Full-time wakeup-free method and device, equipment and medium

Abstract

The invention provides a full-time wakeup-free method and device, equipment and a medium. The method comprises the following steps: acquiring user voice; performing intention recognition on the user voice to obtain intention information, and performing voice recognition on the user voice to obtain voice recognition confidence; and determining whether to trigger full-time wakeup-free based on whether the intention information belongs to an intention effective range and a size relationship between the voice recognition confidence and a preset threshold. According to the full-time wakeup-free method and device, the equipment and the medium provided by the invention, the wakeup-free operation is triggered only under the condition that the intention is effective and the voice recognition result is credible by judging whether the intention information of the user voice is within the effective range and combining the voice recognition confidence, so that the accuracy of full-time wakeup-free is effectively improved, and the user experience is improved. The false triggering rate is remarkably reduced, so that the stability of voice interaction is improved, and smoother full-time wakeup-free experience is brought.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a full-time wake-up-free method, device, equipment and medium. Background Art

[0002] The full-time wake-up-free function allows smart devices to directly recognize and execute user commands without the need for a wake-up word, significantly improving the convenience of voice interaction. This function is widely used in smart homes, smart cockpits and other fields. By reducing the need for users to wake up repeatedly, it improves the natural and continuous human-computer interaction experience. However, the control of false trigger rate is still the main difficulty in the development of full-time wake-up-free technology.

[0003] Commonly used full-time wake-up-free technology mainly controls the false trigger rate through three solutions: multi-mode recognition, dynamic wake-up-free word list, and semantic result verification. Among them, multi-mode recognition technology combines non-voice devices such as cameras to assist in determining whether the user has issued a command. Although it effectively controls the false trigger rate, it increases the hardware cost; the dynamic wake-up-free word list solution switches the wake-up-free words in different scenarios, which has poor applicability, limits flexibility, and the false trigger rate cannot be effectively controlled; the semantic result verification solution performs semantic matching on the recognition content to confirm that it is consistent with the current scene, but it relies heavily on the semantic accuracy. Therefore, how to effectively reduce the false trigger rate of the full-time wake-up-free function is an urgent problem to be solved. Summary of the invention

[0004] The present invention provides a full-time wake-up-free method, device, equipment and medium, which are used to solve the defect of high false trigger rate in related technologies.

[0005] The present invention provides a full-time wake-up-free method, comprising: Get user voice; Performing intent recognition on the user's voice to obtain intent information, and performing speech recognition on the user's voice to obtain speech recognition confidence; Based on whether the intention information belongs to the valid range of the intention and the relationship between the voice recognition confidence and the preset threshold, it is determined whether to trigger full-time wake-up-free.

[0006] According to a full-time wake-up-free method provided by the present invention, the preset threshold comprises a first threshold; The determining whether to trigger full-time wake-up-free based on whether the intention information belongs to the effective range of the intention and the size relationship between the voice recognition confidence and the preset threshold includes: When the intention information belongs to the effective range of the intention and the voice recognition confidence is greater than the first threshold, determining to trigger full-time wake-up-free; When the intent information does not belong to the valid range of the intent and the voice recognition confidence is less than or equal to the first threshold, it is determined not to trigger full-time wake-up-free.

[0007] According to a full-time wake-up-free method provided by the present invention, the preset threshold also includes a second threshold, and the second threshold is greater than the first threshold; The determining whether to trigger full-time wake-up-free based on whether the intention information belongs to the effective range of the intention and the size relationship between the voice recognition confidence and the preset threshold value also includes: When the intent information does not fall within the valid range of the intent and the voice recognition confidence is greater than the first threshold, it is determined whether to trigger full-time wake-up-free based on the relationship between the voice recognition confidence and the second threshold.

[0008] According to a full-time wake-up-free method provided by the present invention, the preset threshold value further includes a third threshold value, and the third threshold value is smaller than the first threshold value; The determining whether to trigger full-time wake-up-free based on whether the intention information belongs to the effective range of the intention and the size relationship between the voice recognition confidence and the preset threshold value also includes: When the intent information belongs to the valid range of the intent and the voice recognition confidence is less than or equal to the first threshold, it is determined whether to trigger full-time wake-up-free based on the size relationship between the voice recognition confidence and the third threshold.

