Method and device for predicting silent time window, storage medium and electronic device

Abstract

The invention discloses a silent time window prediction method and device, a storage medium and an electronic device, and relates to the technical field of smart home, the method comprises the following steps: obtaining historical voice interaction data of a target object, the historical voice interaction data being used for indicating an interaction voice of the target object and a voice device; determining a first time difference between voice interaction time corresponding to historical voice interaction data and current time, and determining first sentence segmentation interval time corresponding to the historical voice interaction data; determining a target weight corresponding to the historical voice interaction data according to the first time difference; and predicting a target silence time window of interaction between the target object and the voice device according to the target weight and the first sentence segmentation interval time. Through the method, the problem that the overall voice interaction duration is increased due to the fact that a long silent time window is selected to adapt to different sentence segmentation habits of different user groups during speaking in the related technology can be solved.

Description

Technical Field

[0001] The present application relates to the field of smart home technology, and more specifically, to a method and device for predicting a silent time window, a storage medium, and an electronic device. Background Art

[0002] In the current trend of social life, with the family as the core, voice interaction control based on artificial intelligence (AI) voice technology has become the mainstream demand leading the intelligent home. People are eager to enjoy a more intelligent life. Through simple voice commands, they can easily control various intelligent devices and systems, answer questions, and get operation instructions. This high recognition rate and fast voice interaction response not only enhances the seamless communication between users and smart homes, but also brings people unprecedented convenience and intelligent home experience. The voice interaction function in smart homes has become an indispensable part of people's daily lives. Smart homes greatly simplify the control of smart devices and systems, and also provide convenient question-and-answer and operation guidance services. However, different users have different speaking speeds and sentence punctuation habits, especially those with slower speaking speeds and longer sentence punctuation intervals. Traditional voice activity detection (VAD) often needs to set a longer silent time window when processing their voice data to ensure the integrity of the voice input. However, this unified processing approach undoubtedly increases the voice interaction response time of all users and reduces the user's voice interaction experience.

[0003] In the related art, an effective solution has not yet been proposed for the problem of increasing the overall voice interaction duration by selecting a longer silent time window to adapt to the different sentence-breaking habits of different user groups when speaking. Summary of the invention

[0004] The embodiments of the present application provide a method and device for predicting a silence time window, a storage medium, and an electronic device, so as to at least solve the problem in the related art that a longer silence time window is selected to adapt to the different sentence segmentation habits of different user groups when speaking, thereby increasing the overall voice interaction duration.

[0005] According to an embodiment of the present application, a method for predicting a silence time window is provided, comprising: acquiring historical voice interaction data of a target object, wherein the historical voice interaction data is used to indicate the interactive voice between the target object and a voice device; determining a first time difference between a voice interaction time corresponding to the historical voice interaction data and a current time, and determining a first sentence interval time corresponding to the historical voice interaction data; determining a target weight corresponding to the historical voice interaction data based on the first time difference; and predicting a target silence time window for the interaction between the target object and the voice device based on the target weight and the first sentence interval time.

[0006] In an exemplary embodiment, determining the target weight corresponding to the historical voice interaction data according to the first time difference includes: obtaining multiple voice interaction data corresponding to the target object, and determining the second time difference between the voice interaction time corresponding to each voice interaction data and the current time; dividing the multiple voice interaction data and the second time difference corresponding to each voice interaction data into training set data and test set data, and obtaining an initial attenuation coefficient; training a target model according to the training set data and the initial attenuation coefficient to obtain a trained target model, wherein the target model is used to determine the weight corresponding to each voice interaction data; verifying the trained target model according to the test set data; when it is determined that the trained target model has passed the verification, inputting the first time difference into the trained target model so that the trained training model outputs the target weight.

[0007] In an exemplary embodiment, determining a first sentence-break interval time corresponding to the historical voice interaction data includes: determining whether there is a first voice interruption event in the historical voice interaction data, wherein the first voice interruption event includes: a voice disappearance event or a silence event; in the case of determining that there is the first voice interruption event in the historical voice interaction data, determining a first time when the first voice interruption event starts, and determining whether there is a voice occurrence event after the first voice interruption event; in the case of determining that there is the voice occurrence event after the first voice interruption event, determining a second time when the voice occurrence event starts, and determining the first sentence-break interval time according to the difference between the second time and the first time; in the case of determining that there is no voice occurrence event after the first voice interruption event, determining a third time when the historical voice interaction data ends, and determining the first sentence-break interval time according to the difference between the third time and the first time.

[0008] In an exemplary embodiment, after determining whether there is a first voice interruption event in the historical voice interaction data, the method further includes: in the case where it is determined that the first voice interruption event exists in the historical voice interaction data, determining whether there is the voice occurrence event within a target time period after the first time; in the case where it is determined that the voice occurrence event exists within the target time period, determining that the voice interruption event is the voice disappearance event; in the case where it is determined that the voice occurrence event does not exist within the target time period, determining that the voice interruption event is the silence event.

