Signal correction model training method, signal correction method, device and equipment

Abstract

An embodiment of the present application provides a method for training a signal correction model, a signal correction method, a device, and a device. The method for training a signal correction model includes: obtaining training signals, where the training signals include: multiple groups of corresponding signals, and the correlation coefficient of the corresponding signals is greater than or equal to a first preset threshold; respectively adding a preset noise signal to one of the signals in each group of corresponding signals to obtain target training signals; training an initial signal correction model based on the target training signals to obtain a signal correction model. By implementing the present invention, a correction model can be obtained by training a neural network based on signals that meet the correlation coefficient condition, and the model can be used to correct a signal to be corrected that does not meet the correlation coefficient condition, so that the corrected signal meets the correlation coefficient condition, avoiding inaccurate processing results obtained when using the correlation coefficient for subsequent signal processing.

Description

Technical Field

[0001] This application belongs to the technical field of signal processing, and particularly relates to a method for training a signal correction model, a signal correction method, a device, and equipment. Background Art

[0002] Currently, when collecting speech signals, they are mainly collected through voice interaction devices. However, in the actual collection process, due to differences in the quality of each microphone in the voice interaction device, the collected signal data often has errors, and the signals that are theoretically strongly correlated have poor correlation in actual calculations. For the errors in the collected data, during subsequent signal processing, it will further affect the processing results, and the obtained results have low accuracy. Summary of the Invention

[0003] Embodiments of this application provide a method for training a signal correction model, a signal correction method, a device, and equipment, which can correct signals so that the corrected signals have strong correlation and obtain more accurate analysis and processing results.

[0004] In a first aspect, embodiments of this application provide a method for training a signal correction model, including:

[0005] Obtain training signals, where the training signals include: multiple groups of corresponding signals, and the correlation coefficient of the corresponding signals is greater than or equal to a first preset threshold;

[0006] Add a preset noise signal to one of the signals in each group of corresponding signals respectively to obtain target training signals;

[0007] Train an initial signal correction model based on the target training signals to obtain a signal correction model.

[0008] In a second aspect, embodiments of this application provide a signal correction method, including:

[0009] Obtain a signal to be corrected, where the signal to be corrected is a signal whose correlation coefficient with a reference signal is less than the first preset threshold;

[0010] Input the signal to be corrected into the signal correction model to obtain a corrected signal, and the signal correction model is trained according to the signal correction model training method in the first aspect.

[0011] In a third aspect, embodiments of this application provide a signal correction model training device, including:

[0012] An obtaining module, configured to obtain training signals, where the training signals include: multiple groups of corresponding signals, and the correlation coefficient of the corresponding signals is greater than or equal to a first preset threshold;

[0013] An adding module, configured to add a preset noise signal to one of the signals corresponding to each group respectively, to obtain a target training signal;

[0014] A training module, configured to train an initial signal correction model based on the target training signal, to obtain a signal correction model.

[0015] In a fourth aspect, an embodiment of the present application provides a signal correction device, including:

[0016] An obtaining module, configured to obtain a signal to be corrected, where the signal to be corrected is a signal whose correlation coefficient with a reference signal is less than a first preset threshold;

[0017] An input module, configured to input the signal to be corrected into a correction model, to obtain a corrected signal, where the correction model is trained according to the signal correction model training method in the first aspect.

[0018] In a fifth aspect, an embodiment of the present application provides a signal correction device, including:

[0019] A processor, and a memory storing computer program instructions;

[0020] The processor reads and executes the computer program instructions to implement the signal correction model training method in the first aspect, or the processor reads and executes the computer program instructions to implement the signal correction method in the second aspect.

[0021] In a sixth aspect, an embodiment of the present application provides a computer storage medium, on which computer program instructions are stored, and when the computer program instructions are executed by a processor, the signal correction model training method in the first aspect is implemented, or when the computer program instructions are executed by a processor, the signal correction method in the second aspect is implemented.

