Auditory bionic voiceprint recognition method and device, electronic equipment and storage medium

Abstract

The invention provides an auditory bionic voiceprint recognition method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining a to-be-recognized voice signal, and carrying out the cutting processing of the to-be-recognized voice signal, and obtaining a plurality of to-be-recognized voice segments; inputting the to-be-recognized voice segments into a cochlea bionic filter to obtain voice frequency characteristics; inputting the voice frequency characteristics into a pre-trained voiceprint recognition model to obtain a voiceprint recognition result; wherein the voiceprint recognition model is obtained based on a cochlea core bionic network. Through the cochlea bionic filter and the voiceprint recognition model provided by the invention, higher adaptive capacity to noise and complex environments is realized, so that the reliability and efficiency of voiceprint recognition are improved.

Description

Technical Field

[0001] The present invention relates to the field of biometric identification technology, and in particular to an auditory bionic voiceprint recognition method, device, electronic equipment and storage medium. Background Art

[0002] Biometric recognition technology is a technology that detects and analyzes the unique biological characteristics of the human body (such as fingerprints, faces, irises, voiceprints, etc.) to identify the person. It is highly secure and convenient. Voiceprint is a unique biological feature, which is the characteristic information contained in human voice signals. Identity authentication is achieved by analyzing the voice characteristics of individuals. With the rapid development of biometric recognition technology, voiceprint recognition has gradually become an important means of biometric authentication.

[0003] In recent years, deep learning has made breakthrough progress in many fields and has been widely used. However, the existing deep learning-based methods have the following shortcomings in the application of voiceprint recognition: First, the interpretability is poor, which makes it difficult to trace the decision-making process of the model, reducing the credibility of the system and the security of the application; second, the existing deep learning methods usually use acoustic features such as filter banks, Mel cepstral coefficients and spectrograms as model inputs, and lack the ability to directly process raw audio signals. In addition, some end-to-end voiceprint recognition systems fail to apply the characteristics of the speech signal processing pathway in the biological auditory system to the design of the neural network model structure, resulting in the model being unable to fully utilize these characteristics to enhance the extraction of speech features, affecting its adaptability and performance in practical applications. Summary of the invention

[0004] The present invention provides an auditory bionic voiceprint recognition method, device, electronic device and storage medium to solve the defects in the prior art and achieve higher adaptability to noise and complex environments, thereby improving the reliability and efficiency of voiceprint recognition.

[0005] The present invention provides an auditory bionic voiceprint recognition method, comprising: Acquire a speech signal to be recognized, and segment the speech signal to be recognized to obtain a plurality of speech segments to be recognized; Inputting each of the speech segments to be recognized into a cochlear bionic filter to obtain speech frequency characteristics; The speech frequency feature is input into a pre-trained voiceprint recognition model to obtain a voiceprint recognition result; wherein the voiceprint recognition model is obtained based on a cochlear nucleus bionic network.

[0006] According to an auditory bionic voiceprint recognition method provided by the present invention, the voiceprint recognition model includes a first cochlear nucleus bionic network, a fully connected network and an output layer connected in sequence; The first cochlear nucleus bionic network is used to extract temporal features and spatial features based on the speech frequency features, obtain temporal feature representation based on the temporal features and spatial features, and perform weighted fusion processing on the temporal feature representation and the speech frequency features to obtain speech feature representation; The fully connected network and the output layer are used to process the speech feature representation in sequence to obtain the voiceprint recognition result.

[0007] According to an auditory bionic voiceprint recognition method provided by the present invention, the first cochlear nucleus bionic network includes a spiral ganglion neuron layer, a stellate cell layer, an attention mechanism and a plexus cell layer, the attention mechanism is respectively connected to the spiral ganglion neuron layer, the stellate cell layer and the plexus cell layer, and the spiral ganglion neuron layer is connected to the stellate cell layer; The spiral ganglion neuron layer is used to perform a convolution operation in the time direction based on the speech frequency feature to obtain a first bionic feature; The stellate cell layer is used to perform a convolution operation in the time direction based on the first bionic feature to obtain a second bionic feature; The attention mechanism is used to obtain an intermediate feature by performing weighted fusion based on the speech frequency feature and the second bionic feature through an attention parameter; The cluster cell layer is used to perform a convolution operation in the time direction based on the intermediate features to obtain the speech feature representation.

