Financial data identity recognition method based on biometric recognition
By analyzing the degree of interference and adhesion of the peak data segment of the voice ripple waveform, and evaluating the complexity of the voice ripple scene, the problem that the identity recognition system cannot effectively judge the security of the environment is solved, and the accuracy and efficiency of identity recognition are improved.
Patent Information
- Application Number
- CN202410704819.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-03
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2044-06-03
AI Technical Summary
When the identity identification system verifies the voice of a person, the entered sound information usually contains a large amount of external environment and irrelevant personnel's voice information, which makes it impossible to effectively judge the security of the verification environment and reduces the accuracy and efficiency of identity identification.
By acquiring and analyzing the interference and adhesion of the voiceprint information in the voiceprint peak data segment, the clarity of the voiceprint information is obtained, and the complexity of the voiceprint scene is evaluated to judge the security of the identity verification environment. Specific methods include calculating the degree of interference, adhesion and clarity of voiceprint information, and identifying the identity based on the complexity of voiceprint scene.
It improves the accuracy of judging and verifying the security of the environment, enhances the accuracy and efficiency of identity identification, and ensures the efficiency of real-time call of financial data.
Smart Images

Figure CN118568698B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of identity recognition technology, and in particular to a financial data identity recognition method based on biometric recognition. Background Art
[0002] Biometric identification technology is an identity authentication method based on the combination of biological principles and computer science and technology. Its main purpose is to extract the unique independent characteristics of organisms for the authentication of non-public data or file usage rights. When the system identifies financial data related to the enterprise, in order to ensure a high degree of convenience, the identification system is usually set up in a multi-person environment; at the same time, in order to ensure a high degree of security, voiceprint biometrics are usually used, that is, after verifying the environmental security of the person's voice information, the person's voice information is used for identity identification, thereby further ensuring the security of financial data.
[0003] During the process of the system verifying the identity of the person using their voice, the sound information recorded by the identity recognition system usually contains a large amount of sound information from the external environment and other irrelevant people, which results in the identity recognition system being unable to effectively judge the security of the verification environment in which the person is located, thereby reducing the accuracy of identity recognition, reducing the efficiency of identity recognition, and reducing the efficiency of real-time calls to financial data by authorized personnel. Summary of the invention
[0004] The present invention provides a financial data identity recognition method based on biometric recognition to solve the existing problem that the sound information entered by the identity recognition system usually contains a large amount of sound information of the external environment and other irrelevant persons, resulting in the identity recognition system being unable to effectively judge the security of the verification environment in which the verification person is located.
[0005] The financial data identification method based on biometric identification of the present invention adopts the following technical solutions:
[0006] The following steps are involved:
[0007] Acquire a plurality of voiceprint waveform peak data segments of the person to be verified, wherein the voiceprint waveform peak data segments contain a plurality of voiceprint data;
[0008] Analyze the smoothness of the overall waveform profile of the voiceprint data in the same voiceprint waveform peak data segment to obtain the interference degree of the voiceprint information of the voiceprint waveform peak data segment; compare the smooth adhesion of the waveform profile changes between adjacent voiceprint waveform peak data segments to obtain the adhesion degree of the voiceprint information of the voiceprint waveform peak data segment; combine the interference degree of the voiceprint information with the adhesion degree of the voiceprint information to obtain the clarity of the voiceprint information of the voiceprint waveform peak data segment;
[0009] According to the clarity of the voiceprint information, the information distribution contained in the voiceprint waveform peak data segment is analyzed to obtain the voiceprint scene complexity of the person to be verified; the identity of the financial data is recognized based on the voiceprint scene complexity.
[0010] Preferably, the method for obtaining the interference degree of the voiceprint information is:
[0011] For any voiceprint waveform peak data segment, obtain the mean value of the adjacent data differences of the voiceprint waveform peak data segment;
[0012] According to the mean value of the difference between adjacent data, the interference degree of the voiceprint information of the voiceprint waveform peak data segment is obtained; the interference degree of the voiceprint information is positively correlated with the mean value of the difference between adjacent data.
