Method and device for determining identity of target object, storage medium and electronic device

Abstract

The present invention discloses a method and device for determining the identity of a target object, a storage medium and an electronic device, wherein the method is applied to an intelligent voice device, and the intelligent voice device is equipped with a clustering classifier. The method comprises: extracting features of first voice data emitted by a first target object by a voice feature extractor to obtain a first voice feature of the first voice data; inputting the first voice feature into the clustering classifier to obtain a target identity type of the first target object corresponding to the first voice feature; sending the target identity type to a cloud server, and receiving the identity information of the first target object returned by the cloud server according to the target identity type. The above technical solution is adopted to solve the problem that the existing identity recognition system requires users to register in advance, which will lead to the complexity of the device and the unintelligence of the device.

Description

Technical Field

[0001] The present invention relates to the field of communications, and in particular to a method and device for determining the identity of a target object, a storage medium and an electronic device. Background Art

[0002] With the advancement of science and technology and the development of artificial intelligence, intelligent algorithms are increasingly used in daily life. Especially for home appliances, their intelligent development is crucial. The most critical issue of intelligence is to intelligently solve the urgent needs of users.

[0003] Existing identity recognition systems that recognize user identities through voice usually require users to pre-register voice data, extract user voice features (such as voiceprint features) based on the voice data, and then, when the identity recognition system obtains the voice data of the first user, compare the voice features corresponding to the voice data of the first user with the voice features of registered users, take the closest result as the recognition result, and then determine the identity information of the first user. However, based on the existing identity recognition system, users are required to pre-register, which will bring complexity and unintelligence to the device design.

[0004] Regarding the related technologies, the existing identity recognition system requires users to register in advance, which will lead to the complexity and unintelligence of the equipment. No effective solution has been proposed yet. Summary of the invention

[0005] The embodiments of the present invention provide a method and device for determining the identity of a target object, a storage medium and an electronic device, so as to at least solve the problem in the related art that the existing identity recognition system requires users to register in advance, which leads to the complexity and unintelligence of the device.

[0006] According to an embodiment of the present invention, a method for determining the identity of a target object is provided, which is applied to an intelligent voice device, wherein the intelligent voice device is configured with a cluster classifier, including: performing feature extraction on first voice data emitted by a first target object through a voice feature extractor to obtain a first voice feature of the first voice data; inputting the first voice feature into the cluster classifier to obtain a target identity type of the first target object corresponding to the first voice feature; sending the target identity type to a cloud server, and receiving the identity information of the first target object returned by the cloud server according to the target identity type.

[0007] In an exemplary embodiment, a plurality of second voice data emitted by a second target object is obtained, and a plurality of second voice features of the plurality of second voice data are respectively extracted, wherein the second target object includes one or more objects; the plurality of second voice features are clustered by an initial clusterer to determine one or more clustering categories, wherein the clustering category is used to indicate the identity type of the second target object corresponding to any second voice feature among the plurality of second voice features; the parameters of the initial clusterer are iteratively updated according to the clustering category until a training end condition is met, thereby obtaining the clustering classifier.

[0008] In an exemplary embodiment, the multiple second speech features are clustered by an initial clusterer to determine one or more clustering categories, including: clustering the multiple second speech features by an initial clusterer to determine the feature distance between any two second speech features among the multiple second speech features; when the feature distance is less than or equal to a preset threshold, determining that the any two second speech features are in the same clustering category.

[0009] In an exemplary embodiment, the number of clustering categories is determined based on a received first target instruction, wherein the first target instruction is used to indicate the number of identity types that the intelligent voice device can recognize, and the first target instruction is an instruction sent by the second target object through a terminal device, and the second target object is an object that has not registered a voiceprint on the voiceprint recognition system corresponding to the intelligent voice device.

