Account abnormity determination method and device, storage medium and electronic equipment

Abstract

The embodiment of the invention provides an account abnormity determination method and device, a storage medium and electronic equipment, and the method comprises the steps: obtaining connection data and voice data in the process of voice communication with a voice account of a target account; under the condition that the account label of the target account is the first label, target historical voice data of the target account are obtained, and the target historical voice data are sound data sent by a historical object through the voice account when the voice account is connected in a historical time period; and when it is determined that the voice feature matching degree between the target historical voice data and the voice data is smaller than a preset threshold value and the connection data is abnormal, determining that the target account is an abnormal account. The problem that whether the account is abnormal or not cannot be effectively determined in the prior art is solved, and the effect of effectively determining the account abnormality is achieved.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of computer technology, and more specifically, to a method and device for determining an abnormality in an account, a storage medium, and an electronic device. Background Art

[0002] In financial services, especially in the credit industry, customer identity confirmation is a key link to ensure business compliance and operational security. When a customer fails to repay a loan on time, the financial institution will contact the customer by telephone to remind and negotiate repayment. However, in recent years, with the rise of anti-collection companies, the credit industry faces new challenges, namely that customers may entrust business personnel in these companies to answer collection calls, and these business personnel often use professional strategies and rhetoric to evade repayment responsibilities or demand more favorable repayment terms. In this case, the traditional strong identity determination method usually involves collecting and verifying the customer's specific biometric information, such as voiceprints, fingerprints or facial recognition, to determine whether the account is abnormal by confirming a specific personal identity. This process often requires customers to actively register before using the service. With the development of technologies such as AI and speech synthesis, fraud and impersonation technologies are also iteratively updated. The strong identity determination method lacks a certain generalization ability and there is a problem of being unable to effectively determine whether the account is abnormal. Summary of the invention

[0003] The embodiments of the present application provide a method and device for determining whether an account is abnormal, a storage medium, and an electronic device to solve the problem that the related art cannot effectively determine whether an account is abnormal.

[0004] According to one embodiment of the present application, a method for determining account abnormality is provided, comprising: obtaining connection data and voice data during a voice call with a voice account of a target account, wherein the connection data is data generated during the process of connecting to the voice account, and the voice data is sound data emitted by a first object through the voice account when the voice account is connected; in a case where the account tag of the target account is a first tag, obtaining target historical voice data of the target account, wherein the target historical voice data is sound data emitted by a historical object through the voice account when the voice account is connected in a historical time period; in a case where it is determined that the voice feature matching degree between the target historical voice data and the voice data is less than a preset threshold, and there is an abnormality in the connection data, determining that the target account is an abnormal account.

[0005] In an exemplary embodiment, during a voice call with a voice account of a target account, connection data and voice data are obtained, including: during the voice call with the voice account, metadata and voice quality of the voice call are detected, wherein the metadata includes a connection method used during the voice call; when it is determined from the metadata that the connection method is a voice transfer method, the metadata is determined as the connection data, wherein the voice transfer method includes a method of transferring the voice call to another object for answering; during the voice call with the voice account, voice having a voice quality greater than a preset quality threshold is intercepted from the generated voice to obtain the voice data.

[0006] In an exemplary embodiment, when the account tag of the target account is a first tag, before obtaining the historical voice data of the target account, the method further includes: searching for the account tag of the target account from a tag database according to the account information of the target account; when the account tag of the target account is a first tag, determining that the target account is an account in a normal state; when the account tag of the target account is a second tag, determining that the target account is an account in an abnormal state; wherein the target account is an account that conducts financial transactions with other accounts.

[0007] In an exemplary embodiment, when the account tag of the target account is the first tag, obtaining the historical voice data of the target account includes: obtaining N historical voice data of the target account from a voiceprint database according to the account information of the target account, wherein N is a natural number greater than or equal to 1; selecting historical voice data having a voice quality greater than a preset quality threshold from the N historical voice data to obtain the target historical voice data.

[0008] In an exemplary embodiment, when it is determined that the voice feature matching degree between the above-mentioned target historical voice data and the above-mentioned voice data is less than a preset threshold, and the above-mentioned connection data has an abnormality, the above-mentioned target account is determined to be an abnormal account, including: extracting the voice features of the above-mentioned target historical voice data and the voice features of the above-mentioned voice data respectively to obtain the first target voice feature and the second target voice feature; calculating the matching degree between the above-mentioned first target voice feature and the above-mentioned second target voice feature to obtain the above-mentioned voice feature matching degree; when abnormal data is parsed from the above-mentioned connection data, and the above-mentioned voice feature matching degree is less than a preset threshold, the above-mentioned target account is determined to be the above-mentioned abnormal account, wherein the connection method used in the above-mentioned voice call process included in the above-mentioned abnormal data is a voice transfer method.

[0009] In an exemplary embodiment, the speech features of the target historical speech data and the speech features of the speech data are extracted respectively to obtain a first target speech feature and a second target speech feature, including: performing semantic filtering operations on the target historical speech data and the speech data respectively to remove background speech data in the target historical data and the speech data respectively to obtain first speech data and second speech data; performing standardization operations on the first speech data and the second speech data to unify the audio references of the first speech data and the second speech data respectively to obtain third speech data and fourth speech data; extracting speech features in the third speech data and the fourth speech data in segments respectively to obtain a plurality of first speech features corresponding to the third speech data and a plurality of second speech features corresponding to the fourth speech data; fusing a plurality of the first speech features to obtain the first target speech feature; fusing a plurality of the second speech features to obtain the second target speech feature.

