Phone card fraud detection method, system, computer equipment and storage medium

Abstract

The present scheme relates to a method, system, computer equipment and storage medium for detecting phone card fraud. The method comprises: obtaining a suspect list containing phone numbers of phone cards suspected of being fraudulently used; obtaining a phone recording of the owner corresponding to the phone number, and adjusting the collection frequency of the phone recording to a target collection frequency through a resampling algorithm; inputting the phone recording containing the target collection frequency into a voiceprint recognition algorithm model to obtain a voiceprint feature corresponding to the phone recording; searching for a registered voiceprint feature corresponding to the phone number in a registered recording base; comparing the voiceprint feature with the registered voiceprint feature to obtain a comparison result, and determining whether the phone card corresponding to the phone number is fraudulently used based on the comparison result. By performing voiceprint recognition on the collected phone recording and the target phone recording in the database, and determining whether the owner corresponding to the phone recording is the registered owner based on the voiceprint recognition result, it is possible to accurately detect whether the phone card is fraudulently used.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to a method, system, computer equipment and storage medium for detecting fraudulent use of telephone cards. Background Art

[0002] Telecom fraud refers to the criminal behavior of fabricating false information and setting up scams through telephone, Internet and text messages to commit remote and contactless fraud against victims, and inducing victims to make payments or transfers. The purpose of deception is usually achieved by impersonating others and counterfeiting and forging various legal cloaks and forms. With the development of science and technology, the fraudulent means of fraud have become more covert and tracing has become more difficult. When committing fraud, scammers usually use other people's phone cards. After using other people's phone cards to commit fraud, victims and law enforcement officers usually call the owner of the phone card to check, but the owner is not the fraud perpetrator, which makes the fraud success rate of fraudsters continue to increase, making it more difficult for law enforcement officers to supervise and crack down on fraudsters.

[0003] The traditional way to intercept fraudulent calls is usually to deploy an interception platform at the call port to screen the passing calls one by one. However, the traditional way to intercept fraudulent calls cannot intercept fraudulent calling cards and has the problem of difficulty in detection. Summary of the invention

[0004] Based on this, in order to solve the above technical problems, a method, system, computer device and storage medium for detecting fraudulent use of a telephone card are provided, which can accurately detect whether a telephone card is being used fraudulently.

[0005] A method for detecting fraudulent use of a telephone card, the method comprising:

[0006] Obtaining a suspect list, wherein the suspect list includes telephone numbers suspected of being fraudulently used for calling cards;

[0007] Acquire a telephone recording of the owner corresponding to the telephone number, and adjust the collection frequency of the telephone recording to a target collection frequency through a resampling algorithm;

[0008] Inputting the telephone recording containing the target acquisition frequency into a voiceprint recognition algorithm model to obtain a voiceprint feature corresponding to the telephone recording;

[0009] Searching for a registered voiceprint feature corresponding to the telephone number in a registered recording database;

[0010] The voiceprint feature is compared with the registered voiceprint feature to obtain a comparison result, and it is determined based on the comparison result whether the phone card corresponding to the phone number is used fraudulently.

[0011] In one embodiment, before adjusting the acquisition frequency of the telephone recording to the target acquisition frequency by a resampling algorithm, the method further includes:

[0012] Intercepting each voice segment in the telephone recording, and calculating each recording duration corresponding to each voice segment;

[0013] Counting the duration distribution of each of the voice durations, and determining whether the telephone recording is coherent according to each of the recording durations and the duration distribution;

[0014] When the telephone recording is continuous, the collection frequency of the telephone recording is adjusted to a target collection frequency through a resampling algorithm.

[0015] In one embodiment, the method further comprises:

[0016] Collect the registration voice of the phone card account opening user;

[0017] Intercepting each registration voice segment in the account opening registration voice, and calculating each registration recording duration corresponding to each registration voice segment;

[0018] Counting the standard duration distribution of each registration voice duration, and determining whether the account opening registration voice is coherent according to each registration recording duration and the standard duration distribution;

[0019] When the account opening and registration recording is coherent, the account opening and registration voice is input into the voiceprint recognition algorithm model to obtain the registration voiceprint feature corresponding to the account opening and registration recording, and the registration voiceprint feature is stored in the registration recording base library.

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

[0021] Collecting the telephone recording via a first frequency;

[0022] Collecting the account opening and registration voice through a second frequency;

[0023] The first frequency is adjusted to a target acquisition frequency through a resampling algorithm, and the second frequency is adjusted to a target acquisition frequency through the resampling algorithm.