[0010] In an exemplary embodiment, after receiving the identity information of the first target object returned by the cloud server according to the target identity type, the method further includes: playing a voice registration request based on the identity information obtained from the cloud; and upon receiving a confirmation instruction from the first target object in response to the voice registration request, registering the first target object on the smart voice device based on the identity information and the confirmation instruction.

[0011] In an exemplary embodiment, the cloud server is configured with identity information of M user objects, the identity information of the first target object is obtained by matching the target identity type of the first target object with the identity information of the M user objects, and the identity information of the M user objects is determined based on the user object bound to the terminal device.

[0012] In an exemplary embodiment, after sending the target identity type to the cloud server, the method further includes: receiving a target instruction sent by the cloud server based on the identity information of the first target object when the cloud server determines the identity information of the first target object based on the target identity type; and responding to the target instruction so that the target home appliance performs a target operation corresponding to the identity information according to the target instruction.

[0013] According to another embodiment of the embodiments of the present invention, there is also provided an identity determination device for a target object, comprising: an acquisition module, used to perform feature extraction on first voice data emitted by a first target object through a voice feature extractor, and obtain a first voice feature of the first voice data; a determination module, used to input the first voice feature into the clustering classifier, and obtain a target identity type of the first target object corresponding to the first voice feature; a receiving module, used to send the target identity type to a cloud server, and receive the identity information of the first target object returned by the cloud server according to the target identity type.

[0014] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is provided, in which a computer program is stored, wherein the computer program is configured to execute the above-mentioned method for determining the identity of the target object when running.

[0015] According to another aspect of an embodiment of the present invention, there is provided an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the target object identity determination method through the computer program.

[0016] In an embodiment of the present invention, a voice feature extractor is used to perform feature extraction on first voice data emitted by a first target object to obtain a first voice feature of the first voice data; the first voice feature is input into the clustering classifier to obtain a target identity type of the first target object corresponding to the first voice feature; the target identity type is sent to a cloud server, and the cloud server receives identity information of the first target object returned according to the target identity type. That is, the embodiment of the present invention does not require the target object to be registered in advance, but only needs to determine the target identity type of the first target object corresponding to the first voice feature through the clustering classifier of the intelligent voice device, and the cloud determines the identity information of the first target object according to the target identity type; the above technical solution is adopted to solve the problem of the existing identity recognition system requiring users to register in advance, which will lead to the complexity and unintelligence of the device, thereby making the intelligent voice device more intelligent and expanding the scope of application. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0018] Figure 1 It is a hardware structure block diagram of a mobile terminal of a method for determining the identity of a target object according to an embodiment of the present invention;

[0019] Figure 2 is a flow chart of a method for determining the identity of a target object according to an embodiment of the present invention;

[0020] Figure 3 is a schematic diagram of a method for determining the identity of a target object according to an optional embodiment of the present invention;

[0021] Figure 4 is a structural block diagram of a device for determining the identity of a target object according to an embodiment of the present invention (I);

[0022] Figure 5 2 is a structural block diagram of a device for determining the identity of a target object according to an embodiment of the present invention. DETAILED DESCRIPTION

[0023] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0024] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0025] The method embodiments provided in the embodiments of the present application can be executed in a mobile terminal or a similar computing device. Taking running on a mobile terminal as an example, Figure 1 FIG. 1 is a hardware structure block diagram of a mobile terminal of a method for determining the identity of a target object according to an embodiment of the present invention. Figure 1 As shown, the mobile terminal may include one or more ( Figure 1 Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data. In an exemplary embodiment, the mobile terminal may also include a transmission device 106 and an input / output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the mobile terminal. Figure 1 More or fewer components as shown, or with Figure 1 Equivalent functions or comparisons shown Figure 1 A different configuration with more features is shown.

