Noise reduction processing method and device for voice call, electronic equipment and chip

Abstract

The invention discloses a noise reduction processing method and device for a voice call, electronic equipment and a chip, and relates to the technical field of communication, and the method comprises the steps: firstly recording first voice data during the voice call of a target user; sending the recorded first voice data to a server, wherein the first voice data is used for training a first voice noise reduction model corresponding to the target user; and receiving a trained first voice noise reduction model sent by the server, wherein the first voice noise reduction model is used for noise reduction processing during voice communication of the target user. Compared with the prior art, the voice data of the target user can be recorded and sent to the server side, so that the server side can train the voice noise reduction model corresponding to the target user, and in the process of noise reduction of the voice data of the target user, the voice noise reduction model corresponding to the target user sent by the server side is used to reduce the noise of the voice data of the target user. Noise reduction processing can be carried out more accurately, and the noise reduction effect of voice communication is improved.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method, device, electronic device and chip for noise reduction processing of voice calls. Background Art

[0002] When a user uses a terminal to make a voice call, it is possible that the user is in a noisy environment, and the voice uttered by the user will be sent to the other party's terminal together with the environmental noise.

[0003] Currently, a speech noise reduction model can be used for processing to achieve the purpose of noise reduction for voice calls, wherein the speech noise reduction model adopts a universal model, that is, a unified noise reduction model is adopted for processing voice calls of different users.

[0004] However, the voices emitted by different users are generally different. If a unified noise reduction model is used for processing, the noise reduction effect of voice calls will be affected. Summary of the invention

[0005] In view of this, the present application provides a noise reduction processing method, device, electronic device and chip for voice calls, the main purpose of which is to improve the current technical problem that the noise reduction effect of voice calls will be affected if a unified noise reduction model is used for processing because the voices emitted by different users are generally different.

[0006] In a first aspect, the present application provides a method for noise reduction processing of a voice call, comprising:

[0007] Recording the first voice data of the target user during the voice call;

[0008] Sending the first recorded voice data to a server, where the first voice data is used to train a first voice noise reduction model corresponding to the target user;

[0009] The first speech noise reduction model that has been trained and sent by the server is received, where the first speech noise reduction model is used for noise reduction processing during a voice call of the target user.

[0010] In a second aspect, the present application provides a noise reduction processing device for voice calls, comprising:

[0011] A recording module, configured to record first voice data of a target user during a voice call;

[0012] A sending module, configured to send the first recorded voice data to a server, wherein the first voice data is used to train a first voice noise reduction model corresponding to the target user;

[0013] The receiving module is configured to receive the first speech noise reduction model that has been trained and sent by the server, and the first speech noise reduction model is used for noise reduction processing during the voice call of the target user.

[0014] In a third aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the noise reduction processing method for voice calls described in the first aspect is implemented.

[0015] In a fourth aspect, the present application provides an electronic device, comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor implements the noise reduction processing method for voice calls described in the first aspect when executing the computer program.

[0016] In a fifth aspect, the present application provides a chip comprising one or more interface circuits and one or more processors; the interface circuit is used to receive a signal from a memory of an electronic device and send the signal to the processor, wherein the signal comprises a computer instruction stored in the memory; when the processor executes the computer instruction, the electronic device executes the noise reduction processing method for voice calls described in the first aspect.

[0017] By means of the above technical scheme, the present application provides a method, device, electronic device and chip for noise reduction processing of voice calls, which specifically first records the first voice data of the target user during the voice call; sends the recorded first voice data to the server, and the first voice data is used to train the first voice noise reduction model corresponding to the target user; receives the trained first voice noise reduction model sent by the server, and the first voice noise reduction model is used for noise reduction processing during the voice call of the target user. Compared with the current existing technology, the present application can record the voice data of the target user and send it to the server, so that the server can train the voice noise reduction model corresponding to the target user. In the process of noise reduction of the voice data of the target user, the voice noise reduction model corresponding to the target user sent by the server can be used to perform noise reduction processing more accurately, thereby improving the noise reduction effect of the voice call.

[0018] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] 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.

