A voice wake-up method, apparatus, device, and medium

Abstract

The application discloses a voice wake-up method, device, equipment and medium, and relates to the field of electronic intelligence, which is applied to an electronic device and comprises the following steps: recording first audio data collected by an audio input device within a preset time period after starting up, and calculating a first voice wake-up volume threshold by using a pre-designed calculation rule based on volume information in the audio data; determining a second threshold based on the first voice wake-up volume threshold and a preset initial voice wake-up volume threshold after the preset time period, determining target voice data greater than the second threshold from second audio data collected in real time, and then inputting the target voice data into a voice wake-up judgment model that has been trained; and when the model determines that the target voice data meets a target wake-up condition, a corresponding wake-up operation is performed. The application can reduce the occupation of hardware resources by the voice wake-up function, enhance the stability of the voice wake-up model, and reduce the risk of misjudgment and miswakeup.

Description

Technical Field

[0001] This invention relates to the field of electronic intelligence, and in particular to a voice wake-up method, device, equipment, and medium. Background Technology

[0002] Currently, both smart speakers and mobile phones include voice wake-up functionality for interaction with their built-in voice assistant modules. This functionality is provided by hardware manufacturers and sold after integration. Manufacturers can provide comprehensive adaptation based on the microphones and audio decoding hardware in smart speakers and mobile phones. With the booming development of e-sports hotels, some businesses also offer features such as digital human assistants running on PCs. For the convenience of gamers, especially when interacting with digital human assistants during gameplay, voice wake-up is the most suitable interaction point.

[0003] The computers in esports hotels may have various configurations and assembly schemes, and most rely on external headsets to collect audio signals. The wide variety of computer hardware and external headset brands and types, coupled with the inability to obtain relevant information in advance for targeted optimization, makes it difficult to address these differences. Microphone variations directly impact the quality of the collected audio signals and the magnitude of data amplification, thus affecting feature extraction and posing a significant challenge to subsequent voice wake-up algorithms, greatly impacting their accuracy and applicability. Furthermore, differences in environmental background, user voices, and volume also affect model stability. Moreover, since the wake-up algorithm runs on the user's PC, and esports hotel customers primarily engage in gaming and other activities on PCs, they are highly sensitive to machine performance. Voice wake-up needs to minimize hardware resource consumption and reduce disruption to the user's gaming experience. Additionally, frequent communication during gaming sessions places high demands on the wake-up algorithm's ability to discern wake words, increasing the probability of misjudgments. False wake-ups severely interfere with the user experience and must be avoided as much as possible.

[0004] As can be seen from the above, in the existing voice wake-up process, how to avoid interference with the stability of the voice wake-up model due to differences in voice devices, environmental backgrounds, user voices, and volume is a problem that needs to be solved in this field. Summary of the Invention

[0005] In view of this, the purpose of this invention is to provide a voice wake-up method, apparatus, device, and medium, and to provide a data filtering operation that filters data based on the user's volume, reducing the amount of data entering the wake-up algorithm. This reduces hardware resource consumption and mitigates the instability of the voice wake-up model due to variations in different implementation environments, while also lowering the risk of algorithm misjudgment and false wake-up. The specific solution is as follows:

[0006] Firstly, this application discloses a voice wake-up method, including:

[0007] Optionally, the step of calculating the first voice wake-up volume threshold based on the volume information in the first audio data using a preset calculation rule includes:

[0008] Based on the volume information in the first audio data and the preset initial voice wake-up volume threshold, the first voice wake-up volume threshold is calculated using the preset threshold calculation formula.

[0009] Optionally, after recording the first audio data collected by the audio input device used for sound acquisition during a preset audio acquisition time period after power-on, the method further includes:

[0010] A preset data feature extraction algorithm is used to determine the first data feature corresponding to the first audio data, and the first feature parameter corresponding to the first data feature is determined.

