Model training method, sound recognition method, equipment and storage medium
By extracting and training the animal's call, a call recognition model is generated, which solves the problem of low accuracy in humans' emotional recognition of animals and improves communication and interaction between humans and animals.
Patent Information
- Application Number
- CN202411495053.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-24
- Publication Date
- 2025-03-21
AI Technical Summary
The prior art cannot ensure that humans can accurately identify the emotions of animals, resulting in poor communication and interaction between humans and animals.
By obtaining the audio parameters and image information of multiple sample animals and performing model training based on the annotation information, a call recognition model is generated. This model can identify the emotions and call meanings of the target animal, improving the accuracy of emotion recognition.
By identifying the emotions and meanings of animals, the accuracy and interaction between humans and animals are improved, and misunderstandings and misjudgments are reduced.
Smart Images
Figure CN119673175A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of emotion recognition, and in particular to a model training method, a call recognition method, a device and a storage medium. Background Art
[0002] With the development of social economy and the improvement of living standards, people have gradually developed a strong interest in animals to add more fun and vitality to their daily lives. As a result, people's demand for emotional communication and understanding of animals is growing. However, there are communication barriers between human language and animal language, and effective interaction between humans and animals has become a difficult problem that needs to be solved.
[0003] Most existing technologies use methods such as training animals to issue commands that humans can understand by pressing keys, helping humans understand what the animals want to express, or for humans to judge the emotions corresponding to the sounds made by animals based on their own experience.
[0004] However, existing technology cannot ensure that humans can accurately identify animal emotions, and there is still a lack of true communication and interaction between humans and animals. Summary of the invention
[0005] The present application provides a model training method, a call recognition method, a device and a storage medium to solve the problems in the prior art of low accuracy in human emotion recognition of animals and poor interactivity between humans and animals.
[0006] In a first aspect, the present application provides a model training method, comprising:
[0007] Acquire sample animal information of a plurality of sample animals, wherein the sample animal information of each sample animal includes audio parameters and a sample animal image when the sample animal makes a sound;
[0008] Acquire labeling information of each sample animal information, wherein the labeling information is used to indicate the emotion and meaning of the call of the sample animal;
[0009] Model training is performed based on the sample animal information of the multiple sample animals and the annotation information of each sample animal information to generate a call recognition model.
[0010] In a possible design, the model training is performed according to the sample animal information of the multiple sample animals and the labeling information of each sample animal information to generate a call recognition model, including:
[0011] Performing cluster analysis on the multiple sample animals according to the sample animal information of the multiple sample animals to determine multiple animal categories;
[0012] Obtaining a training set corresponding to each animal category, wherein the training set corresponding to the animal category includes sample animal information corresponding to the animal category and labeling information corresponding to each sample animal information;
[0013] For any animal category, model training is performed according to the training set corresponding to the animal category to generate a call sub-recognition model corresponding to the animal category, and the call sub-recognition model belongs to the call recognition model.
[0014] In one possible design, the method further includes:
[0015] According to the sample animal information of the multiple sample animals and the labeling information of each sample animal information, for each animal category, call audio corresponding to different call meanings and / or soothing call audio corresponding to different emotions are determined from the sample call audios emitted by the multiple sample animals.
[0016] In a second aspect, the present application provides a call recognition method, comprising:
[0017] When the target animal makes a sound, obtaining the sound audio and the animal image of the target animal;
[0018] Extracting features of the calling audio to obtain audio parameters of the calling audio;
[0019] The audio parameters and the animal image are input into a call recognition model to obtain the emotion and call meaning of the target animal output by the call recognition model, wherein the call recognition model is obtained by model training according to the model training method described in any one of the first aspects above.
[0020] In a possible design, inputting the audio parameters and the animal image into a call recognition model to obtain the emotion and call meaning of the target animal output by the call recognition model includes:
[0021] Determining a target animal category of the target animal based on the audio parameters and / or the animal image;
[0022] The audio parameters and the animal image are input into a target call sub-recognition model corresponding to the target animal category, and the emotion and call meaning of the target animal output by the target call sub-recognition model are obtained. The target call sub-recognition model belongs to the call recognition model.
[0023] In one possible design, the method further includes:
[0024] If the emotion of the target animal is a preset emotion, a soothing call audio corresponding to the emotion of the target animal is output.
[0025] In one possible design, the method further includes:
[0026] If the call of the target animal means a preset health status, a reminder message is output, and the reminder message is used to remind the user to pay attention to the health status of the target animal.