[0009] In an exemplary embodiment, after predicting the target silent time window for the interaction between the target object and the voice device according to the target weight and the first sentence interval time, the method further includes: determining whether there is a second voice interruption event in the first interactive voice when it is determined that the voice device receives the first interactive voice issued by the target object; determining whether the second voice interruption event is a silent event when it is determined that the first interactive voice exists; determining the sum of the target silent time window and the preset extension time when it is determined that the second voice interruption event is the silent event, and determining a first size relationship between the second sentence interval time corresponding to the first interactive voice and the sum; when the first size relationship indicates that the second sentence interval time is greater than or equal to the sum, ending the pickup of the first interactive voice, and determining a first weight corresponding to the first interactive voice after the pickup is ended; predicting the first silent time window for the interaction between the target object and the voice device according to the first weight and the second sentence interval time, and updating the target silent time window according to the first silent time window.

[0010] In an exemplary embodiment, after determining the first size relationship between the second sentence interval time corresponding to the first interactive voice and the sum value, the method also includes: when the first size relationship indicates that the second sentence interval time is less than the sum value, performing endpoint detection on the first interactive voice to obtain a voice occurrence event after the second sentence interval; determining whether the silent event exists in the second interactive voice corresponding to the voice occurrence event; when it is determined that the silent event exists in the second interactive voice, determining the second size relationship between the third sentence interval time corresponding to the second interactive voice and the sum value; when the second size relationship indicates that the third sentence interval time is greater than or equal to the sum value, ending the pickup of the second interactive voice, and merging the first interactive voice and the second interactive voice to update the first interactive voice; determining the second weight corresponding to the updated first interactive voice; predicting the second silent time window for the interaction between the target object and the voice device based on the second weight, the second sentence interval time and the third sentence interval time, and updating the target silent time window based on the second silent time window.

[0011] In an exemplary embodiment, predicting a second silent time window for interaction between the target object and the voice device based on the second weight, the second sentence interval time and the third sentence interval time includes: determining a third size relationship between the second sentence interval time and the third sentence interval time; determining the maximum sentence interval time between the second sentence interval time and the third sentence interval time based on the third size relationship; and determining the second silent time window based on the second weight and the maximum sentence interval time.

[0012] In an exemplary embodiment, after predicting the target silent time window for the interaction between the target object and the voice device based on the target weight and the first sentence interval time, the method also includes: when it is determined that the voice device has received a third interactive voice, determining whether the sound-emitting object corresponding to the third interactive voice is the target object; when it is determined that the sound-emitting object is not the target object, determining whether there is a third silent time window for the interaction between the sound-emitting object and the voice device in the silent time window library, wherein the silent time window library includes: the target silent time window corresponding to the target object; when it is determined that the third silent time window does not exist in the silent time window library, determining the third weight corresponding to the third interactive voice and the fourth sentence interval time corresponding to the third interactive voice; predicting the third silent time window based on the third weight and the fourth sentence interval time, and storing the third silent time window corresponding to the sound-emitting object in the silent time window library.

[0013] According to another embodiment of the embodiments of the present application, a prediction device for a silent time window is also provided, including: an acquisition module, used to acquire historical voice interaction data of a target object, wherein the historical voice interaction data is used to indicate the interactive voice between the target object and a voice device; a first determination module, used to determine a first time difference between the voice interaction time corresponding to the historical voice interaction data and the current time, and determine a first sentence interval time corresponding to the historical voice interaction data; a second determination module, used to determine a target weight corresponding to the historical voice interaction data based on the first time difference; and a prediction module, used to predict a target silent time window for the interaction between the target object and the voice device based on the target weight and the first sentence interval time.

[0014] According to another aspect of the embodiments of the present application, a computer-readable storage medium is provided, in which a computer program is stored, wherein the computer program is configured to execute the above-mentioned method for predicting the silence time window when running.

[0015] According to another aspect of an embodiment of the present application, an electronic device is also provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the above-mentioned method for predicting the silence time window through the computer program.

[0016] According to another aspect of the embodiments of the present application, a computer program product is provided, including a computer program, wherein the computer program is executed by a processor to perform the above-mentioned method for predicting a silence time window.