[0009] According to a full-time wake-up-free method provided by the present invention, the preset threshold is determined based on the speech recognition confidence of the positive sample and the speech recognition confidence of the reverse sample, the positive sample is the sample speech determined to trigger the full-time wake-up-free, and the reverse sample is the sample speech determined not to trigger the full-time wake-up-free.

[0010] A full-time wake-up-free method according to the present invention also includes: When it is determined that full-time wake-up-free is triggered, respond to the user voice.

[0011] The present invention also provides a full-time wake-up-free device, comprising the following modules: A voice acquisition unit, used to acquire user voice; A model processing unit, configured to perform intent recognition on the user's voice to obtain intent information, and perform speech recognition on the user's voice to obtain speech recognition confidence; The trigger judgment unit is used to determine whether to trigger full-time wake-up-free based on whether the intention information belongs to the effective range of the intention and the relationship between the voice recognition confidence and the preset threshold.

[0012] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the full-time wake-up-free method as described above is implemented.

[0013] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the full-time wake-up-free method as described above is implemented.

[0014] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements any of the above-mentioned full-time wake-up-free methods.

[0015] The full-time wake-up-free method, device, equipment and medium provided by the present invention determine whether the intention information of the user's voice is within a valid range, and combined with the voice recognition confidence, trigger the wake-up-free operation only when the intention is valid and the voice recognition result is credible, thereby effectively improving the accuracy of the full-time wake-up-free and significantly reducing the false trigger rate, thereby improving the stability of voice interaction and bringing a smoother full-time wake-up-free experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the present invention or related technologies, the following briefly introduces the drawings required for use in the embodiments or related technical descriptions. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 This is one of the flow charts of the full-time wake-up-free method provided by the present invention.

[0018] Figure 2 This is the second flow chart of the full-time wake-up-free method provided by the present invention.

[0019] Figure 3 It is a structural schematic diagram of the full-time wake-up-free device provided by the present invention.

[0020] Figure 4 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are 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 creative work are within the scope of protection of the present invention.

[0022] With the popularization of smart devices and voice interaction technology, the full-time wake-up-free function has become a key requirement for improving user experience in human-computer interaction. The full-time wake-up-free function allows smart devices to directly recognize user voice commands without using wake-up words, providing users with a more natural and smooth voice interaction experience. It is widely used in smart homes, car systems and other fields. However, the problem of false triggering of devices in the full-time wake-up-free state has always been a bottleneck in the development of technology. The high false trigger rate not only affects the user experience, but may also cause the device to perform incorrect operations.

[0023] The existing full-time wake-up-free technology mainly controls the false trigger rate through three schemes: multi-mode recognition, dynamic wake-up-free word list and semantic result verification. Multi-mode recognition technology combines cameras, sensors and other devices with voice signals to jointly judge user intentions, which significantly improves accuracy, but has high hardware requirements and increases equipment costs. The dynamic wake-up-free word list solution switches specific command words according to different scenarios to limit the system recognition range. It is suitable for certain specific scenarios, but its flexibility is limited, and it is easy to cause false triggers when the user's expression changes slightly. In addition, the reduction of the false trigger rate cannot be guaranteed. The semantic result verification solution adds a semantic analysis step after voice recognition to ensure that the instructions are consistent with the current operating context and further reduce false triggers, but it relies on the generalization ability and accuracy of the semantic model. Once the recognition deviates, the false trigger problem is still serious.

[0024] In view of the above problems, an embodiment of the present invention provides a full-time wake-up-free method. Figure 1 FIG. 1 is one of the flow charts of the full-time wake-up-free method provided by the present invention, such as Figure 1 As shown, the method includes: Step 110, obtaining user voice.

