End-to-end multi-device false wakeup prevention method based on UDP local area network communication mechanism

Abstract

The invention discloses an end-to-end multi-device false wake-up prevention method based on a UDP local area network communication mechanism, and the method comprises the steps: judging the distance based on the size difference of wake-up word sound energy received by each STA, transmitting a calculated sound signal size value to an AP end in a UDP broadcast mode, and enabling the AP end to transmit the received UDP information to all STAs in a receiving-transmitting mode, the invention relates to the technical field of local area network communication. According to the end-to-end multi-device false wake-up prevention method based on the UDP local area network communication mechanism, an existing voice wake-up recognition framework does not need to be damaged, too complex algorithm design is not added, and the problem that similar devices are mistakenly waken up in a certain area is solved under the condition that hardware design is not changed; the method has good transportability when solving the problems, the method can be completed on a device capable of carrying out network communication, and no requirement for the computing power of a network access point exists in a public network, a local area network or other communication modes.

Description

Technical Field

[0001] The present invention relates to the technical field of local area network communication, and in particular to an end-to-end multi-device false awakening prevention method based on a UDP local area network communication mechanism. Background Art

[0002] When multiple smart devices with voice recognition capabilities (especially those of the same model) are in the same place, they can easily trigger false wakeups. For example, when a worker is working with multiple teammates on a mission, the wakeup word may accidentally wake up a teammate's device. To address similar issues, the following are currently mainstream solutions that may solve voice recognition-related problems:

[0003] ① Customized wake-up words and voiceprint recognition: Users can customize differentiated wake-up words and bind their identities through voiceprint recognition technology to reduce the probability of accidental touches of common wake-up words.

[0004] ② Set confidence level: Current mainstream near-field voice wake-up systems use acoustic models to calculate the confidence level of the wake-up word and set a fixed threshold for preliminary judgment. If the confidence level exceeds the threshold, a secondary verification is performed based on contextual audio features. For example, the system checks whether the wake-up word is delivered within a normal speaking speed range; if it exceeds the time limit, it is considered a false trigger.

[0005] ③ Hybrid local and cloud verification architecture: A lightweight wake-up model is deployed locally to handle high-confidence commands, while low-confidence scenarios are uploaded to the cloud for secondary verification of the complex model. For example, Amazon Alexa uses a hybrid architecture of local first-level wake-up and cloud-based second-level intent recognition.

[0006] The closest prior art to our invention is a hybrid local and cloud-based authentication architecture, but the only similarity lies in its use of a combination of local and network authentication to confirm the wake-up state. This solution addresses the confidence issue of wake-up word recognition, improving wake-up accuracy through secondary cloud-based verification, but does not address the issue of multiple devices being mistakenly woken up. Currently, no better prior art addresses this issue for industrial voice communication devices.

[0007] Custom wake-up words and voiceprint recognition require users to record the custom wake-up words 3-5 times to train the voiceprint model. The system will then only respond to wake-up requests that match the voiceprint. In such industries, communication equipment is often exchanged between multiple members and front-line workers change frequently. Therefore, using voiceprint recognition is not suitable for these operating scenarios.

[0008] Setting up a secondary confidence check can improve recognition capabilities and effectively reduce false triggering of a single device, but it cannot solve the problem of false wakeups when multiple devices are present. The characteristics used for wake-up word recognition during sound transmission do not vary much, so differentiating between multiple devices will not solve the existing problem.

[0009] The hybrid local and cloud-based verification architecture places high demands on network signals and platform computing power. Its workflow involves processing high-confidence instructions based on local deployment, uploading low-confidence information to the platform, and leveraging the platform's higher computing power for high-complexity verification to ensure correct processing of low-confidence information. This solution cannot resolve the issue of false wakeups when multiple devices are present. When multiple devices are present, even if they receive high-confidence information, wakeups will occur. Industrial communication equipment is often used in areas without public network coverage, where network bandwidth and channels are limited and computing power cannot be provided. Therefore, this scenario cannot resolve the existing false wakeup issue when devices are present.