[0008] First, according to an auditory bionic voiceprint recognition method provided by the present invention, the output layer includes a Softmax function or an additive residual Softmax function.

[0009] According to an auditory bionic voiceprint recognition method provided by the present invention, the voiceprint recognition model also includes a second cochlear nucleus bionic network, and the second cochlear nucleus bionic network is connected between the first cochlear nucleus bionic network and the fully connected network.

[0010] According to an auditory bionic voiceprint recognition method provided by the present invention, the cochlear bionic filter includes a filter group composed of multiple filters with different response frequencies; wherein each of the filters respectively extracts the speech frequency features of the corresponding response frequency from the speech segment to be recognized.

[0011] According to an auditory bionic voiceprint recognition method provided by the present invention, the training process of the voiceprint recognition model includes: Acquire a training set; wherein the training set includes a plurality of training samples; Constructing a loss function based on the training samples; The training samples in the training set are input into a basic voiceprint recognition model, and iterative training is performed with minimizing the loss function as the training goal until the training is completed to obtain the voiceprint recognition model; wherein the basic voiceprint recognition model is obtained based on the cochlear nucleus bionic network.

[0012] The present invention also provides an auditory bionic voiceprint recognition device, comprising: A first acquisition module is configured to acquire a speech signal to be recognized, and segment the speech signal to be recognized to obtain a plurality of speech segments to be recognized; A first input module is configured to input each of the speech segments to be recognized into a cochlear bionic filter to obtain speech frequency characteristics; The second input module is configured to input the speech frequency feature into a pre-trained voiceprint recognition model to obtain a voiceprint recognition result; wherein the voiceprint recognition model is obtained based on a cochlear nucleus bionic network.

[0013] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, any of the above-mentioned auditory bionic voiceprint recognition methods is implemented.

[0014] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the auditory bionic voiceprint recognition method as described in any one of the above is implemented.

[0015] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the auditory bionic voiceprint recognition method described above is implemented.

[0016] The auditory bionic voiceprint recognition method, device, electronic device and storage medium provided by the present invention construct an initial feature set related to the original voice signal to be recognized by dividing the voice signal to be recognized into voice segments to be recognized. These voice segments to be recognized are input into the cochlear bionic filter. The cochlear bionic filter is based on the characteristics of the cochlear basilar membrane and simulates the response of hair cells to specific frequencies to directly extract the voice frequency features from the original voice waveform efficiently and accurately, and then the voiceprint recognition model provided by the present invention is used to perform voiceprint recognition. A higher adaptability to noise and complex environments is achieved, thereby improving the reliability and efficiency of voiceprint recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 It is a schematic diagram of the flow of the auditory bionic voiceprint recognition method provided by the present invention.

[0019] Figure 2 This is one of the structural diagrams of the voiceprint recognition model provided by the embodiment of the present invention.

[0020] Figure 3 It is a schematic diagram of the local regulatory circuit of the ventral cochlear nucleus in an embodiment provided by the present invention.

[0021] Figure 4 Schematic diagram of the connection of neurons in the cochlear nucleus bionic network in the embodiment provided by the present invention.

[0022] Figure 5 This is the second structural diagram of the voiceprint recognition model provided by the embodiment of the present invention.

[0023] Figure 6 This is the third structural diagram of the voiceprint recognition model provided by the embodiment of the present invention.

[0024] Figure 7 It is a structural schematic diagram of the auditory bionic voiceprint recognition device provided by the present invention.

[0025] Figure 8 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0026] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are 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.

[0027] Combine the following Figure 1-Figure 8 The present invention describes the auditory bionic voiceprint recognition method, device, electronic device and storage medium.

