User identity determination method based on voiceprint recognition, electronic equipment and storage medium
Through the user identity determination method based on voiceprint recognition and combined with user behavior characteristics, the problems of low manual review efficiency and poor accuracy in the prior art are solved, and efficient and accurate user identity recognition is achieved.
Patent Information
- Application Number
- CN202411736150.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-29
- Publication Date
- 2025-02-28
AI Technical Summary
In the prior art, the identification of abnormal rights protection agency behaviors by manually reviewing user information is low in efficiency and accuracy, and there is a problem of inconsistent standards.
The user identity determination method based on voiceprint recognition is adopted, and the user identity is determined by obtaining the voice feature vector of the current user and the preset voiceprint model list, and the user's behavior characteristics are combined to calculate the target anomaly and determine the user's identity.
It improves the efficiency and accuracy of user identity identification, reduces the risk of manual auditing, eliminates the possibility of different auditors listening to recordings to give different conclusions, and enhances the reliability of identification results.
Smart Images
Figure CN119541502A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of user identity determination, and in particular to a user identity determination method based on voiceprint recognition, an electronic device and a storage medium. Background Art
[0002] As financial institutions continue to pay attention to consumer rights, a large number of abnormal professional agency rights protection phenomena have emerged in the financial field. Anti-collection alliances provide abnormal agency rights protection services in the name of clearing debts and repairing credit for overdue users. Agents use illegal methods such as impersonating lawyers or legal workers to make complaints and visits to regulatory agencies, and repeatedly complain and report lending financial institutions, which seriously damages the legitimate rights and interests of consumers and affects the normal business order of the financial industry. The industry is increasingly showing a trend of specialization, organization, scale and industrialization. In the existing technology, in order to identify abnormal rights protection agency behavior, it is usually confirmed whether the complainant is the user himself by manually reviewing user information. The efficiency and accuracy of this method are low. Summary of the invention
[0003] In view of the above technical problems, the technical solution adopted by the present invention is: According to a first aspect of the present application, a method for determining a user identity based on voiceprint recognition is provided, the method comprising the following steps: S100, obtaining a recognition speech feature vector QR corresponding to the speech to be recognized issued by the current user and each preset voiceprint model corresponding to the specified user information provided by the current user, so as to obtain a preset voiceprint model list W=(W1, W2, ..., W i , …, W n ), i=1, 2, …, n; where W i is the i-th preset voiceprint model corresponding to the specified user information, and n is the number of preset voiceprint models corresponding to the specified user information; W i It is obtained by training with the voice file of the user corresponding to the specified user information; S200, input QR to each preset voiceprint model in W respectively, so as to obtain a first confidence that the speech to be recognized output by each preset voiceprint model is actually uttered by the user corresponding to the specified user information, so as to obtain a first confidence list λ=(λ1, λ2, …, λ i , …, λ n ); where λ i W i The first confidence of the output; S300, if each first confidence in λ is greater than a preset confidence threshold, it is determined that the current user is the user that actually corresponds to the specified user information; otherwise, the first confidences in λ are sorted to obtain a sorted first confidence list λ'=(λ'1, λ'2, ..., λ' i , …, λ' n ); where λ' i is the i-th first confidence obtained by sorting the first confidences in λ; λ' r The generation time of the corresponding preset voiceprint model is earlier than λ' r+1 The generation time of the corresponding preset voiceprint model; r=1, 2, ..., n-1; S400, determining, based on λ', a target confidence γ that the speech to be recognized is actually uttered by the user corresponding to the designated user information; S500, determining an initial abnormality degree δ corresponding to the current user according to a number of behavioral characteristics of the user actually corresponding to the specified user information; wherein δ is used to characterize the abnormality of the current user; S600, according to γ and δ, determine the target abnormality degree θ=α1×γ+α2×δ of the current user; wherein α1 is a preset first weight, α2 is a preset second weight; α1+α2=1; S700, if θ>S1, it is determined that the current user is not the user actually corresponding to the specified user information; wherein S1 is a preset first abnormality threshold.
[0004] Furthermore, after step S700, the method further includes the following steps: S800, if θ<S2, it is determined that the current user is the user that actually corresponds to the specified user information; wherein S2 is a preset second abnormality threshold; S1>S2; S900: If S1≤θ≤S2, determine manually whether the current user is the user that actually corresponds to the specified user information.