[0025] Specifically, the surrounding sound environment is continuously monitored in full-time standby mode through hardware facilities such as built-in microphones to capture possible user voices. In the process of obtaining user voices, noise reduction technology can be used to process the received voice signals, filter out environmental noise, and retain the main features of the user's voice, thereby improving the clarity and accuracy of the user's voice input. This process ensures that the user's true intentions can be effectively identified and reduces misjudgments caused by environmental interference. For example, when a user says "turn on navigation" in the car, the user's voice signal is extracted under the interference of background noise. After noise reduction processing, the "turn on navigation" instruction is accurately obtained, and the instruction is analyzed and executed in subsequent steps to achieve operations without wake-up words.

[0026] Step 120, performing intent recognition on the user voice to obtain intent information, and performing speech recognition on the user voice to obtain speech recognition confidence.

[0027] Specifically, the acquired user voice is subjected to intent recognition and voice recognition to ensure the accuracy and effectiveness of subsequent operations. First of all, intent recognition is a technology that extracts the user's real needs from the user's voice through natural language processing and deep learning technology. The core of intent recognition is to understand the operational intention behind the user's speech. It is not just to recognize specific words or phrases, but to connect these words with the context to determine the function the user wants to achieve. For example, when a user says "I want to listen to some light music" or "play some soothing songs", although the specific expressions are different, the intent recognition technology can recognize these instructions as the "play music" intention. Intent recognition technology generally uses a deep learning model that has been pre-trained with a large amount of voice data and instruction samples to identify the same intent in different forms of expression. The output of intent recognition is intent information, which contains the classification of the user's real needs and specific operation instructions. Intent information is not just a simple summary of the user's voice, but a deep analysis of the user's request, which can associate complex sentences with corresponding functions, thereby helping the device to respond quickly according to the user's intention. For example, in a car system, intent information may include specific functions such as vehicle control, navigation, and playing music.

[0028] In addition to the intention recognition of the user's voice, speech recognition is also performed, that is, converting the audio signal into text, etc., for subsequent analysis and execution. For the text converted from the user's voice, a speech recognition confidence is generated to measure the trustworthiness of the recognized text. The speech recognition confidence provides a probabilistic score that represents the reliability judgment of the recognized text result. This score is usually based on a variety of factors, including the quality of the audio signal, the matching degree of the language model, the decoding path during the recognition process, and background noise. In different environments and conditions, the speech recognition confidence will be different, which assists in determining whether to take further actions on the recognition results. Speech recognition confidence is particularly important in practical applications. For example, when the device recognizes the instruction "Navigate to the company", if the recognized text confidence is high, it means that the recognition accuracy is high, and the navigation operation can be directly executed at this time; if the confidence is low, it means that the device is not confident enough about the recognized text result, and may choose to re-ask the user or not execute the instruction immediately to reduce the risk of misoperation.

[0029] By combining intent information and voice recognition confidence, user voices can be screened more intelligently to ensure the accuracy of operations. Even in complex environments, the dual standards of intent information and voice recognition confidence can still be used to reasonably judge the effectiveness of user voices, thereby effectively reducing false triggers and improving the reliability and user experience of the full-time wake-up-free function. This dual judgment mechanism not only understands the real needs of users, but also enhances adaptability and accuracy in diverse usage scenarios.

[0030] Step 130, based on whether the intention information belongs to the effective range of the intention and the relationship between the voice recognition confidence and the preset threshold, determine whether to trigger full-time wake-up-free.

[0031] The effective scope of intent is a set of intents set to trigger specific operations, which includes all intent recognitions that are allowed to be directly executed. This scope is usually configured according to the device application scenario and user needs to ensure that only specific command types are responded to without wake-up to avoid unnecessary false triggering. For example, in a car system, the effective scope of intent may include intent categories related to navigation, vehicle control, and communication, such as "navigate to the destination", "adjust the air conditioning temperature", "make a call", etc. In the smart home scenario, the effective scope of intent may include commands closely related to home control, such as "turn on the lights", "adjust the temperature", and "play music". The setting of the effective scope of intent helps filter out the types of commands that should be responded to in specific scenarios, thereby reducing the possibility of false triggering and improving the accuracy of responding to user voice.