[0024] In one embodiment, the training process of the voiceprint recognition algorithm model includes:

[0025] Acquire an initial recognition algorithm model, and extract initial parameters in the initial recognition algorithm model;

[0026] Collecting sample voice data and preprocessing the sample voice data to obtain target sample voice data;

[0027] Extracting acoustic features from the target sample speech data, and calling an encoder to encode the acoustic features into a high-dimensional embedding vector through a neural network;

[0028] Inputting the high-dimensional embedding vector into the initial recognition algorithm model to obtain an initial recognition result, and adjusting the initial parameters according to the initial recognition result to obtain target parameters;

[0029] The initial recognition algorithm model is adjusted according to the target parameters to obtain the voiceprint recognition algorithm model.

[0030] In one embodiment, comparing the voiceprint feature with the registered voiceprint feature to obtain a comparison result includes:

[0031] Comparing the voiceprint feature with the registered voiceprint feature, and calculating a comparison score between the voiceprint feature and the registered voiceprint feature;

[0032] The comparison result is obtained according to the comparison score.

[0033] A telephone card fraud detection system, the system comprising:

[0034] A list acquisition module is used to acquire a suspect list, wherein the suspect list contains phone numbers suspected of being fraudulently used phone cards;

[0035] A frequency adjustment module, used to obtain a telephone recording of the owner corresponding to the telephone number, and adjust the collection frequency of the telephone recording to a target collection frequency through a resampling algorithm;

[0036] A voiceprint feature extraction module, used to input the telephone recording containing the target acquisition frequency into a voiceprint recognition algorithm model to obtain a voiceprint feature corresponding to the telephone recording;

[0037] A voiceprint feature search module, used to search for a registered voiceprint feature corresponding to the telephone number in a registered recording database;

[0038] The fraudulent use detection module is used to compare the voiceprint feature with the registered voiceprint feature to obtain a comparison result, and determine whether the phone card corresponding to the phone number is fraudulently used according to the comparison result.

[0039] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0040] Obtaining a suspect list, wherein the suspect list includes telephone numbers suspected of being fraudulently used for calling cards;

[0041] Acquire a telephone recording of the owner corresponding to the telephone number, and adjust the collection frequency of the telephone recording to a target collection frequency through a resampling algorithm;

[0042] Inputting the telephone recording containing the target acquisition frequency into a voiceprint recognition algorithm model to obtain a voiceprint feature corresponding to the telephone recording;

[0043] Searching for a registered voiceprint feature corresponding to the telephone number in a registered recording database;

[0044] The voiceprint feature is compared with the registered voiceprint feature to obtain a comparison result, and it is determined based on the comparison result whether the phone card corresponding to the phone number is used fraudulently.

[0045] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the following steps:

[0046] Obtaining a suspect list, wherein the suspect list includes telephone numbers suspected of being fraudulently used for calling cards;

[0047] Acquire a telephone recording of the owner corresponding to the telephone number, and adjust the collection frequency of the telephone recording to a target collection frequency through a resampling algorithm;

[0048] Inputting the telephone recording containing the target acquisition frequency into a voiceprint recognition algorithm model to obtain a voiceprint feature corresponding to the telephone recording;

[0049] Searching for a registered voiceprint feature corresponding to the telephone number in a registered recording database;

[0050] The voiceprint feature is compared with the registered voiceprint feature to obtain a comparison result, and it is determined based on the comparison result whether the phone card corresponding to the phone number is used fraudulently.

[0051] The above-mentioned phone card fraud detection method, system, computer device and storage medium obtain a suspect list, which contains the phone number of the phone card suspected of being fraudulently used; obtain the phone recording of the owner corresponding to the phone number, and adjust the collection frequency of the phone recording to the target collection frequency through a resampling algorithm; input the phone recording containing the target collection frequency into the voiceprint recognition algorithm model to obtain the voiceprint feature corresponding to the phone recording; search for the registered voiceprint feature corresponding to the phone number in the registered recording base; compare the voiceprint feature with the registered voiceprint feature to obtain the comparison result, and determine whether the phone card corresponding to the phone number is fraudulently used according to the comparison result. By performing voiceprint recognition on the collected phone recording and the target phone recording in the database, it is possible to accurately detect whether the phone card is fraudulently used. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 A diagram of an application environment of a method for detecting fraudulent use of a telephone card in an embodiment;

[0053] Figure 2 A schematic diagram of a process of detecting fraudulent use of a telephone card in one embodiment;

[0054] Figure 3 is a structural block diagram of a phone card fraud detection system in one embodiment;

[0055] Figure 4 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0057] It is understood that the terms "first", "second", etc. used in this application may be used herein to describe frequencies, but these frequencies are not limited by these terms. These terms are only used to distinguish a first frequency from another frequency. For example, without departing from the scope of this application, a first frequency may be referred to as a second frequency, and similarly, a second frequency may be referred to as a first frequency. Both the first frequency and the second frequency are frequencies, but they are not the same frequency.