[0020] 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.

[0021] Figure 1 A schematic diagram showing a flow chart of a method for noise reduction processing of a voice call provided in an embodiment of the present application is shown;

[0022] Figure 2 A schematic diagram showing a flow chart of a method for noise reduction processing of a voice call provided in an embodiment of the present application is shown;

[0023] Figure 3 A schematic diagram showing an example provided by an embodiment of the present application;

[0024] Figure 4 A schematic diagram showing an example provided by an embodiment of the present application;

[0025] Figure 5 A schematic diagram showing an example provided by an embodiment of the present application;

[0026] Figure 6 A schematic diagram showing an example provided by an embodiment of the present application;

[0027] Figure 7 A schematic diagram showing an example provided by an embodiment of the present application;

[0028] Figure 8 A schematic diagram of the structure of a noise reduction processing device for voice calls provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0029] The embodiments of the present application will be described in more detail below with reference to the accompanying drawings. It should be noted that the embodiments and features in the embodiments of the present application can be combined with each other without conflict.

[0030] In order to improve the current technical problem that the noise reduction effect of voice calls will be affected if a unified noise reduction model is used for processing due to the general differences in the voices emitted by different users. This embodiment provides a noise reduction processing method for voice calls, such as Figure 1 As shown, the method includes:

[0031] Step 101: Record first voice data of a target user during a voice call.

[0032] In some examples, the method shown in this embodiment can be executed by a terminal device, which may include a mobile phone, a tablet computer with a voice call function, a car device, a camera, etc. The embodiments of the present disclosure do not limit the specific technology and specific device form used by the terminal device.

[0033] In an embodiment of the present application, the first voice data may be voice data recognized by the terminal device as containing the voiceprint features of the target user; the target user may be a fixed user of the terminal device, a user who frequently uses the terminal device, a trusted user in the terminal device, etc. Optionally, the target user of a terminal device may be one person or several persons, and the number of target users of the terminal device is not specifically limited herein.

[0034] For example, if the terminal device is a mobile phone, the target user may be the owner of the mobile phone, or may be a second user or a third user who has recorded a voiceprint model in the mobile phone, and so on.

[0035] Step 102: Send the recorded first voice data to the server.

[0036] The first speech data is used to train a first speech noise reduction model corresponding to the target user.

[0037] In an embodiment of the present application, the first speech noise reduction model is a noise reduction model for a target user. Optionally, the server may be a cloud server, etc. The server obtains the first speech noise reduction model through training with the first speech data.

[0038] Optionally, the noise reduction model can be used to remove noise data. For example, during a call or video, there will be external noise when the user is speaking. The noise reduction model can remove a certain amount of noise, making the sound heard by the other user clearer during the call or video, thereby improving the quality of the call or video.

[0039] For example, for target user A, during a call, the voiceprint features of target user A can be identified, and the terminal device sends the voice data containing the voiceprint features of target user A to the server. The server can train a noise reduction model for target user A through the voice data of target user A.

[0040] Step 103: Receive the trained first speech noise reduction model sent by the server.

[0041] Among them, the first speech noise reduction model is used for noise reduction processing during the target user's voice call.

[0042] In the embodiment of the present application, the first speech noise reduction model is a speech noise reduction model for the target user. After the server completes the training of the first speech noise reduction model, the server sends the model to the terminal device for application.

[0043] Compared with the related art, this embodiment can record the voice data of the target user and send it to the server, so that the server can train a voice noise reduction model corresponding to the target user. In the process of denoising the voice data of the target user, the voice noise reduction model corresponding to the target user sent by the server can be used to perform noise reduction processing more accurately, thereby improving the noise reduction effect of voice calls.

[0044] In order to further illustrate the specific implementation process of the method of this embodiment, this embodiment provides the following Figure 2 The specific method shown, wherein the method includes:

[0045] Step 201: Input the voiceprint features of the speaker in the voice call data into the voiceprint recognition model of the target user for calculation.

[0046] Optionally, step 201 may be used to identify whether the voiceprint feature is the voiceprint feature of the target user.

