Noise reduction method and device for voice data

Abstract

The invention discloses a voice data noise reduction method and device. The method comprises the following steps: acquiring a voice data sample, wherein the voice sample data comprises multiple frames of voice data obtained after noise mixing; label information corresponding to the voice data samples is obtained, and the label information is used for marking pure voice features, noise voice features and voice activity features in the voice data samples; determining a preset neural network model; and training a preset neural network model by using the voice data sample and the label information, so that the preset neural network model learns a corresponding relation between the pure voice feature and the noise voice feature and a corresponding relation between the noise voice feature and the voice activity feature, and a voice processing model is obtained. In the model training process, the samples obtained by mixing the noiseless voice data and the noise are used, and the voice activity features in the samples are combined for training, so that the noise reduction performance is better under the condition of the same model calculation amount.

Description

[0001] This invention is a divisional application. The national application number of the parent application is 202210225936.9, the application date is March 7, 2022, and the name of the invention is “Training method of speech processing model, denoising method and device for speech data”. Technical Field

[0002] The present application relates to the field of artificial intelligence technology, and in particular to a method and device for reducing noise in speech data. Background Art

[0003] The noise reduction method based on deep learning usually takes noisy speech and its transformation features as input and clean speech and its transformation features as output. To obtain better noise reduction performance, a complex network model is usually required. In addition, a large amount of calculation and storage resources are consumed during the use of the network model. This reduces the flexibility of the network model and is not conducive to its use on low-resource terminals. Summary of the invention

[0004] In order to solve the above technical problem or at least partially solve the above technical problem, the present application provides a method and device for reducing noise of speech data.

[0005] According to one aspect of an embodiment of the present application, a method for training a speech processing model is provided, comprising:

[0006] Acquire a speech data sample, wherein the speech sample data includes multiple frames of speech data obtained after being mixed with noise;

[0007] Acquire label information corresponding to the speech data sample, wherein the label information is used to mark the pure speech features, noisy speech features and speech activity features in the speech data sample;

[0008] Determine the preset neural network model;

[0009] The preset neural network model is trained using the speech data sample and the label information so that the preset neural network model learns the correspondence between the clean speech features and the noisy speech features, and between the noisy speech features and the speech activity features, to obtain a speech processing model.

[0010] Further, the obtaining of the voice data sample includes:

[0011] Acquiring initial voice data, wherein the initial voice data is voice data carrying pure voice features and voice activity features;

[0012] Determine various types of initial noise characteristics and preset signal-to-noise ratios;

[0013] The initial speech data and the initial noise feature are mixed according to the preset signal-to-noise ratio to obtain the speech data sample.

[0014] Furthermore, the preset neural network model includes: a convolutional neural network, a first recurrent neural network, a second recurrent neural network, a first deep neural network and a second deep neural network;

[0015] The convolutional neural network is connected to the first recurrent neural network and the second recurrent neural network respectively, the first recurrent neural network is connected to the first deep neural network, and the second recurrent neural network is connected to the second deep neural network, wherein the convolutional neural network is connected to the first recurrent neural network through a connection unit, and the second recurrent neural network is also connected to the first recurrent neural network through the connection unit.

[0016] Further, the preset neural network model is trained using the speech data sample and the label information so that the preset neural network model learns the correspondence between the clean speech feature and the noisy speech feature, and between the noisy speech feature and the speech activity feature, to obtain a speech processing model, including:

[0017] Input the speech data sample into the preset neural network model so that the convolutional neural network in the preset neural network model extracts the first feature of the speech data sample, inputs the first feature into the connection unit and the second recurrent neural network respectively, the second recurrent neural network detects the first voice activity feature and the first noise voice feature in the first feature, and extracts the first voice activity feature, inputs the first voice activity feature into the connection unit and the second deep neural network respectively, the connection unit concatenates the first feature and the first voice activity feature to obtain a second feature, inputs the second feature into the first recurrent neural network, and the first recurrent neural network detects the second voice activity feature and the second noise voice feature in the second feature, extracts the second voice activity feature, inputs the second voice activity feature into the first deep neural network for feature superposition to obtain a third feature, and determines a target value output by the second deep neural network at the same time;

[0018] In a case where the target value is used to indicate that the voice activity feature satisfies a preset feature, determining initial voice data corresponding to the voice data sample, and extracting a fourth feature corresponding to the initial voice data;

[0019] A loss function value is calculated based on the third feature and the fourth feature, and when the loss function value is less than a preset threshold, the preset neural network model is determined as the speech processing model.

