An integrated intelligent home control and management method

Abstract

The present invention relates to the technical field of terminal equipment, and in particular to an integrated smart home control and management method, the method comprising: collecting sound signals; preprocessing the sound signals to obtain multiple sound components; reorganizing the sound components according to the weight corresponding to each sound component, the weight being determined according to the probability that the sound component is noise, to obtain a voice instruction; and executing an action according to the voice instruction. Smart home devices collect sound signals and preprocess the sound signals to obtain multiple sound components, and then reorganize the sound components according to the weights corresponding to the sound components to obtain the voice instructions corresponding to the sound signals, thereby improving the accuracy of the smart home devices in executing actions according to the instructions.

Description

Technical Field

[0001] The present invention relates to the technical field of terminal equipment, and in particular to an integrated smart home control and management method. Background Art

[0002] With the continuous development of smart homes, each device in the smart home can perform actions that match the voice commands issued by the user, thereby realizing convenient control of the smart home.

[0003] In the related art, smart home devices can collect sounds made by users and denoise the collected sounds through empirical mode decomposition (EMD) method, so that corresponding instructions can be obtained through the denoised sounds, and then corresponding actions can be performed according to the instructions.

[0004] However, in the process of decomposing and denoising the sound through EMD, some noise is still not filtered, causing the unfiltered noise to interfere with the instructions corresponding to the sound, thereby affecting the accuracy of the smart home devices in executing actions according to the instructions. Summary of the invention

[0005] In order to solve the technical problem that in the process of decomposing and denoising the sound by EMD, some noise is still not filtered, resulting in the interference of the unfiltered noise to the instructions corresponding to the sound, the purpose of the present invention is to provide an integrated smart home control and management method, and the technical solution adopted is as follows:

[0006] In a first aspect, an integrated smart home control and management method is provided, the method comprising:

[0007] Collecting sound signals;

[0008] Preprocessing the sound signal to obtain multiple sound components;

[0009] Recombining the sound components according to the weight corresponding to each sound component to obtain a voice instruction, wherein the weight is determined according to the probability that the sound component is noise;

[0010] An action is performed according to the voice command.

[0011] Optionally, the reorganizing the sound components according to the weight corresponding to each of the sound components to obtain the voice instruction includes:

[0012] Determine the noise probability corresponding to each of the sound components;

[0013] The voice command is obtained by reorganizing the sound components according to the noise probabilities corresponding to each of the sound components.

[0014] Optionally, the determining the noise probability corresponding to each of the sound components includes:

[0015] For each of the sound components, segment the sound component to obtain a plurality of periodic signals, and determine the similarity between any two of the periodic signals;

[0016] Determine the noise probability corresponding to each of the periodic signals according to the similarity between any two of the periodic signals;

[0017] The noise probability of the sound component is determined according to the noise probability corresponding to each of the periodic signals.

[0018] Optionally, segmenting the sound component to obtain a plurality of periodic signals, and determining the similarity between any two of the periodic signals includes:

[0019] Converting the sound component to obtain a frequency signal;

[0020] Segmenting the frequency signal according to a reference frequency in the frequency signal to obtain a plurality of periodic signals, wherein the reference frequency is a maximum frequency in the frequency signal;

[0021] The similarity between any two periodic signals is determined according to the data sequence of each periodic signal.

[0022] Optionally, determining the noise probability of the sound component according to the noise probability corresponding to each of the periodic signals includes:

[0023] Determine the noise probability mean and probability range of each periodic signal according to the noise probability corresponding to each periodic signal;

[0024] Obtaining multiple maximum amplitudes and multiple maximum amplitude difference averages corresponding to the sound component;

[0025] The noise probability of the sound component is obtained by performing calculation according to the noise probability mean, the probability range, the multiple maximum amplitudes, and the mean of the multiple maximum amplitude differences.

[0026] Optionally, the step of reorganizing according to the noise probabilities respectively corresponding to each of the sound components to obtain the voice instruction includes:

[0027] Combining at least two of the sound components according to the noise probability corresponding to each of the sound components to obtain a plurality of combined signals;

[0028] Obtaining a vocal feature value of each of the combined signals;

[0029] According to the plurality of human voice feature values, a voice instruction is obtained by combining them.