[0027] In one possible design, the method further includes:
[0028] In response to the user's operation on the target call meaning control, the call audio corresponding to the target call meaning control is output.
[0029] In a third aspect, an embodiment of the present application provides an electronic device, comprising: at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the method described in the first aspect, the second aspect and various possible designs above.
[0030] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, in which computer-executable instructions are stored. When a processor executes the computer-executable instructions, the methods described in the first aspect, the second aspect and various possible designs are implemented.
[0031] The present application provides a model training method, a call recognition method, a device and a storage medium. First, sample animal information of multiple sample animals is obtained, wherein the sample animal information of each sample animal includes audio parameters and sample animal images when the sample animal makes a call. Secondly, the annotation information of each sample animal information for indicating the emotion and call meaning of the sample animal is obtained. Finally, model training is performed according to the sample animal information of multiple sample animals and the annotation information of each sample animal information to obtain a call recognition model. It should be noted that the sample animal information used for model training is obtained in a situation where two animals interact, and the interaction between the two animals helps to interpret information more accurately and recognize animal emotions. After obtaining the call recognition model, when the target animal makes a call, the call audio and animal image of the target animal are obtained. After that, feature extraction is performed on the call audio to obtain the audio parameters of the call audio. Next, the audio parameters and the animal image are input into the call recognition model to obtain the emotion and call meaning of the target animal output by the call recognition model. The emotion and call meaning of the target animal are recognized by the call recognition model, which further improves the accuracy of emotion recognition of the target animal. People can make correct responses to animals based on the output of the model, avoiding misunderstandings and improving the interactivity between humans and animals. BRIEF DESCRIPTION OF THE DRAWINGS
[0032] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.
[0033] Figure 1 A flowchart of a method for voice recognition provided in an embodiment of the present application;
[0034] Figure 2 A flowchart of a model training method provided in an embodiment of the present application;
[0035] Figure 3 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application.
[0036] The above drawings have shown clear embodiments of the present application, which will be described in more detail later. These drawings and text descriptions are not intended to limit the scope of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION
[0037] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.
[0038] In the technical solution of this application, the collection, storage, use, processing, transmission, provision and disclosure of information such as financial data or user data involved shall comply with the provisions of relevant laws and regulations and shall not violate public order and good morals.
[0039] It should be noted that in the embodiments of the present application, certain software, components, models and other existing solutions in the industry may be mentioned, and they should be regarded as exemplary. Their purpose is only to illustrate the feasibility of implementing the technical solution of the present application, but it does not mean that the applicant has or will necessarily use the solution.
[0040] The development of social economy and the improvement of living standards have increased people's leisure time and disposable income, which in turn gives them more resources and time to care for and raise animals. Animals have thus become an important part of enriching people's lives and providing emotional support.
[0041] People's interest in animals is not limited to viewing and raising them, but also includes understanding animal behavior and emotions. As the relationship between humans and animals becomes closer, people hope to better understand the emotions and needs of animals in order to provide better care and establish a deeper emotional connection.
[0042] However, there are significant differences between human and animal languages, making direct communication difficult. Current technology mainly trains animals to use specific buttons or devices to express their needs, for example, training a dog to press a button to indicate that it wants food or to go for a walk. Although this method is effective, it is not natural and the training process is complicated.
[0043] Another common method is for humans to judge the emotions corresponding to the sounds or behaviors of animals based on their own experience and observation. For example, by observing a dog's barking, tail wagging, etc., one can infer its emotional state. This method relies on human subjective judgment, and because different people may have different understandings of the same animal behavior, the judgment results may be inconsistent and accuracy is difficult to guarantee.
[0044] It can be seen that existing technologies cannot capture and translate the more complex emotions and psychological states of animals, and there are still great obstacles to emotional communication between humans and animals. In addition, existing technologies are more one-way, that is, animals express their needs in a specific way, and humans understand and respond. There is a lack of two-way emotional communication, and it cannot achieve true interaction between humans and animals.
[0045] In response to the above technical problems, the inventor discovered during the research on the emotional states corresponding to animal calls that it is not possible to rely solely on animal behavior training and subjective judgment to identify the emotional states corresponding to animal calls, and that other technical means are needed to accurately identify and translate the emotional states of animals. Based on this, in order to reduce misunderstandings and misjudgments of the emotional states of animals, the inventor thought of obtaining a model that can identify the corresponding emotional states based on animal calls through model training. This model can help humans accurately understand the emotions and meanings of animals' calls, so as to make targeted and correct responses, making communication between humans and animals more natural and two-way.