[0010] In an exemplary embodiment, after determining that the voice feature matching degree between the target historical voice data and the voice data is less than a preset threshold and that the connection data is abnormal, and determining that the target account is an abnormal account, the method further includes: setting the account tag of the target account to a second tag in the tag database according to the account information of the target account to mark the target account as an abnormal account; and triggering a processing operation on the abnormal account based on the second tag.

[0011] According to one embodiment of the present application, a device for determining account abnormality includes: a first acquisition module, used to acquire connection data and voice data during a voice call with a voice account of a target account, wherein the connection data is data generated during the process of connecting to the voice account, and the voice data is sound data emitted by a first object through the voice account when the voice account is connected; a second acquisition module, used to acquire target historical voice data of the target account when the account tag of the target account is a first tag, wherein the target historical voice data is sound data emitted by a historical object through the voice account when the voice account is connected in a historical time period; a first determination module, used to determine that the target account is an abnormal account when it is determined that the voice feature matching degree between the target historical voice data and the voice data is less than a preset threshold and there is an abnormality in the connection data.

[0012] In an exemplary embodiment, the first acquisition module includes: a first detection submodule, used to detect the metadata and voice quality of the voice call during the voice call with the voice account, wherein the metadata includes the connection method used during the voice call; a first determination submodule, used to determine the metadata as the connection data when it is determined from the metadata that the connection method is a voice transfer method, wherein the voice transfer method includes a method of transferring the voice call to other objects for answering; a first interception submodule, used to intercept the voice with a quality greater than a preset quality threshold from the generated voice during the voice call with the voice account to obtain the voice data.

[0013] In an exemplary embodiment, the above-mentioned device also includes: a first search module, which is used to search for the account tag of the above-mentioned target account from the tag database according to the account information of the above-mentioned target account before obtaining the historical voice data of the above-mentioned target account when the account tag of the above-mentioned target account is the first tag; a second determination module, which is used to determine that the above-mentioned target account is an account in a normal state when the account tag of the above-mentioned target account is the first tag; a third determination module, which is used to determine that the above-mentioned target account is an account in an abnormal state when the account tag of the above-mentioned target account is the second tag; wherein the above-mentioned target account is an account that conducts financial transactions with other accounts.

[0014] In an exemplary embodiment, the above-mentioned first acquisition module includes: a first acquisition sub-module, used to obtain N historical voice data of the above-mentioned target account from the voiceprint database according to the account information of the above-mentioned target account, wherein the above-mentioned N is a natural number greater than or equal to 1; a first selection sub-module, used to select historical voice data with voice quality greater than a preset quality threshold from the above-mentioned N historical voice data to obtain the above-mentioned target historical voice data.

[0015] In an exemplary embodiment, the first determination module includes: a first extraction submodule, used to extract the voice features of the target historical voice data and the voice features of the voice data, respectively, to obtain the first target voice features and the second target voice features; a first calculation submodule, used to calculate the matching degree between the first target voice features and the second target voice features, to obtain the voice feature matching degree; a second determination submodule, used to determine that the target account is the abnormal account when abnormal data is parsed from the connection data and the voice feature matching degree is less than a preset threshold, wherein the connection method used in the voice call process included in the abnormal data is a voice transfer method.

[0016] In an exemplary embodiment, the first extraction submodule includes: a first removal unit, which is used to perform semantic filtering operations on the target historical voice data and the voice data, respectively, to remove the background voice data in the target historical data and the voice data, respectively, to obtain first voice data and second voice data; a first unification unit, which is used to perform standardization operations on the first voice data and the second voice data, respectively to unify the audio references of the first voice data and the second voice data, to obtain third voice data and fourth voice data; a first extraction unit, which is used to extract voice features in the third voice data and the fourth voice data, respectively, in segments, to obtain multiple first voice features corresponding to the third voice data, and multiple second voice features corresponding to the fourth voice data; a first fusion unit, which is used to fuse multiple first voice features to obtain the first target voice feature; and a second fusion unit, which is used to fuse multiple second voice features to obtain the second target voice feature.

[0017] In an exemplary embodiment, the above-mentioned device also includes: a first marking module, which is used to determine that the voice feature matching degree between the above-mentioned target historical voice data and the above-mentioned voice data is less than a preset threshold, and the above-mentioned connection data has an abnormality, and after determining that the above-mentioned target account is an abnormal account, set the account label of the above-mentioned target account to a second label in the label database according to the account information of the above-mentioned target account to mark the above-mentioned target account as an abnormal account; based on the above-mentioned second label, trigger the processing operation of the above-mentioned abnormal account.

[0018] According to another embodiment of the present application, a computer program product is provided, including a computer program, wherein the computer program is configured for a processor to execute the steps of any of the above method embodiments.

[0019] According to another embodiment of the present application, a computer-readable storage medium is provided, in which a computer program is stored, wherein the computer program is configured to enable a processor to execute the steps in any of the above method embodiments.

[0020] According to another embodiment of the present application, an electronic device is provided, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor is configured to execute the computer program to perform the steps in any one of the method embodiments.