[0010] In summary, the problem of multiple devices waking up incorrectly mentioned above has not yet been effectively solved. Summary of the Invention

[0011] (1) Technical problems solved

[0012] In response to the deficiencies in the prior art, the present invention provides an end-to-end multi-device anti-false wake-up method based on the UDP local area network communication mechanism. When multiple similar voice interaction products are within a certain spatial range, there will be a "one call, a hundred responses" problem in which multiple devices respond simultaneously after the user shouts the wake-up word. Especially in the field of industrial communication equipment where multiple products of the same type are present, users often only want to wake up the one they are wearing, but end up waking up a large number of them, resulting in confusion in the subsequent voice interaction command recognition and affecting the user experience. The present invention solves this common problem technically in a simple and efficient manner through the wake-up process design, thereby achieving the purpose of waking up only one's own device without waking up other devices, and further ensuring that users can complete instrument operations through voice interaction, thereby freeing their hands to continue working.

[0013] (2) Technical solution

[0014] To achieve the above objectives, the present invention is implemented through the following technical solutions: an end-to-end multi-device anti-false awakening method based on the UDP local area network communication mechanism, specifically comprising the following steps:

[0015] S1. The microphone picks up the voice signal and performs analog-to-digital conversion for subsequent processing. The digital voice signal is used as an input signal to enter the voice recognition wake-up word recognition program module for processing. After completing the traditional wake-up word recognition, it enters the pre-wake-up state;

[0016] S2. While confirming the wake-up word match, calculate the energy of the wake-up word. The calculation of the energy here is related to the bit depth of data transmission. The sound energy is:

[0017]

[0018] Among them, BD is the bit depth, si is the amplitude of the speech signal, and the size of a single si is different due to the waveform of different STA sampling starting points during ADC. Therefore, in order to reflect the energy size of a piece of speech, it is necessary to use the method of calculating the root mean square value of multiple points, that is, the expression The value n is the number of sampling points in the voice segment. The final result is the voice signal energy received by different STAs. Its value is a negative number in dB, and the voice signal energy calculation is now complete.

[0019] S3. This value needs to be sent to the AP via the STA's radio using the UDP protocol. To ensure real-time speech recognition, a time stamp is required to limit the UDP sending and receiving process. When multiple STAs are in pre-wake-up mode, the AP will receive data from multiple STAs. Since the data is sent and received immediately, the AP does not need to perform additional signal processing.

[0020] S4. Each STA will receive voice energy signals from all STAs in the pre-wake-up state within a certain period of time. Each received signal will be written into a defined list. After a certain period of time, a list containing multiple data will be obtained. The maximum value among them will be found and compared with the voice energy value of the STA.

[0021] S5. If the values are equal, the STA enters the wake-up mode and broadcasts a successful wake-up response word such as "Received" or "I'm here". The wake-up process ends and the STA enters the voice recognition stage.

[0022] S6. If the values are not equal, it means that the wake-up word voice energy value received by this STA is not the maximum, the wake-up fails, and the process of this STA ends.

[0023] Preferably, the bit depth in step S2 refers to the number of binary bits used by each audio sampling point when performing analog-to-digital conversion, which determines the amplitude accuracy and dynamic range of the sound signal. Typical bit depths are 8, 16, and 32 bits.

[0024] Preferably, in step S4, the communication time during which each STA receives the voice energy level signals of all STAs entering the pre-awakening state within a certain period of time is 100 ms.

[0025] Preferably, when all n STAs enter the pre-wake-up state and the radio frequency signal is sent successfully, the n STAs will then receive the UDP broadcast signal sent by the AP.

[0026] Preferably, each STA enters the wake-up status judgment process after the UDP sending and receiving delay time. n STAs receive n audio data. After comparing one by one, STA_m is equal to the maximum value, the wake-up is successful, and the voice recognition mode is entered. The remaining STAs are less than the maximum value in the data, and the wake-up failure process ends.

[0027] Preferably, all volume data are sent to all STAs via UDP signals, so that the judgment process is completed in each STA. Now a new solution is proposed to complete the judgment process in the AP, and finally the AP only sends a wake-up success start instruction to the only STA.

[0028] Preferably, the volume value of each STA is obtained and the volume signal and device IP are sent to the AP end through the radio frequency network. The AP has differences in processing the received information. The AP end will set the time of receiving the first signal as 0ms and compare the sound energy value of the UDP broadcast information received within a certain period of time, and select the maximum value and its corresponding STA number.