[0026] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the method for determining the identity of the target object in the embodiment of the present invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, to implement the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the mobile terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0027] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the mobile terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0028] In this embodiment, a method for determining the identity of a target object is provided, which is applied to the above-mentioned mobile terminal, specifically an intelligent voice device, wherein the intelligent voice device is configured with a clustering classifier. Figure 2: is a flow chart of a method for determining the identity of a target object according to an embodiment of the present invention, the flow chart includes the following steps:

[0029] Step S202, extracting features from first speech data emitted by a first target object by a speech feature extractor to obtain first speech features of the first speech data;

[0030] It should be noted that the intelligent voice device is configured with a voice feature extractor. If the intelligent voice device is not configured with a voice feature extractor, the first voice data emitted by the first target object needs to be sent to an external voice feature extractor. The intelligent voice device can be a multimedia playback and display device such as a speaker, television, or iPad.

[0031] Step S204, inputting the first speech feature into the cluster classifier to obtain a target identity type of the first target object corresponding to the first speech feature;

[0032] It should be noted that the target identity types include but are not limited to: age and gender.

[0033] Step S206: sending the target identity type to a cloud server, and receiving identity information of the first target object returned by the cloud server according to the target identity type.

[0034] Through the above steps, the first voice data emitted by the first target object is feature extracted by the voice feature extractor to obtain the first voice feature of the first voice data; the first voice feature is input into the clustering classifier to obtain the target identity type of the first target object corresponding to the first voice feature; the target identity type is sent to the cloud server, and the identity information of the first target object returned by the cloud server according to the target identity type is received, that is, the embodiment of the present invention does not require the target object to be registered in advance, but only needs to determine the target identity type of the first target object corresponding to the first voice feature through the clustering classifier of the intelligent voice device, and determine the identity information of the first target object according to the target identity type; the above technical solution is adopted to solve the problem of the existing identity recognition system, which requires users to register in advance, which will lead to the complexity and unintelligence of the device, thereby making the intelligent voice device more intelligent and expanding the scope of application.

[0035] For example, based on the first voice feature of the first target object, the target identity type of the first target object is determined to be a middle-aged man. The cloud server determines all the corresponding family members in the account information corresponding to the intelligent voice device, and determines the identity information of the first target object among the family members based on the identity category of middle-aged man.

[0036] In an exemplary embodiment, before the first speech feature is input into the clustering classifier to obtain the target identity type of the first target object corresponding to the first speech feature, multiple second speech data emitted by the second target object are obtained, and multiple second speech features of the multiple second speech data are respectively extracted, wherein the second target object includes: one or more objects; clustering the multiple second speech features through an initial clusterer to determine one or more clustering categories, wherein the clustering category is used to indicate the identity type of the second target object corresponding to any second speech feature; iteratively updating the parameters of the initial clustering classifier according to the clustering category until the training end condition is met to obtain the clustering classifier.

[0037] That is to say, the initial clustering classifier is trained by the second voice data. Specifically, when the intelligent voice device is running, the intelligent voice device collects the second voice data of the second target object, and extracts the second voice features corresponding to the second voice data. For example, the second voice features are represented by q0, q1, q2, ..., qt, and the second voice feature vector is clustered and analyzed by a preset algorithm; the initial clustering classifier is trained according to the clustering category. When the initial clustering classifier of the intelligent voice device is maturely trained, the intelligent voice device stops updating the parameters of the clustering classifier to obtain the trained clustering classifier, wherein the preset algorithms include: k-means, k-medoids, k-modes, k-medians, kernelk-means.

[0038] In an exemplary embodiment, the multiple second speech features are clustered by an initial clusterer to determine one or more clustering categories, including: clustering the multiple second speech features by an initial clusterer to determine the feature distance between any two second speech features among the multiple second speech features; when the feature distance is less than or equal to a preset threshold, determining that the any two second speech features are in the same clustering category.

[0039] That is to say, according to the feature distance principle, multiple second voice features are divided into one or more cluster categories. Specifically, the feature distance of any two second voice features among the multiple second voice features is determined according to the feature difference scorer of the intelligent voice device, and whether any two second voice features belong to the same cluster category is determined according to the feature distance, and then one or more cluster categories are determined from the multiple second voice features. It should be noted that the feature distance between any two second voice features is inversely proportional to the probability that any two second voice features belong to the same target object. The smaller the feature distance between any two second voice features, the greater the probability that any two second voice features belong to the same target object.

