Distributed wake-up decision-making methods, electronic devices and software products

By calculating the wake-up decision score and dynamic backoff duration of multi-dimensional wake-up evaluation index parameters, the functional conflict problem of multi-device synchronous response is solved, achieving highly accurate and efficient device wake-up and improving user experience.

CN121510052BActive Publication Date: 2026-11-17HISENSE HOME APPLIANCES GRP CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202511414860.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2025-09-29
Publication Date
2026-11-17
Estimated Expiration
2045-09-29

AI Technical Summary

Technical Problem

When multiple smart devices respond to the same wake word simultaneously, it can easily lead to function trigger conflicts, making it impossible to accurately wake up the device the user expects, thus reducing the user experience of voice wake-up.

Method used

By acquiring multi-dimensional wake-up evaluation index parameters, a wake-up decision score is calculated, the backoff duration is dynamically calculated, and an immediate response is made when the score is higher than that of other devices to avoid accidentally waking up other devices and improve response accuracy and speed.

Benefits of technology

It improves the accuracy of wake-up decisions in multi-device scenarios, reduces the probability of false wake-ups, enhances the user interaction experience, ensures that highly compatible devices respond first, and reduces redundant calculations and power consumption.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121510052B_ABST
    Figure CN121510052B_ABST
Patent Text Reader

Abstract

This application discloses a distributed wake-up decision-making method, electronic device, and program product, relating to the field of electronic device technology. The method includes: after detecting a user voice signal containing a wake-up word, acquiring first wake-up event information; calculating a first wake-up decision score for a first smart device based on multiple wake-up evaluation index parameters; using the first wake-up decision score to quantify the matching degree of the first smart device as the user's desired wake-up device; calculating a first backoff time for the first smart device based on the first wake-up decision score; and, within the first backoff time, if a second wake-up decision score sent by a second smart device is acquired, executing a wake-up response action when the first wake-up decision score is greater than the second wake-up decision score. Using the above method, multiple smart devices can avoid parallel responses based on the same wake-up word, effectively shortening the invalid waiting time of the highly matched first smart device and improving the user experience of voice wake-up.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of electronic device technology, and more particularly to a distributed wake-up decision-making method, electronic device, and program product. Background Technology

[0002] With the deep integration of artificial intelligence and Internet of Things technologies, the smart home market has experienced explosive growth. Smart devices with voice interaction capabilities, such as smart speakers, smart TVs, smart air conditioners, smart door locks, and portable voice devices, have been widely penetrated into homes, offices, and other scenarios, forming a distributed interactive environment with multiple devices working together.

[0003] Currently, multiple smart devices are often deployed simultaneously in a single area such as a home or office. If multiple smart devices are configured with the same wake word, then after the user emits a voice signal containing the wake word, multiple smart devices in the area may respond synchronously.

[0004] However, when multiple smart devices respond in parallel based on the same wake word, it can easily lead to mutual interference (such as function trigger conflicts) in the subsequent execution of instructions. This can prevent the user from waking up only the smart device they want to wake up, resulting in confusing interactive feedback and a reduced user experience of voice wake-up. Summary of the Invention

[0005] This application provides a distributed wake-up decision-making method, electronic device, and program product, which can solve the problem that when multiple smart devices respond in parallel according to the same wake word, the smart devices are prone to interfering with each other in the subsequent instruction execution (such as function trigger conflict), making it impossible to wake up only the smart device that the user wants to wake up, resulting in confusing interactive feedback for the user and reducing the user experience of voice wake-up.

[0006] Firstly, a distributed wake-up decision-making method is provided for application in a first intelligent device, the method comprising:

[0007] After detecting a user voice signal containing a wake word, the first wake-up event information is obtained; the first wake-up event information includes wake-up evaluation index parameters of multiple dimensions;

[0008] The first wake-up decision score of the first smart device is calculated based on multiple wake-up evaluation index parameters; the first wake-up decision score is used to quantify the degree of matching of the first smart device with the user's expected wake-up device.

[0009] The first backoff duration of the first smart device is calculated based on the first wake-up decision score;

[0010] If a second wake-up decision score is obtained from the second smart device within the first backoff period, a wake-up response action is executed if the first wake-up decision score is greater than the second wake-up decision score.

[0011] In the above technical solution, after detecting a user voice signal containing a wake-up word, first wake-up event information including wake-up evaluation index parameters of multiple dimensions can be obtained, and a first wake-up decision score of the first smart device can be calculated based on the multiple wake-up evaluation index parameters. Furthermore, by introducing multi-dimensional wake-up evaluation index parameters to construct a quantitative decision system, the matching degree between the first smart device and the user's expectations is transformed into a calculable first wake-up decision score, replacing the traditional single-index parameter judgment, improving the accuracy of the first wake-up decision score evaluation, and avoiding misjudgments caused by interference from a single parameter. Finally, a first backoff duration is dynamically calculated based on the first wake-up decision score, and if a second wake-up decision score sent by a second smart device is obtained within the first backoff duration, a wake-up response action is executed when the first wake-up decision score is greater than the second wake-up decision score. Therefore, by dynamically designing the backoff duration and executing the wake-up response action when the first wake-up decision score is greater than the second wake-up decision score, the backoff waiting of the first smart device can have differentiated adaptability, and it can ensure that the responding first smart device is the device the user expects to wake up, reducing the probability of mistakenly waking up other smart devices and improving the user experience of voice wake-up. Furthermore, within the first backoff period, once the wake-up decision score comparison confirms that the matching degree of itself (the first smart device) is higher, the wake-up response action is executed directly without waiting for the first backoff period to end. This can also effectively shorten the invalid waiting time of the first smart device with a high matching degree, thereby improving the wake-up response speed and further enhancing the user experience of voice wake-up.

[0012] In one embodiment, multiple wake-up evaluation metrics include the wake-up angle and distance between the user and the device, the signal-to-noise ratio of the user's voice signal, and wake-up confidence. A first wake-up decision score for the first smart device is calculated based on these multiple wake-up evaluation metrics, including:

[0013] The wake-up angle, signal-to-noise ratio, and wake-up confidence are weighted and summed with their respective weights to obtain the first wake-up decision score.

[0014] In the above technical solution, the normalized results of each wake-up evaluation index parameter are systematically integrated by weighted summation. This retains core information such as the user's spatial orientation reflected by wake-up angle and distance, the voice signal quality reflected by signal-to-noise ratio, and the wake-up effectiveness represented by wake-up confidence. At the same time, the weight allocation highlights the impact of key indicators. The final output score can more accurately quantify the matching degree between the device and the user's expectations, providing a reliable basis for subsequent calculation of the first backoff duration and response decision-making. This effectively reduces the problem of false wake-ups or suboptimal device responses caused by isolated judgment of single parameters, and improves the decision-making accuracy and user interaction experience in distributed wake-up scenarios.

[0015] In one embodiment, the weight corresponding to distance is less than or equal to the weights corresponding to the other wake-up evaluation metric parameters.

[0016] In the above technical solution, by setting the weight corresponding to distance to be less than or equal to the weights corresponding to other wake-up evaluation index parameters, decision bias caused by insufficient accuracy of the distance parameter can be effectively avoided. That is, the interference of the low-reliability distance parameter on the total score can be reduced, making the first wake-up decision score more dependent on more reliable indicators such as signal-to-noise ratio, wake-up confidence, and wake-up angle, thereby improving the accuracy of the first decision score.

[0017] In one embodiment, calculating the first backoff duration of the first smart device based on the first wake-up decision score includes:

[0018] The product of the first wake-up decision score and the preset minimum waiting time is determined as the first backoff time.

[0019] In the above technical solution, the first backoff time is determined by multiplying the first wake-up decision score by a preset minimum waiting time. This allows for differentiated and dynamic calculation of the first backoff time, replacing the traditional fixed waiting window. This resolves the contradiction between the fixed short window (missing optimal device detection) and the fixed long window (delayed response). The first intelligent device with a high first wake-up decision score does not need to wait for a uniform long window and can quickly enter the decision-making stage in subsequent steps by comparing wake-up decision scores, ensuring timely wake-up response.

[0020] In one embodiment, calculating the first backoff duration of the first smart device based on the first wake-up decision score includes:

[0021] The product of the first wake-up decision score and the preset minimum waiting time is determined as the decision score backoff time;

[0022] A duration factor for the first smart device is randomly generated, and the product of the duration factor and the preset duration is determined as the random backoff duration;

[0023] The first backoff time is determined by the sum of the decision score backoff time, the random backoff time, and the preset backoff time corresponding to the first intelligent device.