[0058] The method for detecting fraudulent use of a telephone card provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Figure 1As shown, the application environment includes a computer device 110. The computer device 110 can obtain a suspect list, which includes a phone number of a suspected phone card that has been fraudulently used; the computer device 110 can obtain a phone recording of the owner corresponding to the phone number, and adjust the collection frequency of the phone recording to the target collection frequency through a resampling algorithm; the computer device 110 can input the phone recording containing the target collection frequency into the voiceprint recognition algorithm model to obtain the voiceprint feature corresponding to the phone recording; the computer device 110 can search for the registered voiceprint feature corresponding to the phone number in the registered recording base; the computer device 110 can compare the voiceprint feature with the registered voiceprint feature to obtain a comparison result, and determine whether the phone card corresponding to the phone number has been fraudulently used based on the comparison result. Among them, the computer device 110 can be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, portable wearable devices, etc.

[0059] In one embodiment, Figure 2 As shown, a method for detecting phone card fraud is provided, comprising the following steps:

[0060] Step 202, obtaining a suspect list, wherein the suspect list includes telephone numbers of telephone cards suspected of being fraudulently used.

[0061] The suspect list includes the phone number of the suspected fraudulently used calling card. Specifically, the phone number of the suspected fraudulently used calling card can be obtained by the staff through screening, and the staff can input the phone number of the suspected fraudulently used calling card into the computer device, and the computer device generates the suspect list and stores it in the database.

[0062] Step 204, obtaining the telephone recording of the owner corresponding to the telephone number, and adjusting the collection frequency of the telephone recording to the target collection frequency through a resampling algorithm.

[0063] The staff can call the phone number of the suspected fraudulently used calling card and talk to the owner of the phone number. The computer equipment can record the conversation between the staff and the owner through the customer service system.

[0064] The resampling algorithm can be used to unify audio collected at different frequencies to a specified collection frequency. In this embodiment, after obtaining the telephone recording, the computer device can extract the collection frequency of the telephone recording and adjust the collection frequency of the telephone recording to a target collection frequency through the resampling algorithm. The target collection frequency can be pre-set.

[0065] Step 206, input the telephone recording containing the target acquisition frequency into the voiceprint recognition algorithm model to obtain the voiceprint feature corresponding to the telephone recording.

[0066] The voiceprint recognition algorithm model can be a pre-trained model used to recognize the voiceprints of various telephone recordings and obtain the recognition results. The voiceprint recognition algorithm model can be stored in the computer device, and the computer device can input the telephone recording containing the target collection frequency into the voiceprint recognition algorithm model to obtain the voiceprint features corresponding to the telephone recording.

[0067] Step 208: Search the registered voiceprint feature corresponding to the phone number in the registered recording database.

[0068] When each registered user registers a phone card, each registered phone recording is stored. The computer device can input each registered phone recording into the voiceprint recognition algorithm model to obtain each registered voiceprint feature corresponding to each registered phone recording.

[0069] The registered voiceprint features are stored in the registered recording base. Specifically, each registered voiceprint feature can correspond to each phone number, and one phone number can correspond to multiple registered voiceprint features. For example, one phone number can correspond to 500 registered voiceprint features. The registered voiceprint features belonging to the same phone number can be packaged and stored in the registered recording base. The computer device can search for each registered voiceprint feature corresponding to the phone number in the registered recording base.

[0070] Step 210, compare the voiceprint feature with the registered voiceprint feature to obtain a comparison result, and determine whether the phone card corresponding to the phone number is used fraudulently based on the comparison result.