[0022] In a seventh aspect, an embodiment of the present application provides a computer program product, and when the instructions in the computer program product are executed by a processor of an electronic device, the electronic device is enabled to execute the signal correction model training method in the first aspect, or when the instructions in the computer program product are executed by a processor of an electronic device, the electronic device is enabled to execute the signal correction method in the second aspect.

[0023] The signal correction model training method, signal correction method, device and equipment in the embodiments of the present application can train a neural network according to signals that meet the correlation conditions to obtain a signal correction model, and use the correction model to correct the signal to be corrected to obtain a corrected signal, and the corrected signal can meet the correlation coefficient condition, so as to avoid the problem of inaccurate processing results when performing signal processing using the signal correlation coefficient subsequently. Description of the Drawings

[0024] To more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings required for use in the embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0025] Figure 1 is a schematic flowchart of a signal correction model training method provided by an embodiment of the present application;

[0026] Figure 2 is a schematic flowchart of a signal correction method provided by an embodiment of the present application;

[0027] Figure 3 is a schematic structural diagram of a signal correction model training device provided by an embodiment of the present application;

[0028] Figure 4 is a schematic structural diagram of a signal correction device provided by an embodiment of the present application;

[0029] Figure 5 is a schematic structural diagram of a signal correction device provided by an embodiment of the present application. Detailed Embodiments

[0030] The following will describe in detail the features and exemplary embodiments of various aspects of the present application. To make the objectives, technical solutions, and advantages of the present application clearer and more understandable, the following further describes the present application in detail in combination with the drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application, rather than limiting the present application. For those skilled in the art, the present application can be implemented without some of these specific details. The following description of the embodiments is only to provide a better understanding of the present application by showing examples of the present application.

[0031] 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 entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including", or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article, or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or further includes elements inherent to such process, method, article, or device. Without more limitations, the elements defined by the statement "including..." do not exclude the existence of additional identical elements in the process, method, article, or device including the said elements.

[0032] Currently, due to factors such as the quality of microphones in voice interaction devices, when collecting the signals of voice data, signals that are theoretically strongly correlated may have low correlation in actual calculations. When performing filtering operations using the correlation coefficient subsequently, the filtering results may be inaccurate.

[0033] To solve the problems of the prior art, the embodiments of the present application provide a method for training a signal correction model, a signal correction method, a device, and a device. First, the signal correction method provided by the embodiments of the present application will be introduced below.

[0034] Figure 1 The flowchart of the signal correction model training method provided by an embodiment of the present application is shown. As Figure 1 shown, the method may include the following steps:

[0035] S110. Obtain training signals, where the training signals include: multiple groups of corresponding signals, and the correlation coefficient of the corresponding signals is greater than or equal to a first preset threshold.

[0036] Exemplarily, the training signals may be obtained from a database or collected by a voice interaction device. The voice interaction device includes at least two audio collection devices. Specifically, the voice interaction device obtains the signals of at least two audio collection devices in multiple voice interaction devices collecting the same voice data, and determines the signals with a correlation coefficient greater than or equal to the first preset threshold between the corresponding signals as the training signals.

[0037] Among them, the first preset threshold may be set to 0.8. The embodiments of the present invention do not specifically limit the first preset threshold, and those skilled in the art may determine it according to actual situations.

[0038] S120. Add a preset noise signal to one of the signals in each group of corresponding signals to obtain target training signals.

[0039] Exemplarily, the training signals may be represented by A - B. Add a preset noise signal to one of the signals A in the corresponding signals of the training signals to obtain a noise signal A'. The obtained multiple groups of corresponding signals after adding noise are used as the target training signals. Among them, the noise signal may be a Gaussian noise signal. The signal A' after adding noise should satisfy the condition that the correlation coefficient Y (A′,B) with the signal B corresponding to A is less than a second preset threshold.