[0040] In an exemplary embodiment, the number of clustering categories is determined based on a received first target instruction, wherein the first target instruction is used to indicate the number of identity types that the intelligent voice device can recognize; the first target instruction is an instruction sent by the second target object through a terminal device, and the second target object is an object whose voiceprint is not registered on the voiceprint recognition system corresponding to the intelligent voice device.

[0041] Specifically, after the intelligent voice device deploys the non-registered speaker identity recognition algorithm, when the intelligent voice device is running, the second target object sets the maximum number of identity category discriminations supported by the intelligent voice device through the target terminal; thus, the number of cluster categories of the intelligent voice device is the maximum number of identity category discriminations, that is, the maximum number of cluster categories can be determined according to the first target instruction.

[0042] Furthermore, after receiving the identity information of the first target object returned by the cloud server according to the target identity type, a voice registration request is played according to the identity information obtained from the cloud; when a confirmation instruction of the first target object in response to the voice registration request is received, the first target object is registered on the intelligent voice device according to the identity information and the confirmation instruction.

[0043] It should be noted that the intelligent voice device sends a request to the target object to register the first target object on the intelligent voice device based on the acquired identity information, and determines whether to register the first target object on the intelligent voice device based on the response of the first target object.

[0044] Optionally, the cloud server is configured with identity information of M user objects, the identity information of the first target object is obtained by matching the target identity type of the first target object with the identity information of the M user objects, and the identity information of the M user objects is determined according to the user object bound to the terminal device.

[0045] In an exemplary embodiment, before extracting the first voice feature of the first voice data through a trained voice feature extractor, multiple third voice data are obtained from the database of the cloud server, and multiple third voice features corresponding to the multiple third voice data are determined; the voice feature extractor is trained with the multiple third voice data and the multiple third voice features to adjust the parameters of the voice feature extractor to obtain a trained voice feature extractor.

[0046] That is to say, by obtaining multiple third voice data in the cloud server, and respectively determining the third voice features corresponding to the multiple third voice data, the third voice data is used as the input of the voice feature extractor, and the third voice features are used as the output of the voice feature extractor, and the voice feature extractor is trained according to the third voice data and the third voice features, for example, an iverse extractor or a deep learning neural network model.

[0047] In an exemplary embodiment, after obtaining a trained speech feature extractor, the method includes: inputting any third speech data among the multiple third speech data into the trained speech feature extractor to obtain a third speech feature corresponding to the any third speech data; and training a feature difference scorer in the intelligent voice device at least according to the third speech feature corresponding to the any third speech data.

[0048] That is, a feature difference scorer S is trained based on the trained feature extractor, and two speech features extracted by the trained feature extractor are input, and the possibility that the speech features belong to the same speaker is output, that is, S(q0,q1). The higher the value, the greater the possibility that the speech feature q0 and the speech feature q1 belong to the same person.

[0049] In an exemplary embodiment, after the target identity type is sent to a cloud server, when the cloud server determines the identity information of the first target object based on the target identity type, a target instruction sent by the cloud server based on the identity information of the first target object is received; and the target instruction is responded to so that the target home appliance performs a target operation corresponding to the identity information according to the target instruction.

[0050] Specifically, the cloud server stores information about family members under the account corresponding to the smart voice device. The cloud server determines the identity information of the first target object based on the target identity type of the target object. For example, if the first target object is an elderly person in the family, an instruction to increase the weight of opera is pushed to the speaker, so that the target home appliance pushes the opera according to the target instruction.

[0051] In order to better understand the process of the above-mentioned target object identity determination method, the implementation method flow of the above-mentioned target object identity determination is described below in combination with an optional embodiment, but it is not used to limit the technical solution of the embodiment of the present invention.