[0011] Accordingly, the voice wake-up method further includes:

[0012] After the preset audio acquisition time period, a second feature parameter is determined based on the first feature parameter and the preset initial feature parameter, and the second data parameter is updated in the trained voice wake-up determination model.

[0013] Optionally, the first feature parameter, the second feature parameter, and the preset initial feature parameter all include the mean and variance of the audio data.

[0014] Optionally, the voice wake-up method further includes:

[0015] Set a preset initial voice wake-up volume threshold; wherein the preset initial voice wake-up volume threshold is a threshold determined and set based on the quartiles calculated using volume information from pre-collected sample data.

[0016] Optionally, the step of calculating the first voice wake-up volume threshold based on the volume information in the first audio data using a preset calculation rule includes:

[0017] Calculate the full decibel scale information in the first audio data;

[0018] Based on the full decibel scale information, the first voice wake-up volume threshold is calculated using preset calculation rules.

[0019] Optionally, determining the second voice wake-up volume threshold based on the first voice wake-up volume threshold and the preset initial voice wake-up volume threshold includes:

[0020] The second voice wake-up volume threshold is determined based on the first voice wake-up volume threshold and the preset initial voice wake-up volume threshold, and using a preset moving average algorithm.

[0021] Secondly, this application discloses a voice wake-up device, comprising:

[0022] The first threshold calculation module is used to record the first audio data collected by the audio input device for collecting sound during a preset audio acquisition time period after power-on, and to calculate the first voice wake-up volume threshold based on the volume information in the first audio data using preset calculation rules.

[0023] The second threshold calculation module is used to determine a second voice wake-up volume threshold based on the first voice wake-up volume threshold and the preset initial voice wake-up volume threshold after the preset audio acquisition time period, and to determine target voice data that is greater than the second voice wake-up volume threshold from the second audio data acquired in real time by the audio input device, and then input the target voice data into the trained voice wake-up judgment model.

[0024] The voice wake-up module is used to perform a corresponding wake-up operation when the voice wake-up determination model determines that the target voice data meets the target wake-up conditions.

[0025] Thirdly, this application discloses an electronic device, including:

[0026] Memory, used to store computer programs;

[0027] A processor is used to execute the computer program to implement the aforementioned voice wake-up method.

[0028] Fourthly, this application discloses a computer storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the steps of the aforementioned disclosed voice wake-up method.

[0029] This application first records the first audio data collected by the audio input device during a preset audio acquisition period after power-on. Based on the volume information in the first audio data, a first voice wake-up volume threshold is calculated using preset calculation rules. Then, after the preset audio acquisition period, a second voice wake-up volume threshold is determined based on the first voice wake-up volume threshold and a preset initial voice wake-up volume threshold. Target voice data exceeding the second voice wake-up volume threshold is then identified from the second audio data collected in real-time by the audio input device. This target voice data is then input into a trained voice wake-up determination model. When the voice wake-up determination model determines that the target voice data meets the target wake-up condition, the corresponding wake-up operation is executed. In this way, by using a data filtering operation based on the user's volume, a portion of the data entering the wake-up algorithm is reduced. This reduces hardware resource consumption and mitigates the instability of the voice wake-up model due to differences in various implementation environments, while also reducing the risk of algorithm misjudgment and false wake-up. Attached Figure Description

[0030] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0031] Figure 1 A flowchart of a voice wake-up method provided in this application;

[0032] Figure 2 A voice wake-up flowchart is provided for this application;

[0033] Figure 3 A flowchart of a specific voice wake-up method provided in this application;

[0034] Figure 4 A schematic diagram of a parameter replacement process provided in this application;

[0035] Figure 5 A schematic diagram of a voice wake-up device provided in this application;

[0036] Figure 6 This application provides a structural diagram of an electronic device. Detailed Implementation

[0037] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0038] In existing technologies, differences in voice devices, environmental backgrounds, user voices, and volume levels can interfere with the stability of voice wake-up models. This application provides a data filtering operation that filters data based on the user's volume, reducing the amount of data entering the wake-up algorithm. This reduces hardware resource consumption and mitigates the instability of the voice wake-up model caused by variations in implementation environments, while also lowering the risk of algorithm misjudgment and false wake-up.