[0046] The technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.
[0047] An embodiment of the present application provides a method for voice recognition. Figure 1 The flowchart of the method for voice recognition provided in the embodiment of the present application is shown in FIG. Figure 1 As shown, the call recognition method includes:
[0048] S101. When a target animal makes a sound, obtain the sound audio and the animal image of the target animal.
[0049] The execution subject of the embodiment of the present application is an electronic device, which can be a terminal device or a server. In practical applications, whether the electronic device is a terminal device or a server can be determined according to actual conditions and is not specifically limited.
[0050] The target animal refers to the specific animal that you want to monitor or study. The sound made by the target animal is recorded by using an audio capture device (such as a microphone), and the corresponding photo or video of the target animal making the sound is obtained by using an image capture device (such as a camera, sensor).
[0051] Explanatory, the photos or videos obtained include the category, expression and behavior of the target animal. Among them, animal expression refers to the facial features of the animal, such as the size, shape and color of the eyes, the position and angle of the ears, the opening and closing of the mouth, the exposure of teeth and the position of the tongue, etc. Animal behavior refers to the body language of the animal, such as the way the tail shakes, the overall posture such as jumping, lying down, etc.
[0052] S102: extract features of the calling audio to obtain audio parameters of the calling audio.
[0053] In a possible implementation, audio processing software or programming language can be used to analyze the calling audio, and then obtain audio parameters corresponding to the calling audio, wherein the audio parameters specifically include audio duration, characteristic peaks, pitch, and vocalization period.
[0054] For example, assuming that a certain audio recording of a dog barking, the total duration of the barking audio can be measured using audio processing software, and the frequency region with the strongest energy in the barking audio, i.e., the characteristic peak, can be found by analyzing the spectrum graph. For the pitch, the pitch can be determined using a fundamental frequency detection algorithm, and the periodic pattern of the sound can also be determined by analyzing the waveform. This application does not limit the specific method for obtaining the audio parameters of the barking audio.
[0055] It is worth noting that since the calling audio collected by the audio capture device usually contains noise, redundant information or other irrelevant sounds, before feature extraction of the calling audio, it is necessary to remove these useless parts through preprocessing to improve the quality of the calling audio.
[0056] In a possible implementation, preprocessing the calling audio includes the following aspects:
[0057] (1) Noise reduction: Use filters or other noise reduction techniques to remove background noise;
[0058] (2) Normalization: Adjust the audio amplitude corresponding to the calling audio to make it within a standard range to facilitate subsequent analysis;
[0059] (3) Segmentation: Divide the long-duration call audio into multiple shorter segments for easier processing and analysis.
[0060] S103: Input the audio parameters and the animal image into a call recognition model to obtain the emotion and call meaning of the target animal output by the call recognition model.
[0061] This step first determines the target animal category of the target animal based on the audio parameters and / or animal image; secondly, inputs the audio parameters and animal image into the target call sub-recognition model corresponding to the target animal category; finally, obtains the emotion and call meaning of the target animal output by the target call sub-recognition model.
[0062] It should be noted that the call recognition model includes multiple target call sub-recognition models, and each target call sub-recognition model can only be used to process audio parameters and animal images of one category of target animals.
[0063] Among them, the call recognition model is a system based on machine learning and artificial intelligence technology, which can be deployed in practical applications such as smart pet devices, mobile phone applications, etc. When the call of the target animal is detected, its emotion or state can be analyzed in real time, and corresponding feedback or suggestions can be given, thereby helping the owner of the target animal to better understand the needs and emotions of the target animal.
[0064] In one possible implementation, Figure 2 A flow chart of the model training method provided in the embodiment of the present application is as follows: Figure 2 As shown in the figure, the training process of the call recognition model is as follows:
[0065] S201. Obtain sample animal information of multiple sample animals.
[0066] The sample animal information of each sample animal includes audio parameters when the sample animal makes a sound and an image of the sample animal.
[0067] Specifically, a large number of sample call audios of sample animals of different categories are collected, and the audio parameters corresponding to these sample call audios, including audio duration, characteristic peaks, pitch, and vocalization cycle, as well as the sample animal images corresponding to each sample call audio are recorded.
[0068] Explanatory, the sample animal information of this application is obtained during the interaction between two animals, that is, the obtained sample animal image should contain both the sample animal and the other animal interacting with the sample animal. Therefore, the obtained sample animal image will include the category, expression, and behavior of the sample animal corresponding to the sample call audio, as well as the expression and behavior of the other animal interacting with the sample animal.