[0032] In addition to judging by the effective range of intent, the voice recognition confidence can also be combined to further judge whether to trigger full-time wake-up-free. Voice recognition confidence is a credibility score for the text result of user voice recognition, which is used to indicate the accuracy judgment of the recognition result. Compare the voice recognition confidence with the preset threshold. If the confidence is higher than the preset threshold, it means that the recognition result is more reliable and accurate, and meets the full-time wake-up-free triggering condition of the voice recognition confidence; if it is lower than the preset threshold, it means that the credibility of the recognition result is insufficient, and the device may choose not to trigger full-time wake-up-free to avoid causing misoperation.

[0033] By combining the dual judgment of the effective range of intent and the confidence of voice recognition, the accuracy and reliability of the full-time wake-up-free function can be significantly improved. On the one hand, the effective range of intent ensures that only valid commands are responded to, avoiding false triggering of invalid commands; on the other hand, the voice recognition confidence screening mechanism effectively reduces erroneous operations caused by inaccurate recognition. This judgment mechanism greatly reduces the possibility of false triggering, ensures that user commands can be accurately understood and executed in complex environments, and significantly improves the user's interactive experience.

[0034] The full-time wake-up-free method provided in the embodiment of the present invention determines whether the intention information of the user's voice is within a valid range, and combines the voice recognition confidence to trigger the wake-up-free operation only when the intention is valid and the voice recognition result is credible, thereby effectively improving the accuracy of the full-time wake-up-free method and significantly reducing the false trigger rate, thereby improving the stability of voice interaction and bringing a smoother full-time wake-up-free experience.

[0035] Based on the above embodiment, the preset threshold includes a first threshold; The determining whether to trigger full-time wake-up-free based on whether the intention information belongs to the effective range of the intention and the size relationship between the voice recognition confidence and the preset threshold includes: When the intention information belongs to the effective range of the intention and the voice recognition confidence is greater than the first threshold, determining to trigger full-time wake-up-free; When the intent information does not belong to the valid range of the intent and the voice recognition confidence is less than or equal to the first threshold, it is determined not to trigger full-time wake-up-free.

[0036] Specifically, the preset thresholds include a first threshold, which is used to determine whether the voice recognition confidence is high enough to decide whether to trigger full-time wake-up. The significance of the first threshold is to set a reliability standard for the voice recognition results, ensuring that the user's voice is responded to only under conditions of high confidence, thereby reducing misoperations caused by inaccurate or ambiguous voice recognition.

[0037] During the judgment process, if the intent information belongs to the valid range of intent and the voice recognition confidence is greater than the first threshold, it will be determined that full-time wake-up is triggered. In this case, the command content of the user's voice meets the preset valid intent range, and the confidence reaches or exceeds the first threshold, indicating that the voice recognition result has a high degree of credibility and the command content is valid, so it can respond to user needs quickly and accurately. This dual verification of high confidence and effective intent ensures that false triggering of full-time wake-up is reduced, providing users with a smoother and more timely interactive experience. On the contrary, when the intent information is not within the valid range of intent and the voice recognition confidence is lower than the first threshold, full-time wake-up is not triggered. This means that the user's voice command does not meet the preset valid intent category, and the credibility of voice recognition is low, and the command may not be the user's real need or the intention is unclear. In this case, the command will be ignored to avoid unnecessary misoperation and further improve the reliability and stability of the recognition process.

[0038] Through the filtering effect of the first threshold, it can effectively filter out highly credible instructions and ignore low-confidence content, thereby improving the accuracy of voice recognition. The dual screening mechanism based on the effective range of intent and voice recognition confidence judgment can ignore user voices with invalid intent, making the full-time wake-up-free function more intelligent and accurate, and significantly reducing the false trigger rate.

[0039] Based on the above embodiment, the preset threshold value further includes a second threshold value, and the second threshold value is greater than the first threshold value; The determining whether to trigger full-time wake-up-free based on whether the intention information belongs to the effective range of the intention and the size relationship between the voice recognition confidence and the preset threshold value also includes: When the intent information does not fall within the valid range of the intent and the voice recognition confidence is greater than the first threshold, it is determined whether to trigger full-time wake-up-free based on the relationship between the voice recognition confidence and the second threshold.