[0017] In an embodiment of the present application, historical voice interaction data between the target object and the voice device is obtained, the first time difference between the historical voice interaction data and the corresponding voice interaction time and the current time is determined, and the first time interval between the historical voice interaction data is determined, and the target weight corresponding to the historical voice interaction data is determined according to the first time difference; the target silent time window for the interaction between the target object and the voice device is predicted according to the target weight and the first time interval between the sentences. That is, the embodiment of the present application determines the target weight according to the first time difference between the voice interaction time corresponding to the historical voice interaction data and the current time, and then predicts the target silent time window according to the target weight and the first time interval between the sentences. According to the embodiment of the present application, the problem of increasing the overall voice interaction duration caused by selecting a longer silent time window to adapt to the different sentence-breaking habits of different user groups when speaking in the related art can be solved. Then different silent time windows are determined to adapt to the different sentence-breaking habits of different user groups, thereby reducing the overall voice interaction duration. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0019] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0020] Figure 1 It is a hardware environment diagram of a method for predicting a silent time window according to an embodiment of the present application;

[0021] Figure 2 is a flowchart of a method for predicting a silent time window according to an embodiment of the present application;

[0022] Figure 3 This is an overall framework diagram of voice interaction between a user and a large model in an optional embodiment of the present application;

[0023] Figure 4 is a schematic diagram of a voice interaction process according to an embodiment of the present application;

[0024] Figure 5 It is a schematic diagram of a device corresponding to a method for improving the success rate of speech recognition and shortening the duration of speech interaction according to an optional embodiment of the present application;

[0025] Figure 6 is a flow chart of a method for improving the success rate of speech recognition and shortening the duration of speech interaction according to an optional embodiment of the present application;

[0026] Figure 7 It is a structural block diagram of a silent time window prediction device according to an embodiment of the present application. DETAILED DESCRIPTION

[0027] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0028] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0029] According to one aspect of an embodiment of the present application, a method for predicting a silent time window is provided. The method for predicting a silent time window is widely used in smart home, smart home, smart home device ecology, smart house ecology and other whole-house intelligent digital control application scenarios. Optionally, Figure 1 is a hardware environment diagram of a silent time window prediction method of an embodiment of the present application. In this embodiment, the silent time window prediction method can be applied to a module in a home appliance. The module can be applied to Figure 1 In the hardware environment composed of the home appliance 102 and the server 104 shown in FIG. Figure 1 As shown, the server 104 is connected to the home appliance 102 via a network, and can be used to provide services (such as application services, etc.) for the node or the client installed on the node. A database can be set on the server or independently of the server to provide data storage services for the server 104. Cloud computing and / or edge computing services can be configured on the server or independently of the server to provide data computing services for the server 104.

[0030] The network may include but is not limited to at least one of the following: wired network, wireless network. The wired network may include but is not limited to at least one of the following: wide area network, metropolitan area network, local area network, and the wireless network may include but is not limited to at least one of the following: WIFI (Wireless Fidelity), Bluetooth. The home appliance 102 may be but is not limited to smart air conditioners, smart range hoods, smart refrigerators, smart ovens, smart stoves, smart washing machines, smart water heaters, smart washing equipment, smart dishwashers, smart projection equipment, smart TVs, smart clothes hangers, smart curtains, smart audio and video, smart sockets, smart speakers, smart speakers, smart fresh air equipment, smart kitchen and bathroom equipment, smart bathroom equipment, smart sweeping robots, smart window cleaning robots, smart mopping robots, smart air purification equipment, smart steamers, smart microwave ovens, smart kitchen treasures, smart purifiers, smart water dispensers, smart door locks, etc.

[0031] In this embodiment, a method for predicting a silent time window is provided, which is applied to the above module. Figure 2 : is a flowchart of a method for predicting a silent time window according to an embodiment of the present application, and the process includes the following steps:

[0032] Step S202, obtaining historical voice interaction data of the target object, wherein the historical voice interaction data is used to indicate the interactive voice between the target object and the voice device;

[0033] Step S204, determining a first time difference between a voice interaction time corresponding to the historical voice interaction data and the current time, and determining a first sentence segmentation interval time corresponding to the historical voice interaction data;

[0034] Step S206, determining a target weight corresponding to the historical voice interaction data according to the first time difference;

[0035] Step S208: predicting a target silent time window for interaction between the target object and the voice device according to the target weight and the first sentence segmentation interval.

[0036] Through the above steps, the historical voice interaction data between the target object and the voice device is obtained, the first time difference between the historical voice interaction data and the corresponding voice interaction time and the current time is determined, and the first time interval corresponding to the historical voice interaction data is determined. The target weight corresponding to the historical voice interaction data is determined according to the first time difference; the target silent time window for the interaction between the target object and the voice device is predicted according to the target weight and the first time interval. That is to say, the embodiment of the present application determines the target weight according to the first time difference between the voice interaction time corresponding to the historical voice interaction data and the current time, and then predicts the target silent time window according to the target weight and the first time interval. According to the embodiment of the present application, the problem of increasing the overall voice interaction time caused by selecting a longer silent time window to adapt to the different sentence-breaking habits of different user groups when speaking in the related art can be solved. Then different silent time windows are determined to adapt to the different sentence-breaking habits of different user groups, reducing the overall voice interaction time.