[0071] The voiceprint feature is the voiceprint recognition result of the current conversation between the staff and the owner of the phone number, and the registered voiceprint feature is the voiceprint recognition result of the registered user of the phone card when registering. The computer device can compare the voiceprint feature with the registered voiceprint feature to obtain a comparison result. The comparison result may include the same result or different result. The same result may indicate that the voiceprint of the owner of the current phone number is highly similar to the voiceprint of the registered user of the phone card, and they can be identified as the same person, indicating that the phone card has not been used fraudulently; the different result may indicate that the voiceprint of the owner of the current phone number is less similar to the voiceprint of the registered user of the phone card, and the two are not the same person, indicating that the phone card has been used fraudulently.

[0072] In this embodiment, the computer device obtains a suspect list, which includes a phone number of a suspected phone card that has been fraudulently used; obtains a phone recording of the owner corresponding to the phone number, and adjusts the collection frequency of the phone recording to the target collection frequency through a resampling algorithm; inputs the phone recording containing the target collection frequency into the voiceprint recognition algorithm model to obtain the voiceprint feature corresponding to the phone recording; searches for the registered voiceprint feature corresponding to the phone number in the registered recording base; compares the voiceprint feature with the registered voiceprint feature to obtain a comparison result, and determines whether the phone card corresponding to the phone number has been fraudulently used based on the comparison result. By performing voiceprint recognition on the collected phone recording and the target phone recording in the database, it is possible to accurately detect whether the owner of the phone recording is the registered owner based on the voiceprint recognition result, and whether the phone card has been fraudulently used.

[0073] In one embodiment, a provided method for detecting fraudulent use of a telephone card may further include a process for detecting telephone recordings, the specific process including: intercepting each voice segment in the telephone recording, and calculating each recording duration corresponding to each voice segment; counting the duration distribution of each voice duration, and determining whether the telephone recording is coherent based on each recording duration and duration distribution; when the telephone recording is coherent, adjusting the acquisition frequency of the telephone recording to the target acquisition frequency through a resampling algorithm.

[0074] After collecting the telephone recording, the computer device can intercept each voice segment in the telephone recording. Specifically, the computer device can remove the silent or non-voice segments in the telephone recording to obtain each voice segment.

[0075] The computer device can calculate the recording durations corresponding to the various voice segments. Specifically, the computer device can calculate the cumulative duration of the pure voice part as the effective recording duration. The computer device can also count the duration distribution of each voice duration, and determine whether the telephone recording is coherent based on the durations and duration distribution of each recording. When the telephone recording is coherent, it means that the telephone recording has passed the detection. The computer device can further adjust the collection frequency of the telephone recording to the target collection frequency through a resampling algorithm.

[0076] In this embodiment, the computer device can improve the recognition efficiency of telephone recordings by processing telephone recordings.

[0077] In one embodiment, a provided method for detecting fraudulent use of a telephone card may further include a process of collecting an account opening and registration voice, the specific process including: collecting the account opening and registration voice of the telephone card account opening user; intercepting each registration voice segment in the account opening and registration voice, and calculating each registration recording duration corresponding to each registration voice segment; counting the standard duration distribution of each registration voice duration, and determining whether the account opening and registration voice is coherent based on each registration recording duration and the standard duration distribution; when the account opening and registration recording is coherent, inputting the account opening and registration voice into the voiceprint recognition algorithm model to obtain the registration voiceprint feature corresponding to the account opening and registration recording, and storing the registration voiceprint feature in the registration recording base library.

[0078] When a user registers a phone card, the staff can collect the user's registration voice through a computer device. The computer device can process the collected registration voice. Specifically, the computer device can intercept each registration voice segment in the registration voice, that is, remove the silent or non-voice segments in the registration voice, and calculate the cumulative time of the pure voice part as the effective registration recording time.

[0079] The computer device can further count the standard duration distribution of each registration voice duration, and judge whether the account opening registration voice is coherent according to the duration of each registration recording and the standard duration distribution, so as to obtain a judgment result. When the judgment result is that the account opening registration voice is coherent, the computer device can input the account opening registration voice into the voiceprint recognition algorithm model, obtain the registration voiceprint feature corresponding to the account opening registration recording, and store the registration voiceprint feature in the registration recording base library; when the judgment result is that the account opening registration voice is incoherent, the computer device can determine that the voice quality of the account opening registration voice is unqualified, and can return the judgment result of unqualified voice quality, so that the staff can re-collect the account opening registration voice.

[0080] In one embodiment, a provided method for detecting phone card fraud may also include a process of unifying the voice collection frequency, the specific process including: collecting telephone recordings through a first frequency; collecting account opening and registration voices through a second frequency; adjusting the first frequency to a target collection frequency through a resampling algorithm, and adjusting the second frequency to the target collection frequency through the resampling algorithm.