[0020] According to another aspect of the embodiment of the present application, a method for reducing noise of speech data is also provided, comprising:

[0021] Obtaining raw voice data to be processed;

[0022] Inputting the original speech data into a pre-trained speech processing model so that the speech processing model extracts features of the original speech data and outputs a target value and a target speech feature based on the features;

[0023] In a case where the target value is used to indicate that the target speech feature satisfies a preset feature, target speech data is generated based on the target speech feature, and the preset feature is a feature of speech data that does not carry noise.

[0024] Furthermore, the step of inputting the target speech data into a pre-trained speech processing model so that the speech processing model outputs a target value and a target speech feature includes:

[0025] The original speech data is input into the speech processing model so that the convolutional neural network in the speech processing model extracts the original features of the original speech data, and the original features are respectively input into the connection unit and the second recurrent neural network. The second recurrent neural network detects the original speech activity features and the original noise speech features in the first original features, and extracts the original speech activity features. The original speech activity features are respectively input into the connection unit and the second deep neural network. The connection unit splices the original features and the original speech activity features to obtain spliced ​​features. The spliced ​​features are input into the first recurrent neural network, and the first recurrent neural network detects the target speech activity features and the target noise speech features in the spliced ​​features, extracts the target speech activity features, and inputs the target speech activity features into the first deep neural network for feature superposition to obtain the target speech features, and at the same time determines the target value output by the second deep neural network.

[0026] According to another aspect of the embodiment of the present application, there is also provided a training device for a speech processing model, comprising:

[0027] A first acquisition module is used to acquire a speech data sample, wherein the speech sample data includes speech data obtained after multiple frames are mixed with noise;

[0028] A second acquisition module is used to acquire label information corresponding to the speech data sample, wherein the label information is used to mark the pure speech features, noisy speech features and speech activity features in the speech data sample;

[0029] A determination module, used to determine a preset neural network model;

[0030] A training module is used to train the preset neural network model using the speech data sample and the label information, so that the preset neural network model learns the correspondence between the clean speech features and the noisy speech features, and the noisy speech features and the speech activity features, to obtain a speech processing model.

[0031] According to another aspect of the embodiment of the present application, there is also provided a device for reducing noise of speech data, including:

[0032] An acquisition module, used to acquire original voice data to be processed;

[0033] An extraction module, configured to input the original speech data into a pre-trained speech processing model so that the speech processing model extracts features of the original speech data and outputs a target value and a target speech feature based on the features;

[0034] A processing module is used to generate target speech data based on the target speech feature when the target value is used to indicate that the target speech feature satisfies a preset feature, and the preset feature is a feature of speech data without carrying noise.

[0035] According to another aspect of an embodiment of the present application, a storage medium is further provided, which includes a stored program, and the above steps are executed when the program is run.

[0036] According to another aspect of an embodiment of the present application, there is also provided an electronic device, including a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus; wherein: the memory is used to store computer programs; and the processor is used to execute the steps in the above method by running the program stored in the memory.

[0037] The embodiment of the present application also provides a computer program product including instructions, which, when executed on a computer, enables the computer to execute the steps in the above method.

[0038] The above technical solution provided by the embodiment of the present application has the following advantages over the prior art: during the model training process, the present application uses noise-free speech data and samples mixed with noise, and combines the speech activity features in the samples for training, which can achieve better noise reduction performance under the same model calculation conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0040] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0041] Figure 1 A flowchart of a method for training a speech processing model provided in an embodiment of the present application;

[0042] Figure 2 A schematic diagram of the structure of a preset neural network model provided in an embodiment of the present application;

[0043] Figure 3 A flowchart of a method for reducing noise in speech data provided by another embodiment of the present application;

[0044] Figure 4 A block diagram of a training device for a speech processing model provided in an embodiment of the present application;

[0045] Figure 5 A block diagram of a device for reducing noise of speech data provided by another embodiment of the present application;

[0046] Figure 6 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. The schematic embodiments of the present application and their descriptions are used to explain the present application and do not constitute improper limitations on the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0048] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another similar entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0049] The embodiments of the present application provide a method for training a speech processing model, a method and device for reducing noise in speech data. The method provided in the embodiments of the present invention can be applied to any required electronic device, for example, it can be an electronic device such as a server or a terminal, which is not specifically limited here. For the convenience of description, it is referred to as an electronic device in the following.