[0037] Optionally, the above-mentioned step S206 of determining the target weight corresponding to the historical voice interaction data according to the first time difference includes: obtaining multiple voice interaction data corresponding to the target object, and determining the second time difference between the voice interaction time corresponding to each voice interaction data and the current time; dividing the multiple voice interaction data and the second time difference corresponding to each voice interaction data into training set data and test set data, and obtaining an initial attenuation coefficient; training the target model according to the training set data and the initial attenuation coefficient to obtain a trained target model, wherein the target model is used to determine the weight corresponding to each voice interaction data; verifying the trained target model according to the test set data; when it is determined that the trained target model has passed the verification, inputting the first time difference into the trained target model so that the trained training model outputs the target weight.

[0038] It is understood that the specific steps for determining the target weight include:

[0039] Collect a large amount of historical voice interaction data of the target object (which can be a family member), including the time points when the voice starts and ends. For each piece of historical voice interaction data, calculate the second time difference between its voice interaction time and the current time, which reflects the newness of the voice interaction data. Divide the collected multiple voice interaction data and their corresponding second time difference into training set data and test set data. The training set data is used for model training, while the test set data is used to verify the accuracy of the model. Select a preliminary attenuation coefficient as the starting point for model training. The attenuation coefficient determines the degree of influence of historical data on the current prediction. The target model (i.e., the model used to determine the weights) is trained using the training set data and the initial attenuation coefficient. A machine learning algorithm, such as a weighted moving average algorithm, can be used to adjust the model parameters and optimize the weight allocation. The target model learns how to assign different weights according to the newness of the data (i.e., the second time difference), so that the target silence time window can be predicted more accurately.

[0040] After the target model is trained, the test set data is used to verify the performance of the trained target model and evaluate the prediction accuracy of the trained target model. If the model performs well on the test set data, it means that the trained target model can effectively allocate weights according to the newness of the data, thereby accurately predicting the silent time window.

[0041] If the model is verified, the first time difference of the current voice interaction (i.e., the time difference from when the user stops speaking to when the system starts silent timing) is used as input and sent to the trained target model. The target model determines the target weight of the current voice interaction data based on the first time difference.

[0042] Through the above steps, the length of the silent time window can be dynamically adjusted according to the user's (i.e., target object's) most recent interaction mode (e.g., the first time difference), ensuring that each user's voice interaction at different time points can be optimally processed, thereby improving the success rate of voice recognition, shortening the duration of voice interaction, and improving user experience.

[0043] Optionally, the above-mentioned step S204 of determining the first sentence interruption interval corresponding to the historical voice interaction data includes: determining whether there is a first voice interruption event in the historical voice interaction data, wherein the first voice interruption event includes: a voice disappearance event or a silence event; when it is determined that the first voice interruption event exists in the historical voice interaction data, determining the first time when the first voice interruption event starts, and determining whether there is a voice occurrence event after the first voice interruption event; when it is determined that there is a voice occurrence event after the first voice interruption event, determining the second time when the voice occurrence event starts, and determining the first sentence interruption interval according to the difference between the second time and the first time; when it is determined that there is no voice occurrence event after the first voice interruption event, determining the third time when the historical voice interaction data ends, and determining the first sentence interruption interval according to the difference between the third time and the first time.

[0044] It can be understood that the interval between sentence breaks is the time when the target speech is silent.

[0045] Determine the first time of the target object's first speech interruption event: Once the first speech interruption event is identified, the system will record the exact time point when this event begins.

[0046] Detecting a speech occurrence event: Checking whether there is a speech occurrence event after the first speech interruption event (ie, the user starts speaking again). If a speech occurrence event is detected, determining a second time when the speech occurrence event starts.

[0047] Calculate the first inter-sentence interval time (speech occurrence event): If a speech occurrence event is detected after the first speech interruption event, the second time is calculated by calculating the difference between the second time and the first time. This difference represents the duration of a speech pause of the target object and is an important basis for determining the silence time window.

[0048] Calculate the first inter-sentence interval (no voice occurrence event): If no voice occurrence event is detected after the first voice interruption event (i.e., the user does not speak again and the voice interaction ends), the time point at which the historical voice interaction data ends will be recorded, i.e., the third time. Then, calculate the difference between the third time and the first inter-sentence interval of the first time to determine the first inter-sentence interval. In this case, the first inter-sentence interval represents the total duration from when the user stops speaking to when the voice interaction ends.

[0049] Through the above steps, the sentence segmentation habits of each user in historical voice interactions are determined, and the possible sentence segmentation intervals that users may take in different situations are captured. These data are crucial for optimizing the VAD silent time window, because it allows the system to dynamically adjust the waiting time according to the individual differences of users, which not only ensures the complete collection of voice commands, but also avoids unnecessary waiting, thereby improving the accuracy and efficiency of voice recognition, shortening the total duration of voice interaction, and ultimately improving the user's smart home experience. This intelligent optimization method reflects the powerful potential of AI technology in personalized services and real-time response.

[0050] Among them, after determining whether there is a first voice interruption event in the historical voice interaction data, the method also includes: when it is determined that the first voice interruption event exists in the historical voice interaction data, determining whether there is the voice occurrence event within a target time period after the first time; when it is determined that the voice occurrence event exists within the target time period, determining that the voice interruption event is the voice disappearance event; when it is determined that the voice occurrence event does not exist within the target time period, determining that the voice interruption event is the silence event.

