Voiceprint recognition method and device with adaptive registration number, electronic device, and storage medium

Abstract

This application provides a voiceprint recognition method, apparatus, electronic device, and computer-readable storage medium with adaptive registration counts. The method includes: acquiring a Gaussian mixture model as a general background model; acquiring registration audio data of a target user and constructing training samples based on the registration audio data; updating the model parameters of the Gaussian mixture model with the training samples according to a maximum a posteriori probability estimation algorithm to obtain an updated Gaussian mixture model; determining, based on a significant difference algorithm, whether there is a significant difference between the updated Gaussian mixture model parameters during this registration and the model parameters of the initially acquired general background model; and determining whether to execute the next round of registration based on the determination result. This solution avoids problems caused by too many or too few registration counts.

Description

Technical Field

[0001] This application relates to the field of speaker recognition technology, and in particular to a voiceprint recognition method and apparatus, electronic device, and computer-readable storage medium with adaptive registration number. Background Technology

[0002] Speaker recognition (SR), also known as voiceprint recognition (VPR), is a biometric identification technology that identifies a speaker based on their personal information in a speech signal. To provide voiceprint recognition services for a target user, the user's audio data needs to be registered on a general background model to obtain a Gaussian mixture model (GMM) corresponding to the target user. This GMM can represent the target user's voiceprint features. In related technologies, registration is performed by setting a preset number of registrations. However, there is no universally effective number of registrations for different registration processes. If the number of registrations is too low, the resulting GMM cannot accurately represent the target user's voiceprint features, potentially leading to poor voiceprint recognition results. If the number of registrations is too high, unnecessary registrations can lead to problems such as wasted power consumption and memory resources. Summary of the Invention

[0003] The purpose of this application is to provide a voiceprint recognition method, device, electronic device, and computer-readable storage medium with adaptive registration number, which can adaptively adjust the number of registrations according to the actual situation of the registration process, complete the registration with the most appropriate number of registrations, and avoid various problems caused by too few or too many registrations.

[0004] On the one hand, this application provides a voiceprint recognition method with adaptive registration number, including:

[0005] Obtain the Gaussian mixture model as a general background model;

[0006] Obtain the registration audio data of the target user, and construct training samples based on the registration audio data;

[0007] The model parameters of the Gaussian mixture model are updated using the training samples according to the maximum a posteriori probability estimation algorithm to obtain the updated Gaussian mixture model.

[0008] Based on the significant difference algorithm, it is determined whether there is a significant difference between the model parameters of the updated Gaussian mixture model during this registration and the model parameters of the initially obtained general background model;

[0009] Based on the assessment results, determine whether to proceed with the next round of the registration process.

[0010] In one embodiment, before obtaining the Gaussian mixture model as a general background model, the method further includes:

[0011] Acquire sample audio data from multiple non-target users and construct multiple training samples;

[0012] According to the expectation-maximization algorithm, the initial Gaussian mixture model is trained with the multiple training samples to obtain a Gaussian mixture model as a general background model.

[0013] In one embodiment, determining whether to execute the next round of the registration process based on the judgment result includes:

[0014] If the judgment result indicates that there is no significant difference, determine to execute the next round of the registration process, and return to the step of obtaining the target user's registration audio data and constructing training samples based on the registration audio data.

[0015] In one embodiment, obtaining the registration audio data of the target user and constructing training samples based on the registration audio data includes:

[0016] Obtain the registration audio data of the current registration process, and concatenate the registration audio data of the current registration process with all the registration audio data of the historical registration processes to obtain the concatenated audio data; wherein, the historical registration processes are the registration processes in the rounds preceding the current registration process;

[0017] Audio features are extracted from the spliced ​​audio data and used as training samples for the current registration process.

[0018] In one embodiment, determining whether to execute the next round of the registration process based on the judgment result includes:

[0019] If the judgment result indicates a significant difference, it is determined that there is no need to execute the next round of the registration process, and the updated Gaussian mixture model in this registration process is used as the target Gaussian mixture model corresponding to the target user.