[0021] Through this application, during a voice call with the voice account of the target account, the connection data and voice data sent and generated by the first object through the voice account are obtained; through the target historical voice data of the target account, when it is determined that the voice feature matching degree between the target historical voice data and the voice data is less than the preset threshold, and there is an abnormality in the connection data, the target account is determined to be an abnormal account. Because this application integrates the analysis of connection data and voice data, dynamically sets the matching degree threshold, compares historical voice data, and uses account tag information, it can not only effectively detect account anomalies and avoid the limitations of single data source judgment, but also adapt to different call conditions and speaker characteristics, improve detection accuracy and generalization ability, and therefore solve the problem of inability to effectively determine whether an account is abnormal in related technologies, and achieves the effect of effectively determining account anomalies. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a hardware structure block diagram of a mobile terminal of a method for determining an account abnormality according to an embodiment of the present application;

[0023] Figure 2 is a flow chart of a method for determining an abnormality of an account according to an embodiment of the present application;

[0024] Figure 3 It is a schematic diagram of a method for determining an abnormal account in a specific embodiment of the present application;

[0025] Figure 4 It is a flowchart of a method for determining an abnormal account in a specific embodiment of the present application;

[0026] Figure 5 It is a structural block diagram of a device for determining an account anomaly according to an embodiment of the present application. DETAILED DESCRIPTION

[0027] The embodiments of the present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0028] It should be noted that the terms "first", "second", etc. in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0029] The method embodiments provided in the embodiments of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Taking running on a mobile terminal as an example, Figure 1 1 is a hardware structure block diagram of a mobile terminal of a method for determining an abnormal account in an embodiment of the present application. Figure 1 As shown, the mobile terminal may include one or more ( Figure 1Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data, wherein the mobile terminal may also include a transmission device 106 and an input / output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the mobile terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0030] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as a computer program corresponding to a method for determining an abnormal account in an embodiment of the present application. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, to implement the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the mobile terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0031] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the mobile terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0032] In this embodiment, a method for determining an account abnormality is provided. Figure 2 is a flowchart of a method for determining an abnormal account according to an embodiment of the present application, such as Figure 2 As shown, the process includes the following steps:

[0033] Step S202, in the process of making a voice call with the voice account of the target account, obtaining connection data and voice data, wherein the connection data is data generated in the process of connecting to the voice account, and the voice data is sound data emitted by the first object through the voice account when the voice account is connected;

[0034] Optionally, the party having the voice call with the voice account of the target account includes but is not limited to manual customer service and intelligent customer service of the financial institution.

[0035] Optionally, the connection data is used to detect whether a transfer operation has occurred in the voice call.

[0036] Optionally, the target account is an account of an opening bank. The voice account is a phone number bound to the target account. The first object may be an object bound to the target account or an object not bound to the target account.

[0037] Optionally, other voice data of other objects are collected to generate an anti-collection database, where the other objects are anti-collection personnel; when the voice feature matching degree between the voice data and other voice data is less than a preset threshold and there is an abnormality in the connection data, the target account is determined to be an abnormal account.

[0038] Step S204, when the account tag of the target account is the first tag, obtaining target historical voice data of the target account, wherein the target historical voice data is the sound data emitted by the historical object through the voice account when the voice account is connected in the historical time period;

[0039] Optionally, the first tag is used to indicate that the target account is a non-abnormal account.

[0040] Step S206, when it is determined that the voice feature matching degree between the target historical voice data and the voice data is less than a preset threshold and the connection data is abnormal, determining that the target account is an abnormal account;

[0041] Optionally, the abnormal account indicates that multiple objects have voice calls with the staff through the voice account bound to the target account.

[0042] In this embodiment, the execution subject of the above steps can be a terminal, a server, a specific processor set in the terminal or server, or a processor or processing device set relatively independently from the terminal or server, but is not limited thereto.

[0043] Alternatively, if Figure 3 As shown, Figure 3It is a schematic diagram of a method for determining an abnormal account in a specific embodiment of the present application. The communication platform module provides basic call services and collects voice. The user center module obtains the connection data and voice data from the communication platform module through the business application module based on the voice-related interface, and calls the abnormal voice recognition and voice basic services in the voice service module to determine the voice feature matching degree between the target historical voice data and the voice data and detect the connection data, thereby determining the abnormal situation of the target account and marking the target account label. The business application module displays the determined target account label from the user center module to the customer service for viewing based on the voice-related interface.

[0044] Through the above steps, in the process of a voice call with the voice account of the target account, the connection data and voice data sent and generated by the first object through the voice account are obtained; through the target historical voice data of the target account, when it is determined that the voice feature matching degree between the target historical voice data and the voice data is less than the preset threshold, and there is an abnormality in the connection data, the target account is determined to be an abnormal account. Since the present application integrates the analysis of the connection data and the voice data, dynamically sets the matching degree threshold, compares the historical voice data, and utilizes the account tag information, it can not only effectively detect account anomalies and avoid the limitations of single data source judgment, but also adapt to different call conditions and speaker characteristics, improve the accuracy and generalization ability of detection, and therefore solves the problem in the related technology that it is impossible to effectively determine whether the account is abnormal, and achieves the effect of effectively determining account anomalies.

[0045] In an exemplary embodiment, during a voice call with a voice account of a target account, connection data and voice data are obtained, including: during the voice call with the voice account, metadata and voice quality of the voice call are detected, wherein the metadata includes a connection method used during the voice call; when it is determined from the metadata that the connection method is a voice transfer method, the metadata is determined as the connection data, wherein the voice transfer method includes a method of transferring the voice call to another object for answering; during the voice call with the voice account, voice having a voice quality greater than a preset quality threshold is intercepted from the generated voice to obtain the voice data.

[0046] Optionally, metadata includes, but is not limited to, information such as the time, duration, connection method, and call device type of the voice call. The connection method is used to determine whether the voice call is transferred to another party other than the original caller, which may include the situation where the voice account is transferred to an employee of an anti-collection company. Indicators for evaluating voice quality include, but are not limited to, signal strength, clarity, volume level, and other indicators. Only when the voice quality is higher than a preset quality threshold will the voice be intercepted and saved to ensure that the voice data has sufficient clarity and integrity. The preset quality threshold can be dynamically adjusted based on the application scenario or voice scenario. For example, if the environment is noisy and the voice is unclear, the preset quality threshold can be lowered.