[0047] Optionally, before step 201, the method of this embodiment further includes: performing noise reduction processing on the voice call data through a universal second voice noise reduction model.

[0048] Optionally, step 201 may specifically include: inputting the voiceprint features of the speaker in the voice call data after noise reduction processing into a voiceprint recognition model of the target user for calculation.

[0049] Optionally, before step 201, the method of this embodiment further includes: acquiring a voiceprint recognition model from a preset storage directory.

[0050] Among them, the voiceprint recognition model is trained based on the voice information emitted by the target user when entering the voice assistant wake-up word.

[0051] In the embodiment of the present application, the voiceprint features may include the speaker's pitch, volume, duration, timbre, etc. The voice call data may be a call made by an ordinary mobile phone, or a voice call made by some third-party applications, etc., which is not limited in this embodiment.

[0052] Optional, such as Figure 3 As shown, the user can turn on the voice wake-up function of a mobile phone terminal or an intelligent hardware device such as the Internet of Things (IOT), and generate a user voiceprint recognition model by entering a wake-up word, such as "Xiao Ai", etc. After the voiceprint recognition model is trained, the user interface (User Interface, UI) prompts the user whether to turn on the target person's voice noise reduction; if the user chooses to turn it on, the wake-up service (service) sends a message (message) to the call service, and copies the voiceprint recognition model to the corresponding directory.

[0053] In the embodiment of the present application, the Internet of Things can be a huge network formed by combining various information sensing devices, such as radio frequency identification devices, infrared sensors, global positioning systems, laser scanners, etc. with the Internet.

[0054] In the embodiments of the present application, Figure 4 As shown, when a user is speaking during a call, a mobile phone terminal or an intelligent hardware device such as IOT can pick up the user's voice, and obtain the voice data after noise reduction processing through a preset general AI noise reduction model, and send it to the user on the other end of the call through a modem.

[0055] Optionally, a modem is a piece of terminal equipment hardware that can translate the digital signals of a terminal equipment into analog signals that can be transmitted along ordinary telephone lines. These analog signals can then be received by another modem at the other end of the line and translated into the language of the terminal equipment. This simple process completes the communication between the two terminal equipments.

[0056] Optionally, the second speech noise reduction model is a general artificial intelligence (AI) noise reduction model. AI noise reduction is a method of performing noise reduction processing on audio, video and other signals through artificial intelligence, and can improve signal quality and clarity by analyzing and processing the signals.

[0057] In an embodiment of the present application, the preset storage directory may be a storage location in a terminal device for storing voiceprint recognition models. In the preset storage directory, not only the voiceprint recognition model of one user may be stored, but also the voiceprint recognition models of other users who have entered voiceprint features through the voice wake-up function may be stored.

[0058] For example, if user A uses the voice wake-up function of mobile phone a and enters the wake-up word, the voiceprint features corresponding to user A can be generated, and after training, the voiceprint recognition model corresponding to user A can be obtained; mobile phone a can also record the voiceprint features corresponding to user B generated by user B entering the wake-up word, and after training, the voiceprint recognition model corresponding to user B can be obtained. In the preset storage directory, the voiceprint recognition models of user A and user B can exist. As an optional method, for the preset storage directory, a threshold value for the number of voiceprint recognition models can be set, which can be 2 or 3, etc.

[0059] Step 202: extract first voice data containing the voiceprint features of the target user from the voice call data according to the recognition result, and record the first voice data.

[0060] In the embodiments of the present application, Figure 5As shown, during a call, after the call service of the terminal device receives the message, the terminal device opens the target user's voice screening channel; the obtained voice data after noise reduction can be input into the voiceprint recognition model for screening calculation to confirm whether it is the voice of the target user. If so, the voice data before noise reduction will be intercepted and the part containing the voice will be saved.

[0061] Exemplarily, if the voices of user A and user B are present at the same time in the current voice call, the terminal device can respectively identify the voiceprint features of user A and user B, and determine whether there are voiceprint recognition models of user A and user B in the preset storage directory. If only the voiceprint feature recognition model of user A exists in the preset storage directory, the voice data of user A is intercepted and saved; if only the voiceprint feature recognition model of user B exists in the preset storage directory, the voice data of user B is intercepted and saved; if both the voiceprint feature recognition model of user A and the voiceprint feature model of user B exist in the preset storage directory, the voice data of user A and the voice data of user B are respectively intercepted and saved.