[0039] This invention discloses a voice wake-up method applied to electronic devices. See [link to relevant documentation]. Figure 1 The method includes:

[0040] Step S11: Record the first audio data collected by the audio input device used for sound acquisition during the preset audio acquisition time period after power-on, and calculate the first voice wake-up volume threshold based on the volume information in the first audio data using preset calculation rules.

[0041] The voice wake-up method in this embodiment is preferably applied to the PC (i.e., Personal Computer) in an e-sports hotel. First, within a preset audio acquisition time period after the computer device is powered on, the first audio data collected by the audio input device is recorded. The preset audio acquisition time is preferably 10 minutes, and the audio input device is preferably a microphone connected to the PC. Specifically, in this embodiment, the first audio data collected by the external microphone is recorded within 10 minutes after the computer device is powered on.

[0042] In a specific implementation, the first audio data collected by the audio input device used for sound acquisition can also be recorded within a preset audio acquisition time period after the voice wake-up program starts after power-on.

[0043] After recording the first audio data, a first voice wake-up volume threshold can be calculated based on the volume information in the first audio data using a preset calculation rule. In a specific implementation, calculating the first voice wake-up volume threshold based on the volume information in the first audio data using a preset calculation rule may include: calculating the first voice wake-up volume threshold using a preset threshold calculation formula based on the volume information in the first audio data and a preset initial voice wake-up volume threshold.

[0044] In a specific implementation, the preset initial voice wake-up volume threshold can be obtained in advance based on the following steps:

[0045] (1) Collect a first preset number of 1.5s human voice audio data and a second preset number of target audio data containing wake words collected from multiple age groups. The audio data can be recorded in multiple scenarios, including but not limited to conference rooms, hotels, outdoor locations, internet cafes, etc., which are not limited here. The content of the human voice audio data is daily conversation, the first preset number is preferably 100,000, and the second preset number is preferably 20,000.

[0046] (2) Using the human voice audio data and target audio data from step 1, calculate the volume of each data item.

[0047] (3) Calculate the quartiles of the volume data obtained in (2). Based on the quartiles, set the initial volume filtering threshold T. The preferred calculation rule is: A1, A2, and A3 are the first quartile, the second quartile, and the third quartile, respectively. If A2-A1 <= A3-A2-2, then the initial volume filtering threshold is A2+2. Otherwise, the initial volume filtering threshold is A1+1.

[0048] In this embodiment, the voice wake-up method may further include: setting a preset initial voice wake-up volume threshold; wherein the preset initial voice wake-up volume threshold is a threshold determined and set based on the quartiles calculated using volume information from pre-collected sample data.

[0049] In this embodiment, calculating the first voice wake-up volume threshold based on the volume information in the first audio data using a preset calculation rule may include: calculating the full decibel scale information in the first audio data; and calculating the first voice wake-up volume threshold based on the full decibel scale information using a preset calculation rule. That is, in this invention, the volume information in the first audio data can be determined using dBFS (Decibels Full Scale). Similarly, in step (2) above, the volume of each data item can also be calculated using dBFS.

[0050] Step S12: After the preset audio acquisition time period, a second voice wake-up volume threshold is determined based on the first voice wake-up volume threshold and the preset initial voice wake-up volume threshold. Target voice data greater than the second voice wake-up volume threshold is determined from the second audio data acquired in real time by the audio input device. Then, the target voice data is input into the trained voice wake-up determination model.

[0051] In this embodiment, after the preset audio acquisition period following device power-on, a second voice wake-up volume threshold can be determined based on the first voice wake-up volume threshold and the preset initial voice wake-up volume threshold. Specifically, in a concrete implementation, if the preset audio acquisition period is ten minutes after power-on, then ten minutes after power-on, the second voice wake-up volume threshold T2 is determined based on the first voice wake-up volume threshold T1 and the preset initial voice wake-up volume threshold T.