[0047] Optionally, when the collection department of a consumer finance company makes a voice call with the voice account of the target account, the call metadata, including call time, duration, connection method, call device type, etc., begins to be automatically collected the moment the voice call is established. For example, the call started at 14:30 on March 15, 2023 and lasted for 10 minutes. The connection method was direct connection and the call device was a smartphone. During the call, the connection method is detected in real time to confirm whether it is a voice transfer method. If the voice call is transferred to another party other than the original caller, such as being transferred to an employee of the anti-collection company, the system will mark this transfer behavior and identify the relevant metadata (including transfer time, device information before and after the transfer, etc.) as connection data. Continuously evaluate voice quality during the voice call, and check indicators such as signal strength, clarity, and volume level. For example, the preset voice quality threshold is that the signal-to-noise ratio is greater than 70db. After the call starts, the detected signal-to-noise ratio is 75db. 5 seconds after the call starts, the signal-to-noise ratio is stable above 75db. From this time point on, the next 10 seconds of voice are intercepted, that is, from the 5th second to the 15th second. After intercepting 10 seconds of high-quality voice clips, it is determined as voice data.

[0048] This embodiment can quickly capture abnormal situations where an account may be answered by a non-account holder by detecting metadata and voice quality of voice calls, thereby achieving the purpose of detection accuracy and reliability. At the same time, data is intercepted from voices with voice quality higher than a preset quality threshold, avoiding the collection of invalid data when the environment is noisy or the device sound quality is poor at the beginning of the call, ensuring that the voice data is of high quality, and achieving the purpose of reducing recognition errors caused by environmental noise or equipment problems.

[0049] In an exemplary embodiment, when the account tag of the target account is a first tag, before obtaining the historical voice data of the target account, the method further includes: searching for the account tag of the target account from a tag database according to the account information of the target account; when the account tag of the target account is a first tag, determining that the target account is an account in a normal state; when the account tag of the target account is a second tag, determining that the target account is an account in an abnormal state; wherein the target account is an account that conducts financial transactions with other accounts.

[0050] Optionally, the account information of the target account includes but is not limited to the account ID and the voice account number. If the target account is determined to be an account in a normal state, the historical voice data of the target account will be obtained and saved in the next voice call for subsequent voiceprint recognition and matching analysis. For example, all call records of the target account in the past three months are obtained, and the voice segments with high clarity and strong signal strength in each call are extracted and saved to obtain the historical voice data. On the contrary, if the account label of the target account is "second label", the target account will be immediately determined to be an account in an abnormal state.

[0051] This embodiment can quickly determine whether the target account is in a normal state by pre-querying the account label of the target account, thereby deciding whether it is necessary to obtain historical voice data for voiceprint comparison. For normal accounts marked as the "first label", the historical voice data retrieval process can be automatically entered, thereby achieving the purpose of quickly completing the feature matching verification of voice features and improving the efficiency of determining whether the target account is an abnormal account.

[0052] In an exemplary embodiment, when the account tag of the target account is the first tag, obtaining the target historical voice data of the target account includes: obtaining N historical voice data of the target account from a voiceprint database according to the account information of the target account, wherein N is a natural number greater than or equal to 1; selecting historical voice data whose voice quality is greater than a preset quality threshold from the N historical voice data to obtain the target historical voice data.

[0053] Optionally, a classification operation is performed on N historical voice data to obtain M historical classified voice data, where M is a natural number greater than 1; historical classified voice data whose voice quality is greater than a preset quality threshold is selected from the M historical classified voice data to obtain the above-mentioned target historical voice data.

[0054] Optionally, before the collection department of a consumer finance company makes a voice call with a target account, if the account tag of the target account is marked as the "first tag" (i.e., the current status of the account is normal), the collection system will perform the following steps to obtain and filter the historical voice data of the target account for subsequent feature matching: First, the collection system first retrieves all the historical voice data of the target account in the past period from the voiceprint database according to the account information of the target account (such as the account ID), for example, the system sets the retrieval time range to the past 6 months, and the goal of each retrieval is to obtain N historical voice data related to the account, where N is a natural number greater than or equal to 1, such as N = 10. Then, the collection system performs quality detection on the retrieved N historical voice data through the built-in voice quality evaluation module, evaluates key indicators such as signal strength, clarity, and volume level, and filters out all voice data with a signal-to-noise ratio greater than 60db from the N historical voice data. These data are regarded as target historical voice data, that is, high-quality voice segments for subsequent voiceprint comparison. For example, in the case of N=10, after screening, the signal-to-noise ratios of 7 historical speech data satisfy the condition, and these 7 speech data will be determined as high-quality target historical speech data.

[0055] This embodiment ensures that the data used for feature matching is clear, complete, and free of significant noise by only selecting historical speech data with speech quality greater than a preset quality threshold, thereby reducing recognition errors caused by poor speech quality. At the same time, it avoids indiscriminate processing of all historical data, thereby reducing unnecessary consumption of computing resources and improving data processing efficiency. In particular, in the scenario of processing a large amount of historical speech data, it can significantly save computing time and storage space.