[0062] Step 203: Send the recorded first voice data to the server.

[0063] The first speech data is used to train a first speech noise reduction model corresponding to the target user.

[0064] Optionally, step 203 may specifically include: sending the first voice data whose number of recorded voice items reaches a first preset threshold to the server; or sending the first voice data recorded within a preset time period to the server.

[0065] In an embodiment of the present application, the first preset threshold can be an integer greater than 1. If the first preset threshold is 1, the recorded voice data can be uploaded in real time. When the first preset threshold is other values, it is uploaded to the server after accumulating to the first preset threshold.

[0066] Optionally, the preset time period may be 1 day, 1 hour, 1 minute, etc., and the terminal device uploads the voice data recorded within the preset time period to the server.

[0067] Exemplarily, if the preset time period is 1 month, the first preset threshold is 50, and the terminal device records 100 voice data within 1 month, the terminal device will only upload the first 50 recorded voice data to the server. In the next month, the first 50 voice data recorded in the next month will be uploaded to the server. The first preset threshold and preset time period can control the amount of data uploaded by the terminal device to the server from being too large, resulting in a huge amount of data on the server, reducing the number of tasks on the server, and thereby improving the training efficiency of the server.

[0068] Step 204: Receive the trained first speech noise reduction model sent by the server.

[0069] Among them, the first speech noise reduction model is used for noise reduction processing during the target user's voice call.

[0070] Optionally, after step 204, the method of this embodiment further includes: obtaining second speech data after noise reduction processing by the first speech noise reduction model; determining the signal-to-noise ratio of the second speech data; if the signal-to-noise ratio is less than a second preset threshold, sending the third speech data before noise reduction processing of the second speech data to the server. Receive the updated trained first speech noise reduction model sent by the server.

[0071] The third voice data is used to update and train the first voice noise reduction model.

[0072] In the embodiment of the present application, the noise reduction processing may specifically be to calculate the coefficient corresponding to the second voice data in the first voice data through the noise reduction model, and the second voice data may be obtained by multiplying the first voice data by the calculated coefficient.

[0073] In the embodiment of the present application, the signal-to-noise ratio (SNR) may be the ratio of the signal to the noise in an electronic device or electronic system. The signal may be an electronic signal from outside the device that needs to be processed by the device, and the noise may be an irregular additional signal that does not exist in the original signal after passing through the device, and the signal does not change with the change of the original signal.

[0074] Optionally, the first speech noise reduction model is a speech noise reduction model for the target user, the second preset threshold is a threshold corresponding to the signal-to-noise ratio, which can be used to measure the noise reduction effect of the first noise reduction model, and the third voice data can be data in which the number of voice items in the user's call record exceeds the first threshold, or it can be voice data recorded in real time for the current call.

[0075] In the embodiments of the present application, Figure 6 As shown, when the terminal device receives the optimized AI noise reduction model, the target person's voice noise reduction optimization channel is opened. Specifically, when a call is in progress, the user speaks, and the mobile terminal or IOT and other intelligent hardware devices pick up the user's voice. The audio first passes through the voiceprint model. If it is the target user's voice, the audio is processed by the optimized target person AI noise reduction model for noise reduction. If it is not the target user's voice, the audio is processed by the preset general AI noise reduction model for noise reduction.

[0076] In the embodiments of the present application, Figure 7As shown, the audio of the optimized target user AI noise reduction model can be passed through subsequent modules to calculate the SNR of the voice. If the signal-to-noise ratio is poor, such as less than 6dB, it means that the current model has a general noise reduction effect on the scene. The noisy voice of the target user can be uploaded to the server in time for targeted optimization.

[0077] Compared with the current existing technology, this embodiment generates a voiceprint model of the target user through the voice wake-up function, filters the audio data of the call based on the voiceprint model, and after uploading the filtered audio data to the server, updates the noise reduction model of the call, thereby obtaining a better voice noise reduction model and sending it to the terminal device. The terminal device uses the voice noise reduction model corresponding to the voiceprint characteristics of the target user for the voice data of the target user, and can more accurately perform noise reduction processing on the voice data of the target user, thereby improving the noise reduction effect of the voice call.

[0078] In order to illustrate the specific implementation process of this embodiment, the following specific application examples are given, but are not limited thereto:

[0079] User B turns on the voice wake-up function of a mobile phone terminal or smart hardware device such as IOT, and generates a user voiceprint model by entering a wake-up word, such as "Xiao Ai Tongxue". After the training is completed, the UI prompts the user whether to turn on the target person's voice noise reduction; if the user chooses to turn it on, the wake-up service sends a message to the call service and copies the voiceprint model to the corresponding directory. The call service receives the message and turns on the target person's voice screening channel.

[0080] When a call is in progress, the user speaks, and the mobile terminal or IOT and other smart hardware devices pick up the user's voice, and after passing through the preset general AI noise reduction model, the noise-reduced voice data is obtained, and sent to the user on the other end of the call through the modem; at the same time, the obtained noise-reduced voice data is input into the voiceprint model for screening and calculation to confirm whether it is the voice of the target user. If so, the voice data before noise reduction will be intercepted, and the part containing the voice will be saved. When the number of intercepted voice items of the target user reaches a certain number, for example 50, the call service uploads the intercepted audio data to the server.

[0081] The server receives the intercepted audio of the target user, trains and produces a new AI noise reduction model, and sends it to the user's mobile terminal or smart hardware devices such as IOT. The new AI model is a speech noise reduction model for the target user.

[0082] The call service receives the optimized AI noise reduction model and opens the target person's voice noise reduction optimization channel; when the call is in progress, the user speaks, and the mobile terminal or IOT and other smart hardware devices pick up the user's voice. The audio first passes through the voiceprint model. If it is the target user's voice, the audio is passed through the optimized target person AI noise reduction model. If it is not the target user's voice, the audio is passed through the preset general AI noise reduction model to continue optimizing the target person noise reduction model. The update frequency can be set, for example, once a month.

[0083] The audio of the optimized target person AI noise reduction model is used to calculate the SNR of the speech. If the signal-to-noise ratio is low, such as less than 6dB, it means that the current model has a general noise reduction effect on the scene. The noisy speech of the target user can be uploaded to the server in time for targeted optimization.

[0084] The existing technology can obtain the target voice signal and the target speaker voice signal; extract the features of the target speaker voice signal to obtain the voiceprint features of the target speaker; input the frequency spectrum features of the target voice signal and the voiceprint features of the target speaker into the pre-trained noise reduction neural network to obtain the noise reduction parameter features of the target voice signal; or compare and extract the voice after preliminary screening with the specific person voiceprint recognition model to perform noise reduction; or use the user's voiceprint feature information to collect the sound signal when the user is talking, compare the sound signal with the voiceprint feature information, and separate the voice signal and the pure noise signal; or send the noisy voice to the server for noise reduction processing, and then return the processed clean voice to the client, and the server will perform additional training on the voice. However, the noise reduction models all use general models and cannot be customized for specific people.

[0085] Compared with the current existing technology, this embodiment generates a voiceprint model of the target user through the voice wake-up function, filters the audio data of the call based on the voiceprint model, and after uploading the filtered audio data to the server, updates the noise reduction model of the call, thereby obtaining a better voice noise reduction model and sending it to the terminal device. The terminal device uses the voice noise reduction model corresponding to the voiceprint characteristics of the target user for the voice data of the target user, and can more accurately perform noise reduction processing on the voice data of the target user, thereby improving the noise reduction effect of the voice call.

[0086] Further, as Figure 1 and Figure 2 The specific implementation of the method shown in the embodiment provides a noise reduction processing device for voice calls, such as Figure 8 As shown, the device includes: a recording module 31, a sending module 32, and a receiving module 33.