[0029] Preferably, the AP only sends a point-to-point wake-up message to the corresponding STA. When STAx receives the message, it indicates that the wake-up is successful and enters the voice recognition mode. If the STA does not receive the AP's wake-up message within a certain period of time, it automatically determines that the wake-up has failed and the process ends.

[0030] Nouns involved in network data transmission:

[0031] AP: An access point is a hardware device that connects wireless devices to a network via the Wi-Fi protocol. It can be a standalone device (like a traditional wireless router) or integrated into a gateway, serving as the core hub of a wireless local area network (WLAN).

[0032] STA: Full name for Station, refers to any wireless terminal device that supports the IEEE 802.11 protocol, such as smartphones and smart communication devices. It is essentially a Wi-Fi client that accesses the network through an AP.

[0033] UDP: User Datagram Protocol, one of the core protocols of the transport layer. It is lightweight and efficient and is widely used in scenarios with high real-time requirements.

[0034] (3) Beneficial effects

[0035] The present invention provides an end-to-end multi-device anti-false wakeup method based on the UDP local area network communication mechanism. Compared with the existing technology, it has the following advantages:

[0036] (1) This end-to-end multi-device anti-false wake-up method based on the UDP local area network communication mechanism does not need to destroy the existing voice wake-up recognition framework, does not add overly complex algorithm design, and solves the problem of similar devices being falsely awakened in a certain area without changing the hardware design.

[0037] (2) The end-to-end multi-device anti-false wake-up method based on the UDP local area network communication mechanism has good portability in solving such problems. It can be completed on any device that can communicate over the network, regardless of whether it is a public network, a local area network or other communication mode. There is no requirement for the computing power of the network access point, as long as it can normally send and receive UDP broadcasts.

[0038] (3) The end-to-end multi-device anti-false wake-up method based on the UDP local area network communication mechanism has higher stability and certainty. There is no need to increase the wake-up threshold in the original voice wake-up algorithm. Compared with various methods of changing the wake-up threshold, it has higher sensitivity and stability. In addition, because this algorithm uses UDP broadcast and sorting the volume of multiple devices, its accuracy reaches 0.01dB (this accuracy is sufficient to distinguish all devices), so its wake-up state is unique. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 This is a schematic diagram of communication between AP and STA of the present invention;

[0040] Figure 2 This is a schematic diagram of signal transmission according to the present invention;

[0041] Figure 3 is a flow chart of the system of the present invention;

[0042] Figure 4 Schematic diagram of the wake-up word signal waveform received by STA1 and the speech portion recognized by VAD in the present invention;

[0043] Figure 5 Schematic diagram of the wake-up word signal waveform received by STA2 and the speech portion recognized by VAD in the present invention;

[0044] Figure 6 Schematic diagram of STA1 and STA2 sending their own volume values to AP in the present invention;

[0045] Figure 7 Schematic diagram of UDP signals received by STA1 and STA2 from the AP in the present invention;

[0046] Figure 8 Schematic diagram of the awakening state judgment of STA1 and STA2 of the present invention;

[0047] Figure 9This is a schematic diagram of the principle of Example 3 of the present invention;

[0048] Figure 10 Schematic diagram of the wake-up instruction judgment process of AP in embodiment 3 of the present invention. DETAILED DESCRIPTION

[0049] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0050] The basic principle of this invention is to determine distance based on the difference in the energy level of the wake-up word received by each STA. The calculated sound signal strength value is sent to the access point via UDP broadcast. The AP then sends the received UDP information to all STAs using a send-and-receive method. Within a limited time, each STA compares its own value with the maximum value in the received data set. If the two are equal, the STA wakes up; if they are not equal, the STA does not wake up and the process ends. The technical solution is described in detail below.