[0040] In one example, the second preset threshold is set to 0.5. The embodiments of the present invention do not specifically limit the second preset threshold, and those skilled in the art may determine it according to actual situations.

[0041] S130. Train the initial signal correction model based on the target training signal to obtain the signal correction model.

[0042] Train the initial signal correction model using signal A' and a preset condition. When the output signal obtained by the initial signal correction model based on the input signal A' has a correlation coefficient with signal B greater than the third preset threshold, the initial signal correction model is the signal correction model. Herein, the third preset threshold can be set and is not limited in this regard; the initial signal correction model is a neural network model.

[0043] In some embodiments, obtaining the training signal includes: obtaining the signals of at least two audio acquisition devices in the voice interaction device collecting the same voice data; calculating the correlation coefficients between every two signals among the signals collected by the at least two audio acquisition devices; and determining the signals with correlation coefficients greater than or equal to the first preset threshold as the training signals. Respectively obtain the signals of at least two audio acquisition devices in the voice interaction device collecting the same voice data, calculate the correlation coefficients between every two of them, and when the correlation coefficient is greater than the first preset threshold, determine the two signals corresponding to the correlation coefficient as the training signals. Herein, the audio acquisition device can be a microphone.

[0044] In some embodiments, training the initial signal correction model based on the target training signal to obtain the signal correction model includes: inputting the target training signal into the initial signal correction model for weighting to obtain the target output signal; and determining the initial signal correction model whose target output signal meets the preset condition as the signal correction model. When training the initial signal correction model, the neural network weights the signal after adding the noise signal to obtain the output signal. Specifically, weight the features of the signal after adding the noise signal, such as the spectral features of the signal, etc. When the correlation coefficient between the signal output by the trained initial signal correction model and the signal corresponding to the signal after adding the noise signal is greater than the third preset threshold, determine the initial signal correction model as the signal correction model. At this time, the signal output by the signal correction model has a correlation with the signal corresponding to the signal after adding the noise signal. Herein, the strength of the correlation of the signal can be represented by the third preset threshold, and the third preset threshold can be set and is not limited in this regard.

[0045] In some embodiments, the preset condition can be that the error function is less than the fourth preset threshold. Wherein the error function error is:

[0046] error=(Y (A,B) -Y (A′,B) ) 2

[0047] Wherein, A is the signal before adding the noise signal, B is the signal corresponding to signal A, A' is the signal after adding the noise signal, Y (A,B)is the correlation coefficient of signal A and signal B, Y (A′,B) is the correlation coefficient of signal A' and signal B. The fourth preset threshold can be set and is not limited herein.

[0048] The signal correction model training method provided by the embodiments of the present application can train the initial signal correction model based on signals that meet the correlation condition, and by weighting the signals, enable the trained signal correction model to correct the signal to be corrected that does not meet the threshold condition.

[0049] Figure 2 shows a schematic flowchart of a signal correction method provided by an embodiment of the present application. As Figure 2 shown, the method may include the following steps:

[0050] S210. Obtain the signal to be corrected, where the signal to be corrected is a signal whose correlation coefficient with the reference signal is less than the first preset threshold.

[0051] Obtain the signal collected by the audio collection device in the voice interaction device, calculate its correlation coefficient with the reference signal, and when the correlation coefficient is less than the first preset threshold, determine the signal as the signal to be corrected. Among them, the first preset threshold can be set and is not limited herein.

[0052] In one example, the relationship between the correlation of signals and the correlation coefficient Y is as follows: when Y < 0.5, the two signals show weak correlation; when 0.5 ≤ Y ≤ 0.8, the two signals show correlation; when 0.8 ≤ Y ≤ 1, the two signals show strong correlation, and the first preset threshold is taken as 0.8.

[0053] S220. Input the signal to be corrected into the signal correction model to obtain the corrected signal, where the signal correction model is trained according to the above-mentioned signal correction model training method.

[0054] Input the obtained signal to be corrected into the correction model, and the correction model weights the signal to be corrected and outputs the corrected signal. Among them, the correction model is trained according to the above-mentioned signal correction model training method.