[0069] S202: Obtain labeling information of each sample animal information.
[0070] Among them, the annotation information is used to indicate the emotions and meanings of the calls of the sample animals.
[0071] It is understandable that, since the acquired sample animal images contain multiple categories of animals and their corresponding expressions and behaviors, the sample animal images may be labeled to facilitate subsequent analysis, model training or other purposes.
[0072] Specifically, labeling the sample animal images requires defining a labeling system, including all possible animal categories, expressions, and behaviors. Then, according to the defined labeling system, a corresponding plurality of labels are attached to each sample animal image. The labeling process can be completed manually or using semi-automatic tools.
[0073] Furthermore, by analyzing the labeled sample animal images, the emotional state of the sample animals and the specific meaning of their calls are determined, and corresponding annotation information is generated. This process may involve technical means such as image recognition, behavior analysis, and sound analysis.
[0074] S203: Perform model training based on the sample animal information of multiple sample animals and the annotation information of each sample animal information to generate a call recognition model.
[0075] Specifically, a comprehensive data set is constructed based on the sample animal information of multiple sample animals, and the comprehensive data set covers a wide range of audio parameters and sample animal images. Next, a machine learning algorithm, especially cluster analysis, is applied to cluster multiple sample animals, determine multiple animal categories, and cluster similar vocalization situations of animals of the same category to form a sound pattern library.
[0076] Explanatory, cluster analysis is a data analysis technique used to divide a group of objects into multiple clusters, so that objects in the same cluster are similar in some sense, while objects in different clusters are different. This embodiment analyzes the audio parameters of sample animals of the same category and groups them according to the similarity of vocalization to form a sound pattern library. These similar vocalizations may have similarities in duration, pitch, frequency, etc.
[0077] Furthermore, a training set corresponding to each animal category is determined based on the sound pattern library, wherein the training set corresponding to the animal category includes sample animal information corresponding to the animal category (audio parameters, sample animal images) and annotation information corresponding to each sample animal information (emotions of the sample animal, meaning of the call). Afterwards, for any animal category, model training is performed according to the training set corresponding to the animal category to generate a call sub-recognition model corresponding to the animal category.
[0078] It can be seen that each animal category has a corresponding call sub-recognition model, and all the call sub-recognition models together constitute the call recognition model. The trained call recognition model is used for real-time analysis of the call audio of the target animal and outputs the different emotions and call meanings conveyed by the target animal.
[0079] Understandably, obtaining the different emotions and meanings of calls conveyed by the target animals is to facilitate the interaction between humans and animals. Therefore, in order to enable humans to better respond to the calls of the target animals, a user-friendly and easy-to-use software interface can be designed and developed.
[0080] In one possible implementation, the sample audio calls of sample animals collected during the model training process are classified, and the sample audio calls with the same emotions and call meanings corresponding to each animal category are grouped together to form a corresponding pet voice library.
[0081] For example, based on the sample animal information of multiple sample animals and the labeling information of each sample animal information, for each animal category, the call audio corresponding to different call meanings and / or the soothing call audio corresponding to different emotions are determined from the sample call audios emitted by the multiple sample animals. The soothing call audio corresponding to each animal category is grouped.
[0082] Furthermore, the classified pet voice library is integrated into the software interface, and each group of sample call audio is associated with the corresponding button in the software interface. Users can control the software interface to play specific voices and interact with animals by triggering different interactive buttons. Among them, the voices expressing the same information are triggered by one or a small number of interactive buttons. The voices played in the software interface come from the pet voice library, and users can select random voices or specified voices in the pet voice library through settings.
[0083] In addition, the software interface can be set to ensure that after knowing the emotional state of the animal, the software interface automatically plays the corresponding voice to interact with the animal, or prompts the user to pay attention to the current emotion of the animal.
[0084] In one possible implementation, if the animal's emotion is a preset emotion, a soothing call audio corresponding to the animal's emotion is output. For example, when the animal appears restless or anxious, the software interface can automatically play a soothing voice to help the animal relax.
[0085] In one possible implementation, the software interface can also monitor the health status of the animal, and if the animal's call means a preset health status, a reminder message is output to remind the user to pay attention to the animal's health status. For example, once the software interface finds that the animal makes abnormal sounds such as coughing or wheezing, it will promptly remind the user to pay attention.