[0040] Specifically, the preset threshold also includes a higher second threshold, which is used to further confirm the reliability of the speech recognition result and whether to trigger the full-time wake-up-free function in specific situations. Compared with the first threshold, the second threshold sets a higher confidence requirement, thereby ensuring that the full-time wake-up-free function can be conditionally triggered even when the intention is not within the valid range to meet user needs.

[0041] When the intention information of the user's voice does not fall within the valid range of the intention but the voice recognition confidence is higher than the first threshold, it will not decide whether to trigger full-time wake-up-free, but will further compare the relationship between the voice recognition confidence and the second threshold to make a more accurate judgment. If the voice recognition confidence exceeds the second threshold, it means that the recognition result has a very high credibility. Even if the intention is not within the valid range, it can still be regarded as a valid command from the user. Therefore, in this case, full-time wake-up-free is triggered to respond to user needs. In this way, even if the instruction category is not within the preset valid range, if the confidence is high enough, there is still a chance to trigger wake-up-free, ensuring that user needs are not missed when facing high-credibility instructions.

[0042] On the contrary, if the voice recognition placement confidence is between the first threshold and the second threshold, it means that although the credibility of the recognition result exceeds the basic judgment standard, it has not reached an extremely high confidence level. In this case, although the confidence of voice recognition is high, the full-time wake-up-free function will not be triggered because the intention information of the user's voice is not within the valid range, so as to avoid executing instructions that may not meet the user's actual needs. This design ensures that when the intention does not meet the valid range, the full-time wake-up-free function will only be allowed to be triggered when the recognition result reaches an extremely high credibility, thereby minimizing the false trigger rate and ensuring the accuracy of the interactive response without affecting the user experience.

[0043] The purpose of this dual-threshold design is to introduce a more refined voice recognition confidence screening mechanism, so that the judgment process can not only judge the validity of the command based on the effective range of the intent, but also respond to commands in the non-effective range with extremely high voice recognition confidence under special circumstances. This mechanism not only improves the accuracy of full-time wake-up-free, but also significantly enhances the flexibility of interaction, further improving the user experience.

[0044] Based on the above embodiment, the preset threshold value further includes a third threshold value, and the third threshold value is smaller than the first threshold value; The determining whether to trigger full-time wake-up-free based on whether the intention information belongs to the effective range of the intention and the size relationship between the voice recognition confidence and the preset threshold value also includes: When the intent information belongs to the valid range of the intent and the voice recognition confidence is less than or equal to the first threshold, it is determined whether to trigger full-time wake-up-free based on the size relationship between the voice recognition confidence and the third threshold.

[0045] Specifically, the preset threshold also includes a third threshold, and the third threshold is less than the first threshold. The third threshold is used to provide a minimum standard when the voice recognition confidence is low to determine whether to trigger full-time wake-up-free. It introduces a more tolerant judgment condition to ensure that even if the voice confidence does not reach the ideal standard, it is still possible to trigger full-time wake-up-free to meet the needs of users.

[0046] When the intent information belongs to the valid range of intent, but the voice recognition confidence is lower than the first threshold, it will not directly decide not to trigger full-time wake-up, but further compare the size relationship between the voice recognition confidence and the third threshold. If the voice recognition confidence is higher than the third threshold, even if the higher standard of the first threshold is not reached, full-time wake-up will still be triggered. The purpose of this is to allow instructions that meet the valid intent to be executed when the confidence is slightly lower, to ensure that the user's valid needs will not be ignored due to the slightly lower voice recognition confidence. This is particularly important in some special scenarios, such as when the ambient noise causes a slight decrease in voice recognition confidence. At this time, the user's instructions can still be responded to, without affecting the continuity and convenience of the operation. However, if the voice recognition confidence is lower than the third threshold, it means that the credibility of the voice recognition result is insufficient. Even if the intent information of the user's voice is within the valid intent range, due to the risk of misrecognition, the full-time wake-up will not be triggered, thereby avoiding misoperation of low-credibility instructions.

[0047] Through this double-layer screening mechanism, the third threshold effectively enhances the flexibility and adaptability of the interaction, making it possible to accurately judge whether to respond to user needs in a more complex environment. The setting of the third threshold not only reduces the strict reliance on high confidence standards, but also ensures timely response to effective user instructions in a changing environment, thus bringing a smarter and smoother interactive experience.

