Multidimensional fusion-based forged voice intelligent detection system

Abstract

The invention relates to the technical field of artificial intelligence, and particularly discloses a forged voice intelligent detection system based on multi-dimensional fusion, and the system comprises an acquisition module which is used for determining the importance score of target features, and screening to-be-selected features from the target features according to the importance score; the feature optimization module is used for establishing a forged voice detection model and determining a performance score of the forged voice detection model after training; sorting the performance scores, and determining a target performance score and a target set according to a performance score difference value; the feature selection module is used for determining a new training set and repeating the steps to obtain a sorting position interval and a target set; and iteration is carried out until the performance score meets a preset requirement, and standard features are determined and used for detecting the forged voice. According to the invention, the number of features required by forged voice detection can be reduced, and the detection efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a forged voice intelligent detection system based on multi-dimensional fusion. Background Art

[0002] Fake voice refers to audio that is artificially synthesized or modified to imitate real human voices. It is usually used to impersonate other people's voices or create fake voice content. Fake voice technology is widely used in fields such as voice assistants, but it may also bring security risks such as fraud and identity impersonation.

[0003] With the rapid development and widespread application of artificial intelligence technology, the technology of forged voice generation is also constantly improving, gradually approaching real voice from timbre, emotion to subtle pronunciation details. This has brought huge challenges to traditional voice detection methods. In some high-risk scenarios, such as identity authentication and voice transmission of important information, forged voice may cause serious security problems.

[0004] The detection method of forged speech usually starts with multi-dimensional information such as speech spectrum, acoustic features, and biometric features. The forged speech recognition model is constructed through multi-dimensional features to identify forgery traces and distinguish between real and fake speech content. However, too many feature inputs often increase the complexity of the model, causing the model to overfit or excessive computational overhead, which in turn affects the detection efficiency in practical applications. Therefore, how to select effective features and build an efficient and accurate forged speech detection model is one of the key issues in current research. Summary of the invention

[0005] The purpose of the present invention is to provide a forged voice intelligent detection system based on multi-dimensional fusion to solve the following technical problems:

[0006] Too many feature inputs often increase the complexity of the model, causing overfitting or excessive computational overhead, which in turn affects the detection efficiency in practical applications.

[0007] The purpose of the present invention can be achieved through the following technical solutions:

[0008] A forged speech intelligent detection system based on multi-dimensional fusion, comprising:

[0009] Acquisition module: pre-processing the features used to detect forged speech to obtain target features, determining the importance score P of the target features based on the gradient boosting tree model, sorting the target features in order of importance score from small to large, and taking the last 40% of the target features in the sorting as candidate features;

[0010] Feature optimization module: establish a forged voice detection model based on a deep learning model, train the forged voice detection model based on a training set, the training set is obtained based on the selected features, determine the performance indicators of the forged voice detection model after training, the performance indicators include accuracy A1, precision A2 and recall A3, and determine the performance score A=A1+A2+A3;

[0011] The performance scores are sorted in descending order, starting from the first performance score in the sorting, and the performance score difference Ai-1=Ai-1-Ai is calculated, where Ai represents the performance score of the i-th position in the sorting. When the performance score difference Ai-1≥Ays, the performance score in the sorting position interval [1, i-1] is used as the target performance score, and Ays represents the preset performance score difference threshold;

[0012] Determine the union of the features to be selected in the training set corresponding to the target performance score, and use it as the target set B;

[0013] Feature selection module: determine the training set based on the selected features in the target set B, determine the performance score, remove the performance score less than or equal to Ai-1, and determine the sorting position interval and the target set B;

[0014] Repeat the above steps until the performance scores corresponding to the target set B are all smaller than the performance scores at the end of the sorting position interval in the previous iteration, determine the training set and performance scores corresponding to the target set and B in the previous iteration, use the selected features in the training set corresponding to the maximum performance score as standard features, and detect forged speech based on the standard features.

[0015] As a further solution of the present invention: in the acquisition module, the process of determining the importance score of the initial feature specifically includes:

[0016] Establishing a database, wherein a preset number of samples are stored in the database, wherein each sample includes an initial feature and a detection label, and the detection label includes real and fake ones;

[0017] A gradient boosting tree model is established based on the gradient boosting tree framework, and the gradient boosting tree model is trained based on the database. The number of times C that a single initial feature is used as a split node during the training process is determined, and the importance score P = C / Ctot of the initial feature is calculated, where Ctot represents the total number of split nodes.