[0052] In this embodiment, determining the second voice wake-up volume threshold based on the first voice wake-up volume threshold and the preset initial voice wake-up volume threshold may include: determining the second voice wake-up volume threshold based on the first voice wake-up volume threshold and the preset initial voice wake-up volume threshold, and using a preset moving average algorithm.

[0053] Specifically, in the process of determining the second voice wake-up volume threshold based on the first voice wake-up volume threshold and the preset initial voice wake-up volume threshold, the threshold can be recalculated according to the volume information of the first audio data, using the calculation rules in step (3) of step S11 above, to obtain the threshold T1, and then the new threshold T2 is calculated according to the moving average method, T2=0.1*T+(1-0.1)*T1.

[0054] In addition, after step (1) in step S11, it may also include: using a convolutional neural network to divide the human voice audio data and target audio data in (1) into positive and negative samples for training, and obtain a voice wake-up determination model (the voice wake-up determination model obtained here is the voice wake-up determination model that has been trained in this step).

[0055] It should be noted that in this embodiment, the voice wake-up determination model uses a default preset initial voice wake-up volume threshold T to filter data during the preset audio collection period after power-on. Data that does not reach the threshold T is directly reported as having no wake-up response, while data that reaches the volume threshold T is passed to the voice wake-up algorithm for wake-up word determination. After the preset audio collection period after power-on, the voice wake-up determination model determines a second voice wake-up volume threshold T2, and uses the second voice wake-up volume threshold to filter data during the subsequent period. Data that does not reach the threshold T2 is directly reported as having no wake-up response, while data that reaches the volume threshold T2 is passed to the voice wake-up algorithm for wake-up word determination.

[0056] Step S13: When the voice wake-up determination model determines that the target voice data meets the target wake-up conditions, the corresponding wake-up operation is executed.

[0057] When the trained voice wake-up determination model determines that the target voice data meets the target wake-up conditions, the corresponding wake-up operation is executed.

[0058] like Figure 2 The diagram shows a voice wake-up flowchart proposed in this invention. When the target voice data is transmitted to the wake-up algorithm in real time, if the wake-up algorithm is passed, a wake-up response operation is performed; if the wake-up algorithm is not passed, a no-wake-up response operation is performed.

[0059] Each time the device is powered on, audio data for a given period is recorded and processed according to a pre-defined method. Based on the characteristics of the audio data during this period, corresponding parameters are calculated. These parameters are then used to adjust the data processing steps of the voice wake-up algorithm. In this way, different hardware devices, users, and background environmental factors are processed separately, reducing the impact of these differences on the voice wake-up algorithm. Furthermore, data filtering principles reduce hardware resource consumption.

[0060] In this embodiment, firstly, during a preset audio acquisition period after power-on, the first audio data collected by the audio input device for sound acquisition is recorded. Based on the volume information in the first audio data, a first voice wake-up volume threshold is calculated using preset calculation rules. Then, after the preset audio acquisition period, a second voice wake-up volume threshold is determined based on the first voice wake-up volume threshold and a preset initial voice wake-up volume threshold. Target voice data exceeding the second voice wake-up volume threshold is then identified from the second audio data collected in real-time by the audio input device. This target voice data is then input into a trained voice wake-up determination model. When the voice wake-up determination model determines that the target voice data meets the target wake-up condition, the corresponding wake-up operation is executed. This reduces the amount of data entering the wake-up algorithm by using a data filtering operation based on the user's volume. This reduces hardware resource consumption and mitigates the instability of the voice wake-up model due to differences in various implementation environments, while also reducing the risk of algorithm misjudgment and false wake-up.