[0048] Based on the above embodiment, the preset threshold is determined based on the speech recognition confidence of the positive sample and the speech recognition confidence of the reverse sample, the positive sample is the sample speech determined to trigger the full-time wake-up-free, and the reverse sample is the sample speech determined not to trigger the full-time wake-up-free.

[0049] Specifically, positive samples refer to user voices that meet the conditions for triggering full-time wake-up-free, that is, voice samples that users hope to be recognized and responded to immediately, and their voice recognition confidence is usually high. Therefore, by analyzing the voice recognition confidence range of positive samples, it is possible to clarify the voice recognition confidence standard required when the trigger conditions are met. This provides a basis for setting a threshold that can reliably trigger full-time wake-up-free, ensuring that a quick and accurate response can be made when a clear and effective user instruction is recognized.

[0050] Reverse samples are user voice samples that meet the conditions for not triggering full-time wake-up, including invalid commands that should be ignored, noise, chatter and other non-command voices, or ambiguous commands with low recognition confidence. The confidence of such samples is usually low. By analyzing the confidence of reverse samples, a lower confidence threshold can be derived to help keep full-time wake-up from being triggered when receiving unclear or unreliable commands. The setting of this bottom line threshold effectively prevents false triggers, allowing the correct distinction between commands that need to be responded to and invalid voices that need to be ignored in complex environments.

[0051] This method of determining the voice confidence threshold based on positive and negative samples provides a clear voice confidence range, which enables effective judgment on whether to trigger full-time wake-up in various situations. The dual analysis of positive and negative samples not only improves the accuracy of responding to effective commands, but also reduces the false trigger rate, enhances adaptability in complex environments, and significantly improves the smoothness and reliability of the user experience.

[0052] Based on the above embodiment, the full-time wake-up-free method further includes: When it is determined that full-time wake-up-free is triggered, respond to the user voice.

[0053] Specifically, when the user's voice is determined to meet the conditions for triggering full-time wake-up-free through dual judgments of intent information and voice recognition confidence, the user's command will be executed directly. For example, when the user makes a voice such as "navigate to the nearest gas station" and it is determined that the user's voice meets the conditions for triggering full-time wake-up-free, the navigation function is automatically started without the user's additional operation. This design not only makes the interaction more natural, but also avoids the tedious steps of multiple wake-ups or confirmations, significantly improving the user experience. This response mechanism is particularly important in complex environments or driving scenarios, allowing users to focus on the current task while the device efficiently completes the command execution in the background.

[0054] Based on the above embodiments, Figure 2 This is the second flow chart of the full-time wake-up-free method provided by the present invention. Figure 2 As shown in the figure, the flowchart shows a full-time wake-up-free decision process based on intent information and speech recognition confidence. Three different speech recognition confidence thresholds (first threshold, second threshold, third threshold) are used to decide whether to trigger the wake-up-free function. The relationship between the three confidence thresholds is: second threshold> first threshold> third threshold. The whole process starts with user voice input, and the voice signal is sent to the intent recognition engine and the speech recognition engine respectively to output intent information and speech recognition confidence for subsequent judgment.

[0055] After summarizing the intention information and voice recognition confidence of the user's voice, in the specific condition judgment, first determine whether the voice recognition confidence is greater than the first threshold and whether the intention information is within the valid range. If the condition is met, the full-time wake-up-free function is directly triggered to respond to the user's voice. If the condition is not met, proceed to the next step of judgment to check whether the voice recognition confidence is greater than the first threshold and the intention information is not within the valid range. At this time, if the voice recognition confidence exceeds the second threshold, the full-time wake-up-free function is still triggered; if the voice recognition confidence does not reach the second threshold, the voice command is ignored.

[0056] When the voice recognition confidence is lower than the first threshold but the intent information is within the valid range, the process will further determine whether the voice recognition confidence is between the third threshold and the first threshold. If the voice recognition confidence falls within the range, the full-time wake-up-free function is triggered; if the confidence is lower than the third threshold, the command is ignored. In the last case, if the confidence is lower than the first threshold and the intent information is not within the valid range, the command is directly ignored and the full-time wake-up-free function is not triggered.