[0030] Optionally, combining at least two of the sound components according to the noise probability corresponding to each of the sound components to obtain a plurality of combined signals includes:

[0031] Classifying each of the sound components according to the noise probability corresponding to each of the sound components, wherein each type of sound component includes at least two of the sound components;

[0032] For each type of sound component, any multiple sound components in each type of sound component are combined to obtain multiple combined signals.

[0033] Optionally, the combining of the plurality of human voice feature values ​​to obtain a voice instruction includes:

[0034] According to the plurality of human voice feature values, selecting a target class sound component corresponding to the human voice feature value having the largest parameter value;

[0035] According to the noise probabilities respectively corresponding to the spare components in the target sound component, each spare component is reorganized in combination with a preset formula to obtain the voice instruction.

[0036] Optionally, the preprocessing of the sound signal to obtain multiple sound components includes:

[0037] The sound signal is decomposed by an empirical mode decomposition method to obtain a plurality of connotation modal components.

[0038] Optionally, executing an action according to the voice instruction includes:

[0039] The voice command is sent to the router, so that the router forwards the voice command to other smart home devices, and through the voice command, the other smart home devices are controlled to perform actions corresponding to the voice command.

[0040] The present invention has the following beneficial effects:

[0041] The present application proposes an integrated smart home control and management method. The smart home device collects sound signals and pre-processes the sound signals to obtain multiple sound components. Then, the sound components are reorganized according to the weights corresponding to the sound components to obtain voice instructions corresponding to the sound signals, thereby improving the accuracy of the smart home device in executing actions according to the instructions. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0043] Figure 1 A system schematic diagram of a smart home system proposed in an embodiment of the present application;

[0044] Figure 2 A schematic flow chart of an integrated smart home control and management method provided in an embodiment of the present application;

[0045] Figure 3 A schematic flow chart of determining the noise probability corresponding to each IMF provided in an embodiment of the present application;

[0046] Figure 4 A schematic flow chart of an embodiment of the present application for obtaining a voice command by reorganizing according to the noise probability corresponding to each IMF. DETAILED DESCRIPTION

[0047] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the integrated smart home control and management method proposed by the present invention, its specific implementation method, structure, features and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0048] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0049] With the continuous development of smart homes, each device in the smart home can perform actions that match the voice commands issued by the user, thereby realizing convenient control of the smart home.

[0050] In the related art, smart home devices can collect sounds made by users and denoise the collected sounds through empirical mode decomposition (EMD) method, so that corresponding instructions can be obtained through the denoised sounds, and then corresponding actions can be performed according to the instructions.

[0051] However, in the process of decomposing and denoising the sound through EMD, some noise is still not filtered, causing the unfiltered noise to interfere with the instructions corresponding to the sound, thereby affecting the accuracy of the smart home devices in executing actions according to the instructions.

[0052] Therefore, the present application proposes an integrated smart home control and management method, in which the smart home device collects sound signals and pre-processes the sound signals to obtain multiple sound components, and then reorganizes the sound components according to the weights corresponding to each sound component to obtain the voice instructions corresponding to the sound signals, thereby improving the accuracy of the smart home device in executing actions according to the instructions.

[0053] The following smart home system established based on multiple smart home devices introduces the integrated smart home control and management method proposed in the embodiment of the present application.

[0054] See also Figure 1 , Figure 1 This is a system schematic diagram of a smart home system proposed in an embodiment of the present application. The smart home system may include: a speaker 110, a projector 120, a camera 130, a TV 140 and a router 150.

[0055] Among them, the speaker 110, the projector 120, the camera 130 and the TV 140 can all be wirelessly connected to the router 150. Moreover, the speaker 110, the projector 120, the camera 130 and the TV 140 can all be smart home devices, and the embodiment of the present application does not specifically limit the smart home devices.

[0056] When any smart home device collects the sound made by the user, it can pre-process and decompose the collected sound, identify the pre-processed IMF components, and determine the weight corresponding to each IMF component, so as to reorganize the collected sound according to different weights to obtain voice commands.

[0057] Accordingly, the smart home device can send the voice command to the router 150, and forward the voice command to another smart home device according to the information carried in the voice command. Then, the other smart home device receiving the voice command can perform a matching action according to the received voice command.

[0058] It should be noted that the above description uses an example of controlling another smart home device through one smart home device. In actual applications, after obtaining a voice command, the smart home device may no longer send a voice command to other smart home devices, but instead perform corresponding actions according to the voice command.