[0051] It can be understood that, in the case of a first voice interruption event, it is necessary to determine whether the first voice interruption event is a silent event or a voice disappearance event. Specifically: in the case of a voice occurrence event within the first time period after the start time of the voice interruption event is determined, the first voice interruption event is determined to be a voice disappearance event; in the case of no voice occurrence event within the first time period after the start time of the voice interruption event is determined, the first voice interruption event is determined to be a silent event.

[0052] Optionally, after predicting the target silent time window for the interaction between the target object and the voice device according to the target weight and the first sentence interval time, the method further includes: determining whether there is a second voice interruption event in the first interactive voice when it is determined that the voice device receives the first interactive voice issued by the target object; determining whether the second voice interruption event is a silent event when it is determined that the first interactive voice exists; determining the sum of the target silent time window and the preset extension time when it is determined that the second voice interruption event is the silent event, and determining a first size relationship between the second sentence interval time corresponding to the first interactive voice and the sum; when the first size relationship indicates that the second sentence interval time is greater than or equal to the sum, ending the pickup of the first interactive voice, and determining a first weight corresponding to the first interactive voice after the pickup is ended; predicting the first silent time window for the interaction between the target object and the voice device according to the first weight and the second sentence interval time, and updating the target silent time window according to the first silent time window.

[0053] It can be understood that, in the case where it is determined that the second voice interruption event is a silent event, the method for determining the target silent time window includes:

[0054] Calculate the sum of the target silent time window and the preset extension time (i.e., the sum value), and determine the relationship between the current second segmentation interval (i.e., the time from when the user stops speaking to when he starts speaking again) and the sum value. If the second segmentation interval is greater than or equal to the sum value, it indicates that the user's segmentation time is long and the voice interaction may have ended.

[0055] If the second segmentation interval is indeed greater than or equal to the preset sum, the system will end the pickup of the first interactive voice, that is, stop recording. Subsequently, the system determines the first weight corresponding to the first interactive voice data obtained after the pickup is completed, and the weight reflects the importance of the current voice data in predicting the silence time window.

[0056] Based on the first weight and the second segmentation interval, the optimization model (which may be a weighted moving average algorithm or other machine learning model) is used again to predict the first silent time window for the user to interact with the voice device. This predicted value more accurately reflects the user's current speaking segmentation habits. Finally, based on the predicted first silent time window, the system will update the target silent time window in real time to adapt to the user's immediate changes and ensure accurate capture and efficient processing of the next voice interaction.

[0057] Through this series of dynamic evaluations and instant adjustments, the embodiments of the present application can accurately capture the personalized speaking habits of different users, optimize the voice interaction performance of smart home voice devices, ensure the integrity and accuracy of voice commands, significantly shorten the interaction time, and enhance the user's voice interaction experience. This real-time optimization mechanism reflects the flexibility and intelligence of AI technology in intelligent voice interaction scenarios, and can continuously adapt to the personalized needs of users and provide a smoother and more natural interaction experience.

[0058] Among them, after determining the first size relationship between the second sentence interval time corresponding to the first interactive voice and the sum value, the method also includes: when the first size relationship indicates that the second sentence interval time is less than the sum value, performing endpoint detection on the first interactive voice to obtain a voice occurrence event after the second sentence interval; determining whether the silent event exists in the second interactive voice corresponding to the voice occurrence event; when it is determined that the silent event exists in the second interactive voice, determining the second size relationship between the third sentence interval time corresponding to the second interactive voice and the sum value; when the second size relationship indicates that the third sentence interval time is greater than or equal to the sum value, ending the pickup of the second interactive voice, and merging the first interactive voice and the second interactive voice to update the first interactive voice; determining the second weight corresponding to the updated first interactive voice; predicting the second silent time window for the interaction between the target object and the voice device according to the second weight, the second sentence interval time and the third sentence interval time, and updating the target silent time window according to the second silent time window.

[0059] It is understandable that when it is determined that the second segmentation interval is less than the preset value, it means that the user may continue to speak soon after a short pause. At this time, the system performs endpoint detection on the first interactive voice to identify the voice occurrence event, that is, the moment when the user starts to speak again. Specifically:

[0060] Determine the silence event in the second interactive voice: Once a voice event is detected, the system will record this new voice data, i.e. the second interactive voice, and check whether there is also a silence event in the second interactive voice. This continuous silence event detection helps the system understand the user's speaking habits more comprehensively.

[0061] The second relationship between the third interval time and the sum value: If a silent event is encountered again in the second interactive voice, the system will calculate the third interval time from the beginning of this silent event to the next voice event, and compare it with the above sum value. This comparison helps to determine whether the user has ended the current voice command or there may be a longer pause.