[0024] In the above technical solution, multiplying the first wake-up decision score by the preset minimum waiting time to determine the decision score backoff time allows this part of the time of multiple intelligent devices to be directly related to their own and the user's expectations, ensuring that intelligent devices with high matching degree have a more reasonable waiting time in this dimension; and, by randomly generating a duration factor and multiplying it by the preset duration to obtain a random backoff time, the randomness can be used to break the situation where the duration of this part of the time of multiple intelligent devices is completely consistent, avoiding response conflicts caused by the end of the synchronous backoff time; and then summing these two parts of the time with the preset backoff time corresponding to the first intelligent device to obtain the first backoff time, so that the final first backoff time can take into account both the differences between multiple intelligent devices and the basic adaptability of intelligent devices, making the calculation of the first backoff time more reasonable.

[0025] In one embodiment, after calculating the first backoff duration of the first smart device based on the first wake-up decision score, the method further includes:

[0026] If a second wake-up decision score is obtained within the first backoff period, the wake-up process will be terminated if the first wake-up decision score is less than the second wake-up decision score.

[0027] If the second wake-up decision score is not obtained within the first backoff period, the wake-up response action is executed at the end of the first backoff period.

[0028] In the above technical solution, when a second wake-up decision score greater than its own first wake-up decision score is obtained, the first intelligent device can immediately and proactively exit the wake-up process. This avoids response conflicts between suboptimal and highly matched intelligent devices (such as simultaneous responses from multiple devices), and also terminates redundant calculations in a timely manner to save computing power and energy consumption. Furthermore, when a second wake-up decision score is not obtained, the first intelligent device can perform a wake-up response action after the backoff period ends. This provides a window for waiting for feedback from other second intelligent devices and avoids the wake-up non-response problem caused by the lack of a comparison object in a single intelligent device scenario. Based on this, through the complementary nature of the above solutions, accurate priority selection is achieved in a multi-intelligent device environment, while ensuring response stability in a single intelligent device environment, thereby improving decision reliability and user interaction experience.

[0029] In one embodiment, the distributed wake-up decision-making method further includes:

[0030] If a second wake-up decision score is obtained during the execution of the wake-up response action, the wake-up response action is terminated if the first wake-up decision score is less than the second wake-up decision score.

[0031] In the above technical solution, during the execution of the wake-up response action, if it is determined that the first wake-up decision score is less than the second wake-up decision score, the first smart device can immediately terminate the currently executing wake-up response action (such as interrupting the response voice playback or turning off the indicator light), and return from the woken-up state to the standby state, avoiding continued occupation of the user's interaction channel. Simultaneously, response space is reserved for the second smart device with the higher score, allowing it to normally initiate the wake-up response, ensuring that the device that ultimately responds to the user is the one with a higher matching degree.

[0032] In one embodiment, the distributed wake-up decision-making method further includes:

[0033] In standby mode, a voice signal containing a wake-up word is broadcast at preset intervals;

[0034] If the second wake-up event information sent after the second smart device responds with a voice signal is not received, then the wake-up response action will be executed after the user's voice signal is received in the next preset period.

[0035] If a second wake-up event message is received after the second smart device responds with a voice signal, then within the next preset period, the steps of obtaining the first wake-up event message after detecting a user voice signal containing a wake word, and subsequent steps, are executed.

[0036] In the above technical solution, in standby mode, a voice signal containing a wake-up word is broadcast at preset intervals. This allows for early detection of the presence of a second, collaborative smart device in the same area, providing environmental information for subsequent decisions and avoiding blind execution of processes. Secondly, regarding the detection results, if the first smart device does not receive a response from the second smart device, it indicates no multi-device competition scenario. Based on this, the first smart device can directly execute a wake-up response action upon receiving the user's voice signal in the next preset interval, eliminating redundant collaboration steps and significantly improving the interaction efficiency in single-device scenarios. Furthermore, if a response is received from the second device, it can be assumed that there is a need for multi-device collaboration in the area. In this case, in the next preset interval, the first smart device can execute the steps of obtaining the first wake-up event information after detecting the user's voice signal containing the wake-up word, as well as subsequent steps, ensuring that no conflicts arise between multiple smart devices due to synchronous responses. Therefore, this method not only guarantees the response speed in single-device scenarios but also ensures decision-making accuracy in multi-device wake-up scenarios, comprehensively optimizing the user's wake-up interaction experience.

[0037] Secondly, embodiments of this application provide a distributed wake-up decision-making device, which is applied to a first intelligent device, and the device includes:

[0038] The first acquisition module is used to acquire first wake-up event information after detecting a user voice signal containing a wake-up word; the first wake-up event information includes wake-up evaluation index parameters of multiple dimensions;

[0039] The first calculation module is used to calculate the first wake-up decision score of the first smart device based on multiple wake-up evaluation index parameters; the first wake-up decision score is used to quantify the degree of matching of the first smart device with the user's expected wake-up device.

[0040] The second calculation module is used to calculate the first backoff duration of the first smart device based on the first wake-up decision score;

[0041] The second acquisition module is used to, within the first backoff time, if it acquires the second wake-up decision score sent by the second smart device, execute a wake-up response action when the first wake-up decision score is greater than the second wake-up decision score.

[0042] Thirdly, embodiments of this application provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method described in the first aspect above.

[0043] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described in the first aspect above.

[0044] Fifthly, embodiments of this application provide a computer program product that, when run on an electronic device, causes the electronic device to execute the method described in the first aspect.

[0045] It is understood that the beneficial effects of the second to fifth aspects mentioned above can be found in the relevant descriptions in the first aspect mentioned above, and will not be repeated here. Attached Figure Description

[0046] Figure 1 This is a schematic diagram illustrating an application scenario for waking up multiple smart devices according to an embodiment of this application;

[0047] Figure 2 This is a configuration block diagram of a control device 100 for a smart device provided in an embodiment of this application;

[0048] Figure 3 This is a configuration block diagram of the controller 110 provided in one embodiment of this application;

[0049] Figure 4 This is a system block diagram of the controller 110 provided in one embodiment of this application;

[0050] Figure 5This is a timing interaction diagram of a distributed wake-up decision method provided in an embodiment of this application;

[0051] Figure 6 This is a schematic diagram illustrating an implementation of a distributed wake-up decision-making method according to an embodiment of this application, which calculates the first backoff duration.

[0052] Figure 7 This is a timing interaction diagram of a distributed wake-up decision method provided in another embodiment of this application;

[0053] Figure 8 This is a timing interaction diagram of a distributed wake-up decision method provided in another embodiment of this application;

[0054] Figure 9 This is a flowchart illustrating a distributed wake-up decision-making method provided in another embodiment of this application;

[0055] Figure 10 This is a schematic diagram of an electronic device provided in one embodiment of this application. Detailed Implementation

[0056] The technical solutions of the embodiments of this application will be described below with reference to the accompanying drawings. In the description of the embodiments of this application, unless otherwise stated, " / " means "or," for example, A / B can mean A or B; "and / or" in this text is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Furthermore, in the description of the embodiments of this application, "multiple" refers to two or more than two.

[0057] Hereinafter, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first," "second," or "third" may explicitly or implicitly include one or more of that feature.

[0058] Specific details, such as particular system architectures and techniques, are set forth for illustrative purposes and not for limitation, to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application can be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods are omitted to avoid unnecessary detail that could obscure the description of this application.

[0059] With the deep integration of artificial intelligence and Internet of Things technologies, the smart home market has experienced explosive growth. Smart devices with voice interaction capabilities, such as smart speakers, smart TVs, smart air conditioners, smart door locks, and portable voice devices, have been widely penetrated into homes, offices, and other scenarios, forming a distributed interactive environment with multiple devices working together.

[0060] Currently, multiple smart devices are often deployed simultaneously in a single area such as a home or office. If multiple smart devices are configured with the same wake word, then after the user emits a voice signal containing the wake word, multiple smart devices in the area may respond synchronously.

[0061] For example, refer to Figure 1 , Figure 1 This is a schematic diagram illustrating an application scenario for multi-device wake-up provided in an embodiment of this application. In a home setting, there may be multiple smart devices, including laptops, smartphones, and smart speakers. When a user emits a voice signal containing a wake word (e.g., "XX, OK"), multiple smart devices (laptops, smartphones) configured with the same wake word in the area may detect the voice signal and attempt to respond synchronously. This can easily lead to mutual interference during subsequent command execution (e.g., audio signal superposition, function trigger conflicts, etc.), resulting in confusing interactive feedback for the user and reducing the user experience of voice wake-up.

[0062] Based on this, in order to quickly enable the smart device expected by the user to respond, reduce the probability of accidentally waking up other smart devices, and improve the user experience of voice wake-up, one embodiment of this application provides a distributed wake-up decision-making method, which is applied to a first smart device. For example, it can be applied to the control device in the first smart device.

[0063] Figure 2 This is a configuration block diagram of a control device 100 for a smart device according to an embodiment of this application. For example... Figure 2 As shown, the control device 100 includes a controller 110, a communication interface 120, a user input / output interface 130, a memory, and a power supply. The control device 100 can receive user input operation commands and convert the operation commands into commands that the controller 110 can recognize and respond to, thus acting as an intermediary for interaction between the user and the controller 110.

[0064] Figure 3 This is a configuration block diagram of the controller 110 provided in one embodiment of this application, as shown below. Figure 3 As shown, the controller 110 includes at least one of the following: a display 260 and a tuner / demodulator 210, a communicator 220, a detector 230, an external device interface 240, an audio output interface 250, a memory, a power supply, and a user interface.

[0065] In some embodiments, the controller includes a processor, a video processor, an audio processor, a graphics processor, RAM, ROM, and a first interface to an nth interface for input / output.

[0066] The display 260 includes a display component for presenting display images, a backlight component for providing a light source, a driving component for driving the display of images, components for displaying video content, image content, and a menu control interface, and a user control UI interface.

[0067] The display component can be a liquid crystal panel, which consists of a liquid crystal molecule layer and electrodes that control the arrangement of the liquid crystal molecules. The liquid crystal panel displays images by controlling the transmission and blocking of light. The backlight component, or backlight source, is located behind or to the side of the liquid crystal panel and provides light. The backlight component can be an LED light source, a miniLED light source, etc., providing higher brightness and better color performance.

[0068] The display 260 may also include a local dimming component, which can realize local dimming; the local dimming component is used to control each backlight area in the backlight assembly, and the local dimming component may be composed of multiple circuits that independently control the backlight areas, and can adjust the brightness of each backlight area in real time according to the content of the displayed image.

[0069] The monitor 260 can be a liquid crystal display (LCD), an LED monitor, a 3D monitor, etc.

[0070] The communicator 220 is a component used to communicate with external devices or servers according to various communication protocol types. For example, the communicator may include at least one of the following: a Wi-Fi module, a Bluetooth module, a wired Ethernet module, other network communication protocol chips or near-field communication protocol chips, and an infrared receiver. The controller 110 can establish the transmission and reception of control signals and data signals with the control device 100 or the server 400 through the communicator 220.

[0071] The user interface can be used to receive control signals from the control device 100 (such as an infrared remote control).

[0072] Detector 230 is used to collect signals from the external environment or to interact with the external environment. For example, detector 230 includes a light receiver, a sensor for collecting ambient light intensity; or, detector 230 includes an image acquisition device, such as a camera, which can be used to collect external environmental scenes, user attributes, or user interaction gestures; or, detector 230 includes a sound acquisition device, such as a microphone, for receiving external sounds.

[0073] The external device interface 240 may include, but is not limited to, one or more of the following: High Definition Multimedia Interface (HDMI), analog or high-definition component input interface (component), composite video input interface (CVBS), USB input interface (USB), RGB port, etc. It may also be a composite input / output interface formed by multiple interfaces mentioned above.

[0074] The tuner / demodulator 210 receives broadcast television signals via wired or wireless means, and demodulates audio and video signals, such as EPG data signals, from multiple wireless or wired broadcast television signals.

[0075] In some embodiments, the controller 110 and the tuner 210 may be located in different separate devices, that is, the tuner 210 may also be located in an external device of the main device where the controller 110 is located, such as an external set-top box.

[0076] The controller 110 controls the operation of the smart device and responds to user operations through various software control programs stored in the memory. The controller 110 controls the overall operation of the smart device. For example, in response to receiving a user command to select a UI object to display on the display 260, the controller 110 can perform operations related to the object selected by the user command.

[0077] In some embodiments, the controller includes at least one of a central processing unit (CPU), a video processor, an audio processor, a graphics processing unit (GPU), RAM (random access memory), ROM (read-only memory), a first to an nth interface for input / output, a communication bus, etc.

[0078] Users can input commands through a graphical user interface (GUI) displayed on the monitor 260, and the user input interface receives the user input commands through the GUI. Alternatively, users can input commands by entering specific sounds or gestures, and the user input interface receives the user input commands by recognizing the sounds or gestures through sensors.

[0079] A "user interface" is the medium through which an application or operating system interacts and exchanges information with the user. It converts information from its internal form to a form that the user can accept. A common form of user interface is the graphical user interface (GUI), which refers to a user interface related to computer operation displayed graphically. It can be an icon, window, control, or other interface element displayed on the screen of an electronic device. Controls can include visual interface elements such as icons, buttons, menus, tabs, text boxes, dialog boxes, status bars, navigation bars, and widgets.

[0080] Figure 4 This is a system block diagram of the controller 110 provided in an embodiment of this application, as shown below. Figure 4 As shown, in some embodiments, the system is divided into four layers, from top to bottom: the Applications layer (referred to as the "Application Layer"), the Application Framework layer (referred to as the "Framework Layer"), the Android runtime and system library layer (referred to as the "System Runtime Layer"), and the kernel layer.

[0081] In some embodiments, at least one application runs in the application layer. These applications may be Windows programs, system settings programs, or clock programs that come with the operating system; they may also be applications developed by third-party developers. In specific implementations, the application packages in the application layer are not limited to the examples above.

[0082] The framework layer provides application programming interfaces (APIs) and a programming framework for applications. The application framework layer includes predefined functions. It acts as a central processing unit, determining the actions taken by applications within the application layer. Through the API, applications can access system resources and obtain system services during execution.

[0083] like Figure 4As shown, the application framework layer in this embodiment includes managers, content providers, etc., wherein the managers include at least one of the following modules: ActivityManager, which interacts with all activities running in the system; LocationManager, which provides access to system location services for system services or applications; PackageManager, which retrieves various information related to application packages currently installed on the device; NotificationManager, which controls the display and clearing of notification messages; and WindowManager, which manages icons, windows, toolbars, wallpapers, and desktop widgets on the user interface.

[0084] In some embodiments, the Activity Manager manages the lifecycle of individual applications and common navigation and back functions, such as controlling application exit, opening, and back actions. The Window Manager manages all window programs, such as obtaining the screen size, determining if a status bar is present, locking the screen, capturing the screen, and controlling display window changes (e.g., shrinking the display window, shaking the display, distorting the display, etc.).

[0085] In some embodiments, the system runtime library layer provides support for the upper layer, namely the framework layer. When the framework layer is used, the Android operating system runs the C / C++ libraries contained in the system runtime library layer to implement the functions that the framework layer needs to perform.

[0086] In some embodiments, the kernel layer is a layer between hardware and software. For example... Figure 4 As shown, the kernel layer includes at least one of the following drivers: audio driver, display driver, Bluetooth driver, camera driver, WIFI driver, USB driver, HDMI driver, sensor driver (such as fingerprint sensor, temperature sensor, pressure sensor, etc.), and power driver.

[0087] The term "first smart device" refers to one of all smart devices that can be woken up by the wake word. This means that all smart devices with the wake-up function of that wake word can be used as the first smart device to execute the method. Furthermore, when a smart device executes the method as the first smart device, other smart devices in the same area that can be woken up by the same wake word can be considered as second smart devices.

[0088] For example, refer to Figure 1When a laptop computer performs the above method, the laptop computer can be considered the first smart device, and the smartphone and smart speaker in the area can be considered the second smart devices. When a smartphone performs the above method, the smartphone can be considered the first smart device, and the laptop computer and smart speaker in the area can be considered the second smart devices; and when a smart speaker performs the above method, the smart speaker can be considered the first smart device, and the laptop computer and smartphone in the area can be considered the second smart devices.

[0089] It is evident that the first intelligent device and the second intelligent device are in a dynamic relationship, with the device executing the method at the core and the rest referring to the intelligent device with the wake word.

[0090] Please see Figure 5 , Figure 5 This is a timing interaction diagram of a distributed wake-up decision method provided in an embodiment of this application. The method includes the following steps:

[0091] S501. After detecting a user voice signal containing a wake-up word, obtain the first wake-up event information.

[0092] The aforementioned first wake-up event information includes wake-up evaluation index parameters from multiple dimensions.

[0093] In one embodiment, the user's voice signal refers to the electrical or digital signal generated by the user's speech, which is converted by the audio acquisition module (such as a microphone) of the smart device. The first smart device can capture the user's voice content containing a pre-set wake-up word (e.g., "Hello, Xiao X") through its own audio acquisition module (such as a microphone). The detection method can employ keyword spotting (KWS) technology. For example, the smart device performs real-time analysis on the audio stream continuously acquired by the microphone, and when it identifies an audio segment that matches the pattern of the preset wake-up word, it determines that a user's voice signal containing the wake-up word has been detected.

[0094] The aforementioned first wake-up event information is a set of comprehensive decision-making information collected by the first smart device after detecting a voice signal containing a wake-up word in order to assess its matching degree with the user's wake-up needs. It includes wake-up evaluation index parameters of multiple dimensions.

[0095] For example, the wake-up evaluation parameters mentioned above include, but are not limited to, the wake-up angle and distance between the user and the wake-up signal, the signal-to-noise ratio of the user's voice signal, and the wake-up confidence level, and there are no limitations on these parameters.

[0096] The wake-up angle can be considered as the angle between the user's voice direction and the smart device's sound pickup direction; the wake-up distance is the straight-line distance between the user and the first smart device. Both can be used to measure the spatial matching degree between the first smart device and the user (e.g., the closer the distance and the more positive the angle, the higher the rationality of the first smart device's response).