[0028] Figure 1 FIG. 1 is a flow chart showing a method for identifying an auditory bionic voiceprint according to an exemplary embodiment. Figure 1As shown, in an exemplary embodiment, the auditory bionic voiceprint recognition method includes steps 110 to 130, which are described in detail as follows.

[0029] Step 110: Acquire a speech signal to be recognized, and segment the speech signal to be recognized to obtain a plurality of speech segments to be recognized.

[0030] In the embodiment of the present invention, the speech signal to be recognized is cut into speech segments to be recognized, so that the voiceprint recognition model can more effectively capture local time and frequency features, which is crucial for tasks such as voiceprint recognition.

[0031] In a possible implementation, the speech signal to be recognized may be divided into speech segments to be recognized of 200 ms, and in the time dimension, two adjacent speech segments to be recognized overlap each other by 10 ms.

[0032] Step 120: input each of the speech segments to be recognized into a cochlear bionic filter to obtain speech frequency characteristics.

[0033] In the embodiment of the present invention, the speech segment to be recognized is input into the cochlear bionic filter, and the cochlear bionic filter simulates the selective response of the cochlear basilar membrane to different frequency signals by learning adjustable parameters. Each filter in the cochlear bionic filter decomposes the speech signal within a specific frequency range to obtain the decomposed speech frequency components, each component corresponding to the speech frequency characteristics of the specific frequency range.

[0034] Step 130, inputting the speech frequency feature into a pre-trained voiceprint recognition model to obtain a voiceprint recognition result; wherein the voiceprint recognition model is obtained based on a cochlear nucleus bionic network.

[0035] The voiceprint recognition model (CNAN, Cochlear Nucleus-inspired Auditory Network) provided in the embodiment of the present invention is a bionic computing model based on deep learning, which is used for voiceprint recognition. The voiceprint recognition model is obtained based on the cochlear nucleus bionic network, which can perform multi-level feature extraction on the input speech frequency features, capture the temporal features and spatial features of the signal, and then obtain the probability distribution of the speaker identity corresponding to the speech signal to realize voiceprint recognition.

[0036] In the embodiment of the present invention, by dividing the speech signal to be recognized into speech segments to be recognized, an initial feature set related to the original speech signal to be recognized is constructed. These speech segments to be recognized are input into the cochlear bionic filter. The cochlear bionic filter is based on the characteristics of the cochlear basilar membrane and simulates the response of hair cells to specific frequencies to directly extract speech frequency features from the original speech waveform efficiently and accurately, and then the voiceprint recognition model provided by the present invention is used to perform voiceprint recognition. A higher adaptability to noise and complex environments is achieved, thereby improving the reliability and efficiency of voiceprint recognition.

[0037] In an exemplary embodiment of the present invention, the voiceprint recognition model includes a first cochlear nucleus bionic network, a fully connected network, and an output layer connected in sequence; The first cochlear nucleus bionic network is used to extract temporal features and spatial features based on the speech frequency features, obtain temporal feature representation based on the temporal features and spatial features, and perform weighted fusion processing on the temporal feature representation and the speech frequency features to obtain speech feature representation; The fully connected network and the output layer are used to process the speech feature representation in sequence to obtain the voiceprint recognition result.

[0038] In the embodiment of the present invention, the structure of the voiceprint recognition model is as follows: Figure 2 As shown, the speech frequency features obtained based on the cochlear bionic filter are input into the cochlear nucleus bionic network for multi-level feature extraction to capture the temporal and spatial features of the signal, thereby obtaining the temporal feature representation of the speech signal, and then weightedly fused the temporal feature representation and the speech frequency feature to obtain the speech feature representation.

[0039] The weighted fused speech feature representation is input into the fully connected network for further processing. Finally, the probability distribution of the speaker identity corresponding to the speech signal is calculated through the output layer, and the maximum value is taken to obtain the speaker identity index, which is the final voiceprint recognition result.