[0062] End voice pickup and merge voice data: If the third segmentation interval is greater than or equal to the sum value, the system will end the second interactive voice pickup and merge the first interactive voice with the second interactive voice to form a complete, updated voice input. This operation ensures that the system can fully capture the user's voice command even if there is a short pause in the user's speech.

[0063] Determine a second weight: Based on the merged voice data, the system determines a second weight, which reflects the importance of the current user voice input in predicting the silence time window.

[0064] The method for predicting the second silent time window for interaction between the target object and the voice device based on the second weight, the second sentence interval time and the third sentence interval time can include determining a third size relationship between the second sentence interval time and the third sentence interval time; determining the maximum sentence interval time between the second sentence interval time and the third sentence interval time based on the third size relationship; and determining the second silent time window based on the second weight and the second sentence interval time and the maximum sentence interval time.

[0065] Through this prediction, the setting of the silent time window can be dynamically adjusted and optimized to adapt to the user's speaking pattern and rhythm, ensuring that the voice commands of the user can be captured in a timely and complete manner regardless of whether the user speaks fast or slow.

[0066] Optionally, after predicting the target silent time window for the interaction between the target object and the voice device according to the target weight and the first sentence interval time in the above step S208, the method further includes: when it is determined that the voice device has received a third interactive voice, determining whether the sound-emitting object corresponding to the third interactive voice is the target object; when it is determined that the sound-emitting object is not the target object, determining whether there is a third silent time window for the interaction between the sound-emitting object and the voice device in the silent time window library, wherein the silent time window library includes: the target silent time window corresponding to the target object; when it is determined that the third silent time window does not exist in the silent time window library, determining the third weight corresponding to the third interactive voice and the fourth sentence interval time corresponding to the third interactive voice; predicting the third silent time window according to the third weight and the fourth sentence interval time, and storing the third silent time window corresponding to the sound-emitting object in the silent time window library.

[0067] It is understandable that the process after step S208 above focuses on processing voice interaction in a multi-person environment, ensuring that the system can accurately distinguish different users and provide each user with an optimized silent time window setting. Specifically:

[0068] When the voice device receives the third interactive voice, the system's first task is to identify the speaker of this voice and determine whether it is the target object analyzed previously. This is achieved through voiceprint recognition technology, which can distinguish the voice characteristics of different family members.

[0069] If the recognition result shows that the speaker is not the target, it means that another family member has started to interact with the voice device. At this time, the system checks the silent time window library to see if a third silent time window has been stored for the newly identified user (i.e., the speaker).

[0070] If the silent time window of the sounding object has not been stored in the silent time window library, the system will process it in the same way as the previous target object, but for the sounding object. First, it will determine the third weight corresponding to the third interactive speech, which reflects the importance of the new user's voice input in speech recognition and silent time window prediction. Then, the system analyzes the sentence interval in the third interactive speech and obtains the fourth sentence interval time. By combining the third weight and the fourth sentence interval time, the system can predict the third silent time window suitable for the new user and store it in the silent time window library for subsequent use.

[0071] The above steps ensure that when smart home voice devices are used by multiple family members, they can quickly adapt to each user's unique speaking habits and provide each user with the most optimized voice interaction experience. Especially in a multi-person environment, this fast and accurate user recognition and personalized setting of the silent time window avoids problems such as inaccurate recognition or long interaction time that may occur when the system uses unified settings, thereby significantly improving the efficiency of voice recognition and the user experience of smart homes.

[0072] In order to better understand the process of the above-mentioned prediction method of the silent time window, the implementation method flow of the above-mentioned prediction of the silent time window is described below in combination with an optional embodiment, but it is not used to limit the technical solution of the embodiment of the present application.

[0073] Voice interaction applications in smart homes are becoming increasingly popular, but different user groups have different sentence segmentation habits when speaking. In order to ensure that the voice of each user can be accurately picked up, the traditional practice often chooses a longer silent time window to adapt to most users. However, this undoubtedly increases the duration of the overall interaction and causes users to be troubled by slow responses. In order to solve this problem, an optional embodiment of the present application proposes a method for improving the success rate of voice recognition and shortening the duration of voice interaction. The above method can adjust the length of the silent time window in real time according to the speaking and sentence segmentation habits of each user. In this way, not only is it ensured that the voice intentions of users who speak slowly and have long sentence segmentation intervals can be fully recognized, but it also significantly shortens the overall interaction time of users with normal sentence segmentation, thereby greatly improving the intelligent voice interaction experience in smart homes.