[0056] In an exemplary embodiment, when it is determined that the voice feature matching degree between the above-mentioned target historical voice data and the above-mentioned voice data is less than a preset threshold, and the above-mentioned connection data has an abnormality, the above-mentioned target account is determined to be an abnormal account, including: extracting the voice features of the above-mentioned target historical voice data and the voice features of the above-mentioned voice data respectively to obtain the first target voice feature and the second target voice feature; calculating the matching degree between the above-mentioned first target voice feature and the above-mentioned second target voice feature to obtain the above-mentioned voice feature matching degree; when abnormal data is parsed from the above-mentioned connection data, and the above-mentioned voice feature matching degree is less than a preset threshold, the above-mentioned target account is determined to be the above-mentioned abnormal account, wherein the connection method used in the above-mentioned voice call process included in the above-mentioned abnormal data is a voice transfer method.

[0057] Optionally, when determining whether the target account is an abnormal account, the collection department of the consumer finance company will analyze the matching degree of the voice features and the abnormal situation of the connection data. The specific process is as follows: First, the voice features of the target historical voice data and the voice features of the voice data in the current voice call are extracted respectively. The voice features include but are not limited to voiceprint features, speech speed features, pitch features, and the frequency of use of specific words. The voice data is analyzed by an artificial intelligence algorithm, such as a deep neural network or a recurrent neural network, to extract the first target voice feature and the second target voice feature. The matching degree between the first target voice feature and the second target voice feature is calculated using a feature matching algorithm to obtain the voice feature matching degree. The voice feature matching degree can be the Euclidean distance between the two feature vectors, the cosine similarity or other applicable metrics. The system will set a preset threshold, such as 0.7 or 80%, to distinguish the matching degree of the voiceprint. If the voice feature matching degree is less than the preset threshold, the system preliminarily determines that the voiceprint of the current call is inconsistent with the voiceprint features in the historical voice data.

[0058] This embodiment effectively identifies whether there is an abnormal situation in which one target account has multiple users by comparing the voice feature matching degree between the target historical voice data and the current call voice data, combined with metadata anomaly detection of the connection data.

[0059] In an exemplary embodiment, the speech features of the target historical speech data and the speech features of the speech data are extracted respectively to obtain a first target speech feature and a second target speech feature, including: performing semantic filtering operations on the target historical speech data and the speech data respectively to remove background speech data in the target historical data and the speech data respectively to obtain first speech data and second speech data; performing standardization operations on the first speech data and the second speech data to unify the audio references of the first speech data and the second speech data respectively to obtain third speech data and fourth speech data; extracting speech features in the third speech data and the fourth speech data in segments respectively to obtain a plurality of first speech features corresponding to the third speech data and a plurality of second speech features corresponding to the fourth speech data; fusing a plurality of the first speech features to obtain the first target speech feature; fusing a plurality of the second speech features to obtain the second target speech feature.

[0060] Optionally, in order to improve the accuracy of voiceprint recognition, the collection system of the consumer finance company takes the following steps to extract the voice features of the target historical voice data and the real-time call voice data respectively, and obtain the first target voice feature and the second target voice feature: first, perform semantic filtering operation on the target historical voice data, and use audio processing algorithms (such as noise threshold, spectrum analysis) to identify and remove background noise therein, such as environmental sound, other people's voices, etc., to ensure that the extracted voice features only come from the first object. Similarly, semantic filtering is also performed on the voice data in the real-time call to remove the background noise, and the first voice data (processed target historical voice data) and the second voice data (processed real-time call voice data) are obtained. For example, there is a call record in the target historical voice data that contains noisy background sound of a cafe. The system uses a spectrum analysis algorithm to identify and filter out the background noise, leaving only the clear human voice part. Then, a standardization operation is performed on the first voice data and the second voice data, and the audio data is preprocessed, for example, the volume and frequency response are adjusted to unify the audio benchmark to ensure that the comparison of voiceprint features between different call records is performed under the same conditions. After processing, the third voice data (the target historical voice data after standardization) and the fourth voice data (the real-time call voice data after standardization) are obtained. For example, the volume in the target historical voice data may be lower than the volume of the real-time call. Through the standardization operation, the volumes of the two are adjusted to the same level to avoid matching errors caused by volume differences. Then, the third voice data and the fourth voice data are segmented, and each segment of the voice data is split into multiple segments of fixed length, such as one segment every 10 seconds, and then the voice features are extracted from each segment using a deep learning model to obtain multiple first voice features corresponding to the third voice data and multiple second voice features corresponding to the fourth voice data. For example, 5 10-second segments are extracted from the target historical voice data, and each segment is processed by the voiceprint recognition model to obtain the corresponding first voice feature. Finally, the multiple first voice features are fused, and the first target voice feature is obtained by averaging, weighted averaging or other feature fusion algorithms; similarly, the multiple second voice features are fused to obtain the second target voice feature, so as to ensure that the features extracted from different voice segments can comprehensively reflect the voiceprint features of the first object, thereby improving the stability and accuracy of recognition.

[0061] In an exemplary embodiment, after determining that the voice feature matching degree between the target historical voice data and the voice data is less than a preset threshold and that the connection data is abnormal, and determining that the target account is an abnormal account, the method further includes: setting the account tag of the target account to a second tag in the tag database according to the account information of the target account to mark the target account as an abnormal account; and triggering a processing operation on the abnormal account based on the second tag.

[0062] Optionally, once the target account is marked as an abnormal account, the risk can be managed through abnormal account processing operations. For example, for manual seats, the system will display the abnormal account mark in real time, and the seat staff can adjust the call strategy accordingly, increase vigilance and perform additional verification steps. For intelligent customer service systems, the system will automatically switch to the preset abnormal account processing process, which may include multiple verifications of voice recognition, requiring customers to provide static identity proof (such as passwords, ID photos, etc.) or transferring the call to a manual seat for manual review to ensure the security of transactions or communications. Optionally, the detailed information of the voice call, including voice characteristics, connection method and abnormal account mark, can be recorded in the history of the target account. This information can be used for subsequent risk analysis and early warning, helping financial institutions to establish a more comprehensive risk management model and prevent future fraud.