[0050] According to one aspect of the embodiments of the present application, a method embodiment of a method for training a speech processing model is provided. Figure 1 A flowchart of a method for training a speech processing model provided in an embodiment of the present application, such as Figure 1 As shown, the method includes:

[0051] Step S11, obtaining a speech data sample, wherein the speech sample data includes a plurality of frames of speech data obtained after being mixed with noise.

[0052] In the embodiment of the present application, step S11, obtaining a voice data sample, includes the following steps A1-A3:

[0053] Step A1, obtaining initial voice data, wherein the initial voice data is voice data carrying pure voice features and voice activity features.

[0054] The method provided in the embodiment of the present application is applied to a speech processing device, which may be a smart phone, a computer, etc. The speech processing device may control a speech acquisition device to acquire initial speech data, and the speech acquisition device may be a recorder, a voice recorder, etc. It is understandable that the initial speech data is speech data without noise (i.e., clean speech data). The initial speech data carries pure speech features and speech activity features, and the speech activity features are used to characterize the sounds present in the initial speech data, etc.

[0055] Step A2: determining multiple types of initial noise characteristics and a preset signal-to-noise ratio.

[0056] In the embodiment of the present application, in order to construct a speech data sample carrying noise, it is also necessary to determine multiple types of noise data, wherein the multiple types of noise data may be: white noise type noise data, additive noise type noise data, multiplicative noise type noise data. The preset signal-to-noise ratio may be a preset noise mixing ratio.

[0057] Step A3: Mix the initial speech data and the initial noise features according to a preset signal-to-noise ratio to obtain a speech data sample.

[0058] In an embodiment of the present application, the speech processing device can mix the initial speech data and noise data according to a preset signal ratio to obtain a speech data sample. The speech data sample obtained at this time is the speech data obtained after multiple frames of clean noise features are mixed with noise.

[0059] Step S12, obtaining label information corresponding to the speech data sample, wherein the label information is used to mark the pure speech features, noisy speech features and speech activity features in the speech data sample.

[0060] In the embodiment of the present application, the label information of the speech data sample is used to mark the speech activity feature and the noise speech feature in the speech data sample. The speech data sample can be a speech code, and the speech activity feature can be a coding feature of the speech code, etc.

[0061] Step S13, determining a preset neural network model.

[0062] In the embodiments of the present application, Figure 2 A schematic diagram of the structure of a preset neural network model provided in an embodiment of the present application, such as Figure 2 As shown, the preset neural network model includes: convolutional neural network CNN (Convolutional Neural Networks), a first recurrent neural network RNN1 (Recurrent Neural Network), a second recurrent neural network RNN2, a first deep neural network DNN1 (Deep Neural Networks) and a second deep neural network DNN2.

[0063] The convolutional neural network (CNN) is respectively connected to the first recurrent neural network (RNN1) and the second recurrent neural network RNN2, the first recurrent neural network (RNN1) is connected to the first deep neural network (DNN1), and the second recurrent neural network (RNN2) is connected to the second deep neural network (DNN2), wherein the convolutional neural network is connected to the first recurrent neural network (RNN1) through a connection unit (Cat), and the second recurrent neural network (RNN2) is also connected to the first recurrent neural network (RNN1) through a connection unit (Cat).

[0064] It should be noted that CNN is used for deep feature extraction. Through a multi-layer CNN network, deep features of speech are extracted. In the embodiment of the present application, a 7-layer CNN network is used, Batch Norm is added between each layer, and a PReLU activation function is used.

[0065] RNN2: RNN2 includes LSTM (Long short-term memory) units (or GRU units) to extract speech activity features. The embodiment of the present application uses two layers of LSTM units to extract speech activity features.

[0066] Cat: The connection unit concatenates the features output by CNN with the speech activity features to form a single feature.

[0067] RNN1, DNN1: The denoising branch outputs mask information. The activation function of DNN1 uses Sigmoid, which is multiplied by the original noisy feature to obtain the estimated clean feature, and then the clean speech is obtained through inverse transformation and overlap-addition operations.