[0005] Further, γ is determined by the following steps: S410, obtaining the generation time of the preset voiceprint model corresponding to each first confidence level in λ' to obtain a generation time list T=(T1, T2, ..., T i ,…,T n ); where T i is λ' i The generation time of the corresponding preset voiceprint model; S410, according to the current time T now and T, determine the weight of each first confidence in λ' to obtain a confidence weight list β = (β1, β2, ..., β i , …, β n); where β i is λ' i The corresponding weight; β i =1-(T now -T i ) / ∑ n j=1 (T now -T i ); S420, according to λ' and β, determine γ=∑ n j=1 (β i ×λ' i ).
[0006] Furthermore, the initial abnormality δ is obtained by a preset expert scoring algorithm.
[0007] Further, W is obtained by the following steps: S110, extracting MFCC features of each voice file corresponding to the specified user information; S120, using the MFCC features of each voice file to train the initial voiceprint model to obtain W.
[0008] Furthermore, the preset voiceprint model is a GMM voiceprint model.
[0009] Furthermore, the preset voiceprint model is stored in a distributed voiceprint model library.
[0010] According to another aspect of the present application, a non-transitory computer-readable storage medium is also provided, in which at least one instruction or at least one program is stored, and the at least one instruction or at least one program is loaded and executed by a processor to implement the above-mentioned user identity determination method based on voiceprint recognition.
[0011] According to another aspect of the present application, an electronic device is provided, including a processor and the above-mentioned non-transitory computer-readable storage medium.
[0012] The present invention has at least the following beneficial effects: The user identity determination method based on voiceprint recognition of the present invention obtains a recognition voice feature vector QR corresponding to the voice to be recognized emitted by the current user and each preset voiceprint model corresponding to the specified user information provided by the current user, so as to obtain a preset voiceprint model list W corresponding to the specified user information; QR is respectively input into each preset voiceprint model in W, so as to obtain a first confidence that the voice to be recognized output by each preset voiceprint model is emitted by the user who actually corresponds to the specified user information, so as to obtain a first confidence list λ corresponding to W; if each first confidence in λ is greater than a preset confidence threshold, it is determined that the current user is the user who actually corresponds to the specified user information; otherwise, the first confidences in λ are sorted to obtain to the first sorted confidence list λ' corresponding to W; according to λ', determine the target confidence γ that the voice to be recognized is emitted by the user who actually corresponds to the specified user information; according to several behavioral characteristics of the user who actually corresponds to the specified user information, determine the initial abnormality δ corresponding to the current user; if θ>S1, it is determined that the current user is not the user who actually corresponds to the specified user information; the method of the present invention replaces the process of manual listening to recordings to distinguish, which can improve the recognition efficiency; the voiceprint comparison results are displayed in the system, and no manual labeling is required, which can reduce operational risks; improve the efficiency and accuracy of recognition; at the same time, the voiceprint comparison model can eliminate different auditors giving different conclusions when listening to the same recording, and reduce the influence of inconsistent standards on the conclusions. BRIEF DESCRIPTION OF THE DRAWINGS
[0013] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0014] Figure 1 A flowchart of a method for determining user identity based on voiceprint recognition provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0015] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.
[0016] It should be noted that, based on the present disclosure, those skilled in the art should understand that an aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number of aspects described herein can be used to implement a device and / or practice a method. In addition, other structures and / or functionalities other than one or more of the aspects described herein can be used to implement this device and / or practice this method.
[0017] The following will refer to Figure 1 The flowchart of the method for determining user identity based on voiceprint recognition is shown, which introduces a method for determining user identity based on voiceprint recognition.
[0018] The method for determining user identity based on voiceprint recognition may include the following steps: S100, obtaining a recognition speech feature vector QR corresponding to the speech to be recognized issued by the current user and each preset voiceprint model corresponding to the specified user information provided by the current user, so as to obtain a preset voiceprint model list W=(W1, W2, ..., W i , …, W n ), i=1, 2, …, n; where W i is the i-th preset voiceprint model corresponding to the specified user information, and n is the number of preset voiceprint models corresponding to the specified user information; W i It is obtained by training with the voice file of the user that actually corresponds to the specified user information.