[0040] Specifically, the weighted fused speech feature representation is flattened into a one-dimensional feature vector, and the fully connected network calculates the response score of each speaker based on the one-dimensional feature vector. Based on the response score, the output layer is used to calculate the probability distribution of each speaker, and then the final voiceprint recognition result is obtained.

[0041] After being integrated in the inferior colliculus, the speech signal continues to be transmitted along the auditory pathway to the medial geniculate body and the auditory cortex. Therefore, in order to simulate this process, the embodiment of the present invention uses a multi-layer fully connected network to complete the final classification task of the auditory cortex.

[0042] Specifically, the fully connected network is calculated as follows: ; ; ; ; ; ; in, , , , , , , , is the trainable weight matrix and error term, N represents the total number of categories in voiceprint recognition, A response score for each speaker.

[0043] In an exemplary embodiment of the present invention, the first cochlear nucleus bionic network includes a spiral ganglion neuron layer, a stellate cell layer, an attention mechanism and a plexus cell layer, the attention mechanism is connected to the spiral ganglion neuron layer, the stellate cell layer and the plexus cell layer respectively, and the spiral ganglion neuron layer is connected to the stellate cell layer; The spiral ganglion neuron layer is used to perform a convolution operation in the time direction based on the speech frequency feature to obtain a first bionic feature; The stellate cell layer is used to perform a convolution operation in the time direction based on the first bionic feature to obtain a second bionic feature; The attention mechanism is used to obtain an intermediate feature by performing weighted fusion based on the speech frequency feature and the second bionic feature through an attention parameter; The cluster cell layer is used to perform a convolution operation in the time direction based on the intermediate features to obtain the speech feature representation.

[0044] In the embodiment of the present invention, since the auditory nerve fibers are in the ventral cochlear nucleus, they control the main cell types in the core area, namely, globular plexus cells, globular plexus cells, T stellate cells and D stellate cells. In the core area of ​​the ventral cochlear nucleus, D stellate cells control plexus cells and T stellate cells to form a local regulatory loop.

[0045] like Figure 3 As shown, Figure 3The schematic diagram of the local regulatory circuit of the ventral cochlear nucleus provided by the present invention, wherein Sp is a spherical plexus cell, Ds is a D stellate cell, Ts is a T stellate cell, I-SG is a spiral ganglion neuron, and IHC is an inner hair cell. The main exogenous excitatory input of the ventral cochlear nucleus is the nerve fiber emitted by the spiral ganglion neurons connected to the inner hair cells of the cochlea, and the neurotransmitter released is glutamate. The cochlear bionic filter is based on the characteristics of the cochlear basilar membrane, simulates the response of hair cells to specific frequencies, and directly extracts frequency features from the original speech waveform efficiently and accurately, with bionic interpretability. Based on this, the embodiment of the present invention proposes a cochlear nucleus bionic network, which adds cochlear nucleus signal processing before performing the next layer of processing to increase the interpretability of the model signal processing.

[0046] In a specific embodiment, the connection mode between neurons can be divided into three categories, namely, full connection mode, multi-synaptic connection projection mode and monosynaptic connection mode. In the full connection mode, each stellate cell receives the synaptic projections of all spiral ganglion neurons, and each plexus cell receives the synaptic projections of all stellate cells. This method can be used as a transition when the specific connection mode has not been dissected out in the current biological field; in the multi-synaptic connection projection mode, a connection mode with a smaller receptive field is achieved; in the monosynaptic connection mode, each stellate cell is projected by a single spiral ganglion neuron, and each plexus cell also receives the projection of a single stellate cell. Finally, each neuron in the plexus cell layer receives the projection of a single spiral neuron and fuses it with the projection of the stellate cell, as shown in the following figure. Figure 4 As shown, Figure 4 Schematic diagram of three connection modes of neurons in the cochlear nucleus bionic network provided by the present invention.