[0013] Preferably, after calculating the interference degree of the voiceprint information, the method further includes:
[0014] Normalize the degree of interference of voiceprint information.
[0015] Preferably, the method for obtaining the adhesion of the voiceprint information is:
[0016] Record any voiceprint waveform peak data segment as a target voiceprint waveform peak data segment, and obtain the endpoint connection distribution distance of the target voiceprint waveform peak data segment;
[0017] According to the endpoint connection distribution distance, the voiceprint information adhesion of the voiceprint waveform peak data segment is obtained; the voiceprint information adhesion is negatively correlated with the endpoint connection distribution distance.
[0018] Preferably, after calculating the sound information adhesion, the method further includes:
[0019] Normalize the sound information adhesion.
[0020] Preferably, the method for acquiring the clarity of the voiceprint information is:
[0021] For any voiceprint waveform peak data segment, the ratio of the voiceprint information adhesion degree of the voiceprint waveform peak data segment to the interference degree of the voiceprint information is used as the voiceprint information clarity of the voiceprint waveform peak data segment.
[0022] Preferably, the method for acquiring the complexity of the voiceprint scene is:
[0023] For any voiceprint waveform peak data segment, obtain the voiceprint information clarity difference change value between different voiceprint waveform peak data segments;
[0024] According to the voiceprint clarity difference variation value, the voiceprint scene complexity of the voiceprint waveform peak data segment is obtained; the voiceprint scene complexity is positively correlated with the voiceprint clarity difference variation value.
[0025] Preferably, after calculating the voiceprint scene complexity, the method further includes:
[0026] Normalize the complexity of voiceprint scenes.
[0027] Preferably, the identity recognition of financial data according to the complexity of the voiceprint scene includes the following specific methods:
[0028] According to the complexity of the voiceprint scenario, iteratively obtain the authentication environment interference value;
[0029] Iteratively determine the authentication environment interference value and perform authentication.
[0030] Preferably, the method for obtaining the identity authentication environment interference value is:
[0031] Get the before-after ratio of the voiceprint scene complexity;
[0032] According to the before-after ratio value, the identity authentication environment interference value is obtained; the identity authentication environment interference value is positively correlated with the before-after ratio value.
[0033] The beneficial effects of the technical solution of the present invention are: analyzing the smoothness of the waveform contour of the entire voiceprint data within the same voiceprint waveform peak data segment to obtain the degree of interference with the voiceprint information; wherein the degree of interference with the voiceprint information is used to describe the influence of the voiceprint information within the voiceprint waveform peak data segment on the voiceprint information of the external environment, so that the restriction on the expression of the voiceprint information of the core concern to be verified is more obvious; comparing the smooth adhesion of the waveform contour changes between adjacent voiceprint waveform peak data segments to obtain the voiceprint information adhesion degree; wherein the voiceprint information adhesion degree is used to describe the obvious difference in port connections between different voiceprint waveform peak data segments, thereby strengthening the correspondence between the changing rhythm of the voice content of the person to be verified and the data value; combining the degree of interference with the voiceprint information with the voiceprint information adhesion degree to obtain a clear voiceprint information. degree; wherein the clarity of the voiceprint information is used to describe the clarity of the voice content of the person to be verified, which increases the importance of the voice content of the person to be verified relative to other sound information content; according to the clarity of the voiceprint information, the information distribution contained in the voiceprint waveform peak data segment is analyzed to obtain the voiceprint scene complexity; wherein the voiceprint scene complexity is used to describe the complexity of the composition of the interfering sound components in the environment where the person to be verified is located, so that the influence of the sound information in the environment on the security of identity recognition is more obvious; the present invention improves the accuracy of judging the security of the verification environment, improves the accuracy of identity recognition, improves the efficiency of identity recognition, and improves the efficiency of real-time calling of financial data by authorized personnel by analyzing the comprehensive influence of the external environment and the sound information of other irrelevant persons on the verification environment where the verification personnel are located. BRIEF DESCRIPTION OF THE DRAWINGS
[0034] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0035] Figure 1 A flowchart of the steps of the financial data identity recognition method based on biometric recognition of the present invention;
[0036] Figure 2 It is a schematic diagram of the comparison of the voiceprint fluctuation transition of the present invention. DETAILED DESCRIPTION
[0037] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, features and effects of the financial data identity recognition method based on biometric recognition proposed by the present invention in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.