[0059] Moreover, in practical applications, smart home devices can also forward the collected sound to the router, and the router can parse and process the sound. Then, according to the recombined voice command, the smart home device that needs to receive the voice command is determined, and the voice command is sent so that the smart home device can perform corresponding actions according to the received voice command.

[0060] Figure 2 FIG. is a schematic flowchart of an integrated smart home control and management method provided by an embodiment of the present application. As an example but not a limitation, this method can be applied to any smart home device in the above smart home system. Refer to Figure 2 This method includes:

[0061] Step 201, collect sound signals.

[0062] During the working process, the smart home device can continuously collect the surrounding sounds. When the collected sound includes a pre-set trigger word, the sound after the trigger word can be collected and stored so that in subsequent steps, the smart home device can process the collected sound signal and thus perform corresponding operations.

[0063] Specifically, the smart home device can continuously detect the sound signal in the current scene and detect the words included in the sound signal. If it is detected that the sound signal includes a pre-set trigger word, the smart home device can store the sound signal within a pre-set duration after the trigger word.

[0064] The pre-set trigger word can be the name of the smart home device or a word common to multiple smart home devices. For example, the common trigger word can be "Xiaoe Xiaoe" or "Xiaoe Tongxue". The embodiment of the present application does not make specific limitations on the pre-set trigger word.

[0065] Moreover, during the process of the smart home device obtaining the sound signal within the pre-set duration, it can be collected by means of endpoint detection. For example, if the smart home device detects the trigger word at 0.1 second (s) and the pre-set duration is 3 s, the smart home device can start collecting the sound signal according to the trigger word and start timing at 0.1 s. When the timing duration is consistent with the pre-set duration, the smart home device can stop collecting the sound signal and complete the collection of the sound signal.

[0066] Step 202, preprocess the sound signal to obtain multiple sound components.

[0067] The sound component is an intrinsic mode function (IMF) obtained by decomposing the sound signal. Of course, it can also be a component signal in other forms, which is not specifically limited in the embodiment of the present application.

[0068] After collecting the sound signal, the smart home device can pre-process the collected sound signal so that in the subsequent step, the multiple sound components obtained by pre-processing can be reorganized to obtain the reorganized voice command, thereby improving the accuracy and reliability of the smart home device in executing actions according to the voice command.

[0069] Accordingly, the smart home device can first decompose the sound signal by EMD decomposition to obtain multiple IMFs, so that in the subsequent steps, the smart home device can determine the weight corresponding to each IMF, and then obtain the voice command according to the reorganization.

[0070] Step 203: Determine the noise probability corresponding to each IMF.

[0071] Among them, the noise probability corresponding to IMF is used to indicate the probability that IMF is noise in the current environment.

[0072] In order to improve the accuracy of smart home devices in executing actions according to sound signals, the smart home devices can identify each IMF to determine the noise probability of each IMF being noise, so that in subsequent steps, the smart home devices can reorganize according to the noise probability of each IMF to obtain voice commands.

[0073] In the process of identifying each IMF, the smart home device can first segment the IMF to obtain multiple periodic signals for each IMF, determine the similarity between the periodic signals, and determine the noise probability of each periodic signal, and then determine the noise probability of the IMF in combination with other parameters corresponding to each periodic signal.

[0074] Accordingly, see Figure 3 , step 203 may include the following steps:

[0075] Step 203a: segment the IMF and determine the similarity between each periodic signal obtained by segmentation.

[0076] Specifically, the smart home device can first convert the form of the IMF signal to obtain a frequency signal, then traverse each frequency in the frequency signal, and select the frequency with the largest parameter value as the reference frequency. After that, the smart home device can calculate the period corresponding to the IMF (such as the inverse of the reference frequency) based on the reference frequency, so that the IMF can be segmented according to the determined period to obtain multiple periodic signals.

[0077] Accordingly, the smart home device can first obtain the data sequence of each periodic signal, and traverse each periodic signal, and calculate the difference between the data sequences of any two periodic signals through a preset formula, so as to determine the similarity between any two periodic signals. The smaller the difference between the data sequences, the higher the similarity between the two periodic signals.