[0020] In one embodiment, the method further includes:

[0021] Acquire test audio data of the user to be identified, and extract test audio features from the test audio data;

[0022] Based on the target Gaussian mixture model, calculate the first probability value corresponding to the test audio feature;

[0023] Based on the initially acquired general background model, calculate the second probability value corresponding to the test audio features;

[0024] Determine whether the difference between the first probability value and the second probability value is greater than a preset difference threshold. If so, determine that the user to be identified is the target user.

[0025] In one embodiment, the method further includes:

[0026] If the difference between the first probability value and the second probability value is not greater than the difference threshold, it is determined that the user to be identified is not the target user.

[0027] On the other hand, this application provides a voiceprint recognition device with adaptive registration times, comprising:

[0028] The first acquisition module is used to acquire the Gaussian mixture model as a general background model;

[0029] The second acquisition module is used to acquire the registration audio data of the target user and construct training samples based on the registration audio data.

[0030] The update module is used to update the model parameters of the Gaussian mixture model with the training samples according to the maximum a posteriori probability estimation algorithm, so as to obtain the updated Gaussian mixture model.

[0031] The judgment module is used to determine, based on the significant difference algorithm, whether there is a significant difference between the model parameters of the updated Gaussian mixture model during this registration and the model parameters of the initially obtained general background model.

[0032] The determination module is used to determine whether to proceed with the next round of the registration process based on the judgment result.

[0033] Furthermore, this application provides an electronic device, the electronic device comprising:

[0034] processor;

[0035] Memory used to store processor-executable instructions;

[0036] The processor is configured to execute the aforementioned voiceprint recognition method with adaptive registration count.

[0037] Furthermore, this application provides a computer-readable storage medium storing a computer program that can be executed by a processor to perform the aforementioned voiceprint recognition method with adaptive registration times.

[0038] In this application, during the registration process using the target user's audio data on a general background model, a significant difference algorithm can be used to evaluate the updated model parameters and the initially acquired general background model parameters. This allows for real-time feedback on whether a Gaussian mixture model accurately represents the target user's voiceprint characteristics has been obtained. Therefore, the registration process can be stopped after obtaining the Gaussian mixture model corresponding to the target user, thus avoiding problems caused by too few or too many registration attempts. Attached Figure Description

[0039] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly described below.

[0040] Figure 1 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application;

[0041] Figure 2 A flowchart illustrating an embodiment of the voiceprint recognition method with adaptive registration times provided in this application;

[0042] Figure 3 A schematic diagram of the training process for voiceprint features provided in an embodiment of this application;

[0043] Figure 4 A schematic flowchart of a voiceprint recognition method provided in an embodiment of this application;

[0044] Figure 5 A block diagram of a voiceprint recognition device with adaptive registration times provided in an embodiment of this application. Detailed Implementation

[0045] The technical solutions in the embodiments of this application will now be described with reference to the accompanying drawings.

[0046] Similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0047] like Figure 1 As shown, this embodiment provides an electronic device 1, including: at least one processor 11 and a memory 12. Figure 1Taking a processor 11 as an example, the processor 11 and memory 12 are connected via a bus 10. The memory 12 stores instructions that can be executed by the processor 11. The instructions are executed by the processor 11 to enable the electronic device 1 to perform all or part of the process of the method in the following embodiments. In one embodiment, the electronic device 1 may be a mobile phone, tablet computer, host, server, or other device, used to execute the voiceprint recognition method with adaptive registration times. In one embodiment, the electronic device may be equipped with a low-power voice recognition chip, thereby enabling the execution of the voiceprint recognition method in this solution using the low-power voice recognition chip.

[0048] The memory 12 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable red-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0049] This application also provides a computer-readable storage medium storing a computer program that can be executed by a processor 11 to perform the voiceprint recognition method with adaptive registration times provided in this application.

[0050] See Figure 2 This is a flowchart illustrating an embodiment of the voiceprint recognition method with adaptive registration times provided in this application. Figure 2 As shown, the method may include steps 210-250.