[0097] First-generation smart devices are typically equipped with multi-microphone arrays. They can use the Sound Source Localization (SSL) algorithm to calculate the direction of arrival (wake-up angle) of the user's voice by utilizing the time difference and phase difference of the voice signals received by different microphones. They can also combine this with a sound wave propagation attenuation model (such as based on the law of sound energy attenuation with distance) to estimate the wake-up distance.

[0098] Furthermore, some devices can typically be linked to camera devices. In this case, when a user's voice signal containing a wake-up word is detected, the camera device can be controlled to capture the user's face, and the user's position can be determined through facial recognition, thereby calculating the angle and distance to the first smart device.

[0099] The signal-to-noise ratio (SNR) is the ratio of the effective power of the user's voice signal to the background noise power, which reflects the clarity of the voice (the higher the SNR, the clearer the voice and the more reliable the wake word recognition).

[0100] The first intelligent device can perform spectrum analysis on the collected audio signal containing the wake word, separate the speech component from the noise component through a noise estimation algorithm (such as noise power spectrum estimation based on a statistical model), and then calculate the power ratio of the two to obtain the signal-to-noise ratio.

[0101] The aforementioned wake-up confidence level can be considered as the degree of confidence that the first intelligent device believes that the detected user voice signal does indeed contain a wake word (the higher the value, the more certain it is that it is a valid wake-up rather than a false trigger).

[0102] For example, the first smart device can use a wake word recognition model (such as a deep neural network model) to reason about audio segments and output the probability value or score of belonging to the wake word (such as a confidence score in the 0-1 range after normalization).

[0103] It should be noted that multiple wake-up evaluation metrics are the basic inputs for calculating the first wake-up decision score. Different parameters can reflect the degree of fit between the first smart device and the user's needs from different dimensions. For example, the signal-to-noise ratio can reflect the clarity of speech, and the wake-up angle can reflect the user's orientation. Combining multiple parameters can avoid the one-sidedness of a single parameter.

[0104] In another embodiment, the first wake-up event information may further include the following information:

[0105] Device Unique Identifier (DEV-UUID): A globally unique identifier for a device, used to accurately distinguish different smart devices in multi-device scenarios and ensure the uniqueness of device identity.

[0106] Wake-up event timestamp (WAKE-UTC): The UTC time at which this wake-up event occurred. It is used to accurately mark the time node when the event occurred and supports time synchronization and event sequence alignment between multiple devices.

[0107] Wake-ID: A unique identifier for this wake-up session, used to distinguish different wake-up events within a single device (such as scenarios where the same device is woken up multiple times), enabling session-level tracking and management.

[0108] Wake-up Word Uniform Identifier (WORD-ID): A globally unique identifier for wake-up words, used to distinguish different wake-up words. Only when wake-up events from multiple devices correspond to the same WORD-ID will they participate in collaborative decision-making, avoiding response conflicts caused by different wake-up words.

[0109] The first wake-up decision score (SCOR-STRA) and the first backoff duration (BK-TIME) are calculated in the following steps.

[0110] WAKE-STATE refers to the current state of a device, including standby (no wake word detected, in sleep or low-power listening state), pre-wake state (a suspected wake word detected, entering a low-power waiting and decision state), and woken state (that is, the state where a response is required and a wake-up response action is executed), which is used to identify the various stages of the device in the wake-up process.

[0111] In this embodiment, the parameters included in the first wake-up event information are not limited.

[0112] It should be added that the first smart device must not be powered off when no user voice signal is detected. For example, it can be in standby mode to detect user voice signals. Then, upon detecting a user voice signal containing a wake-up word, it can enter a pre-wake state. In this pre-wake state, the first smart device can be considered to be operating in a low-power mode, but not yet in the formal response phase.

[0113] S502. Calculate the first wake-up decision score of the first smart device based on multiple wake-up evaluation index parameters.

[0114] The aforementioned first wake-up decision score is used to quantify the degree to which the first intelligent device matches the user's desired wake-up device. Based on this, it can be considered that the first intelligent device obtains a single quantitative value after standardizing (e.g., parameter normalization) and weighting (e.g., dynamically allocating the weights of each parameter according to the scenario) various wake-up evaluation index parameters.

[0115] It should be noted that the first wake-up decision score, which converts multi-dimensional parameters into a uniform scale using a fixed algorithm, is comparable and objective, and is not a subjective judgment. Generally, the higher the score, the more likely the first smart device is to be the target device the user actually wants to wake up; conversely, the lower the score, the less it matches the user's needs, providing data to guide whether to execute a response in the future.

[0116] The matching degree of the user's desired wake-up device refers to the degree of compatibility between the first smart device and the target device the user actually wants to wake up. For example, if a user says a wake-up word in the living room, the smart device they expect to wake up might be the living room smart speaker, rather than the bedroom smart TV. In this case, the matching degree of the living room speaker is usually much higher than that of the bedroom TV.

[0117] Based on this, the matching degree can be considered to be the degree of matching between the user's subjective needs and the objective attributes of the smart device (multi-dimensional wake-up evaluation index parameters). The user's subjective needs (which smart device to wake up) cannot be directly perceived by the first smart device, but need to be indirectly reflected through objective parameters such as wake-up angle, distance, and signal-to-noise ratio, and finally quantified through the first wake-up decision score. This avoids false responses from multiple smart devices and ensures that the optimal device responds first.

[0118] As an example, taking multiple wake-up evaluation parameters, including the wake-up angle and distance between the user and the user, the signal-to-noise ratio of the user's voice signal, and the wake-up confidence, the first smart device can sum up the wake-up angle, distance, signal-to-noise ratio of the user's voice signal, and wake-up confidence to obtain the aforementioned first wake-up decision score.

[0119] In another embodiment, the first electronic device may further sum the wake-up angle, signal-to-noise ratio, and wake-up confidence with their corresponding weights to obtain a first wake-up decision score.

[0120] The weights corresponding to the various wake-up evaluation index parameters can be set in advance based on actual requirements. For example, all weights can be equal, without any limitation.

[0121] As an example, the weight corresponding to distance is less than or equal to the weights corresponding to the other wake-up evaluation metric parameters.

[0122] It should be noted that most current smart devices (such as smart speakers and portable voice terminals) rely on dual-MIC configuration to estimate distance. Due to the small distance between the two MICs (usually only a few centimeters), the time difference measurement error is large, and the algorithm is simplified (no complex environmental compensation model is introduced to control power consumption). The distance estimation result often has an error of more than ±1 meter, and the accuracy is far lower than other wake-up evaluation index parameters.

[0123] For example, the signal-to-noise ratio can be accurately calculated through spectrum analysis, the wake-up confidence is output by the deep model with controllable error, and the wake-up angle is mostly less than ±5° through the sound source localization algorithm.

[0124] Furthermore, distance estimation is easily affected by obstacles, environmental echoes, and other interferences, leading to further inaccuracies. Other wake-up evaluation metrics can mitigate the impact of the environment through algorithmic optimization. If distance is given too high a weight, the low-accuracy distance parameter will excessively dominate the decision-making process, thereby reducing the match between the score and the user's expectations.

[0125] As an example, the weight corresponding to the wake-up angle can be 0.3, the weight corresponding to the distance can be 0.1, the weight corresponding to the signal-to-noise ratio can be 0.3, and the weight corresponding to the wake-up confidence can be 0.3.

[0126] Based on the above explanation, in this embodiment, by setting the weight corresponding to distance to be less than or equal to the weights corresponding to the other wake-up evaluation index parameters, the decision bias caused by insufficient accuracy of the distance parameter can be effectively avoided. That is, the interference of the low-reliability distance parameter on the total score can be reduced, making the first wake-up decision score more dependent on more reliable indicators such as signal-to-noise ratio, wake-up confidence, and wake-up angle, thereby improving the accuracy of the first decision score.

[0127] It should be added that, since the wake-up angle, distance, signal-to-noise ratio of the user's voice signal, and wake-up confidence dimensions are different, the first intelligent device can first normalize the above-mentioned multiple wake-up evaluation index parameters to obtain the normalized wake-up angle, distance, signal-to-noise ratio of the user's voice signal, and wake-up confidence. Finally, the normalized wake-up angle, distance, signal-to-noise ratio of the user's voice signal, and wake-up confidence are summed to obtain the first wake-up decision score.

[0128] The normalization methods include, but are not limited to, linear normalization and standard normalization, and there are no restrictions on these methods.

[0129] As an example, the normalization method for the wake-up angle can be: A = 1 - a / 90°. Where a is the wake-up angle, and A is the normalized score corresponding to the wake-up angle.

[0130] It should be noted that the wake-up angle 'a' refers to the angle (in degrees) between the direction the user speaks and the direction the smart device picks up the sound. Generally, the smaller the angle (e.g., 0°, when the user is facing the smart device directly), the more likely the smart device is to be the one the user wants to wake up; the larger the angle (e.g., 90°, when the user is facing the device from the side), the lower the matching degree.

[0131] In real-world scenarios, the effective interaction angle range is typically 0° to 90°. Beyond 90°, the user is essentially facing away from the device, and the matching degree can be considered zero. Based on the above normalization method, when a = 0° (optimal angle), A = 1 (maximum value); when a = 90° (critical angle), A = 0 (minimum value), accurately reflecting the rule that the smaller the angle, the higher the matching degree.

[0132] The distance normalization method can be: D = 1 - d / 30. Where d is the distance, and D is the normalized score corresponding to the distance. d refers to the straight-line distance between the user and the device (unit: meters). The closer the distance, the clearer the voice signal received by the smart device, and the more likely it is to be the smart device the user wants to wake up; when the distance is too far, the signal attenuation is severe, and the matching degree decreases.

[0133] In real-world scenarios, taking an effective interaction distance of 0-30 meters as an example, beyond 30 meters, the voice signal attenuates to the point of being unreliable, and the matching degree can be considered 0. Based on the above normalization method, when d = 0 (closest distance), D = 1 (maximum value); when d = 30 (critical distance), D = 0 (minimum value), which conforms to the rule that the closer the distance, the higher the matching degree.

[0134] The normalization method for signal-to-noise ratio (SNR) can be: S = s / 30. Here, s is the SNR, and S is the normalized score corresponding to the SNR. The SNR s refers to the ratio of the user's speech signal power to the background noise power (unit: dB). A higher SNR (e.g., 30dB, clearer speech and less noise) makes it easier for smart devices to accurately recognize the wake word, resulting in a higher matching degree; a lower SNR (e.g., 0dB, speech mixed with noise) leads to poorer recognition reliability and a lower matching degree.

[0135] In real-world scenarios, the effective wake-up signal-to-noise ratio (SNR) typically ranges from 0 to 30 dB. Below 0 dB, noise overwhelms the speech, making it virtually unrecognizable; at 30 dB, the speech is clearly identifiable. Based on the above normalization method, calculations show that when s = 30 dB (optimal SNR), S = 1 (maximum value); and when s = 0 dB (critical SNR), S = 0 (minimum value), directly reflecting the principle that a higher SNR results in a higher matching degree.

[0136] The normalization method for wake-up confidence can be: C = c / 1000. Here, c is the wake-up confidence score, and C is the normalized score corresponding to the wake-up confidence score. c is the probability that the device judges the detected signal to contain a wake-up word, usually a score output by the model, ranging from 0 to 1000. A larger value indicates a more certain and effective wake-up.

[0137] Taking a confidence score in the range of 0 to 1000 as an example, 0 indicates no match to the wake word at all, and 1000 indicates a perfect match. By directly converting the raw scores (wake-up confidence scores) of 0 to 1 into normalized values ​​of 0 to 1, we can see that the larger the value, the higher the match.

[0138] In this embodiment, the normalized results of each wake-up evaluation index parameter are systematically integrated by weighted summation. This retains core information such as the user's spatial orientation reflected by wake-up angle and distance, the voice signal quality reflected by signal-to-noise ratio, and the wake-up effectiveness represented by wake-up confidence. At the same time, the weight allocation highlights the impact of key indicators. The final output score can more accurately quantify the matching degree between the device and the user's expectations, providing a reliable basis for subsequent calculation of the first backoff duration and response decision-making. This effectively reduces the problem of false wake-ups or suboptimal device responses caused by isolated judgment of single parameters, and improves the decision-making accuracy and user interaction experience in distributed wake-up scenarios.

[0139] S503. Calculate the first backoff duration of the first smart device based on the first wake-up decision score.

[0140] In one embodiment, the first backoff duration is a waiting time calculated by the first intelligent device based on its own first wake-up decision score. During the first backoff duration, the first intelligent device will not immediately execute a wake-up response, but will wait to receive the wake-up decision scores of other intelligent devices (second intelligent devices) in the same area to confirm whether it is the current optimal response device.

[0141] In one embodiment, the first smart device may determine the first backoff duration based on a preset mapping relationship between wake-up decision score and backoff duration.

[0142] In another embodiment, the first smart device may also determine the first backoff duration by multiplying the first wake-up decision score by a preset minimum waiting time.

[0143] In one embodiment, the minimum waiting time is a pre-set baseline waiting time, which is a quantified parameter for calculating the first backoff time of the first intelligent device. Its value is usually determined based on the actual needs of score interaction between devices. For example, it can cover the communication latency of transmitting wake-up decision scores between intelligent devices, ensuring that even in scenarios with slight network fluctuations or differences in device processing speeds, the first intelligent device can complete score interaction with other devices within the backoff period.

[0144] For example, the minimum waiting time can be 5ms, which avoids the first smart device missing the scores of other devices due to the short time (causing false responses), and also avoids the overall wake-up delay due to the long time.

[0145] It should be noted that by multiplying the first wake-up decision score by a preset minimum waiting time to determine the first backoff time, the first backoff time can be dynamically calculated in a differentiated manner. This replaces the traditional fixed waiting window, resolving the contradiction between the fixed short window missing the optimal device and the fixed long window delaying the response. The first intelligent device with a high first wake-up decision score does not need to wait for a uniform long window and can quickly enter the decision-making stage for whether to respond in subsequent steps by comparing the wake-up decision scores, ensuring the timeliness of the wake-up response.

[0146] S504. If a second wake-up decision score is obtained from the second smart device within the first backoff time, a wake-up response action is executed when the first wake-up decision score is greater than the second wake-up decision score.

[0147] In one embodiment, the aforementioned second wake-up decision score is a score calculated by the second intelligent device (i.e., other intelligent devices that can be woken up by the same wake-up word besides the first intelligent device currently executing the logic) based on the wake-up evaluation index parameters it has collected.

[0148] It is understandable that when the first smart device X executes the above method, it can calculate its own first wake-up decision score X1 and send the first wake-up decision score to other second smart devices, such as second smart device B. At this time, when the second smart device B executes the above method, the second smart device B will be considered the first smart device B, and the first wake-up decision score X1 obtained by the second smart device B is its second wake-up decision score relative to the first smart device B. Furthermore, when the first smart device B calculates its own first wake-up decision score X2 and sends it to the first smart device A, the first wake-up decision score X2 is its second wake-up decision score relative to the first smart device A.

[0149] In another embodiment, the first smart device may also only send the first wake-up event information to other second smart devices. That is, it can be assumed that the first smart device can receive the second wake-up event information sent by other second smart devices. In this case, the first smart device can calculate the second wake-up decision score based on the second wake-up event information.

[0150] It should be noted that the calculation logic of the second wake-up decision score is exactly the same as that of the first wake-up decision score. Its core function is to quantify the matching degree between the second smart device and the device that the user expects to wake up.

[0151] In this embodiment, the first smart device can receive the second wake-up decision score sent by the same second smart device via a local area network (such as Wi-Fi or Bluetooth) or other data transmission methods (e.g., wired or wireless communication protocols such as BLE / Ethernet), ensuring that there is a comparable priority basis. In this embodiment, the communication method between the various smart devices is not limited.

[0152] The aforementioned wake-up response actions are a set of specific behaviors by which the first smart device switches from the pre-wake state to the woken state after confirming that it is the device that the user wants to wake up. It is a direct feedback from the first smart device to the user's wake-up request. In this state, it can respond to the user's voice commands and execute corresponding actions.

[0153] It should be noted that if the first wake-up decision score (e.g., 0.9) is greater than the second wake-up decision score (e.g., 0.6), it can be considered that the first smart device has a higher degree of matching with the user's expectations. In this case, there is no need to wait for the first backoff time to end, but the wake-up response action can be executed directly.

[0154] It is understandable that by replacing the fixed wait for the backoff to end with the larger difference between the first wake-up decision score and the second wake-up decision score within the first backoff period, the first intelligent device with a higher matching degree can confirm the priority and respond in advance. This avoids conflicts caused by multiple intelligent devices responding simultaneously after synchronous waiting, and also shortens the invalid waiting time of the first intelligent device with a high matching degree, thus balancing decision accuracy and wake-up response speed.