[0019] In this embodiment, W can be obtained by the following steps: S110, extracting MFCC features of each voice file corresponding to the designated user information.
[0020] S120, using the MFCC features of each voice file to train the initial voiceprint model to obtain W.
[0021] In this embodiment, the voiceprint recognition technology converts sound signals into electrical signals and then uses a computer to extract the characteristics of the electrical signals to identify individuals. It can identify different speakers by analyzing characteristics such as sound frequency, intensity, pitch and speaking speed.
[0022] When applying for a loan, the user's own recording files are collected through manual voice review, online signature, etc., and transmitted to the interface service corresponding to the voiceprint recognition project in the form of Base64 encryption. The byte data of the user's call recording is then decrypted by Base64, and finally stored in the Base distributed database with the user's identity number as the primary key for subsequent user voiceprint modeling. During the manual voice review, the manual reviewer will group the user's basic information, filled-in form information, and the People's Bank of China credit information into multiple groups of logical questions, and verify whether the person currently being reviewed is the user himself through cross-information.
[0023] Through single-track and double-track recording splitting and VAD silence suppression technology, the user's original .wav format recording is separated, denoised, purified and other pre-operations. The processed continuous recording data is then subjected to pre-emphasis, framing, fast Fourier transform, triangular bandpass filtering, standardization and other operations to extract the MFCC features of the user's recording.
[0024] The specific technical processing process is as follows: 1) Single-track and double-track recording split The user recording data is stored in wav format, which is divided into single-track and dual-track, similar to single-channel and dual-channel in acoustics. When the customer service and the user are talking, the recording is a dual-track recording, one track stores the customer service recording, and the other track stores the user's recording. When the user speaks to the phone alone, for example: the recording of the APP withdrawal process, the recording is a single-track recording. The uploaded recording data is identified by single and dual tracks, and the single-track data corresponding to the user is split out for subsequent voiceprint modeling.
[0025] 2) VAD mute suppression technology After splitting the user single-track recording from the customer service and user call recordings, it is found that the recording is intermittent when only the user single-track recording is available. VAD silence suppression technology can separate silence from actual voice, and remove noise as an interference signal from the original data, eliminating the impact of silence and noise on the voiceprint model.
[0026] After preprocessing and feature extraction, the recorded data obtains an N-dimensional digital feature, which can be used as input data for model training. In this embodiment, the voiceprint model adopts the GMM algorithm. GMM has the characteristics of fast training speed, low resource consumption, and good fitting effect, and can well fit the distribution of voiceprint features. Each user's recorded data will train a corresponding GMM voiceprint model, which is stored in the distributed voiceprint model library for user identity determination; through the above method, each user's recorded data will train a corresponding voiceprint model, which is stored in the distributed voiceprint model library.
[0027] S200, input QR to each preset voiceprint model in W respectively, so as to obtain a first confidence that the speech to be recognized output by each preset voiceprint model is actually uttered by the user corresponding to the specified user information, so as to obtain a first confidence list λ=(λ1, λ2, …, λ i , …, λ n ); where λ i W i The first confidence of the output.
[0028] In this embodiment, the designated user information is provided by the current user. It can be understood that the designated user information may be the user information of other people provided by the current user. By inputting QR into each preset voiceprint model in W respectively, the first confidence level of each preset voiceprint model output that the speech to be recognized is actually emitted by the user corresponding to the designated user information can be obtained. The higher the first confidence level, the more similar it is to the speech sample of the corresponding preset voiceprint model.
[0029] S300, if each first confidence in λ is greater than a preset confidence threshold, it is determined that the current user is the user that actually corresponds to the specified user information; otherwise, the first confidences in λ are sorted to obtain a sorted first confidence list λ'=(λ'1, λ'2, ..., λ' i , …, λ' n ); where λ' i is the i-th first confidence obtained by sorting the first confidences in λ; λ' r The generation time of the corresponding preset voiceprint model is earlier than λ' r+1 The generation time of the corresponding preset voiceprint model; r=1, 2, ..., n-1.
[0030] In this embodiment, if each first confidence in λ is greater than a preset confidence threshold, it means that the designated user information provided by the current user is his / her own information, that is, it is determined that the current user is the user that actually corresponds to the designated user information; otherwise, further judgment is required.