[0018] As a further solution of the present invention: in the acquisition module, the process of determining the initial features specifically includes:

[0019] Using a feature for detecting fake speech as a first feature;

[0020] Standardize the first feature of the numerical type to eliminate the dimension;

[0021] Encode the non-numeric first feature by one-hot encoding;

[0022] The outliers in the first feature are removed by using the isolation forest algorithm.

[0023] As a further solution of the present invention: in the feature optimization module, performance scores of the same size are in the same position in the ranking.

[0024] As a further solution of the present invention: in the feature optimization module, when there is no performance score difference greater than or equal to the performance score difference threshold, starting from the first performance score in the sorting, calculate the first performance score difference ADYi-1=A1-Ai-1, and when the first performance score difference ADYi-1≥η*Ays, use [1, i-1] as the sorting position interval, η is a preset correction coefficient and η>1.

[0025] As a further solution of the present invention: in the feature selection module, when there are two or more performance scores that are the same and are the maximum performance scores, the candidate features in the training set with the smallest number of candidate features are used as standard features.

[0026] As a further solution of the present invention: in the feature optimization module, the process of determining the training set specifically includes:

[0027] The total number n of the features to be selected is determined, and f features to be selected are arbitrarily selected and combined to obtain a training set, where f∈[1,n] and f is an integer.

[0028] As a further solution of the present invention: a single training set is copied, and the number of copies is a preset value.

[0029] The beneficial effects of the present invention are as follows: first, the target features are sorted in order from low to high, and the features at the bottom 40% of the ranking are taken as the selected features. Through preliminary screening, the dimension of the feature set is reduced, so that the subsequent feature optimization is more focused on relatively influential features, thereby improving the efficiency of the detection model; the feature combination is scored based on the performance index, and the feature combination with the best performance is gradually screened. By selecting the performance mutation point, it can converge to the feature combination with higher performance faster, saving the computing cost; the union of the selected features in the training set corresponding to the target performance score is obtained as the target set B, and the optimization process is repeated to screen out the feature set whose current iteration performance is lower than the end point of the previous iteration performance Combine and retain high-performance features; finally, when the performance scores corresponding to the target set B are all lower than the performance of the previous iteration, the target set B of the previous iteration is taken as the final feature set, in which the new sorting position interval and the target set are determined as one iteration; if the number of selected features in two adjacent iterations is the same, a warning message is sent; through multiple rounds of iterations, it gradually converges to the optimal feature set to ensure that the selected features can not only effectively detect forged speech, but also improve the performance of the model, and avoid the influence of redundant features on the model effect; finally, after multiple rounds of optimization, the selected features are selected to select a combination of features with high importance and strong correlation with performance indicators to form standard features for the construction of forged speech detection. This method not only takes into account the importance score of the features, but also integrates multiple performance indicators, making the model more effective and more reliable in detecting forged speech. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] The present invention will be further described below in conjunction with the accompanying drawings.

[0031] Figure 1 It is a flow chart of a forged voice intelligent detection system based on multi-dimensional fusion of the present invention. DETAILED DESCRIPTION

[0032] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0033] See also Figure 1 As shown, the present invention is a forged voice intelligent detection system based on multi-dimensional fusion, comprising:

[0034] Acquisition module: pre-processing the features used to detect forged speech to obtain target features, determining the importance score P of the target features based on the gradient boosting tree model, sorting the target features in order of importance score from small to large, and taking the last 40% of the target features in the sorting as candidate features;

[0035] Feature optimization module: establish a forged voice detection model based on a deep learning model, train the forged voice detection model based on a training set, the training set is obtained based on the selected features, determine the performance indicators of the forged voice detection model after training, the performance indicators include accuracy A1, precision A2 and recall A3, and determine the performance score A=A1+A2+A3;

[0036] The performance scores are sorted in descending order, starting from the first performance score in the sorting, and the performance score difference Ai-1=Ai-1-Ai is calculated, where Ai represents the performance score of the i-th position in the sorting. When the performance score difference Ai-1≥Ays, the performance score in the sorting position interval [1, i-1] is used as the target performance score, and Ays represents the preset performance score difference threshold;

[0037] Determine the union of the features to be selected in the training set corresponding to the target performance score, and use it as the target set B;

[0038] Feature selection module: determine the training set based on the selected features in the target set B, determine the performance score, remove the performance score less than or equal to Ai-1, and determine the sorting position interval and the target set B;

[0039] Repeat the above steps until the performance scores corresponding to the target set B are all smaller than the performance scores at the end of the sorting position interval in the previous iteration, determine the training set and performance scores corresponding to the target set and B in the previous iteration, use the selected features in the training set corresponding to the maximum performance score as standard features, and detect forged speech based on the standard features.

[0040] It should be noted that the target features are sorted in order from low to high, and the features at the bottom 40% of the ranking are taken as the selected features. Through preliminary screening, the dimension of the feature set is reduced, so that the subsequent feature optimization is more concentrated on relatively influential features, thereby improving the efficiency of the detection model; the feature combination is scored based on the performance index, and the feature combination with the best performance is gradually screened. By selecting the performance mutation point, it is possible to converge to the feature combination with higher performance faster, saving the computing cost; the union of the selected features in the training set corresponding to the target performance score is obtained as the target set B, and the optimization process is repeated to screen out the feature set whose performance in the current iteration is lower than the performance end point of the previous iteration, and retain the high-performance features; finally, when the performance scores corresponding to the target set B are all lower than the performance of the previous iteration, the target set B of the previous iteration is taken as the final feature set, in which the new sorting position interval and the target set are determined as one iteration; through multiple rounds of iterations, the optimal feature set is gradually converged to ensure that the selected features can not only effectively detect forged speech, but also improve the performance of the model, and avoid the influence of redundant features on the model effect;

[0041] In another preferred embodiment of the present invention, in the acquisition module, the process of determining the importance score of the initial feature specifically includes:

[0042] Establishing a database, wherein a preset number of samples are stored in the database, wherein each sample includes an initial feature and a detection label, and the detection label includes real and fake ones;

[0043] A gradient boosting tree model is established based on the gradient boosting tree framework, and the gradient boosting tree model is trained based on the database. The number of times C that a single initial feature is used as a split node during the training process is determined, and the importance score P = C / Ctot of the initial feature is calculated, where Ctot represents the total number of split nodes.

[0044] It is worth noting that in the gradient boosting tree model, the feature importance score is usually measured based on the frequency of occurrence of split nodes. A feature with more split nodes has a greater impact on the model's prediction and is usually more important. The importance score of each feature is calculated by using the ratio of the number of splits to the total number of splits. This method can quantify the contribution of each feature to the final classification result (i.e., determining whether the speech is real or fake).

[0045] In another preferred embodiment of the present invention, in the acquisition module, the process of determining the initial features specifically includes:

[0046] Using a feature for detecting fake speech as a first feature;

[0047] Standardize the first feature of the numerical type to eliminate the dimension;

[0048] Encode the non-numeric first feature by one-hot encoding;

[0049] The outliers in the first feature are removed by using the isolation forest algorithm.

[0050] In another preferred embodiment of the present invention, in the feature optimization module, performance scores of the same size are in the same position in the ranking.

[0051] In another preferred embodiment of the present invention, in the feature optimization module, when there is no performance score difference greater than or equal to the performance score difference threshold, starting from the first performance score in the sorting, the first performance score difference ADYi-1=A1-Ai-1 is calculated, and when the first performance score difference ADYi-1≥η*Ays, [1, i-1] is used as the sorting position interval, η is a preset correction coefficient and η>1.

[0052] It should be noted that in the original feature optimization process, if the difference between the sorted performance scores (i.e., the difference between adjacent performance scores) does not reach the preset difference threshold Ays, it is impossible to find a suitable performance score position interval according to the standard steps; the above method can ensure that when the performance score difference between the features is small, the interval with large performance changes can still be found by relaxing the standards, without falling into an infinite loop or failing to complete feature selection due to strict difference restrictions.

[0053] In another preferred embodiment of the present invention, in the feature selection module, when there are two or more performance scores that are the same and are the maximum performance scores, the candidate features in the training set with the smallest number of candidate features are used as standard features.

[0054] It is understandable that, under the premise of ensuring optimal performance, selecting the combination with the least number of features can make the model more concise, while reducing possible redundant features and improving the efficiency and generalization ability of the model; under the condition of consistent performance, selecting the combination with the least number of features can reduce the complexity of the model, avoid overfitting, and improve the operating efficiency of the model. This strategy conforms to the principle of "simplicity first" in feature selection.

[0055] In another preferred embodiment of the present invention, in the feature optimization module, the process of determining the training set specifically includes:

[0056] The total number n of the features to be selected is determined, and f features to be selected are arbitrarily selected and combined to obtain a training set, where f∈[1,n] and f is an integer.