[0052] In this embodiment, a method for determining the identity of a target object is provided. Figure 3 is a schematic diagram of a method for determining the identity of a target object according to an optional embodiment of the present invention. Figure 3 As shown, the specific steps are as follows:

[0053] Step S301: Collect a large amount of third voice data on the same type of device as device A (equivalent to the intelligent voice device in the above embodiment), such as 5,000 people, 1,000 audio sentences per person, and train a basic feature extractor V (equivalent to the voice feature extractor in the above embodiment) (such as an ictor extractor or a deep learning neural network) through the third voice data, and at the same time train a feature difference scorer S based on the basic feature extractor, that is, input the voice features extracted by the two basic feature extractors into the feature difference scorer S, and output the possibility that the two voice features belong to the same speaker, that is, S(q0,q1). The higher the value, the greater the possibility that the voice feature q0 and the voice feature q1 belong to the same person;

[0054] Step S302: After the non-registration speaker identification algorithm is deployed on device A, when device A is running, the user pre-sets the maximum number of user identifications M supported by device A;

[0055] Step S303: During the operation of device A, device A collects the second voice data and extracts the second voice features. Assuming that the feature sequence of the second voice features is represented by q0, q1, q2, ..., qt, a cluster analysis is performed on the second voice features through a preset algorithm, and then a cluster classifier is trained. The default number of cluster categories is M (equivalent to the cluster categories in the above embodiment), and the feature distance is realized by a feature difference scorer S, that is, the larger S(qi, qj), the smaller the feature distance between the voice feature qi and the voice feature qj;

[0056] Step S304: when the cluster classifier of device A is trained to maturity, device A stops updating the parameters of the cluster classifier and obtains cluster classifier Y;

[0057] Step S305: when device A obtains the first voice data, extract the first voice feature qt of the first voice data through the basic feature extractor V, and obtain the target identity category Y(qt) of the target object corresponding to qt using the clustering classifier Y;

[0058] Step S306: The cloud server stores the real identities of M speakers. Device A uploads the target identity category Y (qt) to the cloud server. The cloud server intelligently obtains the identity information of the target object and instructs device A to respond accordingly. For example, if the target object is an elderly person at home, the weight of opera will be increased when pushing songs to be played; if the target object is a child at home, the weight of nursery rhymes will be increased, etc.

[0059] According to the embodiment of the present invention, a voice feature extractor is used to extract features of first voice data emitted by a first target object to obtain a first voice feature of the first voice data; the first voice feature is input into the clustering classifier to obtain a target identity type of the first target object corresponding to the first voice feature; the target identity type is sent to a cloud server, and the cloud server receives identity information of the first target object returned according to the target identity type. That is, the embodiment of the present invention does not require the target object to be registered in advance, but only needs to determine the target identity type of the first target object corresponding to the first voice feature through the clustering classifier of the intelligent voice device, and the cloud determines the identity information of the first target object according to the target identity type; the above technical solution is adopted to solve the problem of the existing identity recognition system requiring users to register in advance, which will lead to the complexity and unintelligence of the device, thereby making the intelligent voice device more intelligent and expanding the scope of application.

[0060] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially 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 device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods of various embodiments of the present invention.

[0061] In this embodiment, a device for determining the identity of a target object is provided. Figure 4 1 is a structural block diagram of a device for determining the identity of a target object according to an embodiment of the present invention (I); Figure 4 As shown, including:

[0062] An acquisition module 42 is used to extract features from first speech data emitted by a first target object through a speech feature extractor to obtain first speech features of the first speech data;

[0063] A determination module 44, configured to input the first speech feature into the cluster classifier to obtain a target identity type of the first target object corresponding to the first speech feature;

[0064] The receiving module 46 is configured to send the target identity type to a cloud server, and receive the identity information of the first target object returned by the cloud server according to the target identity type.