[0078] For example, the preset formula is D ij =dtw(d i ,d j ), where d i represents the data sequence of the ith period signal in the IMF, d j represents the data sequence of the j-th period signal in IMF, D ij Indicates d i and d j The DTW distance between them is used to represent the similarity between periodic signals. The smaller the DTW distance, the higher the similarity between periodic signals.

[0079] Step 203b: determine the noise probability corresponding to each periodic signal.

[0080] After determining the similarity between each periodic signal, the smart home device can determine the noise probability corresponding to each periodic signal based on the difference between the periodic signals, combined with the frequency corresponding to each periodic signal, according to the frequency change and similarity.

[0081] Specifically, for each periodic signal, the smart home device can first obtain the respective maximum frequencies corresponding to the periodic signal, and determine the periodic frequency with the largest parameter value among the multiple maximum frequencies, so that the included maximum frequencies can be normalized by the periodic frequency to obtain multiple normalized frequencies, and then the variance between the normalized frequencies can be calculated based on the multiple normalized frequencies. Correspondingly, the smaller the variance, the smaller the frequency fluctuation of the periodic signal, and the more single the components; the larger the variance, the larger the frequency fluctuation of the periodic signal, and the periodic signal includes data of multiple frequencies.

[0082] Moreover, the smart home device can also calculate the mean similarity between the periodic signal and other periodic signals according to the similarity between the periodic signal and other periodic signals. After that, the smart home device can calculate the noise probability of the periodic signal according to the variance between each normalized frequency and the mean similarity by a preset formula.

[0083] For example, the preset formula is:

[0084] Among them, f j represents the frequency of each maximum value in the periodic signal, f represents the periodic frequency of the periodic signal, represents the variance between each normalized frequency, D ij represents the DTW distance between the i-th periodic signal and the j-th periodic signal, E(D ij ) represents the mean similarity between a periodic signal and other periodic signals, p j represents the noise probability of the j-th period signal, Represents the norm corresponding to the variance between each normalized frequency, norm(E(D ij )) represents the norm corresponding to the mean similarity.

[0085] Step 203c: Determine the noise probability of the corresponding IMF according to the noise probability corresponding to each periodic signal.

[0086] After determining the noise probability corresponding to each periodic signal, the smart home device can calculate in combination with other parameters of each periodic signal to determine the noise probability corresponding to the IMF to which each periodic signal belongs, so that in subsequent steps, the smart home device can reconstruct the speech signal according to the noise probabilities corresponding to each IMF.

[0087] Specifically, the smart home device can first calculate the noise probability of each periodic signal to obtain the mean value of the noise probability of each periodic signal in the IMF. The larger the mean value of the noise probability, the greater the probability that the IMF is noise. At the same time, the smart home device can also calculate the probability range of the noise probability of each periodic signal in the IMF based on the noise probability of each periodic signal, and determine the fluctuation of the noise probability through the parameter value of the probability range. The larger the probability range, the greater the probability that the IMF is noise.

[0088] Moreover, the smart home device can also obtain the amplitude of each periodic signal and determine the maximum amplitude of each periodic signal. Similarly, the smart home device can also segment the IMF signal before converting the frequency signal in a similar manner to segmenting the frequency signal to obtain multiple initial periodic signals, thereby obtaining the initial maximum amplitude of each initial periodic signal.

[0089] Furthermore, the smart home device can calculate the mean of the maximum amplitude difference between each periodic signal and the corresponding initial periodic signal, that is, for each periodic signal, the smart home device can calculate the absolute value of the difference between the two based on the maximum amplitude of the periodic signal and the initial maximum amplitude of the initial periodic signal corresponding to the periodic signal. After that, the mean of the maximum amplitude difference of each periodic signal is calculated based on the absolute value of the difference corresponding to each periodic signal. Similarly, the smart home device can also calculate the mean of the maximum amplitude of the IMF to which it belongs based on the maximum amplitude of each periodic signal.

[0090] Furthermore, the smart home device can calculate the noise probability of the IMF to which it belongs based on the noise probability mean, probability range, maximum amplitude difference mean and maximum amplitude mean of the IMF obtained by the above calculations through a preset formula, that is, determine the possibility of the IMF appearing as environmental noise.