[0031] S400, determining, based on λ', a target confidence γ that the speech to be recognized is actually uttered by the user corresponding to the designated user information.
[0032] Further, γ is determined by the following steps: S410, obtaining the generation time of the preset voiceprint model corresponding to each first confidence level in λ' to obtain a generation time list T=(T1, T2, ..., T i ,…,T n ); where T i is λ' i The generation time of the corresponding preset voiceprint model.
[0033] S410, according to the current time T now and T, determine the weight of each first confidence in λ' to obtain a confidence weight list β = (β1, β2, ..., β i , …, β n ); where β i is λ' i The corresponding weight; β i =1-(T now -T i ) / ∑ n j=1 (T now -T i ).
[0034] S420, according to λ' and β, determine γ=∑ n j=1 (β i ×λ' i ).
[0035] In this embodiment, if there is an inconsistency between the user complaint recording and the comparison result of the user's voiceprint model, in order to improve the accuracy of the model, the comparison results are sorted by time and the weighted average is taken. The closer the time, the higher the weight, to obtain the voiceprint model comparison result score.
[0036] S500, determining an initial abnormality degree δ corresponding to the current user according to a number of behavioral characteristics of the user actually corresponding to the specified user information; wherein δ is used to characterize the abnormal situation of the current user.
[0037] In this embodiment, different expert scores are set for user behaviors that have been verified to be effective through experience, such as changing mobile phone numbers during the complaint process, complaining about different users with the same number, and abnormal behaviors that are consistent with the comparison results of the black voiceprint model library. The scores are weighted and summed with the voiceprint model comparison results to output the final score, i.e., the initial abnormality degree. It should be noted that those skilled in the art can use existing expert scoring algorithms to obtain the initial abnormality degree according to actual needs, which will not be elaborated here.
[0038] S600, according to γ and δ, determine the target abnormality degree θ=α1×γ+α2×δ of the current user; wherein α1 is a preset first weight, α2 is a preset second weight; α1+α2=1.
[0039] In this embodiment, α1 and α2 may be empirical values; the target abnormality is obtained by weighted summation, which comprehensively considers the behavioral characteristics of different dimensions, making the judgment more accurate.
[0040] S700, if θ>S1, it is determined that the current user is not the user actually corresponding to the specified user information; wherein S1 is a preset first abnormality threshold.
[0041] In this embodiment, if θ>S1, it means that the current user is relatively abnormal, and it is determined that the current user is not the user that actually corresponds to the specified user information, and the voice to be recognized is automatically entered into the black voiceprint model library.
[0042] Furthermore, after step S700, the method further includes the following steps: S800, if θ<S2, it is determined that the current user is the user that actually corresponds to the specified user information; wherein S2 is a preset second abnormality threshold; S1>S2.
[0043] In this embodiment, if θ<S2, it means that the current user is relatively normal, and it is determined that the current user is the user that actually corresponds to the designated user information.
[0044] S900: If S1≤θ≤S2, determine manually whether the current user is the user that actually corresponds to the specified user information.
[0045] In this embodiment, if S1≤θ≤S2, it means that the abnormality of the current user is at an intermediate level. At this time, it is necessary to make a manual judgment. Specifically, it can be combined with the online search data of the complaining mobile phone number, such as searching the entire network of social platforms to see whether there is information such as professional complaint agents or legal consultation, and manually judge whether the current complainant is an agent.
[0046] The user identity determination method based on voiceprint recognition in this embodiment has the following advantages: 1. The method of this embodiment is based on the voiceprint model comparison result scoring given by voiceprint comparison technology, combined with the expert scoring of user abnormal behavior, thereby improving the accuracy of identity recognition using voiceprint alone and increasing the logical explainability of illegal agency rights protection.
[0047] 2. The method of this embodiment uses a model scoring method to improve the efficiency of manual review and identification of complaint agents. In most cases, conclusions can be drawn through scoring without the need for manual participation, reducing the risk of manual operation, and turning humans from major participants into auxiliary participants.