[0081] The computer device can use a microphone, a sound card, etc. to collect telephone recordings through a first frequency, wherein the first frequency can be 8KHz, and the computer device can record 8KHz telephone recordings of voice calls through a customer service system.

[0082] In this embodiment, the computer device can use a microphone, a sound card, etc. to collect the account registration voice through the second frequency, wherein the second frequency can be 48kHz, and the computer device can collect the account registration voice through a 48kHz directional microphone.

[0083] The computer device can obtain the first frequency and the second frequency, and perform up-sampling and down-sampling processing using a resampling algorithm to make the two frequencies the same, thereby obtaining a target acquisition frequency. By using a resampling algorithm on voice signals of two different transmission channels to obtain the same acquisition frequency, two voice signals of different channels and different sampling rates can use the same voiceprint recognition algorithm model to obtain voiceprint recognition results, thereby improving the efficiency of voiceprint recognition.

[0084] In one embodiment, a provided method for detecting fraudulent use of a telephone card may further include a training process for a voiceprint recognition algorithm model, the specific process including: obtaining an initial recognition algorithm model and extracting initial parameters in the initial recognition algorithm model; collecting sample voice data and preprocessing the sample voice data to obtain target sample voice data; extracting acoustic features in the target sample voice data, and calling an encoder to encode the acoustic features into a high-dimensional embedding vector through a neural network; inputting the high-dimensional embedding vector into the initial recognition algorithm model to obtain an initial recognition result, and adjusting the initial parameters according to the initial recognition result to obtain the target parameters; adjusting the initial recognition algorithm model according to the target parameters to obtain a voiceprint recognition algorithm model.

[0085] In this embodiment, the voiceprint recognition algorithm model can be divided into modules such as preprocessing module, acoustic feature extractor, encoder, speaker serial number encoder, classifier, trainer, etc. Among them, each module can be provided with relevant interfaces for other modules to call; the trainer can be used as a program entry, calling the preprocessing module to preprocess the sample voice data, calling the acoustic feature extractor to extract relevant acoustic features from the target sample voice data, calling the encoder to encode the acoustic features into a high-dimensional embedding vector through a neural network, calling the speaker serial number encoder to encode the speakers in the data set using digital numbers, calling the classifier to perform multiple classifications on the high-dimensional embedding vector, and the classification error is fed back to the acoustic feature extractor, encoder, and classifier through the trainer for parameter adjustment. Among them, each module can store its internal parameters as an independent file; when parameter adjustment is required, the required module can be switched, and the parameter adjustment and optimization of the specific module can be performed.

[0086] Among them, the data weights in the voiceprint recognition algorithm model can be trained. Specifically, the computer device can extract the loss function in the voiceprint recognition algorithm model, introduce variable weights to the reasoning error of the data sample in the loss function, and count the error rate of each data sample when traversing the data set for parameter optimization each time, and increase the weight of difficult examples with higher error rates to improve the accuracy of the model for difficult data. In addition, users can set different weights for two different error types, false rejection and false acceptance, according to actual needs, to increase the sensitivity of the model to one type of error.

[0087] In one embodiment, a provided method for detecting fraudulent use of a telephone card may further include a voiceprint comparison process, the specific process including: comparing the voiceprint feature with the registered voiceprint feature, and calculating the comparison score between the voiceprint feature and the registered voiceprint feature; and obtaining a comparison result based on the comparison score.

[0088] After obtaining the voiceprint features and the registered voiceprint features, the computer device can compare the voiceprint features with the registered voiceprint features. Specifically, the staff can make a customer service return call to the phone number in the suspected list, compare the voiceprint of the customer service call recording with the account opening and registration recording of the owner of the phone to obtain a comparison score; at the same time, the computer device can use the customer service return call recording to perform voiceprint retrieval in the database containing the account opening and registration recording corresponding to the phone number in the suspected list, and find the most similar one or more users, and calculate the recognition result between the customer service call recording and the most similar user. The computer device can compare the comparison score and recognition result. If the comparison score is "same" and the recognition result is the most similar to the owner of the phone, it is considered that the owner and the user are the same person and there is no fraudulent use; if the comparison score is "different" and the recognition result is not the owner of the phone, it is considered that the owner and the user are not the same person and there is fraudulent use of the phone card.