[0087] The recording module 31 is configured to record the first voice data of the target user during the voice call;

[0088] A sending module 32, configured to send the first recorded voice data to a server, wherein the first voice data is used to train a first voice noise reduction model corresponding to the target user;

[0089] The receiving module 33 is configured to receive the first speech noise reduction model that has been trained and sent by the server, where the first speech noise reduction model is used for noise reduction processing during a voice call of the target user.

[0090] In some examples of this embodiment, the recording module 31 is specifically configured to input the voiceprint features of the speaker in the voice call data into the voiceprint recognition model of the target user for calculation to identify whether the voiceprint features are the voiceprint features of the target user; based on the recognition result, extract the first voice data containing the voiceprint features of the target user from the voice call data, and record the first voice data.

[0091] In some examples of this embodiment, the recording module 31 is also configured to perform noise reduction processing on the voice call data through a general second voice noise reduction model; correspondingly, the recording module 31 is further configured to input the voiceprint features of the speaker in the voice call data after noise reduction processing into the voiceprint recognition model of the target user for calculation.

[0092] In some examples of this embodiment, the recording module 31 is further configured to obtain the voiceprint recognition model from a preset storage directory, where the voiceprint recognition model is trained based on the voice information emitted by the target user when entering the voice assistant wake-up word.

[0093] In some examples of this embodiment, the receiving module 33 is also configured to obtain second speech data after noise reduction processing by the first speech noise reduction model; determine the signal-to-noise ratio of the second speech data; if the signal-to-noise ratio is less than a second preset threshold, send the third speech data before noise reduction processing of the second speech data to the server, and the third speech data is used to update and train the first speech noise reduction model; receive the updated trained first speech noise reduction model sent by the server.

[0094] In some examples of this embodiment, the sending module 32 is specifically configured to send the first voice data when the number of recorded voice items reaches a first preset threshold to the server; or, send the first voice data recorded within a preset time period to the server.

[0095] It should be noted that for other corresponding descriptions of the functional units involved in the noise reduction processing device for voice calls provided in this embodiment, reference can be made to Figure 1 and Figure 2 The corresponding description in will not be repeated here.

[0096] Based on the above Figure 1 and Figure 2 The method shown in the embodiment also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the above-mentioned Figure 1 and Figure 2 The method shown.

[0097] Based on this understanding, the technical solution of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, USB flash drive, mobile hard disk, etc.), including a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods of various implementation scenarios of the present application.

[0098] Based on the above Figure 1 and Figure 2 The method shown, and Figure 8 In order to achieve the above-mentioned purpose, the embodiment of the present application also provides an electronic device, such as a smart phone, a tablet computer, a drone, a smart robot and other smart terminals, which includes a storage medium and a processor; the storage medium is used to store a computer program; the processor is used to execute the computer program to achieve the above-mentioned Figure 1 and Figure 2 The method shown.

[0099] Optionally, the above-mentioned physical device may also include a user interface, a network interface, a camera, a radio frequency (RF) circuit, a sensor, an audio circuit, a WI-FI module, etc. The user interface may include a display, an input unit such as a keyboard, etc., and the optional user interface may also include a USB interface, a card reader interface, etc. The network interface may optionally include a standard wired interface, a wireless interface (such as a WI-FI interface), etc.

[0100] Those skilled in the art will appreciate that the above-mentioned physical device structure provided in this embodiment does not constitute a limitation on the physical device, and may include more or fewer components, or a combination of certain components, or different arrangements of components.

[0101] The storage medium may also include an operating system and a network communication module. The operating system is a program that manages the hardware and software resources of the above-mentioned physical device, and supports the operation of the information processing program and other software and / or programs. The network communication module is used to realize the communication between the components inside the storage medium, and the communication with other hardware and software in the information processing physical device.

[0102] Based on the above Figure 1 and Figure 2 The method shown, and Figure 8 The virtual device embodiment shown in the embodiment further provides a chip, including one or more interface circuits and one or more processors; the interface circuit is used to receive a signal from a memory of an electronic device and send the signal to the processor, the signal including a computer instruction stored in the memory; when the processor executes the computer instruction, the electronic device executes the above-mentioned Figure 1 and Figure 2 The method shown.