[0155] In this embodiment, after detecting a user voice signal containing a wake-up word, first wake-up event information including wake-up evaluation index parameters of multiple dimensions can be obtained, and a first wake-up decision score of the first smart device can be calculated based on the multiple wake-up evaluation index parameters. Furthermore, by introducing multi-dimensional wake-up evaluation index parameters to construct a quantitative decision system, the matching degree between the first smart device and the user's expectations is transformed into a calculable first wake-up decision score, replacing the traditional single-index parameter judgment, improving the accuracy of the first wake-up decision score evaluation, and avoiding misjudgments caused by interference from a single parameter. Finally, a first backoff duration is dynamically calculated based on the first wake-up decision score, and if a second wake-up decision score sent by a second smart device is obtained within the first backoff duration, a wake-up response action is executed when the first wake-up decision score is greater than the second wake-up decision score. Therefore, by dynamically designing the backoff duration and executing the wake-up response action when the first wake-up decision score is greater than the second wake-up decision score, the backoff waiting of the first smart device can have differentiated adaptability, and it can ensure that the responding first smart device is the device the user expects to wake up, reducing the probability of mistakenly waking up other smart devices and improving the user experience of voice wake-up. Furthermore, within the first backoff period, once the wake-up decision score comparison confirms that the matching degree of itself (the first smart device) is higher, the wake-up response action is executed directly without waiting for the first backoff period to end. This can also effectively shorten the invalid waiting time of the first smart device with a high matching degree, thereby improving the wake-up response speed and further enhancing the user experience of voice wake-up.

[0156] In another embodiment, in order to reasonably generate the first backoff duration of the first intelligent device, the first intelligent device may also use, as shown in the example... Figure 6 The steps S601-S603 shown generate the first backoff duration. Details are as follows:

[0157] S601. The product of the first wake-up decision score and the preset minimum waiting time is determined as the decision score backoff time.

[0158] The method for calculating the decision score backoff time can be similar to the method described above, which determines the first backoff time by multiplying the first wake-up decision score by the preset minimum waiting time, and will not be described in detail here.

[0159] S602. Randomly generate the duration factor of the first intelligent device, and determine the random retreat duration by multiplying the duration factor by the preset duration.

[0160] In one embodiment, the aforementioned duration factor is a value randomly generated by the first smart device, typically between 0 and 1, such as 0.3 or 0.72, used to dynamically adjust the preset duration, ensuring that the final generated random backoff duration is differentiated. This randomness ensures that different smart devices will not generate completely identical backoff times under the same conditions.

[0161] The above preset duration is a pre-set base time value (such as 50ms, 100ms), which is the base duration for calculating the random retreat duration.

[0162] It should be noted that the random waiting time obtained by multiplying the duration factor by the preset duration can prevent the backoff cycles of multiple smart devices from being completely synchronized.

[0163] S603. The sum of the decision score backoff time, the random backoff time, and the preset backoff time corresponding to the first intelligent device is determined as the first backoff time.

[0164] In one embodiment, the preset backoff duration is a fixed base waiting time pre-set by the first smart device, and is a component of the first backoff duration. Its value is not randomly set, but can be determined based on the minimum interaction requirements between the smart devices.

[0165] For example, the preset backoff time for central devices such as smart speakers and smart TVs can be 0ms; the preset backoff time for home appliances such as smart air conditioners and smart water dispensers can be 100ms. In this embodiment, the method of setting the preset backoff time is not limited.

[0166] As an example, different preset backoff times can be set for different smart devices based on the number or frequency of times they are woken up.

[0167] In this embodiment, multiplying the first wake-up decision score by a preset minimum waiting time to determine the decision score backoff time allows this portion of the time for multiple smart devices to be directly correlated with their own and the user's expectations, ensuring that smart devices with high matching degrees have a more reasonable waiting time in this dimension. Furthermore, by randomly generating a duration factor and multiplying it by a preset duration to obtain a random backoff time, the randomness can be used to break the situation where the durations of multiple smart devices are completely consistent, avoiding response conflicts caused by the end of the synchronous backoff time. Finally, summing these two portions of the time with the preset backoff time corresponding to the first smart device to obtain the first backoff time allows the final first backoff time to have both the differences between multiple smart devices and the basic adaptability of smart devices, making the calculation of the first backoff time more reasonable.

[0168] In another embodiment, the calculation of the first backoff time can also be based on ambient noise. For example, the first backoff time = preset backoff time + ambient noise decibel value × influence coefficient. The ambient noise decibel value is collected and quantified in real time by the device's microphone (e.g., the sound of kitchen utensils colliding, the sound of a television / conversation in the living room, with noise levels potentially reaching 50-70 dB); the influence coefficient is a preset fixed weight (e.g., 0.5) used to quantify the degree of influence of noise on the backoff time.

[0169] Understandably, the above calculation method is adapted to the actual needs of noisy scenarios. The greater the ambient noise, the more easily the device is interfered with when picking up sound (such as reduced signal-to-noise ratio and fluctuations in wake-up confidence). At this time, by multiplying the noise decibel value by a coefficient to extend the backoff time, the device can have more time to receive the wake-up decision scores of other devices in the same area, avoiding misjudgments caused by noise affecting a single device, and ensuring that the optimal response device can still be locked based on more comprehensive information in an interfering environment.

[0170] In another embodiment, the first backoff time can also be calculated based on the user's historical habits. For example, a pre-trained neural network model learns the user's historical interaction data (such as the frequency of device wake-ups in different time periods / scenes, and the types of devices the user continuously interacts with after wake-up), prioritizing shorter backoff times for devices with high user preference. Specifically, the calculation can be based on the original first backoff time (e.g., decision score time, or decision score backoff time + random backoff time + preset backoff time) by adding a preference coefficient correction term: for high-preference smart devices, a preference coefficient less than 1 (e.g., 0.8) is introduced, and by multiplying it with the original first backoff time, a shorter first backoff time can be obtained; for low-preference smart devices, the original time is maintained or slightly extended.

[0171] Understandably, the above calculation method can focus on personalized experience. For commonly used smart devices, shortening the first backoff time of the first smart device can allow it to finish waiting and respond faster, avoiding low-preference smart devices from triggering the response first due to similar first backoff times. This makes the wake-up decision more in line with the user's potential needs and improves the convenience and satisfaction of the interaction.

[0172] In another embodiment, reference is made to Figure 7 Unlike the scenario corresponding to step S504 above, after calculating the first retreat time, the first intelligent device may also be in various other scenarios, such as S505-S507. Details are as follows:

[0173] S505. If a second wake-up decision score is obtained within the first backoff time, the wake-up process is exited when the first wake-up decision score is less than the second wake-up decision score.

[0174] In one embodiment, the above wake-up process may involve re-entering standby mode and waiting for the next wake-up signal from the user's voice containing the wake word.

[0175] As explained above, the first intelligent device can send its own first wake-up decision score or first wake-up event information to the outside world. Therefore, it can be assumed that the first intelligent device can receive the second wake-up decision score sent by the second intelligent device ("first intelligent device").

[0176] Understandably, when the score of the first wake-up decision is lower than that of the second wake-up decision, the second smart device can be considered to have a higher match with the device the user expects to wake up. In this case, the first smart device can proactively exit the wake-up process, stop subsequent score comparisons and response preparations, and return to standby mode. This not only avoids synchronization conflicts with the second device caused by attempting to respond despite having a lower score, but also reduces unnecessary computational power consumption (as it no longer needs to maintain a pre-wake state).

[0177] S506. If the second wake-up decision score is not obtained within the first backoff duration, the wake-up response action is executed at the end of the first backoff duration.

[0178] In one embodiment, when a second wake-up decision score is not obtained, there may be two scenarios: First, there are no other wakeable second smart devices in the same area (e.g., the user has only placed this one first smart device in the room, and there are no other devices that can receive the same wake word); second, a second smart device exists, but it has not detected the wake word, or has not participated in this wake-up decision, or has not sent the second wake-up decision score to the first smart device within the first backoff period due to network latency. However, regardless of the situation, the essence is that the first smart device does not obtain any external data that can be used for comparison during the backoff period (first backoff period) while waiting for priority comparison with other smart devices, and there is no risk of multi-device response conflict.

[0179] Based on this, it can be argued that if the first smart device continues to wait, there may be no response from the smart device after the user wakes up, resulting in a gap in the user experience. Therefore, taking the end of the first backoff period as the time node, the first smart device can trigger subsequent wake-up response actions, which both follows the preset waiting logic and avoids response delays caused by infinite waiting.

[0180] In this embodiment, when a second wake-up decision score greater than its own first wake-up decision score is obtained, the first intelligent device can immediately and proactively exit the wake-up process. This avoids response conflicts between suboptimal and highly matched intelligent devices (such as simultaneous responses from multiple devices), and also terminates redundant calculations in a timely manner to save computing power and energy consumption. Furthermore, when a second wake-up decision score is not obtained, the first intelligent device can perform a wake-up response action after the backoff period ends. This provides a window for waiting for feedback from other second intelligent devices and avoids the wake-up non-response problem caused by the lack of a comparison object in a single intelligent device scenario. Based on this, through the complementary nature of the above solutions, accurate priority selection is achieved in a multi-intelligent device environment, while ensuring response stability in a single intelligent device environment, thereby improving decision reliability and user interaction experience.