[0051] Step 210: Obtain the Gaussian mixture model as a general background model.

[0052] Among them, the Universal Background Model (UBM) is a Gaussian Mixed Model (GMM) used to characterize the universal features of human voice.

[0053] An electronic device implementing the solution of this application can read a Gaussian mixture model, which serves as a general background model, from a specified storage location.

[0054] In one embodiment, the general background model can be trained before obtaining the general background model.

[0055] The electronic device can acquire sample audio data from multiple non-target users to construct multiple training samples. Here, non-target users refer to any speaker without limitation, and the sample audio data refers to the audio data of non-target users collected for training a general background model. The electronic device can extract audio features from the audio data of each non-target user as training samples, thereby obtaining multiple training samples. Each training sample includes the corresponding speaker's voiceprint features.

[0056] After obtaining multiple training samples, the electronic device can train the initial Gaussian mixture model using the Expectation-Maximization (EM) algorithm with multiple training samples, thereby adjusting the model parameters of the Gaussian mixture model to obtain a trained general background model. The initial Gaussian mixture model can be a Gaussian mixture model with randomly generated model parameters; these parameters can include several types of parameters such as weights, mean, and variance.

[0057] Step 220: Obtain the registration audio data of the target user and construct training samples based on the registration audio data.

[0058] Given a general background model, an electronic device can perform a voiceprint registration process for any target user. The target user is the speaker for whom voiceprint recognition services will subsequently be provided.

[0059] In each round of the registration process, one piece of registration audio data can be obtained. Registration audio data refers to the audio data provided by the target user during the registration process. Electronic devices can collect the target user's registration audio data through an audio capture device (such as a microphone), or obtain previously collected registration audio data of the target user from other devices equipped with audio capture devices. After obtaining the registration audio data, the electronic device can use this registration audio data to construct training samples.

[0060] Step 230: Update the model parameters of the Gaussian mixture model with the training samples according to the maximum a posteriori probability estimation algorithm to obtain the updated Gaussian mixture model.

[0061] After obtaining training samples, the electronic device can update the model parameters of the Gaussian mixture model using these samples. It performs maximum a posteriori (MAP) adaptation based on the training samples, thus completing the model parameter update. The initial update is performed on the Gaussian mixture model, which serves as the general background model; in other words, the model parameters of the general background model are updated. Subsequent updates to the model parameters are iteratively performed on the already updated Gaussian mixture model.

[0062] This update process yields an updated Gaussian mixture model. Because the model parameters are updated using training samples corresponding to the target user, the updated Gaussian mixture model includes model parameters that indicate the voiceprint features of the target user.

[0063] Step 240: Based on the significant difference algorithm, determine whether there is a significant difference between the model parameters of the updated Gaussian mixture model during this registration and the model parameters of the initially obtained general background model.

[0064] Step 250: Based on the judgment result, determine whether to proceed with the next round of the registration process.

[0065] In a registration process, after updating the model parameters, the electronic device can compare the model parameters of the updated general background model with the model parameters of the initially acquired general background model based on a significant difference algorithm to determine whether there is a significant difference between the two.

[0066] Furthermore, based on the judgment result, it can be determined whether the target user's registration process is complete, and the registration process can be terminated if the registration is complete, and the next round of registration process can be executed if the registration is incomplete.

[0067] By employing the above measures, during the registration process on a general background model using the target user's audio data, a saliency difference algorithm can be used to evaluate the updated model parameters and the initially acquired general background model parameters. This allows for real-time feedback on whether a Gaussian mixture model accurately characterizing the target user's voiceprint features has been obtained. Therefore, the registration process can be stopped after obtaining the Gaussian mixture model corresponding to the target user, thus avoiding problems caused by too few or too many registration attempts.