[0063] The present invention will be described below in conjunction with specific embodiments:

[0064] This embodiment is explained by using the confirmation of the target account in a manual customer service or intelligent customer service scenario. Figure 4 This is a flowchart of a method for determining an account abnormality in a specific embodiment of the present application, comprising the following steps:

[0065] S402, starting, the first subject calls in or the artificial or intelligent customer service of the financial institution calls out to enter the customer service workbench page, and the first subject conducts a voice call with the artificial or intelligent customer service through the voice account bound to the target account;

[0066] S404, the user center module feeds back account information and account tags to the business application module, including but not limited to: account ID, voice account number, and first tag;

[0067] S406, the customer service workstation displays to the manual customer service that the target account is a normal account;

[0068] S408, after the voice call is connected, the first party has a voice call with the manual or intelligent customer service through the voice account;

[0069] S410, the communication platform module calls the call extraction interface. If the voice call is transferred to another party other than the original caller, such as being transferred to an employee of an anti-collection company, the transfer behavior is marked, and the relevant metadata (including the transfer time, device information before and after the transfer, etc.) is determined as the connection data. The voice quality is continuously evaluated during the voice call, and indicators such as signal strength, clarity, and volume level are checked. For example, the preset voice quality threshold is a signal-to-noise ratio greater than 70db. After the voice call starts, the detected signal-to-noise ratio is 75db. 5 seconds after the call starts, the signal-to-noise ratio stabilizes above 75db. From this point in time, the next 10 seconds of voice are intercepted, that is, from the 5th second to the 15th second. After intercepting 10 seconds of high-quality voice clips, it is determined as voice data. The connection data and voice data are saved in the recording file under the path specified in the storage module;

[0070] S412, the business application module obtains the connection data and voice data from the storage module based on the path and file name of the recording file returned by the communication platform module;

[0071] S414, the service application module sends the voice account of the target account and the connection data and voice data to the user center module;

[0072] S416, the user center module obtains a first object identifier from a database according to the voice account;

[0073] S418, the user center module obtains the voiceprint registration information table of the target account from the voiceprint database according to the voice account and the first object identifier;

[0074] S420, based on the voiceprint registration information table, determine whether there is historical voice data of the target account in the voiceprint database, if yes, go to S422, otherwise go to S434 to register the voice call;

[0075] S422, obtain a voiceprint registration data list from the voiceprint database, which includes N historical voice data of the target account. First, according to the account information of the target account (such as account ID), retrieve all historical voice data of the target account in the past 6 months from the voiceprint database. Then, through the built-in voice quality assessment module, perform quality detection on the 10 retrieved historical voice data, evaluate key indicators such as signal strength, clarity, and volume level, and screen out all voice data with a signal-to-noise ratio greater than 60db from the 10 historical voice data. These data are regarded as target historical voice data, that is, high-quality voice segments for subsequent voiceprint comparison. For example, in the case of N=10, after screening, the signal-to-noise ratio of 7 historical voice data meets the conditions, and these 7 voice data will be determined as high-quality target historical voice data;

[0076] S424, determine the voice feature matching degree between the target historical voice data and the voice data and whether there is an abnormality in the connection data. First, perform semantic filtering on the target historical voice data, use audio processing algorithms (such as noise threshold, spectrum analysis) to identify and remove background noise, such as environmental sound, other people's voices, etc., to ensure that the extracted voice features only come from the first object. Similarly, semantic filtering is also performed on the voice data in the real-time call to remove background noise, and the first voice data (processed target historical voice data) and the second voice data (processed real-time call voice data) are obtained. For example, there is a call record in the target historical voice data that contains noisy coffee shop background sound. Use the spectrum analysis algorithm to identify and filter out the background noise, and only retain the clear human voice part. Then, perform standardization operations on the first voice data and the second voice data, and pre-process the audio data, for example, adjust the volume and frequency response to unify the audio benchmark to ensure that the voiceprint feature comparison between different call records is carried out under the same conditions. After processing, the third voice data (the target historical voice data after standardization) and the fourth voice data (the real-time call voice data after standardization) are obtained. For example, the volume in the target historical voice data may be lower than the volume of the real-time call. Through the standardization operation, the volumes of the two are adjusted to the same level to avoid matching errors caused by volume differences. Then, the third voice data and the fourth voice data are segmented, and each voice data is split into multiple fixed-length segments, such as one segment every 10 seconds. Then, the voice features are extracted from each segment using a deep learning model to obtain multiple first voice features corresponding to the third voice data and multiple second voice features corresponding to the fourth voice data. For example, 5 10-second segments are extracted from the target historical voice data, and each segment is processed by the voiceprint recognition model to obtain the corresponding first voice feature. Finally, the multiple first voice features are fused, and the first target voice feature is obtained by averaging, weighted averaging or other feature fusion algorithms; similarly, the multiple second voice features are fused to obtain the second target voice feature, and the similarity between the first target voice feature and the second target voice feature is calculated using a feature matching algorithm to obtain the voice feature matching degree, which can be the Euclidean distance between the two feature vectors, cosine similarity or other applicable metrics. If the voice feature matching degree is less than the preset threshold, it is preliminarily determined that the voiceprint of the current call is inconsistent with the voiceprint features in the historical voice data;

[0077] S426, saving the calculated voice feature matching degree;

[0078] S428: If the voice feature matching degree between the target historical voice data and the voice data is less than the preset threshold and the call data is abnormal, the target account is determined to be an abnormal account and there is a situation where one account has multiple users. Feedback the abnormal situation of the target account to the business application module;