[0068] DNN2: Use a 1-layer fully connected network to identify the voice activity features through the fully connected network to obtain a target value, which is 0 or 1, where 0 is used to indicate that the current voice activity features do not match the preset activity features, and 1 is used to indicate that the current voice activity features match the preset activity features.

[0069] Step S14, using the speech data samples and label information to train the preset neural network model, so that the preset neural network model learns the correspondence between the clean speech features and the noisy speech features, and the noisy speech features and the speech activity features, and obtains a speech processing model.

[0070] In the embodiment of the present application, step S14, using the speech data sample and the label information to train the preset neural network model, so that the preset neural network model learns the correspondence between the clean speech features and the noisy speech features, and the noisy speech features and the speech activity features, and obtains the speech processing model, including the following steps B1-B3:

[0071] Step B1, input the speech data sample into a preset neural network model so that the convolutional neural network in the preset neural network model extracts the first feature of the speech data sample, inputs the first feature into the connection unit and the second recurrent neural network respectively, the second recurrent neural network detects the first voice activity feature and the first noise voice feature in the first feature, and extracts the first voice activity feature, inputs the first voice activity feature into the connection unit and the second deep neural network respectively, the connection unit concatenates the first feature and the first voice activity feature to obtain the second feature, inputs the second feature into the first recurrent neural network, and the first recurrent neural network detects the second voice activity feature and the second noise voice feature in the second feature, extracts the second voice activity feature, inputs the second voice activity feature into the first deep neural network for feature superposition to obtain the third feature, and determines the target value output by the second recurrent neural network at the same time.

[0072] It should be noted that the preset neural network model includes two branches. The first branch is used to extract voice activity features, and the first branch includes: a second recurrent neural network and a second deep neural network. The second branch is used to reduce noise on voice data, and the second branch includes: a connection unit, a first recurrent neural network and a first deep neural network, wherein the first recurrent neural network and the second recurrent neural network are both used to detect voice activity features and noisy voice features in the features.

[0073] Step B2, when the target value is used to indicate that the voice activity feature satisfies the preset feature, initial voice data corresponding to the voice data sample is determined, and a fourth feature corresponding to the initial voice data is extracted.

[0074] Step B3, calculating the loss function value based on the third feature and the fourth feature, and when the loss function value is less than a preset threshold, determining the preset neural network model as the speech processing model.

[0075] In the embodiment of the present application, the calculation process of the value of the loss function is as follows:

[0076]

[0077] In the embodiments of the present application, s target is the cross entropy, and e noise is the characteristic variation, <,> represents the dot product of the vector, is the fourth feature, s is the third feature, and SI-SNR is the loss function value.

[0078] In the process of model training, this application uses noise-free speech data mixed with noise samples, and combines the speech activity features in the samples for training, which can improve the noise reduction effect on speech under the same model calculation conditions. In addition, in the process of model training, this application uses different loss functions for training according to different branches, and SI-SNR is used as the loss function for training the first branch. Cross entropy is used as the loss function for training the second branch. In this way, multi-task learning is used to obtain the final speech processing model, which makes the noise reduction performance better.

[0079] Figure 3 A flowchart of a method for reducing noise of speech data provided in an embodiment of the present application is shown in FIG. Figure 3 As shown, the method may include the following steps:

[0080] Step S21, obtaining original voice data to be processed.

[0081] Step S22, inputting the original speech data into a pre-trained speech processing model so that the speech processing model extracts feature information of the original speech data and outputs a target value and a target speech feature based on the feature information.

[0082] In the embodiment of the present application, step S22, inputting the original speech data into a pre-trained speech processing model so that the speech processing model outputs a target value and a target speech feature, includes:

[0083] The original speech data is input into the speech processing model so that the convolutional neural network in the speech processing model extracts the original features of the original speech data, and the original features are respectively input into the connection unit and the second recurrent neural network. The second recurrent neural network detects the original speech activity features and the original noise speech features in the first original features, and extracts the original speech activity features. The original speech activity features are respectively input into the connection unit and the second deep neural network. The connection unit splices the original features and the original speech activity features to obtain spliced ​​features, and the spliced ​​features are input into the first recurrent neural network. The first recurrent neural network detects the target speech activity features and the target noise speech features in the spliced ​​features, extracts the target speech activity features, and inputs the target speech activity features into the first deep neural network for feature superposition to obtain the target speech features, and at the same time determines the target value output by the second deep neural network.