[0089] It should be understood that, although the various steps in the above-mentioned flow chart are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least a part of the steps in the above-mentioned flow chart may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these sub-steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0090] In one embodiment, Figure 3As shown, a phone card fraud detection system is provided, including: a list acquisition module 310, a frequency adjustment module 320, a voiceprint feature extraction module 330, a voiceprint feature search module 340 and a fraud detection module 350, wherein:

[0091] The list acquisition module 310 is used to acquire a suspect list, which includes phone numbers suspected of being fraudulently used phone cards;

[0092] The frequency adjustment module 320 is used to obtain the telephone recording of the phone owner corresponding to the telephone number, and adjust the collection frequency of the telephone recording to the target collection frequency through a resampling algorithm;

[0093] The voiceprint feature extraction module 330 is used to input the telephone recording containing the target collection frequency into the voiceprint recognition algorithm model to obtain the voiceprint feature corresponding to the telephone recording;

[0094] The voiceprint feature search module 340 is used to search for the registered voiceprint feature corresponding to the phone number in the registered recording database;

[0095] The fraudulent use detection module 350 is used to compare the voiceprint feature with the registered voiceprint feature to obtain a comparison result, and determine whether the phone card corresponding to the phone number is fraudulently used based on the comparison result.

[0096] In one embodiment, the frequency adjustment module 320 is also used to intercept each voice segment in the telephone recording, and calculate each recording duration corresponding to each voice segment; count the duration distribution of each voice duration, and determine whether the telephone recording is coherent based on each recording duration and duration distribution; when the telephone recording is coherent, the oversampling algorithm adjusts the acquisition frequency of the telephone recording to the target acquisition frequency.

[0097] In one embodiment, the voiceprint feature search module 340 is also used to collect the account opening registration voice of the phone card account opening user; intercept each registration voice segment in the account opening registration voice, and calculate each registration recording duration corresponding to each registration voice segment; count the standard duration distribution of each registration voice duration, and determine whether the account opening registration voice is coherent based on each registration recording duration and the standard duration distribution; when the account opening registration recording is coherent, input the account opening registration voice into the voiceprint recognition algorithm model to obtain the registration voiceprint feature corresponding to the account opening registration recording, and store the registration voiceprint feature in the registration recording base library.

[0098] In one embodiment, the frequency adjustment module 320 is also used to collect telephone recordings through the first frequency; collect account registration voice through the second frequency; adjust the first frequency to the target collection frequency through the resampling algorithm, and adjust the second frequency to the target collection frequency through the resampling algorithm.

[0099] In one embodiment, the voiceprint feature extraction module 330 is also used to obtain an initial recognition algorithm model and extract initial parameters in the initial recognition algorithm model; collect sample voice data and preprocess the sample voice data to obtain target sample voice data; extract acoustic features in the target sample voice data, and call the encoder to encode the acoustic features into a high-dimensional embedding vector through a neural network; input the high-dimensional embedding vector into the initial recognition algorithm model to obtain an initial recognition result, and adjust the initial parameters according to the initial recognition result to obtain the target parameters; adjust the initial recognition algorithm model according to the target parameters to obtain a voiceprint recognition algorithm model.

[0100] In one embodiment, the fraud detection module 350 is further used to compare the voiceprint feature with the registered voiceprint feature, and calculate the comparison score between the voiceprint feature and the registered voiceprint feature; and obtain the comparison result according to the comparison score.

[0101] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Figure 4 As shown. The computer device includes a processor, a memory, a network interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for detecting fraudulent use of a telephone card is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a key, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0102] Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0103] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0104] Obtain a suspect list, which includes phone numbers suspected of being used fraudulently for calling cards;

[0105] Obtain the phone recording of the owner corresponding to the phone number, and adjust the collection frequency of the phone recording to the target collection frequency through a resampling algorithm;

[0106] Input the telephone recording containing the target collection frequency into the voiceprint recognition algorithm model to obtain the voiceprint features corresponding to the telephone recording;

[0107] Searching for the registered voiceprint feature corresponding to the phone number in the registered recording database;

[0108] The voiceprint feature is compared with the registered voiceprint feature to obtain a comparison result, and based on the comparison result, it is determined whether the phone card corresponding to the phone number has been used fraudulently.