[0065] Through the above-mentioned device, a voice feature extractor is used to extract features of the first voice data emitted by the first target object to obtain the first voice feature of the first voice data; the first voice feature is input into the clustering classifier to obtain the target identity type of the first target object corresponding to the first voice feature; the target identity type is sent to the cloud server, and the identity information of the first target object returned by the cloud server according to the target identity type is received, that is, the embodiment of the present invention does not require the target object to be registered in advance, but only needs to determine the target identity type of the first target object corresponding to the first voice feature through the clustering classifier of the intelligent voice device, and the cloud determines the identity information of the first target object according to the target identity type; the above-mentioned technical solution is adopted to solve the problem of the existing identity recognition system, which requires users to register in advance, which will lead to the complexity and unintelligence of the device, thereby making the intelligent voice device more intelligent and expanding the scope of application.

[0066] For example, based on the first voice feature of the first target object, the target identity type of the first target object is determined to be a middle-aged man. The cloud server determines all the corresponding family members in the account information corresponding to the intelligent voice device, and determines the identity information of the first target object among the family members based on the identity category of middle-aged man.

[0067] In an exemplary embodiment, Figure 5 is a structural block diagram (II) of a device for determining the identity of a target object according to an embodiment of the present invention; Figure 5 As shown, the above-mentioned device also includes: a training module 52, which is used to obtain multiple second voice data emitted by a second target object, and respectively extract multiple second voice features of the multiple second voice data, wherein the second target object includes: one or more objects; clustering the multiple second voice features through an initial clusterer to determine one or more clustering categories, wherein the clustering category is used for the identity type of the second target object corresponding to any second voice feature; iteratively updating the parameters of the initial clustering classifier according to the clustering category until the training end condition is met to obtain the clustering classifier.

[0068] That is to say, the initial clustering classifier is trained by the second voice data. Specifically, when the intelligent voice device is running, the intelligent voice device collects the second voice data of the second target object, and extracts the second voice features corresponding to the second voice data. For example, the second voice features are represented by q0, q1, q2, ..., qt, and the second voice feature vector is clustered and analyzed by a preset algorithm; the initial clustering classifier is trained according to the clustering category. When the initial clustering classifier of the intelligent voice device is maturely trained, the intelligent voice device stops updating the parameters of the clustering classifier to obtain the trained clustering classifier, wherein the preset algorithms include: k-means, k-medoids, k-modes, k-medians, kernel k-means.

[0069] In an exemplary embodiment, the determination module is also used to cluster the multiple second speech features through an initial clusterer to determine the feature distance between any two of the multiple second speech features; when the feature distance is less than or equal to a preset threshold, determine that the any two second speech features are in the same clustering category.

[0070] That is to say, according to the feature distance principle, multiple second voice features are divided into one or more cluster categories. Specifically, the feature distance of any two second voice features among the multiple second voice features is determined according to the feature difference scorer of the intelligent voice device, and whether any two second voice features belong to the same cluster category is determined according to the feature distance, and then one or more cluster categories are determined from the multiple second voice features. It should be noted that the feature distance between any two second voice features is inversely proportional to the probability that any two second voice features belong to the same target object. The smaller the feature distance between any two second voice features, the greater the probability that any two second voice features belong to the same target object.

[0071] In an exemplary embodiment, the determination module is also used to determine the number of clustering categories based on a received first target instruction, wherein the first target instruction is used to indicate the number of identity types that the intelligent voice device can recognize; the first target instruction is an instruction sent by the second target object through a terminal device, and the second target object is an object whose voiceprint is not registered on the voiceprint recognition system corresponding to the intelligent voice device.

[0072] Specifically, after the intelligent voice device deploys the non-registered speaker identity recognition algorithm, when the intelligent voice device is running, the second target object sets the maximum number of identity category discriminations supported by the intelligent voice device through the target terminal; thus, the number of cluster categories of the intelligent voice device is the maximum number of identity category discriminations, that is, the maximum number of cluster categories can be determined according to the first target instruction.