[0061] Figure 3 A flowchart illustrating a specific voice wake-up method provided in this application embodiment. See also... Figure 3 As shown, the method includes:

[0062] Step S21: Record the first audio data collected by the audio input device used for sound acquisition during the preset audio acquisition time period after power-on. Based on the volume information in the first audio data, calculate the first voice wake-up volume threshold using preset calculation rules, and determine the first data feature corresponding to the first audio data using a preset data feature extraction algorithm. Then, determine the first feature parameter corresponding to the first data feature.

[0063] In this embodiment, while determining the first voice wake-up volume threshold, a preset data feature extraction algorithm is also used to determine the first data feature corresponding to the first audio data, and the first feature parameter corresponding to the first data feature is determined. It is understood that the data feature is continuously recorded during a preset audio acquisition time period after power-on. The feature parameter includes, but is not limited to, mean and variance.

[0064] Step S22: After the preset audio acquisition time period, determine the second feature parameter based on the first feature parameter and the preset initial feature parameter, update the second data parameter to the trained voice wake-up determination model, determine the second voice wake-up volume threshold based on the first voice wake-up volume threshold and the preset initial voice wake-up volume threshold, determine the target voice data greater than the second voice wake-up volume threshold from the second audio data collected in real time by the audio input device, and then input the target voice data into the trained voice wake-up determination model.

[0065] In this embodiment, the first feature parameter, the second feature parameter, and the preset initial feature parameter can all include the mean and variance of the audio data. Specifically, after obtaining the human voice audio data and the target audio data in step (1) of step S11, their mean and variance can be calculated as the default data standardization parameters M and S. Correspondingly, after the preset audio acquisition time period, in the process of determining the second feature parameter based on the first feature parameter and the preset initial feature parameter, if the calculated data of the first feature parameter is the mean M1 and the variance S1, the second feature parameter can be calculated using the following calculation rules:

[0066] M2 = 0.1*M + (1 - 0.1)*M1;

[0067] S2 = 0.1*S + (1 - 0.1)*S1;

[0068] Understandably, M2 and S2 are the mean and variance of the second characteristic parameter, respectively.

[0069] In this embodiment, after a preset audio acquisition time period, a second feature parameter is determined based on the first feature parameter and the preset initial feature parameter, and the second data parameter is updated in the trained voice wake-up determination model. Then, the subsequent voice wake-up process continues. The above process can be understood as a parameter replacement process, such as... Figure 4 The diagram shows a parameter replacement process. After powering on or the voice wake-up algorithm is started, the voice wake-up determination model runs with default parameters. Audio data is recorded during the preset audio acquisition period after powering on, and parameters used to replace the default parameters are calculated. After the preset audio acquisition period, the default parameters are replaced, and the process continues.

[0070] Step S23: When the voice wake-up determination model determines that the target voice data meets the target wake-up conditions, the corresponding wake-up operation is executed.

[0071] Furthermore, this invention can also pre-set a preset adjustment period to periodically adjust the parameters and volume threshold in the voice wake-up determination model. Specifically, it can be set to re-record the volume and data characteristics of the audio data during the preset audio acquisition time period every hour, and then calculate the volume threshold and data standardization parameters. A new volume threshold is calculated using the moving average method, and the newly obtained feature parameters are then used to replace the original threshold.

[0072] This embodiment simultaneously records audio volume and audio data characteristics, which greatly reduces interference to the algorithm caused by differences in audio acquisition devices and external ambient noise. It significantly improves the applicability of the algorithm model, eliminating the need for separate optimization for specific devices. This allows the voice wake-up algorithm to be used in scenarios with diverse hardware and complex background environments, such as e-sports hotels. It is adaptable to different environmental noise scenarios, dynamically adjusting as the environment changes, filtering out irrelevant data and reducing the probability of wake-up data being filtered out. It fully considers the impact of scene changes and environmental differences, allowing for adaptive adjustments based on environmental variations and fully utilizing prior information to make the data filtering strategy more accurate. Through these improvements, the invention possesses strong robustness and can operate on various hardware devices and in noisy environments, greatly enhancing the user experience on PCs in e-sports hotels.