[0086] In a possible implementation, the software interface can also respond to the user's operation on the target call meaning control and output the call audio corresponding to the target call meaning control. For example, when an animal makes a hungry call, the software interface can control the intelligent feeding machine to feed the animal and play a voice that expresses a similar meaning to "come to eat" to attract the animal.
[0087] Thus, the software interface can not only respond to the animal's calls, but also induce the animal to perform certain behaviors. For example, the software interface can play specific voices according to user instructions, such as voices expressing meanings such as "invitation to play", "where are you, kid", "I'm angry", "don't mess with me", "come to eat", etc., to guide the animal to perform corresponding behaviors.
[0088] It is worth noting that the software interface usually has an intuitive layout, clear buttons and easy-to-understand operation procedures. The design forms include but are not limited to mobile applications, applets, or computer software, web pages, or remote control.
[0089] Easy to understand, the software interface can control smart feeding machines, interactive toys, smart cat litter boxes and other equipment in the form of mobile phone applications, mini-programs, computer software, web pages, or remote controls.
[0090] Specifically, the software interface can be used to control the smart feeder to feed and play voice messages with similar meanings to "come to eat" to induce animals to eat; the interactive toys can also be controlled to play voice messages expressing meanings such as "invitation to play" and "provocation"; the smart cat litter box can also be controlled to play voice messages expressing meanings such as "induce defecation" to induce animal behavior.
[0091] The present application provides a model training method and a call recognition method. First, sample animal information of multiple sample animals is obtained, wherein the sample animal information of each sample animal includes audio parameters and sample animal images when the sample animal makes a call. Secondly, the annotation information for indicating the emotion and call meaning of each sample animal information is obtained. Afterwards, according to the sample animal information of multiple sample animals, a cluster analysis is performed on multiple sample animals, multiple animal categories are determined, and a training set corresponding to each animal category is obtained. Finally, for any animal category, model training is performed according to the training set corresponding to the animal category, and a call sub-recognition model corresponding to the animal category is generated, and then a call recognition model is composed of all call sub-recognition models. It should be noted that the sample animal information used for model training is obtained in a situation where two animals interact, and the interaction between the two animals helps to interpret information more accurately and recognize animal emotions. In addition, model training is performed on the training set corresponding to each animal category separately, rather than training the training set of all animal categories at the same time, which can further improve the accurate recognition of animal emotions. After obtaining the call recognition model, when the target animal makes a call, the call audio and animal image of the target animal are obtained. Afterwards, feature extraction is performed on the call audio to obtain the audio parameters of the call audio. Next, the target animal category of the target animal is determined based on the audio parameters and / or animal image, and the audio parameters and animal image are input into the target call sub-recognition model corresponding to the target animal category. Finally, the emotion and call meaning of the target animal output by the target call sub-recognition model are obtained, which further improves the accuracy of emotion recognition of the target animal. People can make correct responses to the animals based on the output results of the model to avoid misunderstandings. The present application also helps users respond by designing a software interface, and users can also actively interact with animals through the software interface, thereby improving the interactivity between humans and animals.
[0092] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Figure 3 As shown, the electronic device may include: a transceiver 31 , a processor 32 , and a memory 33 .
[0093] The processor 32 executes the computer execution instructions stored in the memory, so that the processor 32 executes the scheme in the above embodiment. The processor 32 can be a general-purpose processor, including a central processing unit CPU, a network processor (NP), etc.; it can also be a digital signal processor DSP, an application-specific integrated circuit ASIC, a field programmable gate array FPGA or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components.
[0094] The memory 33 is connected to the processor 32 via a system bus and completes communication between them. The memory 43 is used to store computer program instructions.
[0095] The transceiver 31 may be used to communicate with other devices.
[0096] The system bus can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The system bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus. The transceiver is used to realize the communication between the database access device and other computers (such as clients, read-write libraries, and read-only libraries). The memory may include random access memory (RAM) and may also include non-volatile memory.
[0097] The electronic device provided in the embodiment of the present application may be the terminal device of the above embodiment.
[0098] The electronic device provided in the embodiment of the present application is used to execute the technical solutions of the model training method and the call recognition method in the above-mentioned embodiment.
[0099] An embodiment of the present application also provides a computer-readable storage medium, which stores computer instructions. When the computer instructions are executed on a computer, the computer executes the technical solutions of the model training method and the call recognition method in the above embodiments.
[0100] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of modules is only a logical function division, and there may be other division methods in actual implementation, such as multiple modules can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or modules, which can be electrical, mechanical or other forms.
[0101] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to implement the solution of this embodiment.