[0109] In one embodiment, the processor also implements the following steps when executing the computer program: intercepting each voice segment in the telephone recording, and calculating each recording duration corresponding to each voice segment; counting the duration distribution of each voice duration, and determining whether the telephone recording is coherent based on each recording duration and duration distribution; when the telephone recording is coherent, adjusting the acquisition frequency of the telephone recording to the target acquisition frequency through a resampling algorithm.

[0110] In one embodiment, the processor also implements the following steps when executing the computer program: collecting the account opening registration voice of the phone card account opening user; intercepting each registration voice segment in the account opening registration voice, and calculating each registration recording duration corresponding to each registration voice segment; counting the standard duration distribution of each registration voice duration, and determining whether the account opening registration voice is coherent based on each registration recording duration and the standard duration distribution; when the account opening registration recording is coherent, inputting the account opening registration voice into the voiceprint recognition algorithm model to obtain the registration voiceprint feature corresponding to the account opening registration recording, and storing the registration voiceprint feature in the registration recording base library.

[0111] In one embodiment, when the processor executes the computer program, the following steps are also implemented: collecting telephone recordings through a first frequency; collecting account registration voice through a second frequency; adjusting the first frequency to a target collection frequency through a resampling algorithm, and adjusting the second frequency to the target collection frequency through a resampling algorithm.

[0112] In one embodiment, when the processor executes the computer program, the following steps are also implemented: obtaining an initial recognition algorithm model and extracting initial parameters in the initial recognition algorithm model; collecting sample voice data and preprocessing the sample voice data to obtain target sample voice data; extracting acoustic features in the target sample voice data, and calling the encoder to encode the acoustic features into a high-dimensional embedding vector through a neural network; inputting the high-dimensional embedding vector into the initial recognition algorithm model to obtain an initial recognition result, and adjusting the initial parameters according to the initial recognition result to obtain the target parameters; adjusting the initial recognition algorithm model according to the target parameters to obtain a voiceprint recognition algorithm model.

[0113] In one embodiment, when the processor executes the computer program, the following steps are also implemented: comparing the voiceprint feature with the registered voiceprint feature, and calculating the comparison score between the voiceprint feature and the registered voiceprint feature; and obtaining the comparison result according to the comparison score.

[0114] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0115] Obtain a suspect list, which includes phone numbers suspected of being used fraudulently for calling cards;

[0116] Obtain the phone recording of the owner corresponding to the phone number, and adjust the collection frequency of the phone recording to the target collection frequency through a resampling algorithm;

[0117] Input the telephone recording containing the target collection frequency into the voiceprint recognition algorithm model to obtain the voiceprint features corresponding to the telephone recording;

[0118] Searching for the registered voiceprint feature corresponding to the phone number in the registered recording database;

[0119] The voiceprint feature is compared with the registered voiceprint feature to obtain a comparison result, and based on the comparison result, it is determined whether the phone card corresponding to the phone number has been used fraudulently.

[0120] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: intercepting each voice segment in the telephone recording, and calculating each recording duration corresponding to each voice segment; counting the duration distribution of each voice duration, and determining whether the telephone recording is coherent based on each recording duration and duration distribution; when the telephone recording is coherent, adjusting the acquisition frequency of the telephone recording to the target acquisition frequency through a resampling algorithm.

[0121] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: collecting the account opening registration voice of the phone card account opening user; intercepting each registration voice segment in the account opening registration voice, and calculating each registration recording duration corresponding to each registration voice segment; counting the standard duration distribution of each registration voice duration, and determining whether the account opening registration voice is coherent based on each registration recording duration and the standard duration distribution; when the account opening registration recording is coherent, inputting the account opening registration voice into the voiceprint recognition algorithm model to obtain the registration voiceprint feature corresponding to the account opening registration recording, and storing the registration voiceprint feature in the registration recording base library.

[0122] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: collecting telephone recordings through a first frequency; collecting account registration voice through a second frequency; adjusting the first frequency to a target collection frequency through a resampling algorithm, and adjusting the second frequency to the target collection frequency through a resampling algorithm.

[0123] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: obtaining an initial recognition algorithm model and extracting initial parameters in the initial recognition algorithm model; collecting sample voice data and preprocessing the sample voice data to obtain target sample voice data; extracting acoustic features in the target sample voice data, and calling the encoder to encode the acoustic features into a high-dimensional embedding vector through a neural network; inputting the high-dimensional embedding vector into the initial recognition algorithm model to obtain an initial recognition result, and adjusting the initial parameters according to the initial recognition result to obtain the target parameters; adjusting the initial recognition algorithm model according to the target parameters to obtain a voiceprint recognition algorithm model.