[0055] The signal correction method provided by the embodiments of the present application can determine the signal to be corrected based on the correlation coefficient between signals, and train the signal correction model by using signals that meet the correlation coefficient condition. The signal to be corrected is corrected by the trained correction model, so that the corrected signal meets the corresponding correlation coefficient condition, and accurate results can be obtained when using the correlation coefficient for signal processing subsequently.

[0056] In some embodiments, obtaining the signal to be corrected includes: obtaining signals of the same voice data collected by at least two audio acquisition devices in a voice interaction device; calculating the correlation coefficients of the signals collected by every two audio acquisition devices; and determining the signal to be corrected according to the correlation coefficients. Obtaining signals of the same voice data collected by at least two audio acquisition devices in multiple voice interaction devices, calculating the correlation coefficients of every two signals among the signals collected by at least two audio acquisition devices in each voice interaction device, and determining the signal whose correlation coefficient with a reference signal is less than a first preset threshold as the signal to be corrected.

[0057] In some embodiments, determining the signal to be corrected according to the correlation coefficients includes: determining a signal whose correlation coefficient with at least one signal is greater than a first preset threshold as a reference signal according to the correlation coefficients; and determining a signal whose correlation coefficient with the reference signal is less than the first preset threshold as the signal to be corrected. According to the calculated correlation coefficients of every two signals among the signals collected by at least two audio acquisition devices in each voice interaction device, determining a signal whose correlation coefficient with at least one signal is greater than a first preset threshold as a reference signal, and further determining a signal whose correlation coefficient with the reference signal is less than the first preset threshold as the signal to be corrected.

[0058] In some embodiments, when it is determined that the Mic_1 signal and the Mic_2 signal have strong correlation and the Mic_1 signal and the Mic_3 signal do not have strong correlation, it is determined that the Mic_3 signal is the signal to be corrected.

[0059] The signal correction method provided by the embodiments of the present application can determine the signal to be corrected according to the correlation coefficients between signals, and correct the signal to be corrected through a correction model, so that the corrected model meets the correlation coefficient condition, and when using the signal correlation coefficients for signal processing subsequently, the problem of inaccurate processing results can be avoided.

[0060] Figure 3 It is a schematic structural diagram of a signal correction model training device 300 provided by the embodiments of the present application. As Figure 3 shown, the device may include an acquisition module 310, an addition module 320, and a training module 330.

[0061] The acquisition module 310 is configured to acquire training signals, where the training signals include: multiple groups of corresponding signals, and the correlation coefficients of the corresponding signals are greater than or equal to a first preset threshold;

[0062] The addition module 320 is configured to add a preset noise signal to one of the signals in each group of corresponding signals respectively to obtain target training signals;

[0063] The training module 330 is configured to train an initial signal correction model based on the target training signals to obtain a signal correction model.

[0064] The signal correction model training device provided by the embodiment of the present application can train an initial signal correction model based on signals that meet the threshold conditions. By weighting the signals, the trained signal correction model can correct signals that do not meet the threshold conditions.

[0065] In some embodiments, an acquisition module 310 is configured to acquire training signals, including: the acquisition module 310 is configured to acquire signals of the same voice data collected by at least two audio acquisition devices in a voice interaction device; a calculation module 340 is configured to calculate the correlation coefficients between every two of the signals collected by the at least two audio acquisition devices; a determination module 350 is configured to determine the signals with correlation coefficients greater than or equal to a first preset threshold as training signals.

[0066] In some embodiments, a training module 330 is configured to train an initial signal correction model based on target training signals to obtain a signal correction model, including: a weighting module 360 is configured to input the target training signals into the initial signal model for weighting to obtain target output signals; a determination module 350 is configured to determine the initial signal correction model whose target output signals meet the preset conditions as the signal correction model.