[0038] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
[0039] The specific scheme of the financial data identity recognition method based on biometric recognition provided by the present invention is described in detail below with reference to the accompanying drawings.
[0040] See also Figure 1 , which shows a flowchart of a method for identifying financial data based on biometrics provided by an embodiment of the present invention, the method comprising the following steps:
[0041] Step S001: Acquire several voiceprint waveform peak data segments of the person to be verified.
[0042] It should be noted that in the process of identity recognition by the system verification personnel's voice, the sound information recorded by the identity recognition system usually contains a large amount of sound information from the external environment and other irrelevant people, which causes the identity recognition system to be unable to effectively judge the security of the verification environment in which the verification personnel are located, thereby reducing the accuracy of identity recognition, reducing the efficiency of identity recognition, and reducing the efficiency of real-time call-up of financial data by authorized personnel.
[0043] It should be further explained that after the voice information of the verification personnel is digitally converted into a data sequence, the different voice contents of the verification personnel are mainly reflected through the different waveform peak structures presented by different data in the data sequence. Therefore, by analyzing the different band changes in the data sequence, the data sequence can be divided into several voiceprint waveform peak data segments, so as to accurately identify the security of the environmental content.
[0044] In a specific implementation of an embodiment of the present invention, a method for acquiring a voiceprint waveform peak data segment is as follows: a voice recording module of an identity recognition system is used to record the voice of a person to be verified, and a data sequence obtained by digitally converting the voice information of the person to be verified by a voice data conversion module is used as a voiceprint data sequence, wherein the voiceprint data sequence includes multiple voiceprint data.
[0045] Furthermore, the maximum value of all voiceprint data in the voiceprint data sequence is taken as voiceprint waveform peak data, and the minimum value of all voiceprint data in the voiceprint data sequence is taken as voiceprint trough data; taking any voiceprint waveform peak data as an example, on the left side of the voiceprint waveform peak data, the voiceprint trough data with the shortest distance from the voiceprint waveform peak data is taken as the starting port voiceprint data; on the right side of the voiceprint waveform peak data, the voiceprint trough data with the shortest distance from the voiceprint waveform peak data is taken as the ending port voiceprint data; the data segment composed of all voiceprint data between the starting port voiceprint data and the ending port voiceprint data is recorded as a voiceprint waveform peak data segment; and several voiceprint waveform peak data segments are obtained.
[0046] It is particularly noted that if in the voiceprint data sequence other than the divided voiceprint waveform peak data segment, there is a voiceprint waveform peak data with no voiceprint valley data on the left or right side, then the remaining data segment will be regarded as a voiceprint waveform peak data segment.
[0047] It should be noted that the identity recognition system in this embodiment includes four main modules: a voice input module, a voice data conversion module, a voice environment recognition module, and an identity recognition module. The function of the voice input module is to input the voice of the person to be verified and the sound of the surrounding environment; the function of the voice data conversion module is to digitize the sound information input by the voice input module into a digital data sequence; the function of the voice environment recognition module is to identify the security judgment of the environmental sound contained in the digital data sequence on the identity recognition, and push the digital data sequence with higher security to the identity recognition module; the function of the identity recognition module is to perform identity recognition on the digital data sequence.
[0048] So far, several voiceprint waveform peak data segments are obtained through the above method.
[0049] Step S002: Analyze the smoothness of the overall waveform contour of the voiceprint data within the same voiceprint waveform peak data segment to obtain the degree of interference of the voiceprint information of the voiceprint waveform peak data segment; compare the smoothness and adhesion of the waveform contour changes between adjacent voiceprint waveform peak data segments to obtain the adhesion of the voiceprint information of the voiceprint waveform peak data segment; combine the degree of interference of the voiceprint information with the adhesion of the voiceprint information to obtain the clarity of the voiceprint information of the voiceprint waveform peak data segment.
[0050] It should be noted that, ideally, the voiceprint data only contains the voice information content of the person to be verified, and the corresponding voiceprint data as a whole constitutes a relatively smooth waveform contour; in actual practice, in addition to the voice information content of the person to be verified, the voiceprint data will also contain an indefinite amount of environmental sound information, which will interfere with the expression of the voice information of the person to be verified, thereby affecting the smoothness of the formed waveform contour; therefore, by analyzing the smoothness of the overall waveform contour of the voiceprint data within the same voiceprint waveform peak data segment, the degree of interference of the voiceprint information of the voiceprint waveform peak data segment can be obtained.
[0051] Preferably, in some specific implementations of the embodiments of the present invention, the method for obtaining the interference degree of the voiceprint information is: for any voiceprint waveform peak data segment, obtain the mean difference of adjacent data of the voiceprint waveform peak data segment; according to the mean difference of adjacent data, obtain the interference degree of the voiceprint information of the voiceprint waveform peak data segment; the interference degree of the voiceprint information is positively correlated with the mean difference of adjacent data. As an example, the interference degree of the voiceprint information of the voiceprint waveform peak data segment can be calculated by the following formula:
[0052]
[0053] Where Q represents the interference degree of the voiceprint information in the voiceprint waveform peak data segment; N represents the number of all voiceprint data in the voiceprint waveform peak data segment; ΔA n-1,n Indicates the absolute value of the difference between the n-1th voiceprint data and the nth voiceprint data in the voiceprint waveform peak data segment; ΔA n,n+1 It represents the absolute value of the difference between the nth voiceprint data and the n+1th voiceprint data in the voiceprint waveform peak data segment; || represents taking the absolute value; norm[] represents the linear normalization function.
[0054] It should be noted that It represents the mean difference of adjacent data in the voiceprint waveform peak data segment. If the mean difference of adjacent data is larger, it means that the amplitude of voiceprint information change in the voiceprint waveform peak data segment is more obvious, and the corresponding waveform contour is less smooth. In summary, if the voiceprint information in the voiceprint waveform peak data segment is more disturbed, it means that the voiceprint information in the voiceprint waveform peak data segment is more affected by the voiceprint information of the external environment, which makes the expression of the core voiceprint information of the person to be verified more restricted.
[0055] It should be noted that, ideally, the voice information of the person to be verified has clear and regular changes in the pronunciation of words, which leads to regular changes in the peaks of the waveform contour constructed by the corresponding voiceprint data. At the same time, the transition connection between these different peaks will also show obvious adhesion, making the connection transition between the peaks smoother. Therefore, the smooth adhesion of the waveform contour changes between adjacent voiceprint waveform peak data segments can be compared to obtain the voiceprint information adhesion of the voiceprint waveform peak data segment. Please refer to Figure 2 , which shows a schematic diagram of voiceprint fluctuation transition comparison. The curve in the figure represents a waveform schematic curve constructed by part of the voiceprint waveform peak data. Box 1 represents a relatively obvious adhesion between the peaks, and box 2 represents an inconspicuous adhesion between the peaks.
[0056] Preferably, in some specific implementations of the embodiments of the present invention, the method for obtaining the voiceprint information adhesion is: record any voiceprint waveform peak data segment as the target voiceprint waveform peak data segment, obtain the endpoint connection distribution distance of the target voiceprint waveform peak data segment; obtain the voiceprint information adhesion of the voiceprint waveform peak data segment according to the endpoint connection distribution distance; the voiceprint information adhesion is negatively correlated with the endpoint connection distribution distance. The specific process is:
[0057] On the left side of the target voiceprint waveform peak data segment, the voiceprint waveform peak data segment with the smallest distance to the target voiceprint waveform peak data segment is recorded as the left adjacent voiceprint waveform peak data segment; on the right side of the target voiceprint waveform peak data segment, the voiceprint waveform peak data segment with the smallest distance to the target voiceprint waveform peak data segment is recorded as the right adjacent voiceprint waveform peak data segment. As an example, the voiceprint information adhesion of the target voiceprint waveform peak data segment can be calculated by the following formula:
[0058]
[0059] Wherein, B represents the adhesion of the voiceprint information of the target voiceprint waveform peak data segment; ΔA1 represents the absolute value of the difference between the first voiceprint data in the target voiceprint waveform peak data segment and the last voiceprint data in the left adjacent voiceprint waveform peak data segment; Δd1 represents the Euclidean distance of the voiceprint waveform peak data between the target voiceprint waveform peak data segment and the left adjacent voiceprint waveform peak data segment; ΔA2 represents the absolute value of the difference between the last voiceprint data in the target voiceprint waveform peak data segment and the first voiceprint data in the right adjacent voiceprint waveform peak data segment; Δd2 represents the Euclidean distance of the voiceprint waveform peak data between the target voiceprint waveform peak data segment and the right adjacent voiceprint waveform peak data segment; exp() represents an exponential function with a natural constant as the base. The embodiment adopts the exp(-x) model to present the inverse proportional relationship and normalization processing. x is the input of the model. The implementer can select the inverse proportional function and the normalization function according to the actual situation. The acquisition of the Euclidean distance is a well-known technology and will not be repeated in this embodiment.
[0060] It should be noted that ΔA1×Δd1+ΔA2×Δd2 represents the endpoint connection distribution distance of the target voiceprint waveform peak data segment. If the endpoint connection distribution distance is larger, it means that the difference in the connecting part of the waveform contour between the target voiceprint data segment and the left and right adjacent voiceprint waveform peak data segments is more obvious; in summary, if the voiceprint information adhesion of the target voiceprint waveform peak data segment is greater, it means that the port connection between the target voiceprint waveform peak data segment and the surrounding adjacent voiceprint waveform peak data segments is more different and less obvious, reflecting that the front connection of the sound information contained in different words between the target voiceprint waveform peak data segment and the adjacent voiceprint waveform peak data segments is smoother.
[0061] It should be noted that the degree of interference of voiceprint information reflects the degree to which the voice information of the person to be verified is interfered by the sounds of other objects in the environment and the voice information of other people. The adhesion of voiceprint information reflects the smoothness of the connection between the sound information contained in different words of the person to be verified. Both of them reflect the clarity of the voice information of the person to be verified to a certain extent. Therefore, the degree of interference of voiceprint information and the adhesion of voiceprint information can be combined to obtain the clarity of the voiceprint information of the peak data segment of the voiceprint waveform.
[0062] Preferably, in some specific implementations of the embodiments of the present invention, the method for obtaining the clarity of voiceprint information is as follows: for any voiceprint waveform peak data segment, the ratio of the voiceprint information adhesion degree of the voiceprint waveform peak data segment to the degree of interference of the voiceprint information is used as the clarity of the voiceprint information of the voiceprint waveform peak data segment. The clarity of voiceprint information of several voiceprint waveform peak data segments is obtained.
[0063] So far, the voiceprint information clarity of several voiceprint waveform peak data segments is obtained through the above method.
[0064] Step S003: Analyze the information distribution contained in the voiceprint waveform peak data segment according to the clarity of the voiceprint information to obtain the voiceprint scene complexity of the person to be verified; and perform identity recognition of financial data according to the voiceprint scene complexity.
[0065] It should be noted that after obtaining the clarity of the voiceprint information of each voiceprint waveform peak data segment, the clarity of the voice information of the person to be verified contained in the voiceprint waveform peak data segment can be further judged, so as to judge the specific distribution of the sound information of various machine objects and other people in the corresponding scene. Therefore, the information distribution contained in the voiceprint waveform peak data segment can be analyzed according to the clarity of the voiceprint information to obtain the complexity of the voiceprint scene of the person to be verified.
[0066] Preferably, in some specific implementations of the embodiments of the present invention, the method for obtaining the voiceprint scene complexity is: for any voiceprint waveform peak data segment, obtain the voiceprint information clarity difference change value between different voiceprint waveform peak data segments; according to the voiceprint clarity difference change value, obtain the voiceprint scene complexity of the person to be verified; the voiceprint scene complexity is positively correlated with the voiceprint clarity difference change value. As an example, the voiceprint scene complexity of the person to be verified can be calculated by the following formula:
[0067]
[0068] In the formula, C represents the complexity of the voiceprint scenario of the person to be verified; M represents the number of all voiceprint waveform peak data segments; R m Represents the clarity of the voiceprint information of the mth voiceprint waveform peak data segment; R m+1 Indicates the voiceprint clarity of the m+1th voiceprint waveform peak data segment; || indicates taking the absolute value.
[0069] It should be noted that |R m -R m+1 | represents the change value of the voiceprint information clarity difference. If the change value of the voiceprint information clarity difference is larger, it means that when the person to be verified enters the relevant sound information, the sound information contained in the environment is more complex; in summary, if the voiceprint scene complexity of the person to be verified is greater, it means that the person to be verified has more interference factors for identity recognition in the environment during the entire identity recognition process.
[0070] Preferably, in some specific implementations of the embodiments of the present invention, the process of identity recognition is: iteratively obtaining the identity authentication environment interference value according to the complexity of the voiceprint scene; iteratively determining the identity authentication environment interference value, and performing identity authentication. The method for obtaining the identity authentication environment interference value is: obtaining the before and after ratio value of the voiceprint scene complexity; obtaining the identity authentication environment interference value according to the before and after ratio value; the identity authentication environment interference value is positively correlated with the before and after ratio value. As an example, the identity authentication environment interference value can be obtained through the following process:
[0071] The voiceprint scene complexity of the person to be verified is used as the initial voiceprint scene complexity value; an iteration number T1 is preset, so that the person to be verified repeats the voice verification T1 times, and the voiceprint scene complexity of the person to be verified is obtained each time; the inverse proportional normalized value of the ratio between the voiceprint scene complexity of the person to be verified for the gth time and the voiceprint scene complexity of the person to be verified for the g+1th time is used as the before-after ratio value of the person to be verified for the g+1th time; the average of the before-after ratio values of all the persons to be verified before the g+1th time is used as the identity authentication environment interference value of the person to be verified for the g+1th time; and the identity authentication environment interference value of each person to be verified is obtained. In this embodiment, T1=5 is used as an example for description, and this embodiment is not specifically limited, wherein T1 can be determined according to the specific implementation situation; in this embodiment, the identity authentication environment interference value of the person to be verified for the first time is assumed to be consistent with the identity authentication environment interference value of the person to be verified for the second time; in addition, in this embodiment, the exp(-x) model is used by default to present the inverse proportional relationship and normalization processing, and x is the input of the model. The implementer can select the inverse proportional function and normalization function according to the actual situation.
[0072] As an example, identity identification can be performed through the following specific process:
[0073] Among all the identity verification environment interference values of the persons to be verified, if there is one identity verification environment interference value of the persons to be verified that is less than the preset identity verification environment interference value threshold value T2, then the voice environment recognition module determines that the environment of the persons to be verified is safe, and pushes the corresponding voiceprint data sequence to the identity recognition module for identity recognition. This embodiment is described by taking T2=0.45 as an example, and this embodiment is not specifically limited, and T2 can be determined according to the specific implementation situation.