[0048] The user identity determination method based on voiceprint recognition of this embodiment obtains the recognition voice feature vector QR corresponding to the voice to be recognized emitted by the current user and each preset voiceprint model corresponding to the specified user information provided by the current user, so as to obtain a preset voiceprint model list W corresponding to the specified user information; QR is respectively input into each preset voiceprint model in W to obtain a first confidence that the voice to be recognized output by each preset voiceprint model is emitted by the user who actually corresponds to the specified user information, so as to obtain a first confidence list λ corresponding to W; if each first confidence in λ is greater than a preset confidence threshold, it is determined that the current user is the user who actually corresponds to the specified user information; otherwise, the first confidences in λ are sorted to obtain to the first sorted confidence list λ' corresponding to W; according to λ', determine the target confidence γ that the voice to be recognized is emitted by the user who actually corresponds to the specified user information; according to several behavioral characteristics of the user who actually corresponds to the specified user information, determine the initial abnormality δ corresponding to the current user; if θ>S1, it is determined that the current user is not the user who actually corresponds to the specified user information; the method of the present invention replaces the process of manual listening to recordings to distinguish, which can improve the recognition efficiency; the voiceprint comparison results are displayed in the system, and no manual labeling is required, which can reduce operational risks; improve the efficiency and accuracy of recognition; at the same time, the voiceprint comparison model can eliminate different auditors giving different conclusions when listening to the same recording, and reduce the influence of inconsistent standards on the conclusions.
[0049] In addition, although the steps of the method in the present disclosure are described in a specific order in the drawings, this does not require or imply that the steps must be performed in this specific order, or that all the steps shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps, etc.
[0050] An embodiment of the present invention also provides a non-transitory computer-readable storage medium, which can be set in an electronic device to store at least one instruction or at least one program related to implementing a method in a method embodiment. The at least one instruction or the at least one program is loaded and executed by the processor to implement the method provided in the above embodiment.
[0051] The program product may adopt any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.
[0052] Computer readable signal media may include data signals propagated in baseband or as part of a carrier wave, in which readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Readable signal media may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.
[0053] The program code embodied on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the foregoing.
[0054] Program code for performing the operations of the present application may be written in any combination of one or more programming languages, including object-oriented programming languages such as Java, C++, etc., and conventional procedural programming languages such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., using an Internet service provider to connect through the Internet).
[0055] An embodiment of the present invention further provides an electronic device, comprising a processor and the aforementioned non-transitory computer-readable storage medium.
[0056] The electronic device is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.
[0057] The electronic device is presented in the form of a general-purpose computing device. The components of the electronic device may include, but are not limited to: the at least one processor mentioned above, the at least one memory mentioned above, and a bus connecting different system components (including the memory and the processor).
[0058] The memory stores program codes, which can be executed by the processor, so that the processor executes the steps in various embodiments described in this specification.
[0059] The memory may include readable media in the form of volatile memory, such as random access memory (RAM) and / or cache memory, and may further include read only memory (ROM).
[0060] The memory may also include a program / utility having a set (at least one) of program modules, such program modules including but not limited to: an operating system, one or more application programs, other program modules and program data, each of which or some combination may include an implementation of a network environment.
[0061] The bus may represent one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor, or a local bus using any of a variety of bus architectures.
[0062] The electronic device may also communicate with one or more external devices (e.g., keyboards, pointing devices, Bluetooth devices, etc.), may communicate with one or more devices that enable a user to interact with the electronic device, and / or may communicate with any device (e.g., routers, modems, etc.) that enables the electronic device to communicate with one or more other computing devices. Such communication may be performed through an input / output (I / O) interface. Furthermore, the electronic device may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) through a network adapter. The network adapter communicates with other modules of the electronic device through a bus. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, etc.
[0063] Through the description of the above implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the implementation of the present disclosure 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, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the implementation of the present disclosure.
[0064] An embodiment of the present invention further provides a computer program product, which includes program code. When the program product is run on an electronic device, the program code is used to enable the electronic device to execute the steps of the method according to various exemplary embodiments of the present invention described above in this specification.
[0065] Although some specific embodiments of the present invention have been described in detail by way of example, it should be understood by those skilled in the art that the above examples are only for illustration, not for limiting the scope of the present invention. It should also be understood by those skilled in the art that various modifications may be made to the embodiments without departing from the scope and spirit of the present invention.