[0091] For example, the preset formula could be:

[0092] Among them, E(p j ) represents the noise probability mean of each periodic signal in the IMF, R represents the probability range between each periodic signal in the IMF, A j represents the maximum amplitude of the j-th period signal of the IMF, A 0,j represents the maximum amplitude of the jth initial period signal, E(|A j -A 0j |) represents the average of the maximum amplitude differences between the maximum amplitudes of multiple periodic signals and the maximum amplitudes of multiple initial periodic signals, E(A j ) represents the maximum amplitude mean of the IMF, and p represents the noise probability of the IMF.

[0093] Step 204: Reorganize according to the noise probability corresponding to each IMF to obtain a voice instruction.

[0094] After obtaining the noise probability corresponding to each IMF, the smart home device can classify the IMF according to the noise probability, and combine at least two IMFs in each category of IMFs, and determine the human voice feature value of the combined signal according to the combined signal. After that, the smart home device can select the corresponding IMF according to each human voice feature value for reorganization to obtain a voice command.

[0095] Accordingly, see Figure 4 , this step 204 may include the following steps:

[0096] Step 204a: Combine at least two IMFs according to the noise probability corresponding to each IMF to obtain a plurality of combined signals.

[0097] Specifically, the smart home device can first classify each IMF according to the noise probability corresponding to each IMF, and select IMFs with the same or similar noise probability as IMFs of the same category. Each category of IMFs includes at least two IMFs. Accordingly, for each category of IMFs, the smart home device can combine at least two IMFs in the category of IMFs, thereby obtaining at least one combined signal corresponding to each category of IMFs.

[0098] For example, a smart home device can classify IMF into two categories, one for human voice signals and the other for noise signals.

[0099] Step 204b: Obtain the vocal feature value of each combined signal.

[0100] After obtaining multiple combined signals, the smart home device can identify each combined signal and determine the human voice feature value corresponding to each combined signal, so that in subsequent steps, the IMF classification closest to the human voice can be determined according to each human voice feature value.

[0101] The human voice feature value is used to indicate the degree of closeness between the combined signal and the preset voiceprint feature.

[0102] Specifically, for each combined signal, the smart home device can obtain the mel frequency cepstral coefficient (MFCC) of the combined signal, and then combine it with the MFCC of the pre-stored human voiceprint feature to calculate through a preset formula to obtain the human voice probability of the combined signal.

[0103] Afterwards, the smart home device may process the combined signal in a manner similar to step 203 to obtain the noise probability of the combined signal, and calculate based on the human voice probability and noise probability of the combined signal to obtain the human voice feature value of the combined signal.

[0104] For example, the pre-set formulas in smart home devices are: person =exp(-|m-m0|), and

[0105] Where exp represents the natural exponential function, m represents the MFCC parameter of the combined signal, m0 represents the MFCC corresponding to the pre-stored human voiceprint feature, and p person represents the probability of human voice in the combined signal, p environmentrepresents the noise probability of the combined signal, and t represents the vocal feature value of the combined signal.

[0106] Step 204c: select a backup IMF according to multiple voice feature values, and obtain a voice instruction according to the backup IMF combination.

[0107] After determining the human voice feature values ​​corresponding to each combined signal, the smart home device can determine the target human voice feature value with the largest parameter value according to the parameter values ​​of each human voice feature value, and use each IMF in the IMF classification to which the target human voice feature value belongs as a backup IMF. Furthermore, the smart home device can determine the weight of the IMF to which it belongs according to the noise probability corresponding to each backup IMF, and then reorganize each backup IMF through a preset formula to obtain a voice command.

[0108] For example, the preset formula is:

[0109] Among them, IMF k represents the kth alternate IMF in the IMF classification, p k IMF k The corresponding noise probability, 1-p k IMF k The weight in the reconstruction process, represents the normalization of weights, ‖M‖ represents the number of each backup IMF, and s re Represents the reconstructed speech signal.

[0110] Step 205: Execute an action according to the voice command.

[0111] After reconstructing the voice command, the smart home device can recognize the text in the voice command, and search in the pre-set command-action correspondence based on the recognized text to determine the action corresponding to the voice command. The smart home device can then perform the corresponding action based on the search result, enabling the user to control the smart home device through voice.

[0112] It should be noted that the above-mentioned smart home device performs a corresponding action according to the voice command after obtaining the voice command. In actual applications, after obtaining the voice command, the smart home device can also send the voice command to the router, and forward the voice command to other smart home devices through the router, thereby controlling other smart home devices to perform corresponding actions. The embodiments of the present application do not specifically limit this.