[0102] In addition, each functional module in each embodiment of the present application can be integrated into one processing unit, or each module can exist physically separately, or two or more modules can be integrated into one unit. The above-mentioned module-composed unit can be implemented in the form of hardware or in the form of hardware plus software functional units.
[0103] The above-mentioned integrated module implemented in the form of a software function module can be stored in a computer-readable storage medium. The above-mentioned software function module is stored in a storage medium, including a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor to perform some steps of the methods of various embodiments of the present application.
[0104] It should be understood that the processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), etc. A general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in the invention may be directly implemented as being executed by a hardware processor, or may be implemented by a combination of hardware and software modules in the processor.
[0105] The memory may include a high-speed RAM memory, and may also include a non-volatile storage NVM, such as at least one disk memory, and may also be a USB flash drive, a mobile hard disk, a read-only memory, a magnetic disk or an optical disk, etc.
[0106] The bus may be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, the bus in the drawings of the present application is not limited to only one bus or one type of bus.
[0107] The above storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. The storage medium can be any available medium that can be accessed by a general or special purpose computer.
[0108] An exemplary storage medium is coupled to a processor so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can be located in an application specific integrated circuit (ASIC). Of course, the processor and the storage medium can also exist as discrete components in an electronic control unit or a main control device.
[0109] Those skilled in the art can understand that all or part of the steps of implementing the above-mentioned method embodiments can be completed by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, the steps of the above-mentioned method embodiments are executed; and the aforementioned storage medium includes: ROM, RAM, disk or optical disk and other media that can store program codes.
[0110] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application 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 replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.
Claims
1. A model training method, characterized in that: include: Acquire sample animal information of a plurality of sample animals, wherein the sample animal information of each sample animal includes audio parameters and a sample animal image when the sample animal makes a sound; Acquire labeling information of each sample animal information, wherein the labeling information is used to indicate the emotion and meaning of the call of the sample animal; Model training is performed based on the sample animal information of the multiple sample animals and the annotation information of each sample animal information to generate a call recognition model.
2. The method according to claim 1, characterized in that The method of performing model training according to the sample animal information of the plurality of sample animals and the labeling information of each sample animal information to generate a call recognition model includes: Performing cluster analysis on the multiple sample animals according to the sample animal information of the multiple sample animals to determine multiple animal categories; Obtaining a training set corresponding to each animal category, wherein the training set corresponding to the animal category includes sample animal information corresponding to the animal category and labeling information corresponding to each sample animal information; For any animal category, model training is performed according to the training set corresponding to the animal category to generate a call sub-recognition model corresponding to the animal category, and the call sub-recognition model belongs to the call recognition model.
3. The method according to claim 2, characterized in that The method further comprises: According to the sample animal information of the multiple sample animals and the labeling information of each sample animal information, for each animal category, call audio corresponding to different call meanings and / or soothing call audio corresponding to different emotions are determined from the sample call audios emitted by the multiple sample animals.
4. A method for voice recognition, characterized in that: include: When the target animal makes a sound, obtaining the sound audio and the animal image of the target animal; Extracting features of the calling audio to obtain audio parameters of the calling audio; The audio parameters and the animal image are input into a call recognition model to obtain the emotion and call meaning of the target animal output by the call recognition model, wherein the call recognition model is obtained by model training according to the model training method according to any one of claims 1 to 3.
5. The method according to claim 4, characterized in that The step of inputting the audio parameters and the animal image into a call recognition model to obtain the emotion and call meaning of the target animal output by the call recognition model includes: Determining a target animal category of the target animal based on the audio parameters and / or the animal image; The audio parameters and the animal image are input into a target call sub-recognition model corresponding to the target animal category, and the emotion and call meaning of the target animal output by the target call sub-recognition model are obtained. The target call sub-recognition model belongs to the call recognition model.
6. The method according to claim 4 or 5, characterized in that: The method further comprises: If the emotion of the target animal is a preset emotion, a soothing call audio corresponding to the emotion of the target animal is output.
7. The method according to claim 4 or 5, characterized in that: The method further comprises: If the sound of the target animal means a preset health status, a reminder message is output, and the reminder message is used to remind the user to pay attention to the health status of the target animal.
8. The method according to claim 4 or 5, characterized in that: The method further comprises: In response to the user's operation on the target call meaning control, the call audio corresponding to the target call meaning control is output.
9. An electronic device, characterized in that: include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the method according to any one of claims 1 to 8.
10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 8 when executed by a processor.