Voiceprint recognition model training method, voiceprint recognition method and related equipment

Abstract

The present application provides a voiceprint recognition model training method, a voiceprint recognition method and related equipment. The voiceprint recognition model training method comprises: inputting voiceprint data into a model to be trained in sequence for multiple iterations of training to obtain a voiceprint recognition model; wherein the model to be trained comprises L layers of network layers, L being an integer greater than 1; in the process of the Nth iteration training, the following steps are included: the voiceprint data is passed through the L layers of network layers in sequence for feature mapping, and a voiceprint vector is output; when the output value corresponding to the voiceprint vector meets a preset condition, the model to be trained after the Nth iteration training is determined as a voiceprint recognition model, wherein N is a positive integer. In this way, the utilization rate of samples is improved, and the accuracy of the trained voiceprint recognition model is improved.

Description

Technical Field

[0001] The present application relates to the technical field of voiceprint recognition, and in particular to a voiceprint recognition model training method, a voiceprint recognition method and related equipment. Background Art

[0002] As a reliable voiceprint feature authentication technology, voiceprint recognition has broad application prospects in many fields and scenarios such as identity authentication and security verification. However, voice is easily affected by various external factors such as noise environment, emotions, physical condition and internal factors. Therefore, improving the accuracy of voiceprint recognition is of great practical significance. At present, in the training process of voiceprint recognition model, the utilization rate of sample data is low, resulting in low accuracy of the trained voiceprint recognition model. Summary of the invention

[0003] The embodiments of the present application provide a voiceprint recognition model training method, a voiceprint recognition method and related equipment to solve the problem of low accuracy of the trained voiceprint recognition model.

[0004] In order to solve the above technical problems, this application is implemented as follows:

[0005] In a first aspect, an embodiment of the present application provides a voiceprint recognition model training method, comprising:

[0006] Input the voiceprint data into the model to be trained in sequence for multiple iterations of training to obtain a voiceprint recognition model;

[0007] The model to be trained includes L network layers, where L is an integer greater than 1; during the Nth iterative training process, the following steps are included:

[0008] The voiceprint data is sequentially passed through the L-layer network layers for feature mapping, and a voiceprint vector is output;

[0009] When the output value corresponding to the voiceprint vector meets a preset condition, the model to be trained after the Nth iterative training is determined as the voiceprint recognition model, where N is a positive integer.

[0010] It can be seen that in this embodiment, the voiceprint data is sequentially subjected to feature mapping through L layers of network layers, and a voiceprint vector is output, that is, the sample is sequentially subjected to multiple feature mappings, thereby improving the utilization rate of the sample. At the same time, since the feature vector obtained by feature mapping of the next layer of network layers needs to be based on the feature vector output by the previous layer of network layers, multiple feature vectors can be obtained through layer-by-layer feature mapping of L layers of network layers, and the trained voiceprint recognition model can recognize the above multiple feature vectors, thereby improving the accuracy of the voiceprint recognition model finally trained to recognize the voiceprint data. In addition, since the voiceprint data can be converted into voiceprint vectors through layer-by-layer feature mapping of L layers of network layers, the voiceprint vectors can be more easily and accurately recognized by the voiceprint recognition model, thereby further improving the accuracy of the voiceprint recognition model finally trained to recognize the voiceprint.

[0011] In a second aspect, an embodiment of the present application provides a voiceprint recognition method, the method comprising:

[0012] Obtain target voiceprint data of the user to be identified;

[0013] Input the target voiceprint data into the L-layer network layer included in the voiceprint recognition model for feature mapping, and output a target voiceprint vector, where L is an integer greater than 1;

[0014] Voiceprint recognition is performed according to the target voiceprint vector.

[0015] It can be seen that in the embodiment of the present application, the target voiceprint data is input into the L-layer network layer included in the voiceprint recognition model for feature mapping, and the target voiceprint vector is output. In this way, through the layer-by-layer feature mapping of the L-layer network layer, a variety of feature vectors can be obtained, and the voiceprint recognition model can recognize the above-mentioned multiple feature vectors, thereby improving the accuracy of the voiceprint recognition model in recognizing the voiceprint data. In addition, because the target voiceprint data can be converted into a target voiceprint vector through the layer-by-layer feature mapping of the L-layer network layer, and the target voiceprint vector is more easily and accurately recognized by the voiceprint recognition model, thereby further improving the accuracy of the voiceprint recognition model in recognizing the target voiceprint.

[0016] In a third aspect, the present application also provides a voiceprint recognition model training device, including:

[0017] The training module is used to input the voiceprint data into the model to be trained in sequence for multiple iterations of training to obtain a voiceprint recognition model;

[0018] The model to be trained includes L network layers, where L is an integer greater than 1; during the Nth iterative training process, the following steps are included:

[0019] The voiceprint data is sequentially passed through the L-layer network layers for feature mapping, and a voiceprint vector is output;

[0020] When the output value corresponding to the voiceprint vector meets a preset condition, the model to be trained after the Nth iterative training is determined as the voiceprint recognition model, where N is a positive integer.

[0021] In a fourth aspect, the present application also provides a voiceprint recognition device, including:

[0022] An acquisition module is used to acquire the target voiceprint data of the user to be identified;

[0023] An output module, used for inputting the target voiceprint data into the L-layer network layer included in the voiceprint recognition model for feature mapping, and outputting a target voiceprint vector, where L is an integer greater than 1;

[0024] A voiceprint recognition module is used to perform voiceprint recognition according to the target voiceprint vector.

[0025] In a fifth aspect, an embodiment of the present application further provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps in the above-mentioned voiceprint recognition model training method or voiceprint recognition method when executing the computer program.