[0113] In summary, the present application proposes an integrated smart home control and management method. The smart home device collects sound signals and pre-processes the sound signals to obtain multiple sound components. Then, the sound components are reorganized according to the weights corresponding to each sound component to obtain the voice instructions corresponding to the sound signals, thereby improving the accuracy of the smart home device in executing actions according to the instructions.

[0114] Moreover, the smart home device decomposes the sound signal by EMD to obtain IMF components, and then identifies the IMF components to determine the noise part in the sound signal, so as to assign different weights to each IMF component. In this way, the voice commands can be reorganized according to the IMF components with different weights, thereby improving the accuracy of the voice commands reorganized by the smart home device.

[0115] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0116] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. An integrated smart home control and management method, characterized in that: The method comprises: Collecting sound signals; Preprocessing the sound signal to obtain multiple sound components; Recombining the sound components according to the weight corresponding to each sound component to obtain a voice instruction, wherein the weight is determined according to the probability that the sound component is noise; Execute an action according to the voice command; The step of reorganizing the sound components according to the weight corresponding to each sound component to obtain the voice instruction includes: Determine the noise probability corresponding to each of the sound components; Recombining according to the noise probability corresponding to each of the sound components to obtain the voice instruction; The determining of the noise probability corresponding to each of the sound components comprises: For each of the sound components, segment the sound component to obtain a plurality of periodic signals, and determine the similarity between any two of the periodic signals; Determine the noise probability corresponding to each of the periodic signals according to the similarity between any two of the periodic signals; Determining the noise probability of the sound component according to the noise probability corresponding to each of the periodic signals; The step of determining the noise probability of the sound component according to the noise probability corresponding to each of the periodic signals comprises: Determine the noise probability mean and probability range of each periodic signal according to the noise probability corresponding to each periodic signal; Obtaining multiple maximum amplitudes and multiple maximum amplitude difference averages corresponding to the sound component; The noise probability of the sound component is obtained by performing calculation according to the noise probability mean, the probability range, the multiple maximum amplitudes, and the mean of the multiple maximum amplitude differences.

2. The method according to claim 1, characterized in that The step of segmenting the sound component to obtain a plurality of periodic signals and determining the similarity between any two of the periodic signals comprises: Converting the sound component to obtain a frequency signal; Segmenting the frequency signal according to a reference frequency in the frequency signal to obtain a plurality of periodic signals, wherein the reference frequency is a maximum frequency in the frequency signal; The similarity between any two periodic signals is determined according to the data sequence of each periodic signal.

3. The method according to claim 1, characterized in that The step of reorganizing the sound components according to the noise probabilities corresponding to each of the sound components to obtain the voice command includes: Combining at least two of the sound components according to the noise probability corresponding to each of the sound components to obtain a plurality of combined signals; Obtaining a vocal feature value of each of the combined signals; According to the plurality of human voice feature values, a voice instruction is obtained by combining them.

4. The method according to claim 3, characterized in that The step of combining at least two of the sound components according to the noise probability corresponding to each of the sound components to obtain a plurality of combined signals includes: Classifying each of the sound components according to the noise probability corresponding to each of the sound components, wherein each type of sound component includes at least two of the sound components; For each type of sound component, any multiple sound components in each type of sound component are combined to obtain multiple combined signals.

5. The method according to claim 3, characterized in that: The step of combining the plurality of human voice feature values ​​to obtain a voice instruction comprises: According to the plurality of human voice feature values, selecting a target class sound component corresponding to the human voice feature value having the largest parameter value; According to the noise probabilities respectively corresponding to the spare components in the target sound component, each spare component is reorganized in combination with a preset formula to obtain the voice instruction.

6. The method according to any one of claims 1 to 5, characterized in that: The preprocessing of the sound signal to obtain multiple sound components includes: The sound signal is decomposed by an empirical mode decomposition method to obtain a plurality of connotation modal components.

7. The method according to any one of claims 1 to 5, characterized in that: The performing of an action according to the voice instruction comprises: The voice command is sent to the router, so that the router forwards the voice command to other smart home devices, and through the voice command, the other smart home devices are controlled to perform actions corresponding to the voice command.

Images