[0124] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: comparing the voiceprint feature with the registered voiceprint feature, and calculating the comparison score between the voiceprint feature and the registered voiceprint feature; and obtaining the comparison result according to the comparison score.

[0125] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0126] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0127] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A method for detecting fraudulent use of a telephone card, characterized in that: The method comprises: Obtaining a suspect list, wherein the suspect list includes telephone numbers suspected of being fraudulently used for calling cards; Obtain a telephone recording of the phone owner corresponding to the telephone number, intercept each voice segment in the telephone recording, and respectively calculate each recording duration corresponding to each voice segment; count the duration distribution of each recording duration, and determine whether the telephone recording is coherent according to each recording duration and the duration distribution; when the telephone recording is coherent, adjust the collection frequency of the telephone recording to the target collection frequency through a resampling algorithm; Inputting the telephone recording containing the target acquisition frequency into a voiceprint recognition algorithm model to obtain a voiceprint feature corresponding to the telephone recording; Searching for a registered voiceprint feature corresponding to the telephone number in a registered recording database; The voiceprint feature is compared with the registered voiceprint feature to obtain a comparison result, and it is determined based on the comparison result whether the phone card corresponding to the phone number is used fraudulently.

2. The method for detecting fraudulent use of a telephone card according to claim 1, characterized in that: The method further comprises: Collect the registration voice of the phone card account opening user; Intercepting each registration voice segment in the account opening registration voice, and calculating each registration recording duration corresponding to each registration voice segment; Counting the standard duration distribution of each registration recording duration, and determining whether the account opening registration voice is coherent according to the duration of each registration recording and the standard duration distribution; When the account opening and registration voice is coherent, the account opening and registration voice is input into the voiceprint recognition algorithm model to obtain the registration voiceprint feature corresponding to the account opening and registration voice, and the registration voiceprint feature is stored in the registration recording base library.

3. The method for detecting fraudulent use of a telephone card according to claim 2, characterized in that: The method further comprises: Collecting the telephone recording via a first frequency; Collecting the account opening and registration voice through a second frequency; The first frequency is adjusted to a target acquisition frequency through a resampling algorithm, and the second frequency is adjusted to a target acquisition frequency through the resampling algorithm.

4. The method for detecting fraudulent use of a telephone card according to claim 1, characterized in that: The training process of the voiceprint recognition algorithm model includes: Acquire an initial recognition algorithm model, and extract initial parameters in the initial recognition algorithm model; Collecting sample voice data and preprocessing the sample voice data to obtain target sample voice data; Extracting acoustic features from the target sample speech data, and calling an encoder to encode the acoustic features into a high-dimensional embedding vector through a neural network; Inputting the high-dimensional embedding vector into the initial recognition algorithm model to obtain an initial recognition result, and adjusting the initial parameters according to the initial recognition result to obtain target parameters; The initial recognition algorithm model is adjusted according to the target parameters to obtain the voiceprint recognition algorithm model.

5. The method for detecting fraudulent use of a telephone card according to claim 1, characterized in that: The comparing the voiceprint feature with the registered voiceprint feature to obtain a comparison result includes: Comparing the voiceprint feature with the registered voiceprint feature, and calculating a comparison score between the voiceprint feature and the registered voiceprint feature; The comparison result is obtained according to the comparison score.

6. A phone card fraud detection system, characterized in that: The system comprises: A list acquisition module is used to acquire a suspect list, wherein the suspect list contains phone numbers suspected of being fraudulently used phone cards; The frequency adjustment module is used to obtain the telephone recording of the phone owner corresponding to the telephone number, intercept each voice segment in the telephone recording, and respectively calculate the recording duration corresponding to each voice segment; count the duration distribution of each recording duration, and determine whether the telephone recording is coherent according to each recording duration and the duration distribution; when the telephone recording is coherent, adjust the collection frequency of the telephone recording to the target collection frequency through a resampling algorithm; A voiceprint feature extraction module, used to input the telephone recording containing the target acquisition frequency into a voiceprint recognition algorithm model to obtain a voiceprint feature corresponding to the telephone recording; A voiceprint feature search module, used to search for a registered voiceprint feature corresponding to the telephone number in a registered recording database; The fraudulent use detection module is used to compare the voiceprint feature with the registered voiceprint feature to obtain a comparison result, and determine whether the phone card corresponding to the phone number is fraudulently used according to the comparison result.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images