[0074] In smart home scenarios, a convenient entry point for voice interaction is usually provided by advanced devices such as smart home brain screens or smart speakers with high-sensitivity recording functions. The entire voice interaction process is efficient and smooth: users only need to activate the interactive device with a simple wake-up word, and the device immediately enters an efficient sound pickup state and continuously captures the user's voice commands. These audio data are then quickly uploaded to the cloud for high-precision audio recognition and NLP (natural language processing) analysis. With the powerful capabilities of the large model, the system can accurately identify and output the user's intentions, then execute the corresponding voice commands, and give the user clear and intuitive feedback in real time. This process not only improves the level of intelligence in home life, but also ensures a seamless and natural interaction experience between users and smart home devices. Figure 3 This is an overall framework diagram of voice interaction between a user and a large model in an optional embodiment of the present application, such as Figure 3 As shown:

[0075] When it is detected that the user is speaking, the user's words are picked up, the picked-up results are subjected to noise reduction processing, and VAD detection is performed on the noise-reduced speech to obtain audio data; the audio data is sent to the cloud so that the cloud processes the audio data. Specifically: the intelligent cloud service performs automatic speech recognition (Automatic Speech Recognition, abbreviated as ASR) on the audio data, and sends the recognition result to the big model so that the big model can understand the user's intention, and performs text-to-speech (Text-to-Speech, abbreviated as TTS) conversion, and returns the converted audio broadcast data to the smart device so that the smart device can perform voice broadcast.

[0076] according to Figure 3 The framework diagram in the figure can determine the calculation rules of the voice interaction duration, specifically:

[0077] Figure 4 is a schematic diagram of the voice interaction process according to an embodiment of the present application, such as Figure 4 As shown:

[0078] The user wakes up the device, for example: Xiaoyou Xiaoyou;

[0079] User input voice interaction control, for example: turning on the light;

[0080] Voice interaction equipment Figure 3 The noise reduction, VAD detection, ASR speech recognition, large model intent understanding and then TTS conversion steps are carried out, and then the interactive device broadcasts the TTS audio response, for example: OK, it’s turned on.

[0081] That is, the interaction duration is the interval from the time when the user finishes saying "turn on the light" to the time when the TTS voice "OK, it's turned on" starts to be broadcasted back.

[0082] The optional embodiment of the present application significantly optimizes the voice VAD process in the smart home scenario. Specifically, the optional embodiment of the present application uses the reported audio data to accurately identify different user members in the family through voiceprint recognition technology, and customizes a suitable silent time window for each user accordingly. Furthermore, by carefully analyzing the sentence intervals of the user audio and combining the prediction model of the weighted moving average algorithm, the optional embodiment of the present application can dynamically adjust and output a new, more accurate silent time window to adapt to the unique speaking habits of each user. This optimization not only improves the accuracy and efficiency of voice interaction, but also brings a more personalized and smooth voice interaction experience to smart home users. Specifically: Figure 5 is a schematic diagram of a device corresponding to a method for improving the success rate of speech recognition and shortening the duration of speech interaction according to an optional embodiment of the present application, such as Figure 5 As shown, it specifically includes the following modules:

[0083] (1) Silent time window selection module.

[0084] The core function of the silent time window selection module is to accurately identify the user's voice and use advanced voiceprint recognition technology to identify which family member is currently speaking. Once the user's identity is identified, the system will quickly retrieve and obtain the member's personalized silent time window value. The personalized silent time window value is then sent to the terminal device as a key parameter for VAD (voice activity detection) to ensure the accuracy and integrity of voice interaction.

[0085] (2) Sentence interval data collection module.

[0086] The sentence interval data collection module focuses on collecting and analyzing all the sentence interval data of users during voice interaction. The sentence interval data collection module records the speech sentence intervals in each interaction in detail and forms a collection. By conducting in-depth analysis of this collection, the system can accurately capture the user's sentence interval habits and extract the maximum sentence interval time. This key data is then sent to the cloud to provide strong support for subsequent model predictions.

[0087] (3) Adaptive model prediction module.

[0088] The adaptive model prediction module is the core intelligent component. It uses the weighted moving average algorithm to perform in-depth analysis and prediction based on the current user data. Through continuous learning and adaptation, the system can output a personalized silent time window value suitable for the next interaction. In addition, the adaptive model prediction module can also make dynamic adjustments based on possible changes in the user, such as changes in sentence segmentation habits, to ensure the real-time and accuracy of voice interaction. This adaptive capability enables the system to continuously provide users with a more intelligent and convenient home experience.

[0089] Based on the above three modules, a specific process of improving the success rate of speech recognition and shortening the duration of speech interaction can be determined. Figure 6 is a flow chart of a method for improving the success rate of speech recognition and shortening the duration of speech interaction according to an optional embodiment of the present application. Figure 6 As shown:

[0090] Step S601, endpoint detection.

[0091] Receive the denoised audio data and perform VAD endpoint detection.

[0092] Step S602, determining whether voice appears, voice disappears or is silent.

[0093] When it is determined that the voice disappears (i.e., a voice disappearance event), execute step S603-1; when it is determined that the voice appears (i.e., a voice appearance event), execute step S603-2; when it is determined that there is silence (a silence event), execute step S603-3.

[0094] Step S603-1, record time T2 (ie, the first time).

[0095] If the event of the user's speech disappearing is detected, it means that the current user's speech is interrupted or the conversation is about to end, and the time T2 at this time is recorded. Endpoint detection continues.