[0181] Based on the above explanation, it should be added that within the first backoff period of the first smart device, it may only receive the lower second wake-up decision score from the second smart device. Based on this, the first smart device may mistakenly identify itself as the optimal smart device and prematurely enter the wake-up response phase (e.g., it has already started playing the response voice or activated the command receiving module). However, in actual scenarios, two types of delays may exist: First, there may be other second smart devices (with higher second wake-up decision scores), but due to network fluctuations increasing the time spent transmitting scores, or the device itself processing the wake-up signal slowly, the time it sends its second wake-up decision score to the first smart device is later than the time it receives the second wake-up decision score from the previously lower-scoring second smart device; second, the total time from the second smart device calculating the second wake-up decision score to sending it to the first smart device directly exceeds the first smart device's first backoff period, arriving late after the first smart device has already initiated the wake-up response action.

[0182] To address the two types of delays mentioned above, in order for the highly compatible smart device to respond, the first smart device can further execute the following step S507. Details are as follows:

[0183] S507. If a second wake-up decision score is obtained during the execution of the wake-up response action, the wake-up response action shall be terminated if the first wake-up decision score is less than the second wake-up decision score.

[0184] Understandably, when executing a wake-up response, if the first smart device receives a second wake-up decision score from a second smart device, and this second wake-up decision score is higher than its own first wake-up decision score, then the previous decision can be considered biased. Due to network latency, differences in device processing speed, etc., the second wake-up decision score of the second smart device, which better matches the user's expected wake-up, may not arrive within the first smart device's first backoff time, causing the first smart device to falsely initiate a response.

[0185] Therefore, in this embodiment, during the execution of the wake-up response action, if it is determined that the first wake-up decision score is less than the second wake-up decision score, the first smart device can immediately terminate the currently executing wake-up response action (such as interrupting the response voice playback or turning off the indicator light), and return from the woken-up state to the standby state, avoiding continued occupation of the user's interaction channel. Simultaneously, response space is reserved for the second smart device with the higher score, allowing it to normally initiate the wake-up response, ensuring that the device that ultimately responds to the user has a higher matching degree.

[0186] It should be added that, compared with related technologies, this embodiment has significant differences in the functional design after the first smart device is woken up. In related technologies, once the first smart device enters the woken-up state, it usually only focuses on performing wake-up response actions (such as voice response, activation command reception), no longer participates in collaborative decision-making among multiple smart devices, and also closes the wake-up information interaction channel with other devices.

[0187] However, in this embodiment, even when the first smart device is awakened, it retains the ability to make collaborative decisions with other devices. On the one hand, it can proactively send its own first wake-up event information or first wake-up decision score to provide decision-making reference for other second smart devices; on the other hand, it can still continuously receive second wake-up decision scores sent by other second smart devices and perform priority comparisons again based on the scores. This design breaks the limitation that wake-up terminates collaboration, allowing the awakened first smart device to still participate in subsequent decision corrections, further ensuring the matching degree between the final response device and the user's expectations.

[0188] In another embodiment, it should be further explained that if the first wake-up decision score and the second wake-up decision score are exactly the same, based on the calculation logic of the first backoff duration, the first backoff duration of the first smart device and the second backoff duration of the second smart device will usually not be the same. Therefore, they will not end their backoff simultaneously, but will enter the woken-up state sequentially after their respective backoff durations have ended. Based on this characteristic, even if the scores are the same, the problem of multiple smart devices responding simultaneously will not be triggered, thus effectively avoiding interaction conflicts (such as synchronous voice responses and simultaneous preemption of interaction channels).

[0189] In another embodiment, in order to further improve the response rate of the first smart device and enhance the user experience, the first smart device may also perform actions such as Figure 8 The steps S801-S803 shown below determine whether to execute the wake-up response action, and are detailed below:

[0190] S801. In standby mode, a voice signal containing a wake-up word is broadcast at preset intervals.

[0191] In one embodiment, the aforementioned preset period is a fixed time interval (e.g., 1 hour) preset by the first smart device, which is the time basis for triggering the broadcast voice signal action, and its value needs to be set according to the actual scenario requirements.

[0192] Broadcasting refers to the act of a first smart device actively sending audio signals to other second smart devices or the environment within a specific range (such as the coverage area of ​​the same local area network or the Bluetooth communication radius) through its own speaker or communication module. It is different from point-to-point transmission and has the characteristics of wide coverage and multiple receivers.

[0193] In addition, the voice signal containing the wake word is a specific signal that exists in audio form and has a built-in system preset wake word. The signal must meet the device's voice recognition standards (such as clear audio segments and conforming to the voiceprint characteristics of the wake word) to ensure that other devices or the device itself can accurately recognize the wake word information.

[0194] In the above steps, the first smart device can initiate the broadcast only when it is in standby mode. If the first smart device has entered a pre-wake-up state or a wake-up state, the broadcast will be paused. This allows the device to focus solely on user interaction during the wake-up process, as broadcasting containing a wake-up word could interfere with its own operation or that of other smart devices.

[0195] S802. If the second wake-up event information sent after the second smart device responds with a voice signal is not received, then after receiving the user's voice signal in the next preset period, the wake-up response action is executed.

[0196] S803. If the second wake-up event information sent by the second smart device after responding to the voice signal is received, then in the next preset period, the steps of obtaining the first wake-up event information after detecting the user voice signal containing the wake word and subsequent steps are executed.

[0197] Understandably, if the first smart device broadcasts a voice signal containing a wake-up word while in standby mode, and does not receive any second wake-up event information from the second smart device within a preset period (e.g., within 1-3 seconds after the broadcast), it can be assumed that there are no normally online second smart devices in the same area (e.g., only the first smart device is connected to the network, other smart devices are offline / faulty); or, the second smart device did not recognize the voice signal broadcast by the first device (e.g., the signal is blocked, the second device's microphone module is malfunctioning). Essentially, there are currently no other devices that can collaborate, so there is no need to execute a complex multi-device decision-making process.

[0198] Based on this, when entering the next preset cycle, if the first smart device detects a user voice signal containing a wake-up word, it can directly execute the wake-up response action without initiating collaborative schemes such as first backoff duration calculation and score comparison. This simplifies the process in scenarios without multi-device competition, avoids response delays after user wake-up due to redundant decisions, and improves interaction efficiency in single-device scenarios.

[0199] Furthermore, if the second wake-up event information sent by the second intelligent device is successfully received, it can be assumed that there is a second intelligent device in the same area that can cooperate normally. That is, the second intelligent device can recognize broadcast signals and is in a wake-up state, possessing the basic conditions for multi-device decision-making.

[0200] Therefore, when entering the next preset cycle, if the first smart device detects a user's voice signal containing a wake word, it needs to execute a complete multi-device collaborative decision-making process, rather than responding directly. For example, it executes steps S501-S504 as described above.

[0201] In this embodiment, in standby mode, a voice signal containing a wake-up word is broadcast every preset period. This allows for early detection of the presence of a second intelligent device in the same area that can collaborate, providing an environmental basis for subsequent decisions and avoiding blind execution of processes. Secondly, regarding the detection results, if the first intelligent device does not receive a response from the second intelligent device, it indicates that there is no multi-intelligent device competition scenario. Based on this, the first intelligent device can directly execute the wake-up response action after receiving the user's voice signal in the next preset period, eliminating redundant collaboration steps and significantly improving the interaction efficiency in single-intelligent-device scenarios. Furthermore, if a response is received from the second device, it can be considered that there is a need for multi-intelligent-device collaboration in the area. At this time, in the next preset period, the first intelligent device can execute the steps of obtaining the first wake-up event information after detecting the user's voice signal containing the wake-up word, as well as subsequent steps, ensuring that there are no conflicts caused by synchronous responses between multiple intelligent devices. Based on this, the above method not only ensures the response speed in single-intelligent-device scenarios but also takes into account the decision-making accuracy in multi-intelligent-device wake-up scenarios, comprehensively optimizing the user wake-up interaction experience.

[0202] To more clearly illustrate the solutions in this application, specific embodiments are used below to explain the solutions. See details below. Figure 9 , Figure 9 This is a flowchart illustrating a distributed wake-up decision-making method provided in another embodiment of this application.

[0203] Take, for example, a family living room environment with two smart devices (smart speaker and smart TV).

[0204] First, the user stands in the living room and emits a voice signal containing the wake word "Hello, Xiao X". Each of the first smart devices (smart speakers and smart TVs need to execute the above method separately; therefore, both smart speakers and smart TVs can be considered "first smart devices") can obtain their respective first wake-up event information and enter a pre-wake state after detecting the user's voice signal containing the wake word. This information includes the wake-up angle and distance between the first wake-up event information and the user, the signal-to-noise ratio of the user's voice signal, and the wake-up confidence level.

[0205] Specifically, the information for each first wake-up event can be as follows:

[0206] Smart speaker: wake-up angle 30°, distance 1.5m, signal-to-noise ratio 30dB, wake-up confidence score 920 (out of 1000); Smart TV: wake-up angle 60°, distance 3m, signal-to-noise ratio 25dB, confidence score 850.

[0207] The scores for each wake-up evaluation parameter in the smart speaker can be as follows: Wake-up angle score: A = 1 - 30 / 90 = ≈0.667; Distance score: D = 1 - 1.5 / 30 = 1 - 0.05 = 0.95; Signal-to-noise ratio score: S = 30 / 30 = 1; Wake-up confidence score: C = 920 / 1000 = 0.92.

[0208] The scores for the wake-up evaluation parameters in a smart TV can be as follows: Wake-up angle score: A = 1 - 60 / 90 = ≈0.333; Distance score: D = 1 - 3 / 30 = 0.9; Signal-to-noise ratio score: S = 25 / 30 ≈ 0.833; Wake-up confidence score: C = 850 / 1000 = 0.85.

[0209] Then, the first intelligent device can perform a weighted summation of the scores corresponding to various wake-up evaluation index parameters to obtain the aforementioned first wake-up decision score. For example, the weights corresponding to the wake-up angle, distance, signal-to-noise ratio, and wake-up confidence are all set at 0.2.

[0210] The first wake-up decision score for the smart speaker is: 0.667×0.2+0.95×0.3+1×0.3+0.92×0.2=0.902;

[0211] The first wake-up decision score for the smart TV is: 0.333×0.2+0.9×0.3+0.833×0.3+0.85×0.2=0.757.

[0212] Subsequently, each first intelligent device can determine the decision score backoff duration by multiplying its corresponding first wake-up decision score by a preset minimum waiting time, and randomly generate a duration factor for the first intelligent device. The product of this duration factor and a preset duration is then determined as the random backoff duration. Finally, the sum of the decision score backoff duration, the random backoff duration, and the preset backoff duration corresponding to the first intelligent device is determined as the first backoff duration.

[0213] For example, taking a smart speaker with a preset minimum waiting time of 5ms, a preset duration of 20ms, a random factor of 0.3, and a preset backoff time of 0ms as an example, its corresponding first backoff time can be: 0.902×5ms+0.3×20ms+0ms≈10.51ms.

[0214] Furthermore, taking a smart TV with a preset minimum waiting time of 5ms, a preset duration of 20ms, a random factor of 0.6, and a preset backoff time of 10ms as an example, its corresponding first backoff time can be: 0.757×5ms+0.6×20+10ms≈25.785ms.

[0215] At the same time, each of the first intelligent devices can broadcast its own first wake-up event information or first wake-up decision score. Taking the broadcasting of the first wake-up decision score as an example...

[0216] Within the first backoff period of the first smart TV (smart speaker), it can be determined whether the second wake-up decision score of the second smart device (smart TV) has been received.

[0217] Taking a smart speaker as an example of implementing the above method, the smart TV then becomes the second smart device.

[0218] If the second wake-up decision score of the second smart device is received within the first backoff time, it can be determined whether the first wake-up decision score is greater than the second wake-up decision score.

[0219] If the first wake-up decision score is greater than the second wake-up decision score, the first smart device can immediately execute a wake-up response action; if the first wake-up decision score is less than the second wake-up decision score, the first smart device can exit the wake-up process, for example, by entering standby mode.

[0220] Furthermore, if the second wake-up decision score of the second smart device is not received within the first backoff period, the first smart device may also perform a wake-up response action after the first backoff period ends.

[0221] Subsequently, during the wake-up response process, it can also broadcast its own first wake-up decision score every preset period to instruct other second smart devices (smartphones) to execute the above method to determine whether to exit the wake-up process when they receive the "first wake-up decision score".

[0222] Furthermore, during the execution of the wake-up response action, if the second wake-up decision score of another second smart device (smartphone) is obtained, it is also determined whether the first wake-up decision score is greater than the second wake-up decision score. If it is less than the second wake-up decision score, the wake-up response action is terminated (e.g., entering standby mode). Otherwise, if it is greater than the second wake-up decision score, the wake-up response action is still executed.

[0223] Figure 10 This is a schematic diagram of the structure of an electronic device provided in one embodiment of this application. Figure 10As shown, the electronic device 1000 of this embodiment includes: a processor 1010, a memory 1020, and a computer program 1030 stored in the memory 1020 and executable on the processor 1010, such as a program for a distributed wake-up decision method. When the processor 1010 executes the computer program 1030, it implements the steps of each embodiment of the distributed wake-up decision method described above, for example... Figure 5 S501 to S504 are shown.

[0224] For example, the computer program 1030 can be divided into one or more modules, one or more of which are stored in the memory 1020 and executed by the processor 1010 to implement the distributed wake-up decision-making method provided in this application embodiment. One or more modules can be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program 1030 in the electronic device 1000. For example, the computer program 1030 can implement the distributed wake-up decision-making method provided in this application embodiment.

[0225] Electronic device 1000 may include, but is not limited to, processor 1010 and memory 1020. Those skilled in the art will understand that... Figure 10 This is merely an example of electronic device 1000 and does not constitute a limitation on electronic device 1000. It may include more or fewer components than shown, or combine certain components, or different components. For example, electronic device may also include input / output devices, network access devices, buses, etc.

[0226] The processor 1010 may be a central processing unit, or it may be other general-purpose processors, digital signal processors, application-specific integrated circuits, off-the-shelf programmable gate arrays or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0227] The memory 1020 can be an internal storage unit of the electronic device 1000, such as a hard disk or memory of the electronic device 1000. The memory 1020 can also be an external storage device of the electronic device 1000, such as a plug-in hard disk, smart memory card, flash memory card, etc. equipped on the electronic device 1000. Furthermore, the memory 1020 can include both internal storage units and external storage devices of the electronic device 1000.

[0228] This application provides a computer-readable storage medium, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the distributed wake-up decision method as described in the above embodiments.

[0229] This application provides a computer program product that, when run on an electronic device, causes the electronic device to execute the distributed wake-up decision method described in the above embodiments.

[0230] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A distributed wake-up decision-making method, characterized in that, Applied to the first intelligent device, the method includes: After detecting a user voice signal containing a wake word, first wake-up event information is acquired; the first wake-up event information includes wake-up evaluation index parameters of multiple dimensions; The first wake-up decision score of the first smart device is calculated based on various wake-up evaluation index parameters; the first wake-up decision score is used to quantify the degree of matching of the first smart device with the user's expected wake-up device. The first backoff duration of the first smart device is calculated based on the first wake-up decision score; If a second wake-up decision score is obtained from the second smart device within the first backoff time, a wake-up response action is executed when the first wake-up decision score is greater than the second wake-up decision score. The calculation of the first backoff duration of the first smart device based on the first wake-up decision score includes: The product of the first wake-up decision score and the preset minimum waiting time is determined as the decision score backoff time; A duration factor for the first smart device is randomly generated, and the product of the duration factor and a preset duration is determined as the random backoff duration; The first backoff time is determined by the sum of the decision score backoff time, the random backoff time, and the preset backoff time corresponding to the first smart device.

2. The distributed wake-up decision method according to claim 1, characterized in that, The various wake-up evaluation metrics include the wake-up angle and distance between the user and the device, the signal-to-noise ratio of the user's voice signal, and the wake-up confidence level; the calculation of the first wake-up decision score of the first smart device based on the various wake-up evaluation metrics includes: The wake-up angle, distance, signal-to-noise ratio, and wake-up confidence are weighted and summed with their corresponding weights to obtain the first wake-up decision score.

3. The distributed wake-up decision method according to claim 2, characterized in that, The weight corresponding to the distance is less than or equal to the weights corresponding to the other wake-up evaluation index parameters.

4. The distributed wake-up decision method according to claim 1, characterized in that, After calculating the first backoff duration of the first smart device based on the first wake-up decision score, the method further includes: If the second wake-up decision score is obtained within the first backoff time, the wake-up process is exited when the first wake-up decision score is less than the second wake-up decision score. If the second wake-up decision score is not obtained within the first backoff period, the wake-up response action is executed at the end of the first backoff period.

5. The distributed wake-up decision method according to any one of claims 1-4, characterized in that, The distributed wake-up decision method further includes: If the second wake-up decision score is obtained during the execution of the wake-up response action, the wake-up response action is terminated when the first wake-up decision score is less than the second wake-up decision score.

6. The distributed wake-up decision method according to any one of claims 1-4, characterized in that, The distributed wake-up decision method further includes: In standby mode, a voice signal containing the wake-up word is broadcast at preset intervals; If the second wake-up event information sent by the second smart device after responding to the voice signal is not received, the wake-up response action will be executed after the user's voice signal is received in the next preset period. If the second wake-up event information sent by the second smart device after responding to the voice signal is received, then in the next preset period, the steps of obtaining the first wake-up event information after detecting the user voice signal containing the wake word and subsequent steps are executed.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method as described in any one of claims 1 to 6.

8. A computer program product, characterized in that, When the computer program product is run on an electronic device, the electronic device executes the method as described in any one of claims 1 to 6.

Citation Information

Patent Citations

  • Intelligent optimization method and device for wake-up decision, electronic equipment and storage medium

    CN113608792A

  • Device wakeup method, electronic device and storage medium

    CN116453516A