Claims
1. A method for determining user identity based on voiceprint recognition, characterized in that: The method comprises the following steps: S100, obtaining a recognition speech feature vector QR corresponding to the speech to be recognized issued by the current user and each preset voiceprint model corresponding to the specified user information provided by the current user, so as to obtain a preset voiceprint model list W=(W1, W2, ..., W i , …, W n ), i=1, 2, …, n; where W i is the i-th preset voiceprint model corresponding to the specified user information, and n is the number of preset voiceprint models corresponding to the specified user information; W i It is obtained by training with the voice file of the user corresponding to the specified user information; S200, input QR to each preset voiceprint model in W respectively, so as to obtain a first confidence that the speech to be recognized output by each preset voiceprint model is actually uttered by the user corresponding to the specified user information, so as to obtain a first confidence list λ=(λ1, λ2, …, λ i , …, λ n ); where λ i W i The first confidence of the output; S300, if each first confidence in λ is greater than a preset confidence threshold, it is determined that the current user is the user that actually corresponds to the specified user information; otherwise, the first confidences in λ are sorted to obtain a sorted first confidence list λ'=(λ'1, λ'2, ..., λ' i , …, λ' n ); where λ' i is the i-th first confidence obtained by sorting the first confidences in λ; λ' r The generation time of the corresponding preset voiceprint model is earlier than λ' r+1 The generation time of the corresponding preset voiceprint model; r=1, 2, ..., n-1; S400, determining, based on λ', a target confidence γ that the speech to be recognized is actually uttered by the user corresponding to the designated user information; S500, determining an initial abnormality degree δ corresponding to the current user according to a number of behavioral characteristics of the user actually corresponding to the specified user information; wherein δ is used to characterize the abnormality of the current user; S600, according to γ and δ, determine the target abnormality degree θ=α1×γ+α2×δ of the current user; wherein α1 is a preset first weight, α2 is a preset second weight; α1+α2=1; S700, if θ>S1, it is determined that the current user is not the user actually corresponding to the specified user information; wherein S1 is a preset first abnormality threshold.
2. The method for determining user identity based on voiceprint recognition according to claim 1, characterized in that: After step S700, the method further includes the following steps: S800, if θ<S2, it is determined that the current user is the user that actually corresponds to the specified user information; wherein S2 is a preset second abnormality threshold; S1>S2; S900: If S1≤θ≤S2, determine manually whether the current user is the user that actually corresponds to the specified user information.
3. The method for determining user identity based on voiceprint recognition according to claim 1, characterized in that: γ is determined by the following steps: S410, obtaining the generation time of the preset voiceprint model corresponding to each first confidence level in λ' to obtain a generation time list T=(T1, T2, ..., T i , …, T n ); where T i is λ' i The generation time of the corresponding preset voiceprint model; S410, according to the current time T now and T, determine the weight of each first confidence in λ' to obtain a confidence weight list β = (β1, β2, ..., β i , …, β n ); where β i is λ' i The corresponding weight; β i =1-(T now -T i ) / ∑ n j=1 (T now -T i ); S420, according to λ' and β, determine γ=∑ n j=1 (β i ×λ' i ).
4. The method for determining user identity based on voiceprint recognition according to claim 1, characterized in that: The initial abnormality δ is obtained through a preset expert scoring algorithm.
5. The method for determining user identity based on voiceprint recognition according to claim 1, characterized in that: W is obtained by the following steps: S110, extracting MFCC features of each voice file corresponding to the specified user information; S120, using the MFCC features of each voice file to train the initial voiceprint model to obtain W.
6. The method for determining user identity based on voiceprint recognition according to claim 5, characterized in that: The default voiceprint model is the GMM voiceprint model.
7. The method for determining user identity based on voiceprint recognition according to claim 1, characterized in that: The preset voiceprint models are stored in the distributed voiceprint model library.
8. A non-transitory computer-readable storage medium, wherein at least one instruction or at least one program is stored in the storage medium, characterized in that: The at least one instruction or the at least one program is loaded and executed by the processor to implement the method for determining user identity based on voiceprint recognition as described in any one of claims 1 to 7.
9. An electronic device, characterized in that: Includes a processor and the non-transitory computer-readable storage medium of claim 8.