[0103] Through the description of the above implementation methods, those skilled in the art can clearly understand that the present application can be implemented by means of software plus the necessary general hardware platform, or by hardware. By applying the solution of this embodiment, compared with the current prior art, this embodiment generates a voiceprint model of the target user through a voice wake-up function, filters the audio data of the call based on the voiceprint model, and after uploading the filtered audio data to the server, updates the noise reduction model of the call, thereby obtaining a better voice noise reduction model and sending it to the terminal device. The terminal device uses the voice noise reduction model corresponding to the voiceprint feature of the target user for the voice data of the target user, and can more accurately perform noise reduction processing on the voice data of the target user, thereby improving the noise reduction effect of the voice call.

[0104] 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 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.

[0105] The above 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 described herein, but will conform to the widest scope consistent with the principles and novel features applied for herein.

Claims

1. A method for noise reduction processing of voice calls, characterized in that: include: Recording the first voice data of the target user during the voice call; Sending the first recorded voice data to a server, where the first voice data is used to train a first voice noise reduction model corresponding to the target user; The first speech noise reduction model that has been trained and sent by the server is received, where the first speech noise reduction model is used for noise reduction processing during a voice call of the target user.

2. The method according to claim 1, characterized in that The first voice data recorded during the voice call of the target user includes: Inputting the voiceprint feature of the speaker in the voice call data into the voiceprint recognition model of the target user for calculation to identify whether the voiceprint feature is the voiceprint feature of the target user; According to the recognition result, the first voice data including the voiceprint feature of the target user is extracted from the voice call data, and the first voice data is recorded.

3. The method according to claim 2, characterized in that Before inputting the voiceprint features of the speaker in the voice call data into the voiceprint recognition model of the target user for calculation, the method further includes: Performing noise reduction processing on the voice call data through a universal second voice noise reduction model; The step of inputting the voiceprint features of the speaker in the voice call data into the voiceprint recognition model of the target user for calculation includes: The voiceprint features of the speaker in the voice call data after noise reduction processing are input into the voiceprint recognition model of the target user for calculation.

4. The method according to claim 2, characterized in that: Before inputting the voiceprint features of the speaker in the voice call data into the voiceprint recognition model of the target user for calculation, the method further includes: The voiceprint recognition model is obtained from a preset storage directory, where the voiceprint recognition model is trained based on the voice assistant wake-up word entered by the target user.

5. The method according to claim 1, characterized in that After receiving the trained first speech noise reduction model sent by the server, the method further includes: Acquire second speech data after noise reduction processing by the first speech noise reduction model; determining a signal-to-noise ratio of the second speech data; If the signal-to-noise ratio is less than a second preset threshold, sending the third voice data before the second voice data is subjected to noise reduction processing to the server, and the third voice data is used to update and train the first voice noise reduction model; Receive the updated trained first speech noise reduction model sent by the server.

6. The method according to any one of claims 1 to 5, characterized in that The step of sending the first recorded voice data to the server includes: Sending the first voice data whose recorded voice items reach a first preset threshold to the server; or, The first voice data recorded within a preset time period is sent to a server.

7. A noise reduction processing device for voice calls, characterized in that: include: A recording module, configured to record first voice data of a target user during a voice call; A sending module, configured to send the first recorded voice data to a server, wherein the first voice data is used to train a first voice noise reduction model corresponding to the target user; The receiving module is configured to receive the first speech noise reduction model that has been trained and sent by the server, and the first speech noise reduction model is used for noise reduction processing during the voice call of the target user.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

9. An electronic device comprising a storage medium, a processor, and a computer program stored in the storage medium and executable on the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 6 is implemented.

10. A chip, characterized in that: It comprises one or more interface circuits and one or more processors; the interface circuit is used to receive a signal from a memory of an electronic device and send the signal to the processor, the signal comprising a computer instruction stored in the memory; when the processor executes the computer instruction, the electronic device executes the method described in any one of claims 1 to 6.

Images