[0079] S430, the customer service workstation displays the abnormal situation of the target account to the human customer service, and the human customer service can adjust the call strategy accordingly, increase vigilance and perform additional verification steps;

[0080] S432, saving the abnormal situation of the target account in the record of the voice call, where the fields in the record include but are not limited to: identification information of the target account, voice account number, voice call record number, voice feature matching degree, abnormal label of the target account, and effective time of the abnormal label;

[0081] S434, the user center module determines that there is no historical voice data of the target account in the voiceprint database based on the voiceprint registration information table, and feeds back a voiceprint unregistered prompt to the manual customer service through the business application module, and then asynchronously processes the registration process;

[0082] S436, the user center module calls the voiceprint service interface to trigger the voiceprint registration of the voice service module;

[0083] S438, the voice service module uses the user center voiceprint identification interface to obtain the voice account number of the target account and the connection data and voice data for voiceprint registration;

[0084] S440, the registration is successful, and the user center module saves the voiceprint registration information to the voiceprint registration information table of the target account. The fields in the voiceprint registration table include but are not limited to: identification information of the target account, voice account number, voice call record number, file name of the recording file, path of the recording file, and first object identifier;

[0085] S442, end.

[0086] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods of each embodiment of the present application.

[0087] In this embodiment, a device for determining an abnormal account is also provided, which is used to implement the above embodiments and preferred implementations, and will not be repeated here. As used below, the term "module" can implement a combination of software and / or hardware for a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0088] Figure 5 is a structural block diagram of a device for determining an abnormal account according to an embodiment of the present application, such as Figure 5 As shown, the device comprises:

[0089] A first acquisition module 502 is used to acquire connection data and voice data during a voice call with a voice account of a target account, wherein the connection data is data generated during the process of connecting to the voice account, and the voice data is sound data emitted by the first object through the voice account when the voice account is connected;

[0090] The second acquisition module 504 is used to acquire target historical voice data of the target account when the account tag of the target account is the first tag, wherein the target historical voice data is the sound data emitted by the historical object through the voice account when the voice account is connected in the historical time period;

[0091] The first determination module 506 is used to determine that the target account is an abnormal account when it is determined that the voice feature matching degree between the target historical voice data and the voice data is less than a preset threshold and the connection data is abnormal.

[0092] In an exemplary embodiment, the first acquisition module includes: a first detection submodule, used to detect the metadata and voice quality of the voice call during the voice call with the voice account, wherein the metadata includes the connection method used during the voice call; a first determination submodule, used to determine the metadata as the connection data when it is determined from the metadata that the connection method is a voice transfer method, wherein the voice transfer method includes a method of transferring the voice call to other objects for answering; a first interception submodule, used to intercept the voice with a quality greater than a preset quality threshold from the generated voice during the voice call with the voice account to obtain the voice data.

[0093] In an exemplary embodiment, the above-mentioned device also includes: a first search module, which is used to search for the account tag of the above-mentioned target account from the tag database according to the account information of the above-mentioned target account before obtaining the historical voice data of the above-mentioned target account when the account tag of the above-mentioned target account is the first tag; a second determination module, which is used to determine that the above-mentioned target account is an account in a normal state when the account tag of the above-mentioned target account is the first tag; a third determination module, which is used to determine that the above-mentioned target account is an account in an abnormal state when the account tag of the above-mentioned target account is the second tag; wherein the above-mentioned target account is an account that conducts financial transactions with other accounts.

[0094] In an exemplary embodiment, the above-mentioned first acquisition module includes: a first acquisition sub-module, used to obtain N historical voice data of the above-mentioned target account from the voiceprint database according to the account information of the above-mentioned target account, wherein the above-mentioned N is a natural number greater than or equal to 1; a first selection sub-module, used to select historical voice data with voice quality greater than a preset quality threshold from the above-mentioned N historical voice data to obtain the above-mentioned target historical voice data.

[0095] In an exemplary embodiment, the first determination module includes: a first extraction submodule, used to extract the voice features of the target historical voice data and the voice features of the voice data, respectively, to obtain the first target voice features and the second target voice features; a first calculation submodule, used to calculate the matching degree between the first target voice features and the second target voice features, to obtain the voice feature matching degree; a second determination submodule, used to determine that the target account is the abnormal account when abnormal data is parsed from the connection data and the voice feature matching degree is less than a preset threshold, wherein the connection method used in the voice call process included in the abnormal data is a voice transfer method.

[0096] In an exemplary embodiment, the first extraction submodule includes: a first removal unit, which is used to perform semantic filtering operations on the target historical voice data and the voice data, respectively, to remove the background voice data in the target historical data and the voice data, respectively, to obtain first voice data and second voice data; a first unification unit, which is used to perform standardization operations on the first voice data and the second voice data, respectively to unify the audio references of the first voice data and the second voice data, to obtain third voice data and fourth voice data; a first extraction unit, which is used to extract voice features in the third voice data and the fourth voice data, respectively, in segments, to obtain multiple first voice features corresponding to the third voice data, and multiple second voice features corresponding to the fourth voice data; a first fusion unit, which is used to fuse multiple first voice features to obtain the first target voice feature; and a second fusion unit, which is used to fuse multiple second voice features to obtain the second target voice feature.

[0097] In an exemplary embodiment, the above-mentioned device also includes: a first marking module, which is used to determine that the voice feature matching degree between the above-mentioned target historical voice data and the above-mentioned voice data is less than a preset threshold, and the above-mentioned connection data has an abnormality, and after determining that the above-mentioned target account is an abnormal account, set the account label of the above-mentioned target account to a second label in the label database according to the account information of the above-mentioned target account to mark the above-mentioned target account as an abnormal account; based on the above-mentioned second label, trigger the processing operation of the above-mentioned abnormal account.