[0084] It should be noted that in the embodiment of the present application, two branches appear after the convolutional neural network through the speech processing model. The first branch is used to extract speech activity features, and the first branch includes: a second recurrent neural network and a second deep neural network. The second branch is used to reduce noise on speech data, and the second branch includes: a connection unit, a first recurrent neural network, and a first deep neural network.

[0085] Specifically, the convolutional neural network is used to extract the original features of the original speech data, and then the convolutional neural network inputs the original features into the connection unit and the second recurrent neural network respectively. The second recurrent neural network extracts the original speech activity features from the first features, and then inputs the original speech activity features into the second deep neural network. The second deep neural network determines the authenticity of the original speech activity features and obtains the target value. At the same time, the second recurrent neural network also gives the original speech activity features to the connection unit.

[0086] The connection unit concatenates the original speech activity feature with the first feature to obtain a concatenated feature, and then the concatenated feature is denoised by the first recurrent neural network and the first deep neural network to obtain the target speech feature.

[0087] Step S23, when the target value is used to indicate that the target speech feature satisfies a preset feature, target speech data is generated based on the target speech feature, where the preset feature is a feature of speech data that does not carry noise.

[0088] In the embodiment of the present application, when the target value is 1, determining the target value is used to indicate that the target speech feature meets the preset feature. At this time, the target speech data can be generated according to the target speech feature.

[0089] The embodiment of the present application processes the speech data by using two branches. On the one hand, the target value can be used to directly determine whether there is a speech active feature in the original speech data. If there is no speech active feature, no processing is performed, thereby improving the efficiency of the speech processing model. On the other hand, the model structure used by the branch for extracting speech active features is simple, and the number of parameters is much smaller than that of the noise reduction branch, which reduces the amount of model calculation and the consumption of storage resources.

[0090] Figure 4 This is a block diagram of a speech processing model training device provided in an embodiment of the present application. The device can be implemented as part or all of an electronic device through software, hardware, or a combination of both. Figure 4 As shown, the device comprises:

[0091] A first acquisition module 41 is used to acquire a speech data sample, wherein the speech sample data includes speech data obtained after multiple frames are mixed with noise;

[0092] The second acquisition module 42 is used to acquire label information corresponding to the speech data sample, wherein the label information is used to mark the pure speech features, noisy speech features and speech activity features in the speech data sample.

[0093] The determination module 43 is used to determine a preset neural network model.

[0094] The training module 44 is used to train the preset neural network model using the speech data samples and the label information, so that the preset neural network model learns the correspondence between the clean speech features and the noisy speech features, and the noisy speech features and the speech activity features, and obtains a speech processing model.

[0095] In an embodiment of the present application, the first acquisition module 41 is used to acquire initial speech data, wherein the initial speech data is speech data carrying pure speech features and speech activity features; determine multiple types of initial noise features and a preset signal-to-noise ratio; and mix the initial speech data and the initial noise features according to the preset signal-to-noise ratio to obtain the speech data sample.

[0096] In an embodiment of the present application, the preset neural network model includes: a convolutional neural network, a first recurrent neural network, a second recurrent neural network, a first deep neural network and a second deep neural network; the convolutional neural network is respectively connected to the first recurrent neural network and the second recurrent neural network, the first recurrent neural network is connected to the first deep neural network, and the second recurrent neural network is connected to the second deep neural network, wherein the convolutional neural network is connected to the first recurrent neural network through a connecting unit, and the second recurrent neural network is also connected to the first recurrent neural network through a connecting unit.

[0097] In an embodiment of the present application, a training module 44 is used to input a speech data sample into a preset neural network model so that a convolutional neural network in the preset neural network model extracts a first feature of the speech data sample, inputs the first feature into a connection unit and a second recurrent neural network respectively, the second recurrent neural network detects a first speech activity feature and a first noise speech feature in the first feature, and extracts the first speech activity feature, inputs the first speech activity feature into a connection unit and a second deep neural network respectively, the connection unit concatenates the first feature and the first speech activity feature to obtain a second feature, inputs the second feature into the first recurrent neural network, and the first recurrent neural network detects the second speech activity feature and the second noise speech feature in the second feature, extracts the second speech activity feature, inputs the second speech activity feature into the first deep neural network for feature superposition to obtain a third feature, and determines a target value output by the second deep neural network at the same time.