[0026] In a sixth aspect, an embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps in the above-mentioned voiceprint recognition model training method or voiceprint recognition method are implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments of the present application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0028] Figure 1 It is a flow chart of a voiceprint recognition model training method provided in an embodiment of the present application;

[0029] Figure 2 is a structural diagram of a model to be trained provided in an embodiment of the present application;

[0030] Figure 3 is a schematic diagram of the structure of another model to be trained provided in an embodiment of the present application;

[0031] Figure 4is a flow chart of a voiceprint recognition method provided by an embodiment of the present application;

[0032] Figure 5 It is a structural schematic diagram of a voiceprint recognition model training device provided in an embodiment of the present application;

[0033] Figure 6 It is a structural schematic diagram of a voiceprint recognition device provided in an embodiment of the present application;

[0034] Figure 7 It is a structural schematic diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0035] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0036] See also Figure 1 , Figure 1 is a flow chart of the voiceprint recognition model training method provided in the embodiment of the present application. Figure 1 As shown, the following steps are included:

[0037] Step 101: input the voiceprint data into the model to be trained in sequence for multiple iterations of training to obtain a voiceprint recognition model.

[0038] The model to be trained includes L network layers, where L is an integer greater than 1; during the Nth iterative training process, the following steps are included: the voiceprint data is sequentially passed through the L network layers for feature mapping to output a voiceprint vector; when the output value corresponding to the voiceprint vector meets a preset condition, the model to be trained after the Nth iterative training is determined as a voiceprint recognition model, where N is a positive integer.

[0039] The voiceprint data may be, for example, a voiceprint feature vector. The method for obtaining the voiceprint data is not limited here. For example, the voiceprint data may be directly obtained from the sample pool, that is, the voiceprint data is the pre-existing voiceprint data in the sample pool; in addition, the voiceprint data may also be obtained by extracting the audio.

[0040] In addition, data amplification may be performed on the sample data in the sample pool, and the data amplification includes at least one of adding noise, speeding up the speech speed, increasing data disturbance, and the like.

[0041] In addition, the sample data after the above data amplification can be divided into a training set and a test set, and the above voiceprint data can be the data in the training set, wherein the ratio of the training set to the test set is not limited here, for example: it can be 8:2.

[0042] In addition, the specific type of voiceprint data is not limited here. As an optional implementation, the voiceprint data is fbank80 features, and the recognition accuracy of fbank80 features is relatively high, thereby improving the accuracy and training efficiency of voiceprint recognition model training. At the same time, fbank80 features are features that tend to be low-level, and they reflect voice information more richly, and can effectively match the powerful processing capabilities of the voiceprint recognition model for features, especially when the voiceprint recognition model is a transformer model, the fbank80 features can better match the powerful processing capabilities of the voiceprint recognition model for features.

[0043] It should be noted that the specific content of the preset conditions is not limited here.

[0044] As an optional implementation, the preset condition includes: the output value corresponding to the voiceprint vector is less than a preset value. At this time, it can be considered that the output value of the voiceprint vector meets the preset condition, that is, the model to be trained has converged, so the model to be trained at this time can be determined as a voiceprint recognition model.

[0045] Among them, the output value corresponding to the voiceprint vector can be a specific numerical value, and the output value and the voiceprint vector have a one-to-one correspondence, and the above output value can be obtained by converting the classifier. The specific steps can be: outputting the output value corresponding to the voiceprint vector in the voiceprint vector output classifier, and the above output value can be understood as: the classification prediction probability of the model to be trained and the classifier for the voiceprint data.

[0046] It should be noted that the above-mentioned classifier can be embedded in the voiceprint recognition model, that is, the classifier is a part of the voiceprint recognition model. Of course, the above-mentioned classifier can also be connected to the voiceprint recognition model, that is, it can serve as the next-level network layer structure connected to the voiceprint recognition model.

[0047] According to actual practice, the voiceprint data may also correspond to an actual classification value, and the actual classification value may also be referred to as an actual value, and the actual value may exist in identification information, and the identification information may be identification information of the voiceprint data.

[0048] Therefore, as another optional implementation, the preset condition includes: the difference between the output value corresponding to the voiceprint vector and the actual value corresponding to the voiceprint data is within a preset range.

[0049] In this embodiment, when the difference between the output value corresponding to the voiceprint vector and the actual value corresponding to the voiceprint data is within a preset range, it can be considered that the classification prediction probability of the trained model for the voiceprint data is close to the actual classification value, and the trained model is available, that is, it can be judged that the trained model has converged, so that the trained model after the Nth iterative training can be determined as the voiceprint recognition model, and the accuracy of the trained voiceprint recognition model can be made higher.

[0050] Among them, when the voiceprint data passes through L layers of network layers for feature mapping in sequence, the voiceprint data needs to be feature mapped in each layer of the network layer, and the feature mapping in a certain layer of the network layer needs to be implemented based on the features extracted in the previous layer of the network layer. In this way, through the layer-by-layer feature mapping of L layers of network layers, a variety of feature vectors can be obtained, and the trained voiceprint recognition model can recognize the above-mentioned multiple feature vectors, thereby improving the accuracy of the voiceprint recognition model finally trained to recognize the voiceprint data. In addition, because the voiceprint data can be converted into voiceprint vectors through the layer-by-layer feature mapping of L layers of network layers, and the voiceprint vectors are more easily and accurately recognized by the voiceprint recognition model, the accuracy of the voiceprint recognition model finally trained to recognize the voiceprint is further improved.

[0051] The specific number of L network layers is not limited here. For example, L can be 4, that is, it can include 4 network layers. In addition, L can also be an integer greater than 4.

[0052] As an optional implementation, the L-layer network layer includes: a first network layer, a second network layer, a third network layer and a fourth network layer connected in sequence; the step of performing feature mapping on the voiceprint data through the L-layer network layer in sequence and outputting a voiceprint vector includes:

[0053] Inputting the voiceprint data into the first network layer for feature mapping, and outputting a first feature vector;

[0054] Inputting the first feature vector into the second network layer for feature mapping, and outputting a second feature vector;

[0055] Inputting the second feature vector into the third network layer for feature mapping, and outputting a third feature vector;

[0056] Inputting the third feature vector into the fourth network layer for feature mapping, and outputting the voiceprint vector;

[0057] Among them, the dimensions of the first feature vector, the second feature vector, the third feature vector and the voiceprint vector increase in sequence.

[0058] Among them, the first eigenvector can be called a shallow feature, and the shallow feature can refer to the features that tend to be biased towards the bottom layer, mainly including bottom-level features such as volume, timbre, and pitch.

[0059] The second eigenvector can be called shallow-mid-level features, and shallow-mid-level features are features mapped on basic eigenvectors such as volume, timbre, and pitch. Compared with shallow-level features, they are more focused on voiceprint information, that is, the above information is simplified and closer to voiceprint information, but it is still biased towards the underlying volume, timbre and other feature information, but it is already moving closer to voiceprint information (that is, the voiceprint vector mentioned later).

[0060] The third eigenvector can be called mid-to-high-level features. Compared with shallow features and shallow mid-level features, mid-to-high-level features have gradually approached voiceprint features. Mid-to-high-level features can be considered as distinguishing features that are discernible to the human ear and deep network mapping features. The human ear can no longer distinguish mid-to-high-level features (the human ear is naturally insensitive to voiceprint information, that is, the human ear has a poor ability to distinguish whether it is a person based on sound). The third network layer can obtain a better voiceprint semantic feature mapping relationship by extracting features at a higher level than the human ear, but the third eigenvector extracted by the third network layer cannot yet be expressed as a voiceprint feature.

[0061] Voiceprint vectors can also be called high-level features. High-level features are the highest-level features output by the entire model to be trained. Voiceprint vectors are features output based on feature mapping of shallow features, shallow-middle features, and middle-high-level features. Voiceprint vectors can be better recognized by machines (such as classifiers), resulting in higher recognition accuracy.

[0062] From the above, it can be seen that the dimensions of the first eigenvector, the second eigenvector, the third eigenvector and the voiceprint vector increase successively, that is to say: the vector complexity and the diversity of the information represented by the first eigenvector, the second eigenvector, the third eigenvector and the voiceprint vector are constantly increasing.

[0063] In this embodiment, since the feature vector obtained by feature mapping of the next network layer needs to be based on the feature vector output by the previous network layer, the feature mapping of the first network layer, the second network layer, the third network layer and the fourth network layer can make the accuracy of the voiceprint vector outputted finally higher, and thus make the accuracy of the voiceprint recognition model finally trained higher. In this way, through the layer-by-layer feature mapping of L network layers, a variety of feature vectors can be obtained, and the trained voiceprint recognition model can recognize the above-mentioned multiple feature vectors, thereby improving the accuracy of the voiceprint recognition model finally trained to recognize the voiceprint data. In addition, because the voiceprint data can be converted into voiceprint vectors through the layer-by-layer feature mapping of L network layers, the voiceprint vectors are more easily and accurately recognized by the voiceprint recognition model, thereby further improving the accuracy of the voiceprint recognition model finally trained to recognize the voiceprint.

[0064] As an optional implementation, the first network layer, the second network layer, the third network layer and the fourth network layer all include interconnected linear layers and blocks.

[0065] For example: See Figure 2 , the L-layer network layer includes: a first network layer 201, a second network layer 202, a third network layer 203 and a fourth network layer 204 connected in sequence, and the first network layer 201, the second network layer 202, the third network layer 203 and the fourth network layer 204 all include interconnected linear layers and blocks, see Figure 2 Taking the first network layer 201 as an example, the first network layer 201 includes a linear layer 2011 and a block 2012. In addition, the fourth network layer 204 may also be connected to a target linear layer 207, which may be referred to as a Linear layer. In addition, the L-layer network layer may also include the input layer 206 and the fusion layer 205 of the voiceprint data.

[0066] Among them, the linear layer can be called a linear layer, and the block block can be called a transblock block. The above linear layer can perform feature mapping, the block block can perform feature mapping, and the block block can also perform attention mechanism fitting.

[0067] In this implementation, since each network layer in the first network layer, the second network layer, the third network layer and the fourth network layer includes interconnected linear layers and block blocks, and both the linear layers and the block blocks can perform feature mapping, the effect of feature mapping can be further enhanced.

[0068] The number of blocks included in each network layer is not limited here. For example, the first network layer may include 2 blocks, the second network layer may include 2 blocks, the third network layer may include 6 blocks, and the fourth network layer may include 2 blocks. In this way, the accuracy of the model to be trained can be higher and the training speed can be faster during the training process of the model to be trained.

[0069] That is, as an optional implementation, the first network layer, the second network layer, the third network layer and the fourth network layer all include multiple blocks.

[0070] In this way, since multiple blocks (which can be understood as Transblock blocks) are superimposed in each network layer, the network layer can have different receptive fields, making it easy to superimpose L network layers, and superimposing L network layers can enhance the mapping ability of the model to be trained.

[0071] It should be noted that the voiceprint data can be a multi-dimensional feature vector or a one-dimensional feature vector. When the voiceprint data is a one-dimensional feature vector, it can be directly input into the first network layer. When the voiceprint data is a multi-dimensional feature vector, the multi-dimensional feature vector can be first converted into a one-dimensional feature vector, and then the one-dimensional feature vector can be input into the first network layer.

[0072] In addition, the specific type of the model to be trained is not limited here. As an optional implementation, the model to be trained can be a model improved on the basis of the transformer model.

[0073] Since only the voiceprint vector needs to be obtained in this embodiment, as another optional implementation, the model to be trained is a model improved on the basis of the encoder of the transformer model. For example, the network layer of the encoder of the transformer model is modified to L network layers, and each network layer may include a Transblock layer.

[0074] In this way, the structure of the model to be trained in this implementation only needs to be improved based on the encoder of the transformer model, without the need to use the decoder of the transformer model, thereby reducing the structural layer of the model to be trained and improving the training efficiency of the model to be trained.

[0075] For example: See Figure 3 , Figure 3 is a schematic diagram of a complete transformer model, which includes an encoder 301 and a decoder 302. The structure of the model to be trained in this embodiment can be seen in Figure 3 The structure of the encoder 301 in FIG. 1 includes L network layers, and each network layer includes a Transblock block, which can be seen in FIG. Figure 3 The structure of the Transblock block 3010 in the encoder 301, that is, each Transblock block 3010 in the encoder 301 includes an attention mechanism layer 3011, a first addition layer 3012, a linear layer 3013 and a second addition layer 3014, the above-mentioned attention mechanism layer 3011 can also be called a Multi-Head Attention layer, the first addition layer 3012 and the second addition layer 3014 can be called Add&Norm layers, and the linear layer 3013 can be called a Feed Forward layer.

[0076] Among them, the first initial feature vector can be first input into the attention mechanism layer 301 and the first addition layer 3012 respectively. In the first addition layer 3012, the first initial feature vector and the first target feature vector output by the attention mechanism layer 301 of the first initial feature vector can be added to obtain the second initial feature vector. Then, the second initial feature vector is output to the linear layer 3013 and the second addition layer 3014 respectively. Then, in the second addition layer 3014, the second initial feature vector and the second initial feature vector are added by the second target feature vector output by the linear layer 3013 to obtain the third initial feature vector.

[0077] It should be noted that when Figure 3 When the Transblock block shown is a Transblock block included in the first network layer, the first initial feature vector can be understood as the voiceprint data. The first addition layer 3012 and the second addition layer 3014 can both perform normalization and addition operations, and the linear layer 3013 can perform feature mapping operations.

[0078] As an optional implementation, see Figure 2 , the L-layer network layer further includes: an integration layer 205, the integration layer 205 is connected to the first network layer, and before the voiceprint data is input into the first network layer, feature mapping is performed, and the first feature vector is output, the method further includes:

[0079] The pre-acquired features to be trained are input into the fusion layer for feature extraction to obtain the voiceprint data, wherein the dimension of the features to be trained is greater than the dimension of the voiceprint data.

[0080] Among them, the fusion layer 205 can also be called a merge layer, and the fusion layer 205 can be used to reduce the dimension of the features to be trained, and the above-mentioned features to be trained can be features obtained from the sample pool, or features obtained after feature extraction from sample data obtained from the sample pool.

[0081] In this embodiment, the fusion layer 205 can be used to reduce the dimension of the features to be trained to obtain voiceprint data, so that the voiceprint data can be easily input into the first network layer for feature mapping.

[0082] In this embodiment, through step 101, the voiceprint data is sequentially subjected to feature mapping through L layers of network layers, and a voiceprint vector is output, that is, the sample is sequentially subjected to feature mapping for multiple times, thereby improving the utilization rate of the sample. At the same time, since the feature vector obtained by feature mapping of the next layer of network layers needs to be based on the feature vector output by the previous layer of network layers, multiple feature vectors can be obtained through layer-by-layer feature mapping of L layers of network layers, and the trained voiceprint recognition model can recognize the above multiple feature vectors, thereby improving the accuracy of the voiceprint recognition model finally trained to recognize the voiceprint data. In addition, since the voiceprint data can be converted into voiceprint vectors through layer-by-layer feature mapping of L layers of network layers, the voiceprint vectors can be more easily and accurately recognized by the voiceprint recognition model, thereby further improving the accuracy of the voiceprint recognition model finally trained to recognize the voiceprint.

[0083] See also Figure 4 , Figure 4 is a flow chart of a voiceprint recognition method provided in an embodiment of the present application. The method is applied to a voiceprint recognition model. The voiceprint recognition model is connected to a target classifier, such as Figure 4 As shown, the method comprises the following steps:

[0084] Step 401: Obtain target voiceprint data of the user to be identified.

[0085] Step 402: Input the target voiceprint data into the L-layer network layer included in the voiceprint recognition model for feature mapping, and output a target voiceprint vector, where L is an integer greater than 1.

[0086] Step 403: Perform voiceprint recognition according to the target voiceprint vector.

[0087] In the embodiment of the present application, through steps 401 to 403, the target voiceprint data is input into the L-layer network layer included in the voiceprint recognition model for feature mapping, and the target voiceprint vector is output. In this way, through the layer-by-layer feature mapping of the L-layer network layer, a variety of feature vectors can be obtained, and the trained voiceprint recognition model can recognize the above-mentioned multiple feature vectors, thereby improving the accuracy of the voiceprint recognition model finally trained to recognize the voiceprint data. In addition, because the target voiceprint data can be converted into the target voiceprint vector through the layer-by-layer feature mapping of the L-layer network layer, and the target voiceprint vector is more easily and accurately recognized by the voiceprint recognition model, thereby further improving the accuracy of the voiceprint recognition model to recognize the target voiceprint.

[0088] As an optional implementation, the L-layer network layer includes: a first network layer, a second network layer, a third network layer and a fourth network layer connected in sequence;

[0089] The step of inputting the target voiceprint data into the L-layer network layer included in the voiceprint recognition model for feature mapping and outputting the target voiceprint vector includes:

[0090] Inputting the target voiceprint data into the first network layer for feature mapping, and outputting a first feature vector;

[0091] Inputting the first feature vector into the second network layer for feature mapping, and outputting a second feature vector;

[0092] Inputting the second feature vector into the third network layer for feature mapping, and outputting a third feature vector;

[0093] Inputting the third feature vector into the fourth network layer for feature mapping, and outputting a target voiceprint vector;

[0094] The dimensions of the first feature vector, the second feature vector, the third feature vector and the target voiceprint vector increase in sequence.

[0095] In this way, through the layer-by-layer feature mapping of L network layers, multiple feature vectors can be obtained, and the voiceprint recognition model can recognize all of the above multiple feature vectors, thereby improving the accuracy of the voiceprint recognition model in recognizing voiceprint data. In addition, because the target voiceprint data can be converted into a target voiceprint vector through the layer-by-layer feature mapping of L network layers, and the target voiceprint vector is more easily and accurately recognized by the voiceprint recognition model, the accuracy of the voiceprint recognition model in recognizing the target voiceprint data is further improved.

[0096] As an optional implementation, performing voiceprint recognition according to the target voiceprint vector includes:

[0097] The target voiceprint vector and the pre-stored feature vector are input into the target classifier, and a likelihood distribution value is output; wherein the pre-stored feature vector is output by inputting the second voiceprint data of the target user into the voiceprint recognition model.

[0098] When the likelihood distribution value is greater than a preset threshold, it is determined that the user to be identified and the target user are the same user.

[0099] It should be noted that the target voiceprint vector and the pre-stored feature vector can both be x-vector features, and the x-vector features are neural network features extracted by the deep neural network.

[0100] The pre-stored feature vector may be understood as a feature vector corresponding to the second voiceprint data of the target user collected in advance, and may be stored on a server corresponding to the database.

[0101] Among them, the type of target classifier is not limited here. For example, the target classifier can be a plda classifier, that is, the plda algorithm can be run in the plda classifier. In this way, the accuracy of the likelihood distribution value output by the plda classifier is higher, thereby making the accuracy of the judgment result of whether the user to be identified and the target user are the same user higher.

[0102] The likelihood distribution value can also be understood as similarity, that is, the larger the likelihood distribution value is, the higher the possibility that the user to be identified and the target user are the same user.

[0103] In an embodiment of the present application, a voiceprint recognition model and a target classifier connected to the voiceprint recognition model can be used to determine whether the user to be identified and the target user are the same user, thereby improving the accuracy of the recognition result of the voiceprint data of the user to be identified and reducing the loss caused by the inability to accurately identify the voiceprint data of the user to be identified.

[0104] It should be noted that the voiceprint recognition model in the embodiment of the present application can be trained using the method in the above embodiment. Therefore, each feature in the embodiment of the present application can refer to the corresponding description in the above embodiment, and the details will not be repeated here.

[0105] See also Figure 5 , Figure 5 1 is a structural diagram of the voiceprint recognition model training device provided in the embodiment of the present application, which can implement the details of the voiceprint recognition model training method in the above embodiment and achieve the same effect. Figure 5 As shown, the voiceprint recognition model training device 500 includes:

[0106] The training module 501 is used to sequentially input the voiceprint data into the model to be trained for multiple iterations of training to obtain a voiceprint recognition model;

[0107] The model to be trained includes L network layers, where L is an integer greater than 1; during the Nth iterative training process, the following steps are included:

[0108] The voiceprint data is sequentially passed through the L-layer network layers for feature mapping, and a voiceprint vector is output;

[0109] When the output value corresponding to the voiceprint vector meets the preset conditions, the model to be trained after the Nth iterative training is determined as the voiceprint recognition model, where N is a positive integer.

[0110] Optionally, the L-layer network layer includes: a first network layer, a second network layer, a third network layer and a fourth network layer connected in sequence, and the voiceprint data is sequentially passed through the L-layer network layers for feature mapping, and the voiceprint vector is output, including:

[0111] Inputting the voiceprint data into the first network layer for feature mapping, and outputting a first feature vector;

[0112] Inputting the first feature vector into the second network layer for feature mapping, and outputting a second feature vector;

[0113] Inputting the second feature vector into the third network layer for feature mapping, and outputting a third feature vector;

[0114] Inputting the third feature vector into the fourth network layer for feature mapping, and outputting the voiceprint vector;

[0115] Among them, the dimensions of the first feature vector, the second feature vector, the third feature vector and the voiceprint vector increase in sequence.

[0116] Optionally, the first network layer, the second network layer, the third network layer and the fourth network layer all include interconnected linear layers and block blocks.

[0117] Optionally, the first network layer, the second network layer, the third network layer and the fourth network layer all include a plurality of blocks.

[0118] Optionally, the L-layer network layer further includes: an integration layer, the integration layer is connected to the first network layer, and before inputting the voiceprint data into the first network layer for feature mapping and outputting the first feature vector, the method further includes:

[0119] The pre-acquired features to be trained are input into the fusion layer for feature extraction to obtain the voiceprint data, wherein the dimension of the features to be trained is greater than the dimension of the voiceprint data.

[0120] Optionally, the preset condition includes: a difference between an output value corresponding to the voiceprint vector and an actual value corresponding to the voiceprint data is within a preset range.

[0121] Optionally, the voiceprint data is fbank80 features.

[0122] The voiceprint recognition model training device provided in the embodiment of the present application can achieve Figure 1 In order to avoid repetition, the various processes implemented by the voiceprint recognition model training device in the method embodiment will not be repeated here.

[0123] In this way, the voiceprint data is sequentially mapped through L layers of network layers to output a voiceprint vector, thereby improving the utilization rate of the sample. At the same time, since the feature vector obtained by the feature mapping of the latter network layer needs to be based on the feature vector output by the previous network layer, the accuracy of the trained voiceprint recognition model is improved. The target voiceprint data is input into the L layers of network layers included in the voiceprint recognition model for feature mapping, and the target voiceprint vector is output. In this way, through the layer-by-layer feature mapping of the L layers of network layers, a variety of feature vectors can be obtained, and the voiceprint recognition model can recognize the above-mentioned multiple feature vectors, thereby improving the accuracy of the voiceprint recognition model in recognizing the voiceprint data. In addition, through the layer-by-layer feature mapping of the L layers of network layers, the target voiceprint data can be converted into a target voiceprint vector, and the target voiceprint vector is more easily and accurately recognized by the voiceprint recognition model, thereby further improving the accuracy of the voiceprint recognition model in recognizing the target voiceprint.

[0124] See also Figure 6 , Figure 6 A schematic diagram of the structure of a voiceprint recognition device provided in an embodiment of the present application, wherein the voiceprint recognition device is applied to a voiceprint recognition model, and the voiceprint recognition model is connected to a target classifier, such as Figure 6 As shown, the voiceprint recognition device 600 includes:

[0125] An acquisition module 601 is used to acquire the target voiceprint data of the user to be identified;

[0126] An output module 602 is used to input the target voiceprint data into the L-layer network layer included in the voiceprint recognition model for feature mapping, and output a target voiceprint vector, where L is an integer greater than 1;

[0127] The voiceprint recognition module 603 is used to perform voiceprint recognition according to the target voiceprint vector.

[0128] Optionally, the L-layer network layer includes: a first network layer, a second network layer, a third network layer and a fourth network layer connected in sequence;

[0129] The output module 602 includes:

[0130] A first output submodule, configured to input the target voiceprint data into the first network layer for feature mapping, and output a first feature vector;

[0131] A second output submodule, used for inputting the first feature vector into the second network layer for feature mapping, and outputting a second feature vector;

[0132] A third output submodule, used for inputting the second feature vector into the third network layer for feature mapping, and outputting a third feature vector;

[0133] A fourth output submodule, configured to input the third feature vector into the fourth network layer for feature mapping, and output a target voiceprint vector;

[0134] The dimensions of the first feature vector, the second feature vector, the third feature vector and the target voiceprint vector increase in sequence.

[0135] The voiceprint recognition device provided in the embodiment of the present application can achieve Figure 4 In order to avoid repetition, the various processes implemented by the voiceprint recognition device in the method embodiment will not be described here.

[0136] In this way, the voiceprint recognition model and the target classifier connected to the voiceprint recognition model can be used to determine whether the user to be identified and the target user are the same user, thereby improving the accuracy of the recognition result of the voiceprint data of the user to be identified and reducing the loss caused by the inability to accurately identify the voiceprint data of the user to be identified. Through the layer-by-layer feature mapping of L layers of network layers, a variety of feature vectors can be obtained, and the voiceprint recognition model can identify the above-mentioned multiple feature vectors, thereby improving the accuracy of the voiceprint recognition model in recognizing voiceprint data. In addition, because the target voiceprint data can be converted into a target voiceprint vector through the layer-by-layer feature mapping of L layers of network layers, the target voiceprint vector is more easily and accurately recognized by the voiceprint recognition model, thereby further improving the accuracy of the voiceprint recognition model in recognizing the target voiceprint.

[0137] Figure 7 A schematic diagram of the hardware structure of an electronic device for implementing various embodiments of the present application.

[0138] The electronic device 700 includes but is not limited to: a radio frequency unit 701, a network module 702, an audio output unit 703, an input unit 704, a sensor 705, a display unit 706, a user input unit 707, an interface unit 708, a memory 709, a processor 710, and a power supply 711. Those skilled in the art will appreciate that Figure 7 The electronic device structure shown in the figure does not constitute a limitation on the electronic device, and the electronic device may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently. In the embodiments of the present application, the electronic device includes but is not limited to a mobile phone, a tablet computer, a laptop computer, a PDA, a vehicle-mounted terminal, a wearable device, and a pedometer.

[0139] When the electronic device is used to execute the steps in the voiceprint recognition model training method, the processor 710 is used to perform the following operations:

[0140] Input the voiceprint data into the model to be trained in sequence for multiple iterations of training to obtain a voiceprint recognition model;

[0141] The model to be trained includes L network layers, where L is an integer greater than 1; during the Nth iterative training process, the following steps are included:

[0142] The voiceprint data is sequentially passed through the L-layer network layer for feature mapping to output a voiceprint vector;

[0143] When the output value corresponding to the voiceprint vector meets a preset condition, the model to be trained after the Nth iterative training is determined as the voiceprint recognition model, where N is a positive integer.

[0144] Optionally, the L-layer network layer includes: a first network layer, a second network layer, a third network layer and a fourth network layer connected in sequence, and the processor 710 performs feature mapping of the voiceprint data through the L-layer network layer in sequence to output a voiceprint vector, including:

[0145] Inputting the voiceprint data into the first network layer for feature mapping, and outputting a first feature vector;

[0146] Inputting the first feature vector into the second network layer for feature mapping, and outputting a second feature vector;

[0147] Inputting the second feature vector into the third network layer for feature mapping, and outputting a third feature vector;

[0148] Inputting the third feature vector into the fourth network layer for feature mapping, and outputting the voiceprint vector;

[0149] Among them, the dimensions of the first feature vector, the second feature vector, the third feature vector and the voiceprint vector increase in sequence.

[0150] Optionally, the first network layer, the second network layer, the third network layer and the fourth network layer all include interconnected linear layers and block blocks.

[0151] Optionally, the first network layer, the second network layer, the third network layer and the fourth network layer all include a plurality of blocks.

[0152] Optionally, the L-layer network layer also includes: an integration layer, which is connected to the first network layer, and the processor 710 is also used to input the pre-acquired features to be trained into the integration layer for feature extraction to obtain the voiceprint data, wherein the dimension of the features to be trained is greater than the dimension of the voiceprint data.

[0153] Optionally, the preset condition includes: a difference between an output value corresponding to the voiceprint vector and an actual value corresponding to the voiceprint data is within a preset range.

[0154] Optionally, the voiceprint data is fbank80 features.

[0155] When the electronic device is used to execute the steps in the voiceprint recognition method, the processor 710 is used to perform the following operations:

[0156] Obtain the target voiceprint data of the user to be identified;

[0157] Input the target voiceprint data into the L-layer network layer included in the voiceprint recognition model for feature mapping, and output a target voiceprint vector, where L is an integer greater than 1;

[0158] Voiceprint recognition is performed according to the target voiceprint vector.

[0159] Optionally, the L-layer network layer includes: a first network layer, a second network layer, a third network layer and a fourth network layer connected in sequence;

[0160] The processor 710 performs the step of inputting the target voiceprint data into the L-layer network layer included in the voiceprint recognition model for feature mapping and outputting the target voiceprint vector, including:

[0161] Inputting the target voiceprint data into the first network layer for feature mapping, and outputting a first feature vector;

[0162] Inputting the first feature vector into the second network layer for feature mapping, and outputting a second feature vector;

[0163] Inputting the second feature vector into the third network layer for feature mapping, and outputting a third feature vector;

[0164] Inputting the third feature vector into the fourth network layer for feature mapping, and outputting a target voiceprint vector;

[0165] The dimensions of the first feature vector, the second feature vector, the third feature vector and the target voiceprint vector increase in sequence.

[0166] It should be understood that in the embodiment of the present application, the radio frequency unit 701 can be used for receiving and sending signals during information transmission or calls. Specifically, after receiving downlink data from the base station, it is sent to the processor 710 for processing; in addition, uplink data is sent to the base station. Generally, the radio frequency unit 701 includes but is not limited to an antenna, at least one amplifier, a transceiver, a coupler, a low noise amplifier, a duplexer, etc. In addition, the radio frequency unit 701 can also communicate with the network and other devices through a wireless communication system.

[0167] The electronic device provides users with wireless broadband Internet access through the network module 702, such as helping users to send and receive emails, browse web pages, and access streaming media.

[0168] The audio output unit 703 can convert the audio data received by the RF unit 701 or the network module 702 or stored in the memory 709 into an audio signal and output it as sound. Moreover, the audio output unit 703 can also provide audio output related to a specific function performed by the electronic device 700 (for example, a call signal reception sound, a message reception sound, etc.). The audio output unit 703 includes a speaker, a buzzer, a receiver, etc.

[0169] The input unit 704 is used to receive audio or video signals. The input unit 704 may include a graphics processor (GPU) 7041 and a microphone 7042, and the graphics processor 7041 processes the image data of a static picture or video obtained by an image capture device (such as a camera) in a video capture mode or an image capture mode. The processed image frame can be displayed on the display unit 706. The image frame processed by the graphics processor 7041 can be stored in the memory 709 (or other storage medium) or sent via the radio frequency unit 701 or the network module 702. The microphone 7042 can receive sound and can process such sound into audio data. The processed audio data can be converted into a format output that can be sent to a mobile communication base station via the radio frequency unit 701 in the case of a telephone call mode.

[0170] The electronic device 700 also includes at least one sensor 705, such as a light sensor, a motion sensor, and other sensors. Specifically, the light sensor includes an ambient light sensor and a proximity sensor, wherein the ambient light sensor can adjust the brightness of the display panel 7061 according to the brightness of the ambient light, and the proximity sensor can turn off the display panel 7061 and / or the backlight when the electronic device 700 is moved to the ear. As a kind of motion sensor, the accelerometer sensor can detect the magnitude of acceleration in each direction (generally three axes), and can detect the magnitude and direction of gravity when stationary, which can be used to identify the posture of the electronic device (such as horizontal and vertical screen switching, related games, magnetometer posture calibration), vibration recognition related functions (such as pedometer, tapping), etc.; the sensor 705 can also include a fingerprint sensor, a pressure sensor, an iris sensor, a molecular sensor, a gyroscope, a barometer, a hygrometer, a thermometer, an infrared sensor, etc., which will not be repeated here.

[0171] The display unit 706 is used to display information input by the user or information provided to the user. The display unit 706 may include a display panel 7061, which may be configured in the form of a liquid crystal display (LCD), an organic light-emitting diode (OLED), or the like.

[0172] The user input unit 707 can be used to receive input digital or character information, and to generate key signal input related to user settings and function control of the electronic device. Specifically, the user input unit 707 includes a touch panel 7071 and other input devices 7072. The touch panel 7071, also known as a touch screen, can collect the user's touch operation on or near it (such as the user's operation on the touch panel 7071 or near the touch panel 7071 using any suitable object or accessory such as a finger, stylus, etc.). The touch panel 7071 may include two parts: a touch detection device and a touch controller. Among them, the touch detection device detects the user's touch orientation, detects the signal brought by the touch operation, and transmits the signal to the touch controller; the touch controller receives the touch information from the touch detection device, converts it into the contact point coordinates, and then sends it to the processor 710, receives the command sent by the processor 710 and executes it. In addition, the touch panel 7071 can be implemented using multiple types such as resistive, capacitive, infrared and surface acoustic waves. In addition to the touch panel 7071, the user input unit 707 may also include other input devices 7072. Specifically, other input devices 7072 may include but are not limited to a physical keyboard, function keys (such as volume control keys, switch keys, etc.), a trackball, a mouse, and a joystick, which are not described in detail here.

[0173] Furthermore, the touch panel 7071 may be overlaid on the display panel 7061. When the touch panel 7071 detects a touch operation on or near it, it transmits the information to the processor 710 to determine the type of the touch event. Then, the processor 710 provides a corresponding visual output on the display panel 7061 according to the type of the touch event. Figure 7 In the figure, the touch panel 7071 and the display panel 7061 are used as two independent components to realize the input and output functions of the electronic device. However, in some embodiments, the touch panel 7071 and the display panel 7061 can be integrated to realize the input and output functions of the electronic device, which is not limited here.

[0174] The interface unit 708 is an interface for connecting an external device to the electronic device 700. For example, the external device may include a wired or wireless headset port, an external power supply (or battery charger) port, a wired or wireless data port, a memory card port, a port for connecting a device with an identification module, an audio input / output (I / O) port, a video I / O port, a headphone port, etc. The interface unit 708 may be used to receive input (e.g., data information, power, etc.) from an external device and transmit the received input to one or more elements within the electronic device 700 or may be used to transmit data between the electronic device 700 and an external device.

[0175] The memory 709 can be used to store software programs and various data. The memory 709 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory 709 can include a high-speed random access memory, and can also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0176] The processor 710 is the control center of the electronic device. It uses various interfaces and lines to connect various parts of the entire electronic device. It executes various functions of the electronic device and processes data by running or executing software programs and / or modules stored in the memory 709, and calling data stored in the memory 709, so as to monitor the electronic device as a whole. The processor 710 may include one or more processing units; preferably, the processor 710 may integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, user interface, and application programs, and the modem processor mainly processes wireless communications. It is understandable that the above-mentioned modem processor may not be integrated into the processor 710.

[0177] The electronic device 700 may also include a power supply 711 (such as a battery) for supplying power to each component. Preferably, the power supply 711 may be logically connected to the processor 710 through a power management system, thereby implementing functions such as charging, discharging, and power consumption management through the power management system.

[0178] In addition, the electronic device 700 includes some functional modules not shown, which will not be described in detail here.

[0179] Preferably, an embodiment of the present application also provides an electronic device, including a processor 710, a memory 709, and a computer program stored in the memory 709 and executable on the processor 710. When the computer program is executed by the processor 710, the above-mentioned voiceprint recognition model training method or each process of the above-mentioned voiceprint recognition method can be implemented, and the same technical effect can be achieved, which will not be repeated here.

[0180] The embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by the processor 710, each process of the above-mentioned voiceprint recognition model training method or the above-mentioned voiceprint recognition method embodiment is implemented, and the same technical effect can be achieved. To avoid repetition, it is not repeated here. The computer-readable storage medium is, for example, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0181] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0182] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for a terminal (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0183] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the present application and the scope of protection of the claims, all of which are within the protection of the present application.

Claims

1. A voiceprint recognition model training method, characterized in that: include: Input the voiceprint data into the model to be trained in sequence for multiple iterations of training to obtain a voiceprint recognition model; The model to be trained includes L network layers, where L is an integer greater than 1; during the Nth iterative training process, the following steps are included: The voiceprint data is sequentially passed through the L-layer network layers for feature mapping, and a voiceprint vector is output; the input of the i-th network layer is the feature vector obtained by feature mapping of the i-1-th network layer, 2≤i≤L; When the output value corresponding to the voiceprint vector meets the preset conditions, the model to be trained after the Nth iterative training is determined as the voiceprint recognition model, where N is a positive integer.

2. The method according to claim 1, characterized in that: The L-layer network layer includes: a first network layer, a second network layer, a third network layer and a fourth network layer connected in sequence; the voiceprint data is sequentially passed through the L-layer network layer for feature mapping, and the voiceprint vector is output, including: Inputting the voiceprint data into the first network layer for feature mapping, and outputting a first feature vector; Inputting the first feature vector into the second network layer for feature mapping, and outputting a second feature vector; Inputting the second feature vector into the third network layer for feature mapping, and outputting a third feature vector; Inputting the third feature vector into the fourth network layer for feature mapping, and outputting the voiceprint vector; Among them, the dimensions of the first feature vector, the second feature vector, the third feature vector and the voiceprint vector increase in sequence.

3. The method according to claim 2, characterized in that The first network layer, the second network layer, the third network layer and the fourth network layer each include interconnected linear layers and blocks.

4. The method according to claim 3, characterized in that: The first network layer, the second network layer, the third network layer, and the fourth network layer all include a plurality of blocks.

5. The method according to claim 2, characterized in that: The L-layer network layer further includes: an integration layer, the integration layer is connected to the first network layer; the voiceprint data is input into the first network layer for feature mapping, and before outputting the first feature vector, the method further includes: The pre-acquired features to be trained are input into the fusion layer for feature extraction to obtain the voiceprint data, wherein the dimension of the features to be trained is greater than the dimension of the voiceprint data.

6. The method according to claim 1, characterized in that The preset condition includes: the difference between the output value corresponding to the voiceprint vector and the actual value corresponding to the voiceprint data is within a preset range.

7. The method according to any one of claims 1 to 6, characterized in that The voiceprint data is fbank80 features.

8. A voiceprint recognition method, characterized in that: The method comprises: Obtain the target voiceprint data of the user to be identified; The target voiceprint data is input into the L-layer network layer included in the voiceprint recognition model for feature mapping, and a target voiceprint vector is output, where L is an integer greater than 1; the input of the i-th network layer is the feature vector obtained by feature mapping of the i-1-th network layer, 2≤i≤L; Voiceprint recognition is performed according to the target voiceprint vector.

9. The method according to claim 8, characterized in that The L-layer network layer includes: a first network layer, a second network layer, a third network layer and a fourth network layer connected in sequence; The step of inputting the target voiceprint data into the L-layer network layer included in the voiceprint recognition model for feature mapping and outputting the target voiceprint vector includes: Inputting the target voiceprint data into the first network layer for feature mapping, and outputting a first feature vector; Inputting the first feature vector into the second network layer for feature mapping, and outputting a second feature vector; Inputting the second feature vector into the third network layer for feature mapping, and outputting a third feature vector; Inputting the third feature vector into the fourth network layer for feature mapping, and outputting a target voiceprint vector; The dimensions of the first feature vector, the second feature vector, the third feature vector and the target voiceprint vector increase in sequence.

10. A voiceprint recognition model training device, characterized in that: include: The training module is used to input the voiceprint data into the model to be trained in sequence for multiple iterations of training to obtain a voiceprint recognition model; The model to be trained includes L network layers, where L is an integer greater than 1; during the Nth iterative training process, the following steps are included: The voiceprint data is sequentially passed through the L-layer network layers for feature mapping, and a voiceprint vector is output; the input of the i-th network layer is the feature vector obtained by feature mapping of the i-1-th network layer, 2≤i≤L; When the output value corresponding to the voiceprint vector meets a preset condition, the model to be trained after the Nth iterative training is determined as the voiceprint recognition model, where N is a positive integer.

11. A voiceprint recognition device, characterized in that: include: An acquisition module is used to acquire the target voiceprint data of the user to be identified; An output module is used to input the target voiceprint data into the L-layer network layer included in the voiceprint recognition model for feature mapping, and output a target voiceprint vector, where L is an integer greater than 1; the input of the i-th network layer is a feature vector obtained by feature mapping of the i-1-th network layer, 2≤i≤L; A voiceprint recognition module is used to perform voiceprint recognition according to the target voiceprint vector.

12. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps in the voiceprint recognition model training method as described in any one of claims 1 to 7 are implemented, or when the processor executes the computer program, the steps in the voiceprint recognition method as described in claim 8 or 9 are implemented.

13. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps in the voiceprint recognition model training method as described in any one of claims 1 to 7, or, when executed by a processor, implements the steps in the voiceprint recognition method as described in claim 8 or 9.

Images