[0068] In one embodiment, if the determination result indicates that there is no significant difference between the updated model parameters and the initially acquired general background model parameters, it can be determined that the Gaussian mixture model obtained after this registration update is not a Gaussian mixture model that can accurately represent the voiceprint features of the target user. In this case, the electronic device can determine to execute the next round of the registration process and return to step 220 to obtain new registration audio data of the target user, construct training samples based on the newly acquired registration audio data, and continue to execute the registration process of steps 230 to 250 after constructing the training samples.

[0069] By taking the above measures, the next round of registration can be carried out when the difference between the updated model parameters and the model parameters of the initially obtained general background model is not significant.

[0070] In one embodiment, during step 220, the electronic device can acquire the registration audio data of the current registration process and concatenate it with all the registration audio data of the historical registration processes to obtain concatenated audio data. The historical registration processes refer to the registration processes preceding the current registration process.

[0071] During the initial registration, no historical registration process exists; therefore, the audio data from the first registration is used directly as the concatenated audio data. In subsequent registration processes, the audio data from the current process is concatenated with the concatenated audio data from the previous round, and thus with all historical registration audio data. For example, the audio data from the first registration process is denoted as N1, the second as N2, the third as N3, and so on, with the audio data from the m-th registration process denoted as N... m Accordingly, the concatenated audio data for the first registration process is N1, the concatenated audio data for the second registration process includes N1 and N2, the concatenated audio data for the third registration process includes N1, N2, and N3, and so on, the concatenated audio data for the m-th registration process includes N1, N2, ..., N. m-1 N m .

[0072] Electronic devices can extract audio features from the spliced ​​audio data and use them as training samples for the current registration process. Therefore, in each registration process, the constructed training samples can include audio features from all the registration audio data of the target user that have been acquired.

[0073] In one embodiment, if the determination result indicates a significant difference between the updated model parameters and the initially acquired general background model parameters, the electronic device can determine that there is no need to execute the next round of the registration process, and can use the updated Gaussian mixture model from the current registration process as the target Gaussian mixture model for the target user. At this point, the updated Gaussian mixture model from the current registration process already shows a significant difference from the initially acquired general background model, and can accurately characterize the voiceprint features of the target user.

[0074] After obtaining the target Gaussian mixture model, electronic devices can use the target Gaussian mixture model and the initially acquired general background model to provide voiceprint recognition services for target users.

[0075] See Figure 3 This is a schematic diagram of the training process for voiceprint features provided in an embodiment of this application, as shown below. Figure 3As shown, the electronic device first trains a Gaussian Mixture Model (GMM) as the general background model (UBM model) using sample audio data from non-target users. Further, during the target user registration process, the electronic device trains the GMM based on the target user's registration audio data, updating the model parameters to obtain an updated GMM. The electronic device can then use a significant difference algorithm to determine whether the updated GMM parameters are significantly different from the initially acquired general background model parameters. If the difference is significant, the updated GMM can be used as the target GMM for the target user. If the difference is not significant, a new registration process can begin, acquiring new registration audio data from the target user. This new data (all acquired registration audio data) is then used to continue training and updating the updated GMM until the updated model parameters show a significant difference from the initially acquired general background model parameters.

[0076] In one embodiment, see Figure 4 This is a schematic flowchart of a voiceprint recognition method provided in an embodiment of this application, as shown below. Figure 4 As shown, after obtaining the target Gaussian mixture model, the electronic device can acquire test audio data of the user to be identified. Here, the test audio data is the audio data collected when providing voiceprint recognition services. The electronic device can extract test audio features from the test audio data.

[0077] The electronic device can calculate the first probability value corresponding to the test audio feature based on the target Gaussian mixture model. The electronic device can also calculate the second probability value corresponding to the test audio feature based on the initially acquired general background model.

[0078] The electronic device can calculate the difference between a first probability value and a second probability value, and determine whether this difference is greater than a difference threshold. Here, the difference threshold can be set according to requirements. In one case, if the difference is greater than the difference threshold, it can be determined that the voiceprint features of the target user in the test audio features are sufficiently significant; therefore, the user to be identified can be determined as the target user. In another case, if the difference is not greater than the difference threshold, it can be determined that the voiceprint features of the target user in the test audio features are not significant enough; therefore, the user to be identified can be determined as not the target user.