[0098] An embodiment of the present application provides a computer program product, including a computer program, and when the computer program is executed by a processor, the steps in any one of the above method embodiments are implemented.

[0099] An embodiment of the present application further provides a computer-readable storage medium, in which a computer program is stored, wherein the computer program is configured to execute the steps of any of the above method embodiments when run.

[0100] In an exemplary embodiment, the computer-readable storage medium may include, but is not limited to, various media that can store computer programs, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk.

[0101] An embodiment of the present application further provides an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0102] In an exemplary embodiment, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0103] For specific examples in this embodiment, reference may be made to the examples described in the above embodiments and exemplary implementation modes, and this embodiment will not be described in detail herein.

[0104] Obviously, those skilled in the art should understand that the above modules or steps of the present application can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order from that herein, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present application is not limited to any specific combination of hardware and software.

[0105] The above are only preferred embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for determining an abnormal account, characterized in that: include: During a voice call with a voice account of a target account, obtaining connection data and voice data, wherein the connection data is data generated during the process of connecting to the voice account, and the voice data is sound data emitted by the first object through the voice account when the voice account is connected; When the account tag of the target account is the first tag, obtaining target historical voice data of the target account, wherein the target historical voice data is sound data emitted by the historical object through the voice account when the voice account is connected in the historical time period; When it is determined that the voice feature matching degree between the target historical voice data and the voice data is less than a preset threshold and the connection data is abnormal, the target account is determined to be an abnormal account.

2. The method according to claim 1, characterized in that During a voice call with the target account's voice account, obtain the connection data and voice data, including: During the voice call with the voice account, detecting metadata and voice quality of the voice call, wherein the metadata includes a connection method used during the voice call; In the case where it is determined from the metadata that the connection mode is a voice transfer mode, determining the metadata as the connection data, wherein the voice transfer mode includes a mode of transferring the voice call to another object for answering; During the voice call with the voice account, the voice whose voice quality is greater than a preset quality threshold is intercepted from the generated voice to obtain the voice data.

3. The method according to claim 1, characterized in that When the account tag of the target account is the first tag, before acquiring the target historical voice data of the target account, the method further includes: Searching for an account tag of the target account from a tag database according to the account information of the target account; When the account tag of the target account is the first tag, determining that the target account is an account in a normal state; When the account tag of the target account is the second tag, determining that the target account is an account in an abnormal state; The target account is an account that conducts fund transactions with other accounts.

4. The method according to claim 1, characterized in that: When the account tag of the target account is the first tag, obtaining target historical voice data of the target account includes: Acquire N historical voice data of the target account from the voiceprint database according to the account information of the target account, where N is a natural number greater than or equal to 1; Select historical voice data whose voice quality is greater than a preset quality threshold from the N historical voice data to obtain the target historical voice data.

5. The method according to claim 1, characterized in that When it is determined that the voice feature matching degree between the target historical voice data and the voice data is less than a preset threshold, and the connection data is abnormal, determining that the target account is an abnormal account includes: Extracting the speech features of the target historical speech data and the speech features of the speech data respectively to obtain a first target speech feature and a second target speech feature; Calculating a matching degree between the first target speech feature and the second target speech feature to obtain the speech feature matching degree; When abnormal data is parsed from the connection data and the voice feature matching degree is less than a preset threshold, the target account is determined to be the abnormal account, wherein the connection method used in the voice call process included in the abnormal data is a voice transfer method.

6. The method according to claim 5, characterized in that Extracting the speech features of the target historical speech data and the speech features of the speech data respectively to obtain a first target speech feature and a second target speech feature, including: Performing semantic filtering operations on the target historical voice data and the voice data respectively to remove background voice data in the target historical data and the voice data respectively, to obtain first voice data and second voice data; Performing a standardization operation on the first voice data and the second voice data to respectively unify the audio references of the first voice data and the second voice data to obtain third voice data and fourth voice data; Extracting speech features in the third speech data and the fourth speech data in segments respectively, to obtain a plurality of first speech features corresponding to the third speech data and a plurality of second speech features corresponding to the fourth speech data; fusing a plurality of the first speech features to obtain the first target speech feature; The plurality of the second speech features are fused to obtain the second target speech feature.

7. The method according to claim 1, characterized in that After determining that the voice feature matching degree between the target historical voice data and the voice data is less than a preset threshold and the connection data is abnormal, determining that the target account is an abnormal account, the method further includes: According to the account information of the target account, the account tag of the target account is set as a second tag in the tag database to mark the target account as an abnormal account; Based on the second tag, a processing operation on the abnormal account is triggered.

8. A device for determining an abnormal account, characterized in that: include: A first acquisition module is used to acquire connection data and voice data during a voice call with a voice account of a target account, wherein the connection data is data generated during the process of connecting to the voice account, and the voice data is sound data emitted by the first object through the voice account when the voice account is connected; A second acquisition module is used to acquire historical voice data of the target account when the account tag of the target account is the first tag, wherein the historical voice data is sound data emitted by the historical object through the voice account when the voice account is connected in a historical time period; The first determination module is used to determine that the target account is an abnormal account when it is determined that the voice feature matching degree between the historical voice data and the voice data is less than a preset threshold and the connection data is abnormal.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program implements the steps of the method described in any one of claims 1 to 6 when executed by a processor.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method described in any one of claims 1 to 6 are implemented.

Images