[0051] See also Figure 1-10 The present invention provides three technical solutions: an end-to-end multi-device anti-false wakeup method based on the UDP local area network communication mechanism, specifically including the following embodiments:

[0052] Example 1: An end-to-end multi-device anti-false wakeup method based on a UDP local area network communication mechanism, specifically comprising the following steps:

[0053] S1. The microphone picks up the voice signal and performs analog-to-digital conversion for subsequent processing. The digital voice signal is used as an input signal to enter the voice recognition wake-up word recognition program module for processing. If the wake-up word matches, the system wakes up, otherwise it ends (traditional voice wake-up word recognition is a very common technology. This part is not protected by this technical solution. This technical solution has no specificity for the traditional voice recognition in the front part, and all voice recognition solutions can be transplanted into the framework of this technical solution). However, in order to solve the above-mentioned problem of false wake-up when multiple devices exist, this technical solution needs to add a subsequent process after the traditional voice recognition is successful, and enter the pre-wake-up state after completing the traditional wake-up word recognition (traditional recognition is completed but the response task is suspended);

[0054] S2. While confirming the wake-up word match, calculate the energy of the wake-up word. The calculation of the energy here is related to the bit depth of data transmission. The sound energy is:

[0055]

[0056] Among them, BD is the bit depth, si is the amplitude of the speech signal, and the size of a single si is different due to the waveform of different STA sampling starting points during ADC. Therefore, in order to reflect the energy size of a piece of speech, it is necessary to use the method of calculating the root mean square value of multiple points, that is, the expression The value n is the number of sampling points in the speech segment. The final result is the energy of the speech signal received by different STAs. Its value is a negative number in dB (such as -23.05dB). This completes the speech signal energy calculation.

[0057] S3. This value needs to be sent to the AP via the STA's radio using the UDP protocol. To ensure real-time speech recognition, a time stamp is required to limit the UDP transmission and reception time (e.g., 200ms). When multiple STAs are in pre-wake-up mode, the AP will receive data from multiple STAs. Since the data is transmitted and received immediately, the AP does not need to perform additional signal processing.

[0058] S4. Each STA will receive the voice energy signals of all STAs in the pre-wake-up state within a certain period of time. Each received signal will be written into a defined list. Finally, after a certain period of time (such as 200ms), a list containing multiple data will be obtained. The maximum value among them will be found and compared with the voice energy value of the STA.

[0059] S5. If the values are equal, the STA enters the wake-up mode and broadcasts a successful wake-up response word such as "Received" or "I'm here". The wake-up process ends and the STA enters the voice recognition stage.

[0060] S6. If the values are not equal, it means that the wake-up word voice energy value received by this STA is not the maximum, the wake-up fails, and the process of this STA ends.

[0061] In the embodiment of the present invention, the bit depth in step S2 refers to the number of binary bits used at each audio sampling point during analog-to-digital conversion, which determines the amplitude accuracy and dynamic range of the sound signal. Typical bit depths are 8, 16, and 32 bits.

[0062] In the embodiment of the present invention, in step S4, the communication time during which each STA receives the voice energy level signals of all STAs entering the pre-awakening state within a certain period of time is 100 ms.

[0063] Example 2: This example uses an actual case to illustrate. To ensure simplicity, this example uses one AP and two STAs (the principle and process of multiple STAs are exactly the same). The voice signals picked up by the two STAs through their microphones are marked as STA1 and STA2. The two audio signals are about 1.5 seconds long. The waveform and the audio voice part detected by VAD are as follows: Figure 4 and Figure 5 As shown in the figure, the red shaded area is the detected human voice signal. VAD stands for Voice Activity Detection. The VAD process is generally included in the audio front-end algorithm before traditional voice wake-up word recognition. It is not the focus of this patent and will not be detailed here.

[0064] Energy calculation: The speech segment signal is input to the wake-up word recognition module (no special features, general speech recognition modules can be used) and matched with the wake-up word feature. If it does not match, the task ends. If it matches the wake-up word feature, the device enters the pre-wake-up state and requires further confirmation in the subsequent process. Calculate the energy value of the speech segment. In this example, the audio is 16-bit deep, so the calculation method is Si is the amplitude of the i-th data in the speech segment, and n is the number of data in the speech segment. The calculation results are shown in Table 1 below:

[0065] Table 1 Calculated values of the wake-up word signal volume received by STA1 and STA2

[0066]

[0067] There is a significant difference in the voice volume of the two STAs, which are -13.52dB and -19.50dB respectively.