[0079] Because the proposed solution can determine whether to proceed with a new registration process based on the differences in model parameters in real time during the registration of the target user's audio data, it achieves adaptive registration times. This avoids insufficient registration times, which would prevent the target Gaussian mixture model from accurately representing the target user's voiceprint features, thus improving the voiceprint recognition rate. Furthermore, compared to the registration method with a preset number of registrations in related solutions, it avoids excessive registration times that would lead to excessive consumption of memory and power resources, effectively improving the application performance for low-power speech recognition chips.

[0080] Figure 5 This is a block diagram of a voiceprint recognition device with adaptive registration times according to an embodiment of the present invention, as shown below. Figure 5 As shown, the device may include:

[0081] The first acquisition module 510 is used to acquire the Gaussian mixture model as a general background model;

[0082] The second acquisition module 520 is used to acquire the registration audio data of the target user and construct training samples based on the registration audio data.

[0083] The update module 530 is used to update the model parameters of the Gaussian mixture model with the training samples according to the maximum a posteriori probability estimation algorithm, so as to obtain the updated Gaussian mixture model.

[0084] The judgment module 540 is used to determine, based on the significant difference algorithm, whether there is a significant difference between the model parameters of the updated Gaussian mixture model during this registration and the model parameters of the initially obtained general background model.

[0085] The determination module 550 is used to determine whether to proceed with the next round of the registration process based on the judgment result.

[0086] In one embodiment, the device further includes:

[0087] The third acquisition module 560 is used to acquire sample audio data from multiple non-target users and construct multiple training samples.

[0088] The training module 570 is used to train the initial Gaussian mixture model with the multiple training samples according to the expectation-maximization algorithm to obtain a Gaussian mixture model as a general background model.

[0089] In one embodiment, the determining module 550 is further configured to:

[0090] If the judgment result indicates that there is no significant difference, determine to execute the next round of the registration process, and return to the step of obtaining the target user's registration audio data and constructing training samples based on the registration audio data.

[0091] In one embodiment, the second acquisition module 520 is further configured to:

[0092] Obtain the registration audio data of the current registration process, and concatenate the registration audio data of the current registration process with all the registration audio data of the historical registration processes to obtain the concatenated audio data; wherein, the historical registration processes are the registration processes in the rounds preceding the current registration process;

[0093] Audio features are extracted from the spliced ​​audio data and used as training samples for the current registration process.

[0094] In one embodiment, the determining module 550 is further configured to:

[0095] If the judgment result indicates a significant difference, it is determined that there is no need to execute the next round of the registration process, and the updated Gaussian mixture model in this registration process is used as the target Gaussian mixture model corresponding to the target user.

[0096] In one embodiment, the device further includes:

[0097] The identification module 580 is used to acquire test audio data of the user to be identified and extract test audio features from the test audio data; calculate a first probability value corresponding to the test audio features according to the target Gaussian mixture model; calculate a second probability value corresponding to the test audio features according to the initially acquired general background model; determine whether the difference between the first probability value and the second probability value is greater than a preset difference threshold, and if so, determine that the user to be identified is the target user.

[0098] In one embodiment, the identification module 580 is further configured to:

[0099] If the difference between the first probability value and the second probability value is not greater than the difference threshold, it is determined that the user to be identified is not the target user.

[0100] The specific implementation process of the functions and roles of each module in the above-mentioned device can be found in the implementation process of the corresponding steps in the above-mentioned voiceprint recognition method with adaptive registration times, and will not be repeated here.