[0073] Furthermore, the determination module is also used to play a voice registration request based on the identity information obtained from the cloud; when a confirmation instruction from the first target object in response to the voice registration request is received, the first target object is registered on the intelligent voice device based on the identity information and the confirmation instruction.

[0074] Optionally, the determination module is also used to configure the cloud server with identity information of M user objects, the identity information of the first target object is obtained by matching the target identity type of the first target object and the identity information of the M user objects, and the identity information of the M user objects is determined based on the user object bound to the terminal device.

[0075] In an exemplary embodiment, the training module is also used to obtain multiple third voice data from the database of the cloud server, and to determine multiple third voice features corresponding to the multiple third voice data respectively; the multiple third voice data and the multiple third voice features are used to train the voice feature extractor to adjust the parameters of the voice feature extractor to obtain a trained voice feature extractor.

[0076] That is to say, by obtaining multiple third voice data in the cloud server, and respectively determining the third voice features corresponding to the multiple third voice data, the third voice data is used as the input of the voice feature extractor, and the third voice features are used as the output of the voice feature extractor, and the voice feature extractor is trained according to the third voice data and the third voice features, for example, an iverse extractor or a deep learning neural network model.

[0077] In an exemplary embodiment, the training module is also used to input any third voice data among the multiple third voice data into the trained voice feature extractor to obtain the third voice feature corresponding to the any third voice data; and to train the feature difference scorer in the intelligent voice device at least according to the third voice feature corresponding to the any third voice data.

[0078] That is, a feature difference scorer S is trained based on the trained feature extractor, and two speech features extracted by the trained feature extractor are input, and the possibility that the speech features belong to the same speaker is output, that is, S(q0,q1). The higher the value, the greater the possibility that the speech feature q0 and the speech feature q1 belong to the same person.

[0079] In an exemplary embodiment, the acquisition module is also used to receive a target instruction sent by the cloud server based on the identity information of the first target object when the cloud server determines the identity information of the first target object based on the target identity type; and respond to the target instruction so that the target home appliance performs a target operation corresponding to the identity information according to the target instruction.

[0080] Specifically, the cloud server stores information about family members under the account corresponding to the smart voice device. The cloud server determines the identity information of the first target object based on the target identity type of the target object. For example, if the first target object is an elderly person in the family, an instruction to increase the weight of opera is pushed to the speaker, so that the target home appliance pushes the opera according to the target instruction.

[0081] An embodiment of the present invention further provides a storage medium, which includes a stored program, wherein the program executes any of the above methods when it is run.

[0082] Optionally, in this embodiment, the storage medium may be configured to store program codes for executing the following steps:

[0083] S1, extracting features from first speech data emitted by a first target object by a speech feature extractor to obtain first speech features of the first speech data;

[0084] S2, inputting the first speech feature into the cluster classifier to obtain a target identity type of the first target object corresponding to the first speech feature;

[0085] S3, sending the target identity type to a cloud server, and receiving the identity information of the first target object returned by the cloud server according to the target identity type.

[0086] An embodiment of the present invention further provides an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0087] Optionally, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0088] Optionally, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0089] S1, extracting features from first speech data emitted by a first target object by a speech feature extractor to obtain first speech features of the first speech data;

[0090] S2, inputting the first speech feature into the cluster classifier to obtain a target identity type of the first target object corresponding to the first speech feature;

[0091] S3, sending the target identity type to a cloud server, and receiving the identity information of the first target object returned by the cloud server according to the target identity type.

[0092] Optionally, in this embodiment, the above-mentioned storage medium may include but is not limited to: a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk, and other media that can store program codes.

[0093] Optionally, the specific examples in this embodiment may refer to the examples described in the above embodiments and optional implementation modes, and this embodiment will not be described in detail here.