[0057] Through this multi-level voice recognition confidence threshold and intent validity judgment, the process effectively balances the triggering conditions and false triggering control of the full-time wake-up-free function, ensuring that while reducing false triggers, it can flexibly respond to commands with high voice recognition confidence or valid intent, significantly improving the fluency and accuracy of the user interaction experience.

[0058] The full-time wake-up-free device provided by the present invention is described below. The full-time wake-up-free device described below and the full-time wake-up-free method described above can be referenced to each other.

[0059] Figure 3 is a schematic diagram of the structure of the full-time wake-up-free device provided by the present invention, such as Figure 3 As shown, the device comprises: The voice acquisition unit 310 is used to acquire the user's voice; The model processing unit 320 is used to perform intent recognition on the user's voice to obtain intent information, and to perform speech recognition on the user's voice to obtain speech recognition confidence; The trigger judgment unit 330 is used to determine whether to trigger full-time wake-up-free based on whether the intention information belongs to the effective range of the intention and the relationship between the voice recognition confidence and the preset threshold.

[0060] The full-time wake-up-free device provided in the embodiment of the present invention determines whether the intention information of the user's voice is within a valid range, and combines the voice recognition confidence to trigger the wake-up-free operation only when the intention is valid and the voice recognition result is credible, thereby effectively improving the accuracy of the full-time wake-up-free and significantly reducing the false trigger rate, thereby improving the stability of the interaction and bringing a smoother full-time wake-up-free experience.

[0061] Based on any of the above embodiments, the trigger judgment unit is specifically used to: The preset threshold includes a first threshold; When the intention information belongs to the effective range of the intention and the voice recognition confidence is greater than the first threshold, determining to trigger full-time wake-up-free; When the intent information does not belong to the valid range of the intent and the voice recognition confidence is less than or equal to the first threshold, it is determined not to trigger full-time wake-up-free.

[0062] Based on any of the above embodiments, the trigger judgment unit is specifically used to: The preset threshold value also includes a second threshold value, and the second threshold value is greater than the first threshold value; When the intent information does not fall within the valid range of the intent and the voice recognition confidence is greater than the first threshold, it is determined whether to trigger full-time wake-up-free based on the relationship between the voice recognition confidence and the second threshold.

[0063] Based on any of the above embodiments, the trigger judgment unit is specifically used to: The preset threshold value further includes a third threshold value, and the third threshold value is smaller than the first threshold value; When the intent information belongs to the valid range of the intent and the voice recognition confidence is less than or equal to the first threshold, it is determined whether to trigger full-time wake-up-free based on the size relationship between the voice recognition confidence and the third threshold.

[0064] Based on any of the above embodiments, the trigger judgment unit is specifically used to: The preset threshold is determined based on the speech recognition confidence of the positive sample and the speech recognition confidence of the reverse sample. The positive sample is the sample speech that is determined to trigger the full-time wake-up-free function, and the reverse sample is the sample speech that is determined not to trigger the full-time wake-up-free function.

[0065] Based on any of the above embodiments, the trigger judgment unit is specifically used to: When it is determined that full-time wake-up-free is triggered, respond to the user voice.

[0066] Figure 4 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 4 As shown, the electronic device may include: a processor 410, a communication interface 420, a memory 430 and a communication bus 440, wherein the processor 410, the communication interface 420 and the memory 430 communicate with each other through the communication bus 440. The processor 410 may call the logic instructions in the memory 430 to execute the full-time wake-up-free method, which includes: Get user voice; Performing intent recognition on the user's voice to obtain intent information, and performing speech recognition on the user's voice to obtain speech recognition confidence; Based on whether the intention information belongs to the valid range of the intention and the relationship between the voice recognition confidence and the preset threshold, it is determined whether to trigger full-time wake-up-free.

[0067] In addition, the logic instructions in the above-mentioned memory 430 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the relevant technology or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0068] On the other hand, the present invention further provides a computer program product, the computer program product comprising a computer program, the computer program can be stored on a non-transitory computer-readable storage medium, when the computer program is executed by a processor, the computer can execute the full-time wake-up-free method provided by the above methods, the method comprising: Get user voice; Performing intent recognition on the user's voice to obtain intent information, and performing speech recognition on the user's voice to obtain speech recognition confidence; Based on whether the intention information belongs to the valid range of the intention and the relationship between the voice recognition confidence and the preset threshold, it is determined whether to trigger full-time wake-up-free.

[0069] In another aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is implemented when the computer program is executed by a processor to perform the full-time wake-up-free method provided by the above methods, the method comprising: Get user voice; Performing intent recognition on the user's voice to obtain intent information, and performing speech recognition on the user's voice to obtain speech recognition confidence; Based on whether the intention information belongs to the valid range of the intention and the relationship between the voice recognition confidence and the preset threshold, it is determined whether to trigger full-time wake-up-free.

[0070] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0071] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the relevant technology can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiment.

[0072] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A full-time wake-up-free method, characterized in that: include: Get user voice; Performing intent recognition on the user's voice to obtain intent information, and performing speech recognition on the user's voice to obtain speech recognition confidence; Based on whether the intention information belongs to the valid range of the intention and the relationship between the voice recognition confidence and the preset threshold, it is determined whether to trigger full-time wake-up-free.

2. The full-time wake-up-free method according to claim 1, characterized in that: The preset threshold includes a first threshold; The determining whether to trigger full-time wake-up-free based on whether the intention information belongs to the effective range of the intention and the size relationship between the voice recognition confidence and the preset threshold includes: When the intention information belongs to the effective range of the intention and the voice recognition confidence is greater than the first threshold, determining to trigger full-time wake-up-free; When the intent information does not belong to the valid range of the intent and the voice recognition confidence is less than or equal to the first threshold, it is determined not to trigger full-time wake-up-free.

3. The full-time wake-up-free method according to claim 2, characterized in that: The preset threshold value also includes a second threshold value, and the second threshold value is greater than the first threshold value; The determining whether to trigger full-time wake-up-free based on whether the intention information belongs to the effective range of the intention and the size relationship between the voice recognition confidence and the preset threshold value also includes: When the intent information does not fall within the valid range of the intent and the voice recognition confidence is greater than the first threshold, it is determined whether to trigger full-time wake-up-free based on the relationship between the voice recognition confidence and the second threshold.

4. The full-time wake-up-free method according to claim 2, characterized in that: The preset threshold value further includes a third threshold value, and the third threshold value is smaller than the first threshold value; The determining whether to trigger full-time wake-up-free based on whether the intention information belongs to the effective range of the intention and the size relationship between the voice recognition confidence and the preset threshold value also includes: When the intent information belongs to the valid range of the intent and the voice recognition confidence is less than or equal to the first threshold, it is determined whether to trigger full-time wake-up-free based on the size relationship between the voice recognition confidence and the third threshold.

5. The full-time wake-up-free method according to any one of claims 1 to 4, characterized in that: The preset threshold is determined based on the speech recognition confidence of the positive sample and the speech recognition confidence of the reverse sample. The positive sample is the sample speech that is determined to trigger the full-time wake-up-free function, and the reverse sample is the sample speech that is determined not to trigger the full-time wake-up-free function.

6. The full-time wake-up-free method according to any one of claims 1 to 4, characterized in that: Also includes: When it is determined that full-time wake-up-free is triggered, respond to the user voice.

7. A full-time wake-up-free device, characterized in that: include: A voice acquisition unit, used to acquire user voice; A model processing unit, configured to perform intent recognition on the user's voice to obtain intent information, and perform speech recognition on the user's voice to obtain speech recognition confidence; The trigger judgment unit is used to determine whether to trigger full-time wake-up-free based on whether the intention information belongs to the effective range of the intention and the relationship between the voice recognition confidence and the preset threshold.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the full-time wake-up-free method according to any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the full-time wake-up-free method according to any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the full-time wake-up-free method according to any one of claims 1 to 6 is implemented.

Images