[0068] UDP broadcast transmission and reception: UDP (User Datagram Protocol) is one of the core protocols of the transport layer. It is lightweight and efficient and is widely used in scenarios with high real-time requirements. After entering the pre-wake-up state, each STA sends its calculated voice volume value to the AP through the RF module using the UDP protocol. Figure 6 The process of sending the UDP broadcast information to STA1 and STA2.

[0069] It can be seen that both STAs have entered the pre-wake-up state and the RF signal is sent successfully. In the next period of time, the two STAs will receive the UDP broadcast signal sent by the AP. Figure 7 In the process of receiving UDP broadcast signals, since there are only two STAs here, each STA receives two messages, and the delay time is basically within 100ms.

[0070] Wake-up status judgment: Each STA enters the wake-up status judgment process after the UDP sending and receiving delay time (this time is used for sending and receiving radio frequency signals between the STA and the AP, which is set to 200ms here and starts from the time when the STA sends the radio frequency signal). Figure 8 This is the wake-up status judgment process for STA1 and STA2. Both STAs receive two audio data. After comparing them one by one, STA1 is equal to the maximum value, and wake-up is successful, entering the voice recognition mode. STA2 is less than the maximum value in the data, and the wake-up process ends.

[0071] This concludes the entire process. In this example, we leveraged the UDP local area network communication mechanism to prevent two STAs from being woken up simultaneously, resolving the issue of multiple smart devices being woken up by voice. The algorithm remains unchanged when more devices are connected to the AP; however, each STA will receive more volume information. Since the algorithm here only compares a small number of numbers, the computational speed is negligible. Therefore, this solution offers excellent real-time performance, completing the entire process within 300ms, with virtually no noticeable delay during use.

[0072] Example 3: The previous solution sends all volume data to all STAs via UDP signals, allowing the judgment process to be completed in each STA. Now a new solution is proposed to complete the judgment process in the AP. Finally, the AP only sends a wake-up success start instruction to the only STA. The flowchart of the key changes is as follows Figure 9 As shown, the rest is consistent with the above scheme.

[0073] The volume value of each STA is obtained in the same way as the previous solution and the volume signal and device IP are sent to the AP through the RF network. All the steps before this are the same between the two solutions. In the new solution, the AP processes the received information differently. The AP sets the time of receiving the first signal as 0ms and compares the sound energy values of the UDP broadcast information received within a certain period of time (set to 100ms), and selects the maximum value and its corresponding STA number. After that, the AP only sends the corresponding STA (i.e. Figure 9 The AP sends a point-to-point wake-up message to the STA (STAx in the figure). When the STAx receives the message, it is considered that the wake-up is successful and enters the voice recognition mode. If the STA does not receive the wake-up message from the AP within a certain period of time (set to 260ms), the STA automatically determines that the wake-up has failed and the process ends.

[0074] Currently, various industrial sectors are transitioning towards intelligent upgrades, and each industry is building its own IoT system. However, due to industry-specific limitations, IoT systems and solutions in many industries often lag behind those in the consumer market. Consequently, in recent years, some intelligent solutions from the consumer electronics market have been introduced to the industrial sector. However, differences in usage scenarios and industry characteristics have resulted in some solutions failing to adapt to localized environments. This present invention addresses the issue of near-field voice recognition. In the consumer electronics sector, similar issues can be addressed through voiceprint recognition. However, in the industrial sector, devices are often used by multiple personnel, and a single device may be used by multiple people throughout the day. Using voiceprint recognition is cumbersome, requiring re-entry and recognition of voiceprints for each new user. This significantly increases the user barrier to entry, and forgetting to enter voiceprint recognition results in the loss of voice recognition functionality, a significant user pain point. Consumer electronics devices are typically tied to their users on a one-to-one basis. For example, mobile phones only require a single voiceprint recognition for normal use. The solution of the present invention is particularly applicable to the field of industrial communications, especially similar products such as walkie-talkies with voice recognition capabilities, but it is not limited to this field or a fixed product. As long as there is a need for differentiated recognition requirements for voice wake-up on multiple devices and at least local area network communication capabilities are available, this solution can be used.

[0075] Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0076] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0077] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An end-to-end multi-device anti-false wakeup method based on UDP local area network communication mechanism, characterized by: The specific steps include: S1. The microphone picks up the voice signal and performs analog-to-digital conversion for subsequent processing. The digital voice signal is used as an input signal to enter the voice recognition wake-up word recognition program module for processing. After completing the traditional wake-up word recognition, it enters the pre-wake-up state; S2. While confirming the wake-up word match, calculate the energy of the wake-up word. The calculation of the energy here is related to the bit depth of data transmission. The sound energy is: Among them, BD is the bit depth, si is the amplitude of the speech signal, and the size of a single si is different due to the waveform of different STA sampling starting points during ADC. Therefore, in order to reflect the energy size of a piece of speech, it is necessary to use the method of calculating the root mean square value of multiple points, that is, the expression The value n is the number of sampling points in the voice segment. The final result is the voice signal energy received by different STAs. Its value is a negative number in dB, and the voice signal energy calculation is now complete. S3. This value needs to be sent to the AP via the STA's radio using the UDP protocol. To ensure real-time speech recognition, a time stamp is required to limit the UDP sending and receiving process. When multiple STAs are in pre-wake-up mode, the AP will receive data from multiple STAs. Since the data is sent and received immediately, the AP does not need to perform additional signal processing. S4. Each STA will receive voice energy signals from all STAs in the pre-wake-up state within a certain period of time. Each received signal will be written into a defined list. After a certain period of time, a list containing multiple data will be obtained. The maximum value among them will be found and compared with the voice energy value of the STA. S5. If the values are equal, the STA enters the wake-up mode and broadcasts a successful wake-up response word such as "Received" or "I'm here". The wake-up process ends and the STA enters the voice recognition stage. S6. If the values are not equal, it means that the wake-up word voice energy value received by this STA is not the maximum, the wake-up fails, and the process of this STA ends.

2. The end-to-end multi-device anti-false wakeup method based on the UDP local area network communication mechanism according to claim 1 is characterized in that: The bit depth in step S2 refers to the number of binary bits used at each audio sampling point when performing analog-to-digital conversion, which determines the amplitude accuracy and dynamic range of the sound signal. Typical bit depths are 8, 16, and 32 bits.

3. The end-to-end multi-device anti-false wakeup method based on the UDP local area network communication mechanism according to claim 1, characterized in that: In step S4, each STA will receive the voice energy level signals of all STAs entering the pre-awakening state within a certain period of time, and the communication time is 100ms.

4. The end-to-end multi-device anti-false wakeup method based on the UDP local area network communication mechanism according to claim 1, characterized in that: When both STAs enter the pre-wake-up state and the RF signal is sent successfully, the two STAs will then receive the UDP broadcast signal sent by the AP.

5. The end-to-end multi-device anti-false wakeup method based on the UDP local area network communication mechanism according to claim 4, characterized in that: N STAs enter the pre-wake-up state. After the UDP send and receive delay, each STA receives n volume data. Each STA compares the volume with the maximum value in the list. If the STA's volume value is equal to the maximum value in the list, the wake-up is successful and enters speech recognition mode. If the STA's volume value is less than the maximum value in the list, the wake-up fails and the process ends. In this case, it is impossible for the STA to be greater than the data in the list because the UDP data in the list contains the volume values of all STAs in the pre-wake-up state.

6. The end-to-end multi-device anti-false awakening method based on the UDP local area network communication mechanism according to claim 1, characterized in that: All volume data is sent to all STAs via UDP signals, allowing the judgment process to be completed in each STA. Now a new solution is proposed to complete the judgment process in the AP. Finally, the AP only sends a wake-up success start instruction to the only STA.

7. The end-to-end multi-device anti-false awakening method based on the UDP local area network communication mechanism according to claim 1, characterized in that: The volume value of each STA is obtained and the volume signal and device IP are sent to the AP through the RF network. The AP may process the received information differently. The AP will set the time of receiving the first signal to 0ms and compare the sound energy values of UDP broadcast information received within a certain period of time. The maximum value and its corresponding STA number are selected.

8. The end-to-end multi-device anti-false awakening method based on the UDP local area network communication mechanism according to claim 7, characterized in that: The AP only sends a point-to-point wake-up message to the corresponding STA. When STAx receives the message, it indicates that the wake-up is successful and enters the voice recognition mode. If the STA does not receive the AP's wake-up message within a certain period of time, it automatically determines that the wake-up has failed and the process ends.

Images