[0094] Obviously, those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, and optionally, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order than here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0095] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for determining the identity of a target object, characterized in that: Applied to an intelligent voice device, the intelligent voice device is configured with a clustering classifier, and the method comprises: Extracting features of first voice data emitted by a first target object through a voice feature extractor to obtain first voice features of the first voice data, wherein the first target object is an object whose voiceprint has not been registered on the voiceprint recognition system corresponding to the intelligent voice device; Inputting the first speech feature into the clustering classifier to obtain a target identity type of the first target object corresponding to the first speech feature; Sending the target identity type to a cloud server, and receiving identity information of the first target object returned by the cloud server according to the target identity type; Among them, after receiving the identity information of the first target object returned by the cloud server according to the target identity type, the method also includes: playing a voice registration request according to the identity information obtained from the cloud; and when a confirmation instruction of the first target object in response to the voice registration request is received, registering the first target object on the intelligent voice device according to the identity information and the confirmation instruction.

2. The method for determining the identity of a target object according to claim 1, characterized in that: Also includes: Acquire a plurality of second voice data emitted by a second target object, and respectively extract a plurality of second voice features of the plurality of second voice data, wherein the second target object includes one or more objects; Clustering the plurality of second speech features by an initial clusterer to determine one or more cluster categories, wherein the cluster category is used to indicate an identity type of a second target object corresponding to any second speech feature among the plurality of second speech features; The parameters of the initial clusterer are iteratively updated according to the clustering categories until a training end condition is met to obtain the clustering classifier.

3. The method for determining the identity of a target object according to claim 2, characterized in that: Clustering the plurality of second speech features by an initial clusterer to determine one or more clustering categories includes: Clustering the plurality of second speech features by an initial clusterer to determine a feature distance between any two second speech features among the plurality of second speech features; When the feature distance is less than or equal to a preset threshold, it is determined that the any two second speech features belong to the same cluster category.

4. The method for determining the identity of a target object according to claim 2 or 3, characterized in that: The method further comprises: According to the received first target instruction, the number of clustering categories is determined, wherein the first target instruction is used to indicate the number of identity types recognized by the intelligent voice device, the first target instruction is an instruction sent by the second target object through the terminal device, and the second target object is an object whose voiceprint is not registered on the voiceprint recognition system corresponding to the intelligent voice device.

5. The method for determining the identity of a target object according to any one of claims 1 to 3, characterized in that: Also includes: The cloud server is configured with identity information of M user objects, the identity information of the first target object is obtained by matching the target identity type of the first target object with the identity information of the M user objects, and the identity information of the M user objects is determined according to the user object bound to the terminal device.

6. The method for determining the identity of a target object according to any one of claims 1 to 3, characterized in that: After sending the target identity type to the cloud server, the method further includes: In a case where the cloud server determines the identity information of the first target object according to the target identity type, receiving a target instruction sent by the cloud server according to the identity information of the first target object; Respond to the target instruction so that the intelligent voice device performs a target operation corresponding to the identity information according to the target instruction.

7. A device for determining the identity of a target object, characterized in that: include: an acquisition module, configured to extract features from first voice data emitted by a first target object through a voice feature extractor to obtain first voice features of the first voice data, wherein the first target object is an object whose voiceprint has not been registered on the voiceprint recognition system corresponding to the intelligent voice device; a determination module, configured to input the first speech feature into a cluster classifier to obtain a target identity type of the first target object corresponding to the first speech feature; A receiving module, configured to send the target identity type to a cloud server, and receive identity information of the first target object returned by the cloud server according to the target identity type; Among them, the determination module is also used to play a voice registration request based on the identity information obtained from the cloud; when a confirmation instruction from the first target object in response to the voice registration request is received, the first target object is registered on the intelligent voice device based on the identity information and the confirmation instruction.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein the program executes the method described in any one of claims 1 to 6 when executed.

9. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to execute the method according to any one of claims 1 to 6 through the computer program.

Images