[0073] See Figure 5 As shown in the figure, this application discloses a voice wake-up device, which may specifically include:

[0074] The first threshold calculation module 11 is used to record the first audio data collected by the audio input device for collecting sound during a preset audio acquisition time period after power-on, and to calculate the first voice wake-up volume threshold based on the volume information in the first audio data using a preset calculation rule.

[0075] The second threshold calculation module 12 is used to determine a second voice wake-up volume threshold based on the first voice wake-up volume threshold and the preset initial voice wake-up volume threshold after the preset audio acquisition time period, and to determine target voice data greater than the second voice wake-up volume threshold from the second audio data acquired in real time by the audio input device, and then input the target voice data into the trained voice wake-up judgment model.

[0076] The voice wake-up module 13 is used to perform a corresponding wake-up operation when the voice wake-up determination model determines that the target voice data meets the target wake-up conditions.

[0077] In this embodiment, firstly, during a preset audio acquisition period after power-on, the first audio data collected by the audio input device for sound acquisition is recorded. Based on the volume information in the first audio data, a first voice wake-up volume threshold is calculated using preset calculation rules. Then, after the preset audio acquisition period, a second voice wake-up volume threshold is determined based on the first voice wake-up volume threshold and a preset initial voice wake-up volume threshold. Target voice data exceeding the second voice wake-up volume threshold is then identified from the second audio data collected in real-time by the audio input device. This target voice data is then input into a trained voice wake-up determination model. When the voice wake-up determination model determines that the target voice data meets the target wake-up condition, the corresponding wake-up operation is executed. This reduces the amount of data entering the wake-up algorithm by using a data filtering operation based on the user's volume. This reduces hardware resource consumption and mitigates the instability of the voice wake-up model due to differences in various implementation environments, while also reducing the risk of algorithm misjudgment and false wake-up.

[0078] Furthermore, embodiments of this application also disclose an electronic device, Figure 6 This is a structural diagram of an electronic device 20 according to an exemplary embodiment. The content of the diagram should not be construed as limiting the scope of use of this application.

[0079] Figure 6 This is a schematic diagram of the structure of an electronic device 20 provided in an embodiment of this application. Specifically, the electronic device 20 may include: at least one processor 21, at least one memory 22, a power supply 23, a display screen 24, an input / output interface 25, a communication interface 26, and a communication bus 27. The memory 22 stores a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the voice wake-up method disclosed in any of the foregoing embodiments. Alternatively, the electronic device 20 in this embodiment may specifically be an electronic computer.

[0080] In this embodiment, the power supply 23 is used to provide operating voltage for each hardware device on the electronic device 20; the communication interface 26 can create a data transmission channel between the electronic device 20 and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this application, and is not specifically limited here; the input / output interface 25 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

[0081] Furthermore, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk, or optical disk, etc. The resources stored on it can include an operating system 221, computer programs 222, and virtual machine data 223, etc. The virtual machine data 223 can include various types of data. The storage method can be temporary storage or permanent storage.

[0082] The operating system 221 is used to manage and control the various hardware devices on the electronic device 20 and the computer program 222, which may be Windows Server, Netware, Unix, Linux, etc. In addition to including a computer program capable of performing the voice wake-up method executed by the electronic device 20 as disclosed in any of the foregoing embodiments, the computer program 222 may further include a computer program capable of performing other specific tasks.

[0083] Furthermore, this application also discloses a computer-readable storage medium, which includes random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disks, magnetic disks, optical disks, or any other form of storage medium known in the art. When the computer program is executed by a processor, it implements the aforementioned voice wake-up method. Specific steps of this method can be found in the corresponding content disclosed in the foregoing embodiments, and will not be repeated here.

[0084] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatuses disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the method section. Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0085] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0086] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0087] The above provides a detailed description of the voice wake-up method, apparatus, device, and storage medium provided by the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A voice wake-up method, characterized in that, Applied to electronic devices, including: Record the first audio data collected by the audio input device used for sound acquisition during the preset audio acquisition time period after power-on, and calculate the first voice wake-up volume threshold based on the volume information in the first audio data using preset calculation rules. After the preset audio acquisition time period, a second voice wake-up volume threshold is determined based on the first voice wake-up volume threshold and the preset initial voice wake-up volume threshold. Target voice data greater than the second voice wake-up volume threshold is then determined from the second audio data acquired in real time by the audio input device. The target voice data is then input into the trained voice wake-up determination model. When the voice wake-up determination model determines that the target voice data meets the target wake-up conditions, the corresponding wake-up operation is executed. The step of calculating the first voice wake-up volume threshold based on the volume information in the first audio data using a preset calculation rule includes: Based on the volume information in the first audio data and the preset initial voice wake-up volume threshold, the first voice wake-up volume threshold is calculated using the preset threshold calculation formula. The step of determining the second voice wake-up volume threshold based on the first voice wake-up volume threshold and the preset initial voice wake-up volume threshold includes: The second voice wake-up volume threshold is determined based on the first voice wake-up volume threshold and the preset initial voice wake-up volume threshold, and using a preset moving average algorithm.

2. The voice wake-up method according to claim 1, characterized in that, After recording the first audio data collected by the audio input device used for sound acquisition during the preset audio acquisition time period after power-on, the method further includes: A preset data feature extraction algorithm is used to determine the first data feature corresponding to the first audio data, and the first feature parameter corresponding to the first data feature is determined. Accordingly, the voice wake-up method further includes: After the preset audio acquisition time period, a second feature parameter is determined based on the first feature parameter and the preset initial feature parameter, and the second data parameter is updated to the trained voice wake-up determination model.

3. The voice wake-up method according to claim 2, characterized in that, The first feature parameter, the second feature parameter, and the preset initial feature parameter all include the mean and variance of the audio data.

4. The voice wake-up method according to claim 1, characterized in that, Also includes: Set a preset initial voice wake-up volume threshold; wherein the preset initial voice wake-up volume threshold is a threshold determined and set based on the quartiles calculated using volume information from pre-collected sample data.

5. The voice wake-up method according to claim 1, characterized in that, The step of calculating the first voice wake-up volume threshold based on the volume information in the first audio data using a preset calculation rule includes: Calculate the full decibel scale information in the first audio data; Based on the full decibel scale information, the first voice wake-up volume threshold is calculated using preset calculation rules.

6. A voice wake-up device, characterized in that, include: The first threshold calculation module is used to record the first audio data collected by the audio input device for collecting sound during a preset audio acquisition time period after power-on, and calculate the first voice wake-up volume threshold based on the volume information in the first audio data using preset calculation rules. The second threshold calculation module is used to determine a second voice wake-up volume threshold based on the first voice wake-up volume threshold and the preset initial voice wake-up volume threshold after the preset audio acquisition time period, and to determine target voice data that is greater than the second voice wake-up volume threshold from the second audio data acquired in real time by the audio input device, and then input the target voice data into the trained voice wake-up judgment model. The voice wake-up module is used to perform a corresponding wake-up operation when the voice wake-up determination model determines that the target voice data meets the target wake-up conditions. The first threshold calculation module includes: The first threshold calculation unit is used to calculate the first voice wake-up volume threshold based on the volume information in the first audio data and the preset initial voice wake-up volume threshold using a preset threshold calculation formula. The second threshold calculation module includes: The second threshold calculation unit is used to determine the second voice wake-up volume threshold based on the first voice wake-up volume threshold and the preset initial voice wake-up volume threshold, and using a preset moving average algorithm.

7. An electronic device, characterized in that, It includes a processor and a memory; wherein, when the processor executes a computer program stored in the memory, it implements the voice wake-up method as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, Used to store a computer program; wherein, when the computer program is executed by a processor, it implements the voice wake-up method as described in any one of claims 1 to 5.

Images