[0096] Step S603-2, determine whether there is a voice disappearance event before.

[0097] If it is determined that there has been a voice disappearance event before, step S604-1 is executed; if it is determined that there has not been a voice disappearance event before, step S604-2 is executed.

[0098] Step S604-1, record time T1.

[0099] When the start of speaking is detected, if there has been a voice disappearance event in the previous period, the difference between the current time T1 and the time T2 when the voice disappears will be recorded, and the audio data will be sent to the cloud.

[0100] Step S605: update the sentence segmentation interval time set according to the current sentence segmentation interval time.

[0101] Wherein, the current segmentation interval time td n = T1 - T2. The segmentation interval time set TD = {td 1, td 2, ..., tdn}.

[0102] Step S604-2, sending data to the cloud.

[0103] Step S606: distinguish users through voiceprint recognition.

[0104] Step S607: Obtain the silent time window ST.

[0105] After receiving the audio data reported by the smart device, the cloud will perform voiceprint recognition to determine which member of the family the current user is, obtain the silent time window ST (i.e., target silent time window) to which the member belongs, and send it to the smart device.

[0106] Step S603-3, record time T3.

[0107] If no speech is detected continuously and the duration from the end of the session exceeds the silence time window, the current user is considered to have finished speaking. Stop sending voice data to the cloud, and record the time T3 when voice data is stopped.

[0108] Step S608, calculating the interval ΔT=T3-T2 (ie, the second segmentation interval time).

[0109] Step S609: determine whether ΔT is within the silent time window ST.

[0110] If it is determined that the time window is within the silent time window, step S601 is executed; if it is determined that the time window is not within the silent time window, step S610 is executed.

[0111] Step S610, determining whether ΔT<ST+CT is satisfied.

[0112] When ΔT<ST (ie, the target silent time window)+CT (ie, the preset extension time) is satisfied, step S601 is executed; when ΔT<ST+CT is not satisfied, step S611 is executed.

[0113] After the silence time window is exceeded, the delay time CT is increased to perform endpoint detection. If the delay time exceeds the CT, the sound pickup is terminated and the maximum value of the sentence interval time set of this conversation is sent to the cloud.

[0114] Step S611, end sound pickup.

[0115] Step S612: Send the maximum value of the TD set to the cloud.

[0116] Step S613: update the data into the model (ie, the target model).

[0117] Step S614: weighted moving average algorithm prediction model calculation.

[0118] Step S615: output a new silent time window ST1.

[0119] The cloud adds the data to the prediction model and calculates the silent time window suitable for the user through the weighted moving average algorithm for next use.

[0120] In the optional embodiment of the present application, when optimizing the silent time window algorithm model, a weighted moving average algorithm can be selected. This algorithm fully considers the time value of the data, that is, the data at different time points have different influences on the predicted value. By assigning different weights to the data at each time point, the user's silent time window can be predicted more accurately, thereby improving the efficiency and accuracy of voice interaction. The algorithm calculation formula is as follows:

[0121] M t =a1Y1+a2Y2+...+a t Y t .

[0122] Among them, M t is the moving average of the tth period (i.e., the silent time window); Y t is the observed data of the tth period; t is the moving step length; a t is the weight of the tth period. Y1 and Y2 are the observed data of the 1st and 2nd periods; a1 and a2 are the moving averages of the 1st and 2nd periods. t That is, the weight corresponding to each voice interaction data;

[0123] Specifically, the core of the weighted moving average algorithm is to assign different weights to the data in the moving segment. In the optional embodiment of the present application, the data of the last 100 periods are selected as the moving step, which means that the algorithm will consider the sentence interval data of the user's last 100 voice interactions. For these 100 periods of data, different moving weights a are assigned. t , where t represents the period of data. The design of these weights follows a basic principle: the more recent the data, the greater the impact on the predicted value. Therefore, more weights are assigned to recent data, while smaller weights are assigned to more distant data.

[0124] This method can effectively capture changes in user voice interaction habits. For example, if a user has shown a long interval between sentences in recent interactions, the weighted moving average algorithm will quickly increase the time, thereby outputting a longer silent time window. Conversely, if the user's interval between sentences gradually shortens, the algorithm will also adjust the time accordingly and output a shorter silent time window.

[0125] This method of dynamically adjusting the silent time window not only improves the response speed of voice interaction, but also improves the user experience. Users can interact with smart home devices more naturally without having to wait for too long a silent time window to confirm voice commands. At the same time, because the algorithm can learn users' voice habits in real time, it can adapt to the personalized needs of different users and provide users with more accurate and efficient voice services.

[0126] According to an optional embodiment of the present application, first, the optional embodiment of the present application adopts audio data pickup technology to capture the user's voice instructions in real time and report the data to the cloud for processing. This step ensures the accuracy and integrity of the voice data and provides a solid foundation for subsequent analysis and recognition. Then, through voiceprint recognition technology, the system can accurately distinguