medium, which can reduce the calculation time of speech inference and improve the efficiency of speech inference in the financial industry.

[0005] In a first aspect, the embodiments of the present application provide a speech privacy inference method based on a variational quantum circuit, which comprises:

[0006] obtaining a speech signal;

[0007] preprocessing the speech signal to obtain speech data;

[0008] performing quantum computing processing on the speech data based on a first variational quantum circuit model to obtain speech spectrum information, wherein the first variational quantum circuit model is obtained by sparse training processing of a preset second variational quantum circuit model;

[0009] performing identification processing on the speech spectrum information based on a pre-trained cloud network model to obtain speech recognition information, wherein the speech recognition information includes semantic information;

[0010] performing inference processing on the semantic information to obtain a speech inference result.

[0011] According to some embodiments of the present application, the second variational quantum circuit model comprises a loss determination module, a quantum computing module, and an optimization module, the quantum computing module comprises a plurality of gate circuits, and the first variational quantum circuit model can be obtained by the following method:

[0012] Obtaining initial training parameters;

[0013] Inputting the initial training parameters into the loss determination module for calculation and processing to obtain a problem loss value;

[0014] Based on a preset selected probability, a plurality of gate circuits in the quantum computing module are selected and processed, and based on the selected gate circuits, an energy prediction processing is performed on the problem loss value to obtain an energy expectation value;

[0015] Based on the optimization module, the initial training parameters are updated to train the quantum computing module until the energy expectation value is within a preset threshold range.

[0016] According to some embodiments of the present application, the cloud network model comprises a convolution layer, a pooling layer, and a fully connected layer, and the training process of the cloud network model is as follows:

[0017] Obtaining a voice training spectrum set, the voice training spectrum set comprising a plurality of voice training samples and a target voice output value corresponding to each voice training sample;

[0018] Inputting the voice training sample into the convolution layer for feature extraction to obtain voice feature information;

[0019] Inputting the voice feature information into the pooling layer for pooling operation to obtain voice sampling information;

[0020] Inputting the voice sampling information into the fully connected layer for recognition and classification processing to obtain a training voice output value;

[0021] Based on the target voice output value and the training voice output value, an error value is obtained;

[0022] Based on the error value, the network parameters of the convolution layer, the pooling layer, and the fully connected layer are adjusted and processed.

[0023] According to some embodiments of the present application, the pre-processing of the voice signal to obtain voice data comprises:

[0024] Pre-emphasis processing the voice signal to obtain a first voice signal;

[0025] Digital filtering processing the first voice signal to obtain a second voice signal;

[0026] The second speech signal is windowed and framed to obtain the speech data.

[0027] According to some embodiments of the present application, the inference processing on the semantic information obtains a speech inference result, including:

[0028] The semantic information is subjected to keyword recognition processing to obtain keyword information;

[0029] The keyword information is analyzed to obtain the speech inference result.

[0030] According to some embodiments of the present application, the quantum processing on the speech data based on the first variational quantum circuit model obtains speech spectrum information, including:

[0031] The speech data is subjected to quantum computing processing based on the first variational quantum circuit model according to a quantum computing algorithm to obtain the speech spectrum information;

[0032] wherein the quantum computing algorithm is represented as follows:

[0033]

[0034] wherein,

[0035]

[0036]

[0037]

[0038]

[0039]

[0040] R represents a rotation gate, represents a CNOT gate operation, M represents a measurement operation, U i (θ) is U(θ1,…,θ m ), m is 3, x represents the speech data, p represents a preset probability, f(x; θ) represents the speech spectrum information, and θ represents a gate circuit adjustment parameter.

[0041] According to some embodiments of the present application, the network parameter adjustment processing on the convolutional layer, the pooling layer, and the fully connected layer based on the error value, including:

[0042] The fully connected layer is subjected to first network parameter correction processing according to the error value;

[0043] The pooling layer is subjected to second network parameter correction processing according to the error value;

[0044] performing a third network parameter correction process on the convolutional layer according to the error value.

[0045] In a second aspect, the embodiments of the present application also provide a speech privacy inference device based on a variational quantum circuit, the device comprising:

[0046] A first processing module is configured to acquire a speech signal.

[0047] A second processing module is configured to pre-process the speech signal to obtain speech data.

[0048] A third processing module is configured to perform quantum computing processing on the speech data based on a first variational quantum circuit model to obtain speech spectrum information, wherein the first variational quantum circuit model is obtained by performing sparse training processing on a preset second variational quantum circuit model.

[0049] A fourth processing module is configured to perform recognition processing on the speech spectrum information based on a pre-trained cloud network model to obtain speech recognition information, wherein the speech recognition information comprises semantic information.

[0050] A fifth processing module is configured to perform inference processing on the semantic information to obtain a speech inference result.

[0051] In a third aspect, the embodiments of the present application also provide an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the speech privacy inference method based on the variational quantum circuit as described in the first aspect.

[0052] In a fourth aspect, the embodiments of the present application also provide a computer-readable storage medium storing computer-executable instructions for performing the speech privacy inference method based on the variational quantum circuit as described in the first aspect.

[0053] The speech privacy inference method based on the variational quantum circuit according to the embodiments of the present application has at least the following beneficial effects: first, a speech signal is acquired; then, a speech data is obtained by pre-processing the speech signal; then, a speech spectrum information is obtained by performing quantum computing processing on the speech data based on a first variational quantum circuit model, wherein the first variational quantum circuit model is obtained by performing sparse training processing on a preset second variational quantum circuit model; then, a speech recognition information is obtained by performing recognition processing on the speech spectrum information based on a pre-trained cloud network model, wherein the speech recognition information comprises semantic information; finally, a speech inference result is obtained by performing inference processing on the semantic information. Through the above technical solution, the data encryption is improved, and the computing time of the speech inference in the financial industry is reduced, and the efficiency of the speech inference in the financial industry is improved. Attached Figure Description

[0054] The accompanying drawings are used to provide a further understanding of the technical solutions of this application and constitute a part of the specification. They are used together with the embodiments of this application to explain the technical solutions of this application and do not constitute a limitation on the technical solutions of this application.

[0055] Figure 1 This is a flowchart of a voice privacy inference method based on variable quantum circuits provided in one embodiment of this application;

[0056] Figure 2 This is a flowchart of the training process of the second variable quantum circuit model in the voice privacy inference method based on variable quantum circuits provided in one embodiment of this application;

[0057] Figure 3 This is a flowchart of the training process of the cloud network model in the voice privacy inference method based on variable quantum circuits provided in one embodiment of this application;

[0058] Figure 4 This is a flowchart illustrating the preprocessing of a speech signal in a speech privacy inference method based on variable quantum circuits provided in one embodiment of this application.

[0059] Figure 5 This is a flowchart illustrating the semantic information inference process in a voice privacy inference method based on variable quantum circuits provided in one embodiment of this application.

[0060] Figure 6 This is a flowchart illustrating the quantum computing processing of voice data in a voice privacy inference method based on variable quantum circuits provided in one embodiment of this application.

[0061] Figure 7 This is a flowchart illustrating the adjustment of network parameters in a cloud network model within a voice privacy inference method based on variable quantum circuits, provided in one embodiment of this application.

[0062] Figure 8 This is a schematic diagram of a voice privacy inference device based on variable quantum circuits provided in one embodiment of this application;

[0063] Figure 9 This is a schematic diagram of an electronic device provided in one embodiment of this application. Detailed Implementation

[0064] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0065] It is to be noted that, although the functional modules are divided in the device schematic diagram, the logical order is shown in the flowchart, but in some cases, the steps shown or described can be performed in a manner different from the module division in the device, or the order in the flowchart. The terms "first", "second", etc. in the specification and claims and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily describe a specific order or sequence.

[0066] It should be noted that, unless otherwise defined, 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 terms used herein are only for the purpose of describing the embodiments of the present application and are not intended to limit the present application.

[0067] The embodiments of the present application can acquire and process related data based on artificial intelligence technology. Artificial intelligence (AI) is the use of digital computers or computer-controlled machines to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0068] AI is a new technical science that studies and develops theories, methods, technologies and application systems for simulating, extending and expanding human intelligence; artificial intelligence is a branch of computer science, and artificial intelligence aims to understand the essence of intelligence and produce a new intelligent machine that can react in a similar way to human intelligence. The research in this field includes robots, language recognition, image recognition, natural language processing and expert systems. Artificial intelligence can simulate the information process of human consciousness and thinking. Artificial intelligence is also a theory, method, technology and application system that uses digital computers or computer-controlled machines to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0069] The basic technologies of artificial intelligence generally include technologies such as sensors, special artificial intelligence chips, cloud computing, distributed storage, big data processing, operation / interaction systems, mechatronics, etc. Artificial intelligence software technologies mainly include computer vision technology, robot technology, biometric technology, speech processing technology, natural language processing technology, and machine learning / deep learning, etc.

[0070] Artificial intelligence is AI, which is a theory, method, technology and application system that uses digital computers or computer-controlled machines to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0071] The server related to the artificial intelligence technology can be a standalone server or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms.

[0072] The application provides a speech privacy inference method and device based on a variational quantum circuit, an electronic device, and a computer readable storage medium. First, a speech signal is obtained. Then, the speech signal is preprocessed to obtain speech data. Then, the speech data is processed by quantum calculation based on a first variational quantum circuit model to obtain speech spectrum information, wherein the first variational quantum circuit model is obtained by sparse training of a preset second variational quantum circuit model. Then, the speech spectrum information is identified based on a pre-trained cloud network model to obtain speech recognition information, wherein the speech recognition information includes semantic information. Finally, the semantic information is inferred to obtain a speech inference result. Through the above technical solution, the data encryption is improved, the calculation time of speech inference is reduced, and the efficiency of speech inference in the financial industry is improved.

[0073] The speech privacy inference method based on the variational quantum circuit provided by the application embodiments relates to the field of financial technology. The speech privacy inference method based on the variational quantum circuit provided by the application embodiments can be applied in a terminal, can be applied in a server, and can also be software running in the terminal or the server. In some embodiments, the terminal can be a smart phone, a tablet computer, a notebook computer, a desktop computer, etc. The server can be configured as a standalone physical server, a server cluster composed of multiple physical servers, or a distributed system, and can also be configured as a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The software can be an application program for implementing the speech privacy inference method based on the variational quantum circuit, but is not limited to the above forms.

[0074] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0075] It should be noted that in all specific embodiments of this application, when processing data related to user identity or characteristics, such as user information, user behavior data, user historical data, and user location information, user permission or consent is obtained first. Furthermore, the collection, use, and processing of this data comply with relevant laws, regulations, and standards. In addition, when embodiments of this application require access to sensitive personal information of users, separate permission or consent from the user is obtained through pop-ups or redirection to confirmation pages. Only after obtaining the user's separate permission or consent is the necessary user-related data required for the proper functioning of these embodiments acquired.

[0076] The embodiments of this application will be further described below with reference to the accompanying drawings.

[0077] like Figure 1 As shown, Figure 1 This is a flowchart of a voice privacy inference method based on variable quantum circuits provided in one embodiment of this application. The voice privacy inference method based on variable quantum circuits includes, but is not limited to, steps S100 to S500.

[0078] Step S100: Acquire the voice signal;

[0079] Step S200: Preprocess the speech signal to obtain speech data;

[0080] Step S300: Perform quantum computing processing on the speech data based on the first variable quantum circuit model to obtain speech spectrum information. The first variable quantum circuit model is obtained by sparse training processing of the preset second variable quantum circuit model.

[0081] In step S400, the voice spectrum information is recognized based on the pre-trained cloud network model to obtain voice recognition information, wherein the voice recognition information includes semantic information.

[0082] In step S500, the semantic information is inferred to obtain a voice inference result.

[0083] It should be noted that the voice signal is first acquired, then the voice signal is preprocessed to obtain voice data, then the voice data is quantum calculated based on the first variational quantum circuit model to obtain voice spectrum information, wherein the first variational quantum circuit model is obtained by sparsifying the pre-set second variational quantum circuit model, then the voice spectrum information is recognized based on the pre-trained cloud network model to obtain voice recognition information, wherein the voice recognition information includes semantic information, and finally the semantic information is inferred to obtain a voice inference result. Through the above technical solution, the data encryption is improved, and the voice inference calculation time is reduced, and the voice inference efficiency is improved. In the financial industry, for example, for bank business, the voice privacy inference method based on the variational quantum circuit can include acquiring a voice signal, then preprocessing the voice signal to obtain voice data, then quantum calculating the voice data based on the first variational quantum circuit model to obtain voice spectrum information, wherein the first variational quantum circuit model is obtained by sparsifying the pre-set second variational quantum circuit model, then recognizing the voice spectrum information based on the pre-trained cloud network model to obtain voice recognition information, wherein the voice recognition information includes semantic information, and finally inferring the semantic information to obtain a corresponding bank business inference result.

[0084] It should be noted that the acquisition of the voice signal in the embodiment of the present application needs to be determined by the user's consent and is not obtained without the user's knowledge. In the process of acquiring the voice signal, the user's permission or consent will be obtained first, and the collection, use and processing of these data will comply with relevant laws, regulations and standards of relevant countries and regions. When the embodiment of the present application needs to acquire sensitive personal information of the user, the separate permission or separate consent of the user will be obtained through a pop-up window or a jump to a confirmation page, and after obtaining the separate permission or separate consent of the user, the necessary user-related data for the normal operation of the embodiment of the present application will be acquired.

[0085] It should be noted that the first variational quantum circuit model is obtained by sparsifying the pre-set second variational quantum circuit model, and the sparsification processing can well simplify the complexity of subsequent quantum calculation, reduce the voice inference calculation time, and improve the voice inference efficiency.

[0086] It should be noted that the variational quantum algorithm is to train a parameterized quantum circuit with a classical optimizer, and the whole algorithm idea is similar to machine learning; one of the main advantages of variational quantum algorithms is that they provide a general framework that can be used to solve a wide variety of problems; for an optimization problem, the loss function is defined by the expected value of the observable state generated by a parameterized quantum circuit; then the parameter optimization is outsourced to the classical optimizer, and the classical computer trains the quantum circuit by optimizing the expected value of the circuit parameters.

[0087] It should be noted that the cloud network model is a network model located in the cloud server; wherein the cloud server (Elastic Compute Service, ECS) is a simple and efficient, safe and reliable, and elastic computing service with processing capacity; its management is more simple and efficient than physical servers. Users do not need to purchase hardware in advance, and can quickly create or release any number of cloud servers; cloud servers are an important part of cloud computing services, and are service platforms that provide comprehensive business capabilities to all Internet users. The platform integrates the three core elements of Internet applications in the traditional sense: computing, storage, and network, and provides public Internet infrastructure services to users; cloud server services include two core products: cloud server rental services for small and medium-sized enterprise users and high-end users; and elastic computing platform services for large and medium-sized Internet users; each cluster node of the cloud server platform is deployed in the backbone data center of the Internet and can independently provide computing, storage, online backup, hosting, bandwidth, and other Internet infrastructure services.

[0088] It should be noted that the network model can be a neural network model, the neural network model can include a feedforward neural network model and a feedback neural network model, and the feedforward neural network model can include a convolutional neural network model, a fully connected neural network model, and a generative adversarial network model. Among them, neural networks have a wide and attractive prospect in system identification, pattern recognition, intelligent control, etc. In particular, in intelligent control, people are particularly interested in the self-learning function of neural networks, and regard this important feature of neural networks as one of the key keys to solving the problem of controller adaptability in automatic control. Neural networks are complex network systems formed by a large number of simple processing units (called neurons) widely connected with each other, which reflect many basic characteristics of brain function, and are a highly complex nonlinear dynamic learning system. Neural networks have large-scale parallelism, distributed storage and processing, self-organization, self-adaptation, and self-learning ability, and are particularly suitable for processing information processing problems that need to consider many factors and conditions at the same time, and are not accurate and fuzzy.

[0089] In some embodiments, as Figure 2As shown, the second variational quantum circuit model includes a loss determination module, a quantum computing module, and an optimization module, the quantum computing module includes a plurality of gate circuits, and the first variational quantum circuit model can be obtained through but not limited to steps S310 to S340.

[0090] In step S310, an initial training parameter is obtained.

[0091] In step S320, the initial training parameter is input to the loss determination module for calculation processing to obtain a problem loss value.

[0092] In step S330, the plurality of gate circuits in the quantum computing module are selected based on a preset selected probability, and an energy expectation value is obtained by performing energy prediction processing on the problem loss value based on the selected gate circuits.

[0093] In step S340, the initial training parameter is updated based on the optimization module to train the quantum computing module until the energy expectation value is within a preset threshold range.

[0094] It should be noted that the preset second variational quantum circuit model includes a loss determination module, a quantum computing module, and an optimization module, wherein the quantum computing module includes a plurality of gate circuits; during the training of the second variational quantum circuit model, first, an initial training parameter is obtained; then the initial training parameter is input to the loss determination module for calculation processing to obtain a problem loss value; then the gate circuits in the quantum computing module are selected based on a preset selected probability, and an energy expectation value is obtained by performing energy prediction processing on the problem loss value based on the selected gate circuits; finally, the initial training parameter is updated based on the optimization module to train the quantum computing module iteratively until the energy expectation value is within a preset threshold range.

[0095] It is worth noting that the plurality of gate circuits in the quantum computing module are selected based on a preset selected probability, thereby simplifying the subsequent calculation complexity, i.e., sparsifying the quantum computing module.

[0096] It can be understood that the initial training parameter is updated based on the optimization module, and the quantum computing module is retrained based on the updated training parameter until the energy expectation value obtained by the quantum computing module is within a preset threshold range. For example, if the set energy threshold range is 10-15, when the last calculation obtains an energy expectation value of 9, the training parameter is updated and the quantum computing module is retrained, and when this time obtains an energy expectation value of 11, the updating of the training parameter is stopped and the training of the quantum computing module is completed.

[0097] It should be noted that before the privacy inference processing is performed on the voice of the banking industry, the initial training parameters can be obtained, and then the initial training parameters are input to the loss determination module for calculation processing to obtain the problem loss value; then the multiple gate circuits in the quantum computing module are selected based on the preset selected probability, and the energy expectation value is obtained by performing energy prediction processing on the problem loss value based on the selected gate circuit; finally, the initial training parameters are updated based on the optimization module to train the quantum computing module until the energy expectation value is within the preset threshold range.

[0098] In some embodiments, as shown in FIG. 4, the cloud network model includes a convolutional layer, a pooling layer, and a fully connected layer. The training process of the cloud network model can include, but is not limited to, steps S410-S460. Figure 3

[0099] Step S410, obtaining a voice training spectrum set, the voice training spectrum set including a plurality of voice training samples and a target voice output value corresponding to each voice training sample;

[0100] Step S420, inputting the voice training sample to the convolutional layer for feature extraction to obtain voice feature information;

[0101] Step S430, inputting the voice feature information to the pooling layer for pooling operation to obtain voice sampling information;

[0102] Step S440, inputting the voice sampling information to the fully connected layer for recognition and classification processing to obtain a training voice output value;

[0103] Step S450, obtaining an error value based on the target voice output value and the training voice output value;

[0104] Step S460, adjusting the network parameters of the convolutional layer, the pooling layer, and the fully connected layer based on the error value.

[0105] It should be noted that in the process of training the cloud network model, first, a voice training spectrum set is obtained, wherein the voice training spectrum set includes a plurality of voice training samples and a target voice output value corresponding to each voice training sample; then the voice training sample is input to the convolutional layer for feature extraction to obtain voice feature information; then the voice feature information is input to the pooling layer for pooling operation to obtain voice sampling information; then the voice sampling information is input to the fully connected layer for recognition and classification processing to obtain a training voice output value; then an error value is obtained based on the target voice output value and the training voice output value; finally, the network parameters of the convolutional layer, the pooling layer, and the fully connected layer are adjusted based on the error value.

[0106] ​​It is worth noting that the cloud network model includes a convolutional layer, a pooling layer and a fully connected layer, wherein the convolutional layer can perform feature extraction, the pooling layer can perform sampling processing, and the fully connected layer can perform classification processing.

[0107] It is worth noting that in the financial industry, the training process of the cloud network model can obtain a voice training spectrum set in the following manner, and then input the voice training sample into the convolutional layer to perform feature extraction to obtain voice feature information; then input the voice feature information into the pooling layer to perform pooling operation to obtain voice sampling information; then input the voice sampling information into the fully connected layer to perform recognition classification processing to obtain a training voice output value; then obtain an error value based on the target voice output value and the training voice output value; finally, adjust the network parameters of the convolutional layer, the pooling layer and the fully connected layer based on the error value.

[0108] In some embodiments, as shown in Figure 4 The above step S200 can include but is not limited to step S210, step S220 and step S230.

[0109] Step S210, pre-emphasis processing is performed on the voice signal to obtain a first voice signal;

[0110] Step S220, digital filtering processing is performed on the first voice signal to obtain a second voice signal;

[0111] Step S230, windowing and framing processing is performed on the second voice signal to obtain voice data.

[0112] It should be noted that in the process of pre-processing the voice signal, first, pre-emphasis processing is performed on the voice signal to obtain a first voice signal; then, digital filtering processing is performed on the first voice signal to obtain a second voice signal; finally, windowing and framing processing is performed on the second voice signal to obtain voice data.

[0113] It is worth noting that the speech signal is a non-stationary time-varying signal, which carries various information; in speech coding, speech synthesis, speech recognition and speech enhancement and other speech processing, various information contained in the speech needs to be extracted. Generally speaking, the purpose of speech processing is two-fold: one is to analyze the speech signal, extract the feature parameters, and use them for subsequent processing; the other is to process the speech signal, for example, in speech enhancement, the background noise of the noisy speech is suppressed to obtain relatively "clean" speech; in speech synthesis, the segmented speech needs to be spliced and smoothed to obtain synthesized speech with high subjective sound quality, and the application in this regard is also based on the analysis and extraction of speech signal information; in short, the purpose of speech signal analysis is to facilitate the effective extraction and representation of the information carried by the speech signal. Among them, in the insurance industry, the first speech signal can be obtained by pre-emphasizing the insurance speech signal; then the second speech signal can be obtained by performing digital filtering on the first speech signal; finally, the speech data can be obtained by performing windowing and framing on the second speech signal.

[0114] In some embodiments, as shown in Figure 5 The above step S500 can further include but is not limited to step S510 and step S520.

[0115] Step S510, performing keyword recognition processing on the semantic information to obtain keyword information;

[0116] Step S520, analyzing the keyword information to obtain a speech inference result.

[0117] It should be noted that in the process of inferring the semantic information, first, keyword recognition processing is performed on the semantic information to obtain keyword information; then, analyzing the keyword information can obtain a speech inference result.

[0118] It is worth noting that the keyword recognition processing on the semantic information is to extract keywords from the semantic information; then, analyzing the above extracted keywords can obtain the corresponding speech inference result. Among them, the convolutional neural network can be used to perform keyword recognition processing on the semantic information.

[0119] In some embodiments, as shown in Figure 6 The above step S300 can include but is not limited to step S350.

[0120] Step S350, performing quantum computing processing on the speech data according to a quantum computing algorithm through a first variational quantum circuit model to obtain speech spectrum information.

[0121] The quantum computing algorithm is represented as follows:

[0122]

[0123] wherein,

[0124]

[0125]

[0126]

[0127]

[0128] R represents a rotation gate, represents a CNOT gate operation, M represents a measurement operation, U i (θ) is U(θ1,…,θ m ), m is 3, x represents voice data, p represents a preset probability, f(x; θ) represents voice spectrum information, and θ represents a gate circuit adjustment parameter.

[0129] It should be noted that the above quantum computing algorithm can be used to perform quantum computing processing on voice data to obtain corresponding voice spectrum information.

[0130] In some embodiments, as Figure 7 indicated, the step S460 can include but is not limited to steps S461, S462 and S463.

[0131] Step S461, first network parameter correction processing is performed on the full connection layer according to the error value;

[0132] Step S462, second network parameter correction processing is performed on the pooling layer according to the error value;

[0133] Step S463, third network parameter correction processing is performed on the convolution layer according to the error value.

[0134] It should be noted that during the training of the cloud network model, the network parameter adjustment processing needs to be performed on the cloud network model. First, the first network parameter correction processing is performed on the full connection layer according to the error value, then the second network parameter correction processing is performed on the pooling layer according to the error value, and finally the third network parameter correction processing is performed on the convolution layer according to the error value. Through the above method, the back propagation is realized, and the related network parameters are corrected.

[0135] In addition, as Figure 8 indicated, one embodiment of the present application also provides a voice privacy inference device 10 based on a variational quantum circuit, comprising:

[0136] The first processing module 100 is configured to obtain a voice signal.

[0137] The second processing module 200 is configured to pre-process the voice signal to obtain voice data.

[0138] The third processing module 300 is configured to perform quantum computing processing on the voice data based on a first variational quantum circuit model to obtain voice spectrum information, wherein the first variational quantum circuit model is obtained by performing sparse training processing on a preset second variational quantum circuit model.

[0139] The fourth processing module 400 is configured to perform recognition processing on the voice spectrum information based on a pre-trained cloud network model to obtain voice recognition information, wherein the voice recognition information includes semantic information.

[0140] The fifth processing module 500 is configured to perform inference processing on the semantic information to obtain a voice inference result.

[0141] It should be noted that the voice signal is first obtained, then the voice signal is pre-processed to obtain voice data, then the voice data is processed based on the first variational quantum circuit model to obtain voice spectrum information, wherein the first variational quantum circuit model is obtained by performing sparse training processing on the preset second variational quantum circuit model, then the voice spectrum information is processed based on the pre-trained cloud network model to obtain voice recognition information, wherein the voice recognition information includes semantic information, and finally the semantic information is processed to obtain a voice inference result. Through the above technical solution, the data encryption is improved, and the voice inference calculation time is reduced, and the voice inference efficiency is improved.

[0142] It should be noted that the first variational quantum circuit model is obtained by performing sparse training processing on the preset second variational quantum circuit model, and the sparse processing can well simplify the complexity of subsequent quantum computing, reduce the voice inference calculation time, and improve the voice inference efficiency.

[0143] It should be noted that the variational quantum algorithm is to train a parameterized quantum circuit with a classical optimizer, and the whole algorithm idea is similar to machine learning; one of the main advantages of variational quantum algorithms is that they provide a general framework that can be used to solve a wide variety of problems; for an optimization problem, the loss function is defined by the expected value of the observable state generated by a parameterized quantum circuit; then the parameter optimization is outsourced to a classical optimizer, and the classical computer trains the quantum circuit by optimizing the expected value of the circuit parameters.

[0144] The specific embodiments of the voice privacy inference device based on the variational quantum circuit are basically the same as the specific embodiments of the voice privacy inference method based on the variational quantum circuit described above, and will not be repeated here.

[0145] In addition, as Figure 9As shown, one embodiment of the present application further provides an electronic device 700, which comprises a memory 720, a processor 710 and a computer program stored in the memory 720 and executable on the processor 710.

[0146] The processor 710 and the memory 720 can be connected by a bus or other means.

[0147] The non-transitory software program and instructions required for implementing the above-mentioned embodiment of the speech privacy inference method based on variational quantum circuit are stored in the memory 720, and when executed by the processor 710, the speech privacy inference method based on variational quantum circuit of each of the above-mentioned embodiments is performed, for example, the method steps S100 to S500 in Figure 1 , the method steps S310 to S340 in Figure 2 , the method steps S410 to S460 in Figure 3 , the method steps S210 to S230 in Figure 4 , the method steps S510 to S520 in Figure 5 , the method step S350 in Figure 6 , and the method steps S461 to S463 in Figure 7 .

[0148] The above-mentioned device embodiments are only schematic, and the units described as separate components can or can not be physically separate, i.e., can be located in one place or distributed over multiple network units. Part or all of the modules can be selected according to actual needs to achieve the purpose of the present embodiment.

[0149] In addition, one embodiment of the present application further provides a computer readable storage medium storing computer executable instructions, which are executed by a processor 710 or a controller, for example, by a processor 710 in the above-mentioned device embodiment, so that the above-mentioned processor 710 executes the speech privacy inference method based on variational quantum circuit in the above-mentioned embodiments, for example, executes the method steps S100 to S500 in Figure 1 , the method steps S310 to S340 in Figure 2 , the method steps S410 to S460 in Figure 3 , the method steps S210 to S230 in Figure 4 , the method steps S510 to S520 in Figure 5 , the method step S350 in Figure 6 , and the method steps S461 to S463 in Figure 7 .

[0150] The above embodiments can be combined, and the modules with the same names in different embodiments can be the same or different.

[0151] The above describes specific embodiments of the present application, and other embodiments are within the scope of the appended claims. In some cases, the acts or steps recited in the claims can be performed in a different order than the order in which they are recited and still achieve desirable results. In addition, the processes depicted in the figures do not necessarily require the particular order shown or sequential order in order to achieve the desired results. In some implementations, multitasking and parallel processing can be advantageous.

[0152] The embodiments in the present application are described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other. Each embodiment focuses on the difference from other embodiments. In particular, the device, equipment, and computer readable storage medium embodiments are basically similar to the method embodiments, and thus the description is relatively simple, and the relevant parts can be referred to the part of the method embodiment.

[0153] The device, equipment, computer readable storage medium, and method provided by the embodiments of the present application are corresponding, and thus the device, equipment, and computer readable storage medium also have similar beneficial technical effects as the corresponding method. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the corresponding device, equipment, and computer readable storage medium will not be described here.

[0154] In the 1990s, it was quite obvious to distinguish whether an improvement in a technology was in hardware (e.g., improvement in circuit structures of diodes, transistors, switches, etc.) or in software (improvement in method flow). However, as technology has evolved, many improvements in method flow today can be considered as direct improvements in hardware circuit structures. Designers almost always obtain the corresponding hardware circuit structures by programming the improved method flow into hardware circuits. Therefore, it cannot be said that an improvement in a method flow cannot be implemented by hardware entity modules. For example, a programmable logic device (PLD) (e.g., a field programmable gate array (FPGA)) is an integrated circuit whose logic function is determined by user programming of the device. A digital system is "integrated" on a PLD by the designer programming it, rather than by asking a chip manufacturer to design and fabricate a custom integrated circuit chip. Moreover, instead of manually fabricating integrated circuit chips, this programming is now mostly implemented by "logic compiler" software, which is similar to software compilers used in program development, and the original code to be compiled is written in a specific programming language, which is called a hardware description language (HDL), and there are many such languages, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc., and the most commonly used are VHDL (Very-High-Speed Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should be aware that, as long as the method flow is logically programmed in the above-mentioned hardware description languages and programmed into an integrated circuit, a hardware circuit implementing the logical method flow can be easily obtained.

[0155] The controller can be implemented in any suitable way, for example, the controller can take the form of a microprocessor or processor and a computer readable medium storing computer readable program code, such as software or firmware, executable by the (micro)processor, logic gates, switches, an application specific integrated circuit (ASIC), a programmable logic controller and an embedded microcontroller, examples of which include but are not limited to the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20 and Silicone Labs C8051F320, the memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art will also know that, in addition to being implemented in pure computer readable program code, the controller can also be implemented to perform the same functions in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers and embedded microcontrollers, etc. by logically programming the method steps. Therefore, such a controller can be considered as a hardware component, and the means included therein for implementing various functions can also be considered as structures within the hardware component. Alternatively, the means for implementing various functions can even be considered as both a software module implementing a method and a structure within a hardware component.

[0156] The systems, apparatuses, modules or units illustrated by the above embodiments can be specifically implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer can be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, 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.

[0157] For the sake of description, the above apparatuses are described in functional division and are described respectively. Of course, the functions of the units can be implemented in the same or multiple software and / or hardware when implementing the embodiments of the present application.

[0158] Those skilled in the art will understand that the embodiments of the present application can be provided as a method, a system or a computer program product. Therefore, the embodiments of the present application can take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware aspects. Moreover, the embodiments of the present application can take the form of a computer program product implemented on one or more computer usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer usable program code.

[0159] The specification and drawings are illustrative and are not to be construed as limiting. Numerous specific details are described to provide a thorough understanding of the present disclosure. However, in certain instances, well known methods, procedures, components and circuits have not been described in detail so as not to unnecessarily obscure aspects of the present disclosure. The present specification is directed to one or more methods, devices (systems), and computer program products related to the processing of data. The specification and drawings are illustrative and are not to be construed as limiting. Numerous specific details are described to provide a thorough understanding of the present disclosure. However, in certain instances, well known methods, procedures, components and circuits have not been described in detail so as not to unnecessarily obscure aspects of the present disclosure. Figure 1 one or more functions specified in the flow or flows and / or blocks. Figure 2 one or more functions specified in the flow or flows and / or blocks.

[0160] These computer program instructions can 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 instructions which implement the flow Figure 3 one or more functions specified in the flow or flows and / or blocks. Figure 4 one or more functions specified in the flow or flows and / or blocks.

[0161] These computer program instructions can 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 flow Figure 5 one or more functions specified in the flow or flows and / or blocks. Figure 6 Figure 7 Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 one or more functions specified in the flow or flows and / or blocks.

[0162] In one typical configuration, the computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0163] The memory can include non-persistent memory and / or volatile memory, such as random access memory (RAM) and / or cache memory, non-volatile memory, such as read-only memory (ROM), EPROM, and / or flash memory. The memory is an example of computer-readable media.

[0164] Computer-readable media includes permanent and non-permanent, movable and non-movable media, which can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules 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 technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassette, magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information accessible by a computing device. According to the definition herein, computer-readable media does not include transitory media such as modulated data signals and carriers.

[0165] It should also be noted that the terms "comprising", "including", or any other variant thereof are intended to cover non-exclusive inclusion, so that processes, methods, articles or devices including a series of elements not only include those elements, but also include other elements not explicitly listed, or inherent to such processes, methods, articles or devices. Without more limitations, the element defined by the statement "including a" does not exclude the presence of other identical elements in the process, method, article or device including the element.

[0166] In the embodiments of the present application, "at least one" means one or more, and "multiple" means two or more. The association relationship between the associated objects is described by "and / or", which means that there can be three kinds of relationships, for example, A and / or B, which can represent the cases of A alone, A and B together, and B alone. Wherein A, B can be singular or plural. The character " / " generally represents that the associated objects before and after are in an "or" relationship. "At least one of the following" and the like means any combination of these items, including any combination of single or multiple items. For example, at least one of a, b and c can represent: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, c can be single or multiple.

[0167] Embodiments of the present application can 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, and the like, that perform particular tasks or implement particular abstract data types. Embodiments of the present application can 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 can be located in both local and remote computer storage media including memory storage devices.

[0168] Various embodiments in the present application are described in progressive manner, and the same or similar parts between various embodiments can be referred to each other. Each embodiment focuses on the difference from other embodiments. Especially, the system embodiments are described simply because they are basically similar to the method embodiments, and the relevant parts can be referred to the description of the method embodiments.

[0169] The above only describes the embodiments of the present application and is not intended to limit the present application. The present application can have various modifications and changes for those skilled in the art. Any modification, equivalent replacement, improvement, etc. within the spirit and principle of the present application shall be included in the scope of claims of the present application.

Claims

1. A speech privacy inference method based on a variational quantum circuit, characterized by, The method comprises: acquiring a voice signal; preprocessing the voice signal to obtain voice data; performing quantum computing processing on the voice data based on a first variational quantum circuit model to obtain voice spectrum information, wherein the first variational quantum circuit model is obtained by sparse training processing of a preset second variational quantum circuit model; performing identification processing on the voice spectrum information based on a pre-trained cloud network model to obtain voice recognition information, wherein the voice recognition information comprises semantic information; performing inference processing on the semantic information to obtain a voice inference result. The second variational quantum circuit model comprises a quantum computing module, and a plurality of gate circuits in the quantum computing module are selected based on a preset selected frequency to perform sparse processing on the quantum computing module.

2. The variational quantum circuit based speech privacy inference method of claim 1, wherein, The second variational quantum circuit model comprises a loss determination module and an optimization module, and the first variational quantum circuit model can be obtained in the following manner: obtaining initial training parameters; inputting the initial training parameters into the loss determination module to obtain a problem loss value through calculation processing; performing energy prediction processing on the problem loss value based on the selected gate circuit to obtain an energy expectation value; updating the initial training parameters based on the optimization module to train the quantum computing module until the energy expectation value is within a preset threshold range.

3. The variational quantum circuit based speech privacy inference method of claim 1, wherein, The cloud network model comprises a convolution layer, a pooling layer, and a full connection layer, and the training process of the cloud network model is as follows: obtaining a voice training spectrum set, which comprises a plurality of voice training samples and target voice output values corresponding to each voice training sample; inputting the voice training sample into the convolution layer to extract voice feature information; inputting the voice feature information into the pooling layer to perform a pooling operation to obtain voice sampling information; inputting the voice sampling information into the full connection layer to perform identification and classification processing to obtain a training voice output value; obtaining an error value based on the target voice output value and the training voice output value; adjusting the network parameters of the convolution layer, the pooling layer, and the full connection layer based on the error value.

4. The variational quantum circuit based speech privacy inference method of claim 1, wherein, The preprocessing of the voice signal to obtain voice data comprises: performing pre-emphasis processing on the voice signal to obtain a first voice signal; performing digital filtering processing on the first voice signal to obtain a second voice signal; performing windowing and framing processing on the second voice signal to obtain the voice data.

5. The variational quantum circuit based speech privacy inference method of claim 1, wherein, The inference processing of the semantic information to obtain a voice inference result comprises: performing keyword recognition processing on the semantic information to obtain keyword information; analyzing the keyword information to obtain the voice inference result.

6. The variational quantum circuit based speech privacy inference method of claim 1, wherein, The quantum processing of the voice data based on the first variational quantum circuit model to obtain voice spectrum information comprises: performing quantum computing processing on the voice data based on the first variational quantum circuit model according to a quantum computing algorithm to obtain the voice spectrum information; wherein the quantum computing algorithm is represented as follows: wherein, R represents a rotation gate, represents a CNOT gate operation, M represents a measurement operation, U i (θ) is U(θ1,…,θ m ), m is 3, x represents the voice data, p represents a preset probability, f(x; θ) represents the voice spectrum information, and θ represents a gate circuit adjustment parameter.

7. The variational quantum circuit based speech privacy inference method of claim 3, wherein, The adjustment processing of the network parameters of the convolutional layer, the pooling layer and the fully connected layer based on the error value comprises: performing first network parameter correction processing on the fully connected layer according to the error value; performing second network parameter correction processing on the pooling layer according to the error value; performing third network parameter correction processing on the convolutional layer according to the error value.

8. A speech privacy inference apparatus based on a variational quantum circuit, characterized by, The device comprises: a first processing module configured to acquire a voice signal; a second processing module configured to pre-process the voice signal to obtain voice data; a third processing module configured to perform quantum computing processing on the voice data based on a first variational quantum circuit model to obtain voice spectrum information, wherein the first variational quantum circuit model is obtained by performing sparse training processing on a preset second variational quantum circuit model; a fourth processing module configured to perform identification processing on the voice spectrum information based on a pre-trained cloud network model to obtain voice recognition information, wherein the voice recognition information comprises semantic information; a fifth processing module configured to perform inference processing on the semantic information to obtain a voice inference result. The second variational quantum circuit model comprises a quantum computing module, and a plurality of gate circuits in the quantum computing module are selected based on a preset selected frequency to perform sparse processing on the quantum computing module.

9. An electronic device comprising: A memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the voice privacy inference method based on the variational quantum circuit according to any one of claims 1 to 7.

10. A computer-readable storage medium storing computer-executable instructions, the computer-executable instructions comprising: The computer executable instructions are used to execute the voice privacy inference method based on the variational quantum circuit according to any one of claims 1 to 7.

Images