[0098] Figure 5 This is a block diagram of a device for reducing noise of speech data provided in an embodiment of the present application. The device can be implemented as part or all of an electronic device through software, hardware, or a combination of both. Figure 5 As shown, the device comprises:

[0099] The acquisition module 51 is used to acquire the original voice data to be processed.

[0100] The extraction module 52 is used to input the original speech data into a pre-trained speech processing model so that the speech processing model extracts the features of the original speech data and outputs target values ​​and target speech features based on the features.

[0101] The processing module 53 is used to generate target speech data based on the target speech feature when the target value is used to indicate that the target speech feature satisfies a preset feature, where the preset feature is a feature of speech data that does not carry noise.

[0102] The present application also provides an electronic device, such as Figure 6 As shown, the electronic device may include: a processor 1501 , a communication interface 1502 , a memory 1503 and a communication bus 1504 , wherein the processor 1501 , the communication interface 1502 , and the memory 1503 communicate with each other via the communication bus 1504 .

[0103] Memory 1503, used for storing computer programs;

[0104] The processor 1501 is used to implement the steps of the above embodiment when executing the computer program stored in the memory 1503.

[0105] The communication bus mentioned in the above terminal can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

[0106] The communication interface is used for communication between the above terminal and other devices.

[0107] The memory may include a random access memory (RAM) or a non-volatile memory, such as at least one disk memory. Optionally, the memory may also be at least one storage device located away from the aforementioned processor.

[0108] The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0109] In another embodiment provided in the present application, a computer-readable storage medium is also provided, in which instructions are stored. When the computer-readable storage medium is run on a computer, the computer executes the training method of the speech processing model described in any of the above embodiments.

[0110] In another embodiment provided in the present application, a computer program product including instructions is also provided, which, when executed on a computer, enables the computer to execute the training method of the speech processing model described in any of the above embodiments.

[0111] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrated. The available medium can be a magnetic medium, (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state hard disk SolidState Disk), etc.

[0112] The above description is only a preferred embodiment of the present application and is not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application are included in the protection scope of the present application.

[0113] The above description is only a specific implementation of the present application, so that those skilled in the art can understand or implement the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest range consistent with the principles and novel features applied for herein.

Claims

1. A method for reducing noise of speech data, characterized in that: include: Obtaining raw voice data to be processed; Inputting the original speech data into a pre-trained speech processing model so that the speech processing model extracts feature information of the original speech data and outputs a target value and a target speech feature based on the feature information; In a case where the target value is used to indicate that the target speech feature satisfies a preset feature, generating target speech data based on the target speech feature, the preset feature being a feature of speech data that does not carry noise; The step of inputting the original speech data into a pre-trained speech processing model so that the speech processing model extracts feature information of the original speech data and outputs a target value and a target speech feature based on the feature information comprises: The original speech data is input into the speech processing model so that the convolutional neural network in the speech processing model extracts the original features of the original speech data, and the original features are respectively input into the connection unit and the second recurrent neural network. The second recurrent neural network detects the original speech activity features and the original noise speech features in the first original features, and extracts the original speech activity features. The original speech activity features are respectively input into the connection unit and the second deep neural network. The connection unit splices the original features and the original speech activity features to obtain spliced ​​features. The spliced ​​features are input into the first recurrent neural network, and the first recurrent neural network detects the target speech activity features and the target noise speech features in the spliced ​​features, extracts the target speech activity features, and inputs the target speech activity features into the first deep neural network for feature superposition to obtain the target speech features, and at the same time determines the target value output by the second deep neural network.

2. A device for reducing noise of speech data, using the method of claim 1, characterized in that: include: An acquisition module, used to acquire original voice data to be processed; An extraction module, used for inputting the original speech data into a pre-trained speech processing model, so that the speech processing model extracts features of the original speech data and outputs a target value and a target speech feature based on the features; A processing module is used to generate target speech data based on the target speech feature when the target value is used to indicate that the target speech feature satisfies a preset feature, and the preset feature is a feature of speech data without noise.

3. A storage medium, characterized in that: The storage medium includes a stored program, wherein the method steps of claim 1 are executed when the program is executed.

4. An electronic device, characterized in that: It includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus; wherein: Memory, used to store computer programs; A processor, configured to execute the method steps of claim 1 by running a program stored in a memory.

Images