[0057] It is worth noting that among the n candidate features, f features are arbitrarily selected for combination to generate different feature subsets, and these subsets are used to generate corresponding training sets; by generating multiple different feature combinations, the model is trained and tested on training sets of different feature combinations to ensure that the best feature combination can be found.

[0058] In another preferred embodiment of the present invention, a single training set is replicated, and the replication multiple is a preset value.

[0059] It is worth noting that in the training sets of some feature combinations, there may be insufficient number of samples, especially for small-scale feature combinations. By copying, the number of samples in the training set can be increased, thereby balancing the data and ensuring that the model does not have overfitting or underfitting problems due to insufficient data. It also helps the model to have a deeper understanding of specific feature combinations during training, thereby improving the accuracy and robustness of the model.

[0060] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A forged voice intelligent detection system based on multi-dimensional fusion, characterized in that: include: Acquisition module: pre-processing the features used to detect forged speech to obtain target features, determining the importance score P of the target features based on the gradient boosting tree model, sorting the target features in order of importance score from small to large, and taking the last 40% of the target features in the sorting as candidate features; Feature optimization module: establish a forged voice detection model based on a deep learning model, train the forged voice detection model based on a training set, the training set is obtained based on the selected features, determine the performance indicators of the forged voice detection model after training, the performance indicators include accuracy A1, precision A2 and recall A3, and determine the performance score A=A1+A2+A3; The performance scores are sorted in descending order, starting from the first performance score in the sorting, and the performance score difference Ai-1=Ai-1-Ai is calculated, where Ai represents the performance score of the i-th position in the sorting. When the performance score difference Ai-1≥Ays, the performance score in the sorting position interval [1, i-1] is used as the target performance score, and Ays represents the preset performance score difference threshold; Determine the union of the features to be selected in the training set corresponding to the target performance score, and use it as the target set B; Feature selection module: determine the training set based on the selected features in the target set B, determine the performance score, remove the performance score less than or equal to Ai-1, and determine the sorting position interval and the target set B; Repeat the above steps until the performance scores corresponding to the target set B are all smaller than the performance scores at the end of the sorting position interval in the previous iteration, determine the training set and performance scores corresponding to the target set and B in the previous iteration, use the selected features in the training set corresponding to the maximum performance score as standard features, and detect forged speech based on the standard features.

2. According to claim 1, a forged voice intelligent detection system based on multi-dimensional fusion is characterized in that: In the acquisition module, the process of determining the importance score of the initial feature specifically includes: Establishing a database, wherein a preset number of samples are stored in the database, wherein each sample includes an initial feature and a detection label, and the detection label includes real and fake ones; A gradient boosting tree model is established based on the gradient boosting tree framework, and the gradient boosting tree model is trained based on the database. The number of times C that a single initial feature is used as a split node during the training process is determined, and the importance score P = C / Ctot of the initial feature is calculated, where Ctot represents the total number of split nodes.

3. According to the multi-dimensional fusion-based forged voice intelligent detection system of claim 1, it is characterized in that: In the acquisition module, the process of determining the initial features specifically includes: Using a feature for detecting fake speech as a first feature; Standardize the first feature of the numerical type to eliminate the dimension; Encode the non-numeric first feature by one-hot encoding; The outliers in the first feature are removed by using the isolation forest algorithm.

4. According to the multi-dimensional fusion-based forged voice intelligent detection system of claim 1, it is characterized in that: In the feature optimization module, performance scores of the same size are in the same position in the ranking.

5. According to the multi-dimensional fusion-based forged voice intelligent detection system of claim 1, it is characterized in that: In the feature optimization module, when there is no performance score difference greater than or equal to the performance score difference threshold, starting from the first performance score in the sorting, calculate the first performance score difference ADYi-1=A1-Ai-1, and when the first performance score difference ADYi-1≥η*Ays, use [1, i-1] as the sorting position interval, η is the preset correction coefficient and η>1.

6. The forged voice intelligent detection system based on multi-dimensional fusion according to claim 1 is characterized in that: In the feature selection module, when there are two or more performance scores that are the same and are the maximum performance scores, the candidate features in the training set with the smallest number of candidate features are used as standard features.

7. The forged voice intelligent detection system based on multi-dimensional fusion according to claim 1 is characterized in that: In the feature optimization module, the process of determining the training set specifically includes: The total number n of the features to be selected is determined, and f features to be selected are arbitrarily selected and combined to obtain a training set, where f∈[1,n] and f is an integer.

8. The forged voice intelligent detection system based on multi-dimensional fusion according to claim 7 is characterized in that: A single training set is replicated, and the replication multiple is a preset value.

Images