[0047] The spiral ganglion neuron layer receives the output of the filter bank , as the input signal of the cochlear nucleus bionic network, it continues to be transmitted along the auditory pathway. The spiral ganglion neurons in the spiral ganglion neuron layer project synapses to stellate cells and plexus cells. The plexus cells are the main nuclei for ascending information transmission. Before transmitting the received information to the next layer, the plexus cells receive inhibitory information from the stellate cells, and then transmit the information to the next layer. This constitutes the information transmission calculation of the cochlear nucleus bionic network. Therefore, based on the above three connection modes, the connection mode between neurons in the cochlear nucleus bionic network provided in the embodiment of the present invention adopts a full connection mode, a two-synapse connection mode and a monosynapse connection mode, as shown in the specific example. Figure 4As shown in the figure, the fully connected mode and the existing method use the same transfer function, that is, one-dimensional convolution is used to summarize the information of all neurons in the previous layer and perform convolution operations in the time dimension. The two-synaptic connection mode and the monosynaptic connection mode use a more sparse method to transfer information, that is, the two-synaptic connection mode processes the information of two neurons in the previous layer and performs convolution operations in the time dimension, while the monosynaptic connection mode only processes the information of one neuron in the previous layer and performs convolution operations in the time dimension.

[0048] The calculation formula of the cochlear nucleus bionic network is as follows: In fully connected mode (1D convolution): ; ; In bisynaptic connection mode: ; ; In single prominent connection mode: ; ; in, represents the convolution operation in the time direction, Indicates a column-by-column merge operation. represents the activation function, , , , is a trainable multi-channel convolution kernel, is the trainable error term.

[0049] Figure 5 The schematic diagram of the structure of the voiceprint recognition model provided by the present invention is as follows: Figure 5 As shown in the figure, the spiral ganglion neuron layer performs convolution operation in the time direction based on the speech frequency feature to obtain the first bionic feature, and the stellate cell layer performs convolution operation in the time direction based on the first bionic feature to obtain the second bionic feature, thereby further refining the feature information of the speech signal. Through multi-level convolution operations, the features related to the speech signal are updated and integrated. This process can dynamically capture the timing and frequency neighborhood information of the speech signal and gradually strengthen important speech features.

[0050] The embodiment of the present invention simulates the process in which the plexus cells receive inhibitory information from stellate cells before transmitting the information received by the plexus cells to the next layer, and then transmits the information to the next layer. The embodiment of the present invention utilizes the attention mechanism and the convolutional layer to fuse and process the temporal feature representation of the speech signal to obtain a weighted speech feature representation, thereby providing accurate local information for voiceprint recognition. The weighted speech feature representation not only reflects the temporal and spatial features of different frequency components in the speech signal, but also includes the inter-frequency correlation information updated through the aggregation process. This feature extraction method ensures that the unique features of each speech signal can be fully retained, and provides multi-level speech function information for subsequent voiceprint recognition, which enhances the model's perception of key features.

[0051] In this embodiment of the present invention, the calculation method of the fusion of the attention mechanism and the convolutional layer is as follows: In full connection mode: ; ; In bisynaptic connection mode: ; ; In monosynaptic connection mode: ; ; in, , is a trainable attention parameter, is the trainable error term.

[0052] In an exemplary embodiment of the present invention, the output layer includes a Softmax function or an additive residual Softmax function.

[0053] In the embodiment of the present invention, the additive residual Softmax function is a method for improving the inter-class separation by adding a fixed angle residual on the decision boundary based on the traditional Softmax function. This improvement enables the model to better distinguish feature vectors of different categories in the feature space.

[0054] In an exemplary embodiment of the present invention, the voiceprint recognition model further includes a second cochlear nucleus bionic network, and the second cochlear nucleus bionic network is connected between the first cochlear nucleus bionic network and the fully connected network.

[0055] In the embodiment of the present invention, since the superior olivary complex and other nuclei have similar internal neuronal connection patterns, Figure 6As shown in FIG. 1 , a second cochlear nucleus bionic network consistent with the first cochlear nucleus bionic network is added between the first cochlear nucleus bionic network and the fully connected network, and then the output of the first cochlear nucleus bionic network is processed to extract the temporal feature representation, and then perform weighted fusion processing, and the output is recorded as .

[0056] In an exemplary embodiment of the present invention, the cochlear bionic filter comprises a filter group consisting of a plurality of filters with different response frequencies; wherein each of the filters respectively extracts the speech frequency features of the corresponding response frequency from the speech segment to be recognized.

[0057] In the embodiment of the present invention, the cochlear bionic filter is based on the characteristics of the cochlear basilar membrane, simulates the response of hair cells to specific frequencies, and directly extracts frequency features from the original speech waveform efficiently and accurately, which has bionic interpretability. Therefore, the speech segment to be recognized is input into the cochlear bionic filter, and the selective response of the cochlear basilar membrane to different frequency signals is simulated by learning adjustable parameters. Each filter decomposes the speech signal within a specific frequency range to obtain the decomposed speech frequency components, each of which corresponds to the characteristics of a specific frequency range.

[0058] In a possible implementation, two low-pass filters are used to generate a band-pass filter based on the theory of digital signal processing. In actual operation, in the time domain, the band-pass filter function is: ; in, are learnable low and high cutoff frequencies, Represents the time parameter, Given a discrete time speech signal , and a length of Bandpass filter , the output after filtering is: ; Further construct m filters with different response frequencies, where m is a natural number, to obtain a filter bank , the filter bank completes the response of the hair cells to a specific frequency and outputs , It is the length of the frame-level audio data after segmentation.

[0059] In an exemplary embodiment of the present invention, the training process of the voiceprint recognition model includes the following steps, which are described in detail as follows.

[0060] A training set is obtained; wherein the training set includes a plurality of training samples.

[0061] A loss function is constructed based on the training samples.

[0062] The training samples in the training set are input into a basic voiceprint recognition model, and iterative training is performed with minimizing the loss function as the training goal until the training is completed to obtain the voiceprint recognition model; wherein the basic voiceprint recognition model is obtained based on the cochlear nucleus bionic network.

[0063] In the embodiment of the present invention, a loss function of the voiceprint recognition model is set, and the loss function is expressed by the following formula: ; in, represents the total number of training samples in the training set, Indicates the total number of categories in voiceprint recognition. Represents the true probability distribution, that is, if the speech signal is The individual's voice .

[0064] The basic voiceprint recognition model is trained with the minimization of the loss function as the training objective, and the weight matrix of the basic voiceprint recognition model is iteratively updated to obtain the voiceprint recognition model through training.

[0065] The following is a description of the auditory bionic voiceprint recognition device provided by the present invention. The auditory bionic voiceprint recognition device described below and the auditory bionic voiceprint recognition method described above can be referred to each other. It should be noted that the device provided in the following embodiment and the method provided in the above embodiment belong to the same concept, and the specific manner in which each module and unit performs the operation has been described in detail in the method embodiment, which will not be repeated here.

[0066] In an exemplary embodiment of the present invention, see Figure 7 , Figure 7 An auditory bionic voiceprint recognition device according to an exemplary embodiment includes: The first acquisition module 710 is configured to acquire a speech signal to be recognized, and segment the speech signal to be recognized to obtain a plurality of speech segments to be recognized; The first input module 720 is configured to input each of the speech segments to be recognized into the cochlear bionic filter to obtain speech frequency characteristics; The second input module 730 is configured to input the speech frequency feature into a pre-trained voiceprint recognition model to obtain a voiceprint recognition result; wherein the voiceprint recognition model is obtained based on a cochlear nucleus bionic network.

[0067] In an exemplary embodiment of the present invention, the voiceprint recognition model includes a first cochlear nucleus bionic network, a fully connected network, and an output layer connected in sequence; The first cochlear nucleus bionic network is used to extract temporal features and spatial features based on the speech frequency features, obtain temporal feature representation based on the temporal features and spatial features, and perform weighted fusion processing on the temporal feature representation and the speech frequency features to obtain speech feature representation; The fully connected network and the output layer are used to process the speech feature representation in sequence to obtain the voiceprint recognition result.

[0068] In an exemplary embodiment of the present invention, the first cochlear nucleus bionic network includes a spiral ganglion neuron layer, a stellate cell layer, an attention mechanism and a plexus cell layer, the attention mechanism is connected to the spiral ganglion neuron layer, the stellate cell layer and the plexus cell layer respectively, and the spiral ganglion neuron layer is connected to the stellate cell layer; The spiral ganglion neuron layer is used to perform a convolution operation in the time direction based on the speech frequency feature to obtain a first bionic feature; The stellate cell layer is used to perform a convolution operation in the time direction based on the first bionic feature to obtain a second bionic feature; The attention mechanism is used to obtain an intermediate feature by performing weighted fusion based on the speech frequency feature and the second bionic feature through an attention parameter; The cluster cell layer is used to perform a convolution operation in the time direction based on the intermediate features to obtain the speech feature representation.

[0069] In an exemplary embodiment of the present invention, the output layer comprises a Softmax function or an additive residual Softmax function.

[0070] In an exemplary embodiment of the present invention, the voiceprint recognition model further includes a second cochlear nucleus bionic network, and the second cochlear nucleus bionic network is connected between the first cochlear nucleus bionic network and the fully connected network.

[0071] In an exemplary embodiment of the present invention, the cochlear bionic filter comprises a filter group consisting of a plurality of filters with different response frequencies; wherein each of the filters respectively extracts the speech frequency features of the corresponding response frequency from the speech segment to be recognized.

[0072] In an exemplary embodiment of the present invention, the auditory bionic voiceprint recognition device further includes: A second acquisition module is configured to acquire a training set; wherein the training set includes a plurality of training samples; A framework module configured to construct a loss function based on the training sample; An iterative training model is configured to input the training samples in the training set into a basic voiceprint recognition model, and perform iterative training with minimizing the loss function as the training goal until the training is completed to obtain the voiceprint recognition model; wherein the basic voiceprint recognition model is obtained based on the cochlear nucleus bionic network.

[0073] Figure 8 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 8 As shown, the electronic device may include: a processor 810, a communication interface 820, a memory 830 and a communication bus 840, wherein the processor 810, the communication interface 820 and the memory 830 communicate with each other through the communication bus 840. The processor 810 may call the logic instructions in the memory 830 to execute the auditory bionic voiceprint recognition method, which includes: obtaining a voice signal to be recognized, and cutting the voice signal to be recognized to obtain a plurality of voice segments to be recognized; Inputting each of the speech segments to be recognized into a cochlear bionic filter to obtain speech frequency characteristics; The speech frequency feature is input into a pre-trained voiceprint recognition model to obtain a voiceprint recognition result; wherein the voiceprint recognition model is obtained based on a cochlear nucleus bionic network.

[0074] In addition, the logic instructions in the above-mentioned memory 830 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0075] On the other hand, the present invention further provides a computer program product, the computer program product includes a computer program, the computer program can be stored in a non-transitory computer-readable storage medium, when the computer program is executed by a processor, the computer can execute the auditory bionic voiceprint recognition method provided by the above methods, the method includes: obtaining a speech signal to be recognized, and cutting the speech signal to be recognized to obtain a plurality of speech segments to be recognized; Inputting each of the speech segments to be recognized into a cochlear bionic filter to obtain speech frequency characteristics; The speech frequency feature is input into a pre-trained voiceprint recognition model to obtain a voiceprint recognition result; wherein the voiceprint recognition model is obtained based on a cochlear nucleus bionic network.

[0076] In another aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is implemented when the computer program is executed by a processor to perform the auditory bionic voiceprint recognition method provided by the above methods, the method comprising: obtaining a speech signal to be recognized, and performing segmentation processing on the speech signal to be recognized to obtain a plurality of speech segments to be recognized; Inputting each of the speech segments to be recognized into a cochlear bionic filter to obtain speech frequency characteristics; The speech frequency feature is input into a pre-trained voiceprint recognition model to obtain a voiceprint recognition result; wherein the voiceprint recognition model is obtained based on a cochlear nucleus bionic network.

[0077] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0078] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0079] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An auditory bionic voiceprint recognition method, characterized in that: include: Acquire a speech signal to be recognized, and segment the speech signal to be recognized to obtain a plurality of speech segments to be recognized; Inputting each of the speech segments to be recognized into a cochlear bionic filter to obtain speech frequency characteristics; The speech frequency feature is input into a pre-trained voiceprint recognition model to obtain a voiceprint recognition result; wherein the voiceprint recognition model is obtained based on a cochlear nucleus bionic network.

2. The auditory bionic voiceprint recognition method according to claim 1, characterized in that: The voiceprint recognition model includes a first cochlear nucleus bionic network, a fully connected network and an output layer connected in sequence; The first cochlear nucleus bionic network is used to extract temporal features and spatial features based on the speech frequency features, obtain temporal feature representation based on the temporal features and spatial features, and perform weighted fusion processing on the temporal feature representation and the speech frequency features to obtain speech feature representation; The fully connected network and the output layer are used to process the speech feature representation in sequence to obtain the voiceprint recognition result.

3. The auditory bionic voiceprint recognition method according to claim 2, characterized in that: The first cochlear nucleus bionic network includes a spiral ganglion neuron layer, a stellate cell layer, an attention mechanism and a plexus cell layer, wherein the attention mechanism is connected to the spiral ganglion neuron layer, the stellate cell layer and the plexus cell layer respectively, and the spiral ganglion neuron layer is connected to the stellate cell layer; The spiral ganglion neuron layer is used to perform a convolution operation in the time direction based on the speech frequency feature to obtain a first bionic feature; The stellate cell layer is used to perform a convolution operation in the time direction based on the first bionic feature to obtain a second bionic feature; The attention mechanism is used to obtain an intermediate feature by performing weighted fusion based on the speech frequency feature and the second bionic feature through an attention parameter; The cluster cell layer is used to perform a convolution operation in the time direction based on the intermediate features to obtain the speech feature representation.

4. The auditory bionic voiceprint recognition method according to claim 2, characterized in that: The output layer includes a Softmax function or an additive residual Softmax function.

5. The auditory bionic voiceprint recognition method according to claim 2, characterized in that: The voiceprint recognition model also includes a second cochlear nucleus bionic network, which is connected between the first cochlear nucleus bionic network and the fully connected network.

6. The auditory bionic voiceprint recognition method according to claim 1, characterized in that: The cochlear bionic filter comprises a filter group composed of a plurality of filters with different response frequencies; wherein each of the filters respectively extracts the speech frequency feature of the corresponding response frequency from the speech segment to be recognized.

7. The auditory bionic voiceprint recognition method according to any one of claims 1 to 6, characterized in that: The training process of the voiceprint recognition model includes: Acquire a training set; wherein the training set includes a plurality of training samples; Constructing a loss function based on the training samples; The training samples in the training set are input into a basic voiceprint recognition model, and iterative training is performed with minimizing the loss function as the training goal until the training is completed to obtain the voiceprint recognition model; wherein the basic voiceprint recognition model is obtained based on the cochlear nucleus bionic network.

8. An auditory bionic voiceprint recognition device, characterized in that: include: A first acquisition module is configured to acquire a speech signal to be recognized, and segment the speech signal to be recognized to obtain a plurality of speech segments to be recognized; A first input module is configured to input each of the speech segments to be recognized into a cochlear bionic filter to obtain speech frequency characteristics; The second input module is configured to input the speech frequency feature into a pre-trained voiceprint recognition model to obtain a voiceprint recognition result; wherein the voiceprint recognition model is obtained based on a cochlear nucleus bionic network.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the auditory bionic voiceprint recognition method as described in any one of claims 1 to 7 is implemented.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the auditory bionic voiceprint recognition method according to any one of claims 1 to 7 is implemented.

Images