[0074] At this point, this embodiment is completed.
[0075] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A financial data identification method based on biometric identification, characterized in that: The method comprises the following steps: Acquire a plurality of voiceprint waveform peak data segments of the person to be verified, wherein the voiceprint waveform peak data segments contain a plurality of voiceprint data; Analyze the smoothness of the overall waveform profile of the voiceprint data in the same voiceprint waveform peak data segment to obtain the interference degree of the voiceprint information of the voiceprint waveform peak data segment; compare the smooth adhesion of the waveform profile changes between adjacent voiceprint waveform peak data segments to obtain the adhesion degree of the voiceprint information of the voiceprint waveform peak data segment; combine the interference degree of the voiceprint information with the adhesion degree of the voiceprint information to obtain the clarity of the voiceprint information of the voiceprint waveform peak data segment; Analyze the information distribution contained in the voiceprint waveform peak data segment according to the clarity of the voiceprint information to obtain the voiceprint scene complexity of the person to be verified; perform identity recognition of financial data based on the voiceprint scene complexity; The method for obtaining the interference degree of the voiceprint information is as follows: For any voiceprint waveform peak data segment, obtain the mean value of the adjacent data differences of the voiceprint waveform peak data segment; According to the mean value of the difference between adjacent data, the interference degree of the voiceprint information of the voiceprint waveform peak data segment is obtained; the interference degree of the voiceprint information is positively correlated with the mean value of the difference between adjacent data; The method for obtaining the adhesion of the voiceprint information is: Record any voiceprint waveform peak data segment as a target voiceprint waveform peak data segment, and obtain the endpoint connection distribution distance of the target voiceprint waveform peak data segment; According to the endpoint connection distribution distance, the voiceprint information adhesion of the voiceprint waveform peak data segment is obtained; the voiceprint information adhesion is negatively correlated with the endpoint connection distribution distance; The method for obtaining the clarity of the voiceprint information is: For any voiceprint waveform peak data segment, the ratio of the voiceprint information adhesion degree of the voiceprint waveform peak data segment to the degree of interference of the voiceprint information is used as the voiceprint information clarity of the voiceprint waveform peak data segment; The method for obtaining the complexity of the voiceprint scene is: For any voiceprint waveform peak data segment, obtain the voiceprint information clarity difference change value between different voiceprint waveform peak data segments; According to the voiceprint clarity difference variation value, the voiceprint scene complexity of the voiceprint waveform peak data segment is obtained; the voiceprint scene complexity is positively correlated with the voiceprint clarity difference variation value; The specific method of performing identity recognition of financial data based on the complexity of voiceprint scenarios includes: According to the complexity of the voiceprint scenario, iteratively obtain the authentication environment interference value; Iteratively determine the authentication environment interference value and perform authentication; The method for obtaining the identity authentication environment interference value is: Get the before-after ratio of the voiceprint scene complexity; According to the before-after ratio value, the identity authentication environment interference value is obtained; the identity authentication environment interference value is positively correlated with the before-after ratio value.
2. The method for identifying financial data based on biometrics according to claim 1, characterized in that: After calculating the interference degree of the voiceprint information, it also includes: Normalize the degree of interference of voiceprint information.
3. The method for identifying financial data based on biometrics according to claim 1, characterized in that: After calculating the sound information adhesion, it also includes: Normalize the sound information adhesion.
4. The method for identifying financial data based on biometrics according to claim 1, characterized in that: After calculating the complexity of the voiceprint scene, it also includes: Normalize the complexity of voiceprint scenes.
Citation Information
Patent Citations
Phonetic feature processing method of voiceprint identification in noise environment
CN105679312A
Authentication method, terminal equipment and computer readable storage medium
CN109256138A
Identity recognition method and device
CN110880325A
Voiceprint recognition method and device, voiceprint verification method and device, computing equipment and medium
CN111199741A
Identity verification method and device, server and computer readable storage medium
CN115358749A