[0067] The signal correction model training device provided by the embodiment of the present application can train an initial signal correction model based on one-to-one signals and noise signals with strong correlation. By means of weighting, the trained signal correction model can correct signals to be corrected, so that the corrected signals meet the correlation conditions, and accurate results can be obtained in subsequent signal processing.

[0068] Figure 4 It is a schematic structural diagram of a signal correction device 400 provided by the embodiment of the present application. As Figure 4 shown, the device may include an acquisition module 410 and an input module 420.

[0069] The acquisition module 410 is configured to acquire a signal to be corrected, and the signal to be corrected is a signal whose correlation coefficient with a reference signal is less than a first preset threshold.

[0070] The input module 420 is configured to input the signal to be corrected into the signal correction model to obtain a corrected signal, and the correction model is trained according to the above-mentioned signal correction model training method.

[0071] The signal correction device provided by the embodiment of the present application can determine a signal to be corrected based on the correlation coefficient between signals, and train a signal correction model by using signals that meet the correlation coefficient conditions. The signal to be corrected is corrected by the trained correction model, so that the corrected signal meets the corresponding correlation coefficient conditions, and accurate results can be obtained when using the correlation coefficient for signal processing subsequently.

[0072] In some embodiments, an acquisition module 410 is configured to acquire a signal to be corrected, including: the acquisition module 410 is configured to acquire signals of the same voice data collected by at least two audio acquisition devices in a voice interaction device; a calculation module 430 is configured to calculate a correlation coefficient between signals collected by every two audio devices; and a determination module 440 is configured to determine the signal to be corrected according to the correlation coefficient.

[0073] In some embodiments, the determination module 440 is configured to determine the signal to be corrected according to the correlation coefficient, including: the determination module 440 determines, according to the correlation coefficient, a signal whose correlation coefficient with at least one signal is greater than a first preset threshold as a reference signal; and determines a signal whose correlation coefficient with the reference signal is less than the first preset threshold as the signal to be corrected.

[0074] The signal correction device provided by the embodiments of the present application can train a neural network by using the correlation between signals to obtain a signal correction model. At the same time, the signal to be corrected is determined according to the correlation coefficient between signals, and the signal to be corrected is corrected by the correction model, so that the corrected model meets the correlation coefficient condition, and when the signal correlation coefficient is used for signal processing subsequently, the problem of inaccurate processing results can be avoided.

[0075] Figure 3 Each module in the illustrated device has the function of implementing Figure 1 each step therein and can achieve its corresponding technical effect. For the sake of brief description, it will not be elaborated herein.

[0076] Figure 4 Each module in the illustrated device has the function of implementing Figure 2 each step therein and can achieve its corresponding technical effect. For the sake of brief description, it will not be elaborated herein.

[0077] Figure 5 The hardware structure diagram of a signal correction device provided by the embodiments of the present application is shown.

[0078] The signal correction device may include a processor 501 and a memory 502 storing computer program instructions.

[0079] Specifically, the above-mentioned processor 501 may include a Central Processing Unit (CPU), or an Application Specific Integrated Circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0080] The memory 502 may include a mass storage for data or instructions. By way of example and not limitation, the memory 502 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive, or a combination of two or more of these. In one example, the memory 502 may include removable or non-removable (or fixed) media, or the memory 502 is a non-volatile solid-state memory. The memory 502 may be internal or external to the integrated gateway disaster recovery device.

[0081] In one example, the memory 502 may include a read-only memory (ROM), a random access memory (RAM), a magnetic disk storage media device, an optical storage media device, a flash memory device, an electrical, optical, or other physical / tangible memory storage device. Thus, generally, the memory 502 includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to one aspect of the present application.

[0082] The processor 501 reads and executes the computer program instructions stored in the memory 502 to implement Figure 1 the methods / steps S110 to S130 in the illustrated embodiments and achieve Figure 1 the corresponding technical effects achieved by the methods / steps executed in the illustrated examples, or, to implement Figure 2 the methods / steps S110 to S120 in the illustrated embodiments and achieve Figure 2 the corresponding technical effects achieved by the methods / steps executed in the illustrated examples. For the sake of brevity, the description is not repeated here.