[0101] The apparatuses and methods disclosed in the several embodiments provided in this application can also be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of apparatuses, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code, which contains one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0102] In addition, the functional modules in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0103] If a function is implemented as a software module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

Claims

1. A voiceprint recognition method with adaptive registration count, characterized in that, include: Obtain the Gaussian mixture model as a general background model; Obtain the registration audio data of the target user, and construct training samples based on the registration audio data; The model parameters of the Gaussian mixture model are updated using the training samples according to the maximum a posteriori probability estimation algorithm to obtain the updated Gaussian mixture model. Based on the significant difference algorithm, it is determined whether there is a significant difference between the model parameters of the updated Gaussian mixture model during this registration and the model parameters of the initially obtained general background model; Based on the assessment results, determine whether to proceed with the next round of the registration process; If the judgment result indicates a significant difference, it is determined that there is no need to execute the next round of registration process, and the updated Gaussian mixture model in this registration process is used as the target Gaussian mixture model corresponding to the target user. After obtaining the target Gaussian mixture model, voiceprint recognition of the user to be identified is performed, including the following steps: Acquire the test audio data of the user to be identified, and extract test audio features from the test audio data; Based on the target Gaussian mixture model, calculate the first probability value corresponding to the test audio feature; based on the initially acquired general background model, calculate the second probability value corresponding to the test audio feature; determine whether the difference between the first probability value and the second probability value is greater than a preset difference threshold, and if so, determine that the user to be identified is the target user.

2. The method according to claim 1, characterized in that, Before obtaining the Gaussian mixture model as a general background model, the method further includes: Acquire sample audio data from multiple non-target users and construct multiple training samples; According to the expectation-maximization algorithm, the initial Gaussian mixture model is trained with the multiple training samples to obtain a Gaussian mixture model as a general background model.

3. The method according to claim 1, characterized in that, The step of determining whether to proceed with the next round of the registration process based on the judgment result includes: If the judgment result indicates that there is no significant difference, determine to execute the next round of the registration process, and return to the step of obtaining the target user's registration audio data and constructing training samples based on the registration audio data.

4. The method according to claim 3, characterized in that, The process of obtaining the target user's registration audio data and constructing training samples based on the registration audio data includes: Obtain the registration audio data of the current registration process, and concatenate the registration audio data of the current registration process with all the registration audio data of the historical registration processes to obtain the concatenated audio data; wherein, the historical registration processes are the registration processes in the rounds preceding the current registration process; Audio features are extracted from the spliced ​​audio data and used as training samples for the current registration process.

5. The method according to claim 1, characterized in that, The method further includes: If the difference between the first probability value and the second probability value is not greater than the difference threshold, it is determined that the user to be identified is not the target user.

6. A voiceprint recognition device with adaptive registration count, characterized in that, include: The first acquisition module is used to acquire the Gaussian mixture model as a general background model; The second acquisition module is used to acquire the registration audio data of the target user and construct training samples based on the registration audio data. The update module is used to update the model parameters of the Gaussian mixture model with the training samples according to the maximum a posteriori probability estimation algorithm, so as to obtain the updated Gaussian mixture model. The judgment module is used to determine, based on the significant difference algorithm, whether there is a significant difference between the model parameters of the updated Gaussian mixture model during this registration and the model parameters of the initially obtained general background model. The determination module is used to determine whether to proceed with the next round of the registration process based on the judgment result. The determination module is specifically used to: if the judgment result indicates a significant difference, determine that there is no need to execute the next round of registration process, and use the updated Gaussian mixture model in this registration process as the target Gaussian mixture model corresponding to the target user; The recognition module is used to perform voiceprint recognition of the user to be identified after obtaining the target Gaussian mixture model, including: obtaining test audio data of the user to be identified, and extracting test audio features from the test audio data; Based on the target Gaussian mixture model, calculate the first probability value corresponding to the test audio feature; based on the initially acquired general background model, calculate the second probability value corresponding to the test audio feature; determine whether the difference between the first probability value and the second probability value is greater than a preset difference threshold, and if so, determine that the user to be identified is the target user.

7. An electronic device, characterized in that, The electronic device includes: processor; Memory used to store processor-executable instructions; The processor is configured to execute the voiceprint recognition method with adaptive registration number as described in any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The storage medium stores a computer program that can be executed by a processor to perform the voiceprint recognition method with adaptive registration times as described in any one of claims 1-5.

Images