[0083] In one example, the signal correction device may further include a communication interface 503 and a bus 510. Among them, as Figure 5 shown, the processor 501, the memory 502, and the communication interface 503 are connected through the bus 510 and complete communication with each other.

[0084] The communication interface 503 is mainly used to implement communication between various modules, devices, units, and / or devices in the embodiments of the present application.

[0085] The bus 510 includes hardware, software, or both, and couples the components of the signal correction device to each other. By way of example and not limitation, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an InfiniBand interconnect, a Low Pin Count (LPC) bus, a memory bus, a Micro Channel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses or a combination of two or more of these. Where appropriate, the bus 510 may include one or more buses. Although the embodiments of the present application describe and illustrate specific buses, the present application contemplates any suitable bus or interconnect.

[0086] The signal correction device may execute the signal correction method in the embodiments of the present application based on the correlation coefficient between signals and the correction model, so as to implement the combination Figure 1 of the signal correction model training method described, or implement the combination Figure 2 of the signal correction method described.

[0087] In addition, in combination with the signal correction method in the above embodiments, the embodiments of the present application may provide a computer storage medium to implement. Computer program instructions are stored on the computer storage medium; when the computer program instructions are executed by a processor, any one of the signal correction methods in the above embodiments is implemented.

[0088] The embodiments of the present application provide a computer program product, characterized in that when the instructions in the computer program product are executed by a processor of an electronic device, the electronic device executes the above signal correction method.

[0089] It should be clear that the present application is not limited to the specific configurations and processes described above and illustrated in the figures. For the sake of brevity, the detailed description of known methods is omitted here. In the above embodiments, several specific steps are described and illustrated as examples. However, the method process of the present application is not limited to the specific steps described and illustrated, and those skilled in the art can make various changes, modifications, and additions, or change the order between steps after understanding the spirit of the present application.

[0090] The functional blocks shown in the above-described structural block diagrams can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, it can be, for example, an electronic circuit, an application specific integrated circuit (ASIC), appropriate firmware, a plug-in, a functional card, and so on. When implemented in software, the elements of the present application are programs or code segments used to perform the required tasks. The program or code segment can be stored in a machine-readable medium or transmitted over a transmission medium or communication link via a data signal carried in a carrier wave. A "machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical discs, hard disks, fiber optic media, radio frequency (RF) links, and so on. The code segment can be downloaded via a computer network such as the Internet, an intranet, and so on.

[0091] It should also be noted that the exemplary embodiments mentioned in the present application describe some methods or systems based on a series of steps or devices. However, the present application is not limited to the order of the above steps, that is, the steps can be executed in the order mentioned in the embodiments, or different from the order in the embodiments, or several steps can be executed simultaneously.

[0092] Aspects of the present application have been described above with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present application. It should be understood that each block in the flowcharts and / or block diagrams, and the combinations of blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device to produce a machine such that the instructions executed by the processor of the computer or other programmable data processing device enable the implementation of the functions / actions specified in one or more blocks of the flowchart and / or block diagram. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field programmable logic circuit. It should also be understood that each block in the block diagram and / or flowchart, and the combinations of blocks in the block diagram and / or flowchart, can also be implemented by dedicated hardware that performs the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0093] As described above, this is only the specific implementation manner of the present application. Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, modules, and units described above can refer to the corresponding processes in the foregoing method embodiments, and will not be elaborated herein. It should be understood that the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of various equivalent modifications or substitutions, and these modifications or substitutions should all be covered within the protection scope of the present application.

Claims

1. A method for training a signal correction model, characterized in that, it includes: Obtain training signals, where the training signals include: multiple groups of corresponding signals, and the correlation coefficient of the corresponding signals is greater than or equal to a first preset threshold; Add a preset noise signal to one of the signals in each group of corresponding signals respectively to obtain target training signals, where the correlation coefficient between the target training signals and the other signal in the corresponding signals is less than a second preset threshold; Train an initial signal correction model based on the target training signals to obtain a signal correction model, so that the signal correction model corrects the signal to be corrected that does not meet the threshold condition; The obtaining of the training signals includes: Obtain the signals collected by at least two audio collection devices in a voice interaction device respectively for the same voice data; Calculate the correlation coefficients between every two signals among the signals collected by the at least two audio collection devices; Determine the signals with the correlation coefficient greater than or equal to the first preset threshold as the training signals.

2. The method according to claim 1, characterized in that, The training of the initial signal correction model based on the target training signals to obtain a signal correction model includes: Input the target training signals into the initial signal correction model for weighting to obtain target output signals; Determine the initial signal correction model whose target output signals meet the preset conditions as the signal correction model.

3. A signal correction method, characterized in that, it includes: Obtain a signal to be corrected, where the signal to be corrected is a signal whose correlation coefficient with a reference signal is less than a first preset threshold; Input the signal to be corrected into a signal correction model to obtain a corrected signal, and the signal correction model is trained according to the signal correction model training method described in any one of claims 1-2.

4. The method according to claim 3, characterized in that, The obtaining of the signal to be corrected includes: Obtain the signals collected by at least two audio collection devices in a voice interaction device respectively for the same voice data; Calculate the correlation coefficients between the signals collected by every two audio collection devices; Determine the signal to be corrected according to the correlation coefficient.

5. The method according to claim 4, characterized in that, The determining of the signal to be corrected according to the correlation coefficient includes: Determine the signal whose correlation coefficient with at least one signal is greater than the first preset threshold as the reference signal according to the correlation coefficient; Determine the signal whose correlation coefficient with the reference signal is less than the first preset threshold as the signal to be corrected.

6. A signal correction model training device, characterized in that, the device includes: An obtaining module, configured to obtain training signals, where the training signals include: multiple groups of corresponding signals, and the correlation coefficient of the corresponding signals is greater than or equal to a first preset threshold; An adding module, configured to add a preset noise signal to one of the signals in each group of corresponding signals respectively to obtain target training signals, where the correlation coefficient between the target training signals and the other signal in the corresponding signals is less than a second preset threshold; A training module, configured to train an initial signal correction model based on the target training signal to obtain a signal correction model, so that the signal correction model corrects a signal to be corrected that does not meet the threshold condition; The obtaining module is further specifically configured to: Obtain signals of the same voice data collected by at least two audio collection devices in a voice interaction device respectively; Calculate the correlation coefficients between every two of the signals collected by the at least two audio collection devices; Determine the signals with the correlation coefficients greater than or equal to the first preset threshold as the training signals.

7. A signal correction device, Characterized in that, The device includes: An obtaining module, configured to obtain a signal to be corrected, where the signal to be corrected is a signal with a correlation coefficient less than a first preset threshold with a reference signal; An input module, configured to input the signal to be corrected into a signal correction model to obtain a corrected signal, where the signal correction model is trained according to the signal correction model training method described in any one of claims 1-2.

8. A signal correction device, Characterized in that, The signal correction device includes: a processor and a memory storing computer program instructions; The processor reads and executes the computer program instructions to implement the signal correction model training method described in any one of claims 1-2 or the signal correction method described in any one of claims 3-5.

9. A computer storage medium, Characterized in that, Computer program instructions are stored on the computer storage medium, and when the computer program instructions are executed by a processor, the signal correction model training method described in any one of claims 1-2 or the signal correction method described in any one of claims 3-5 is implemented.

10. A computer program product, Characterized in that, When the instructions in the computer program product are executed by a processor of an electronic device, the electronic device is caused to execute the signal correction model training method described in any one of claims 1-2 or the signal correction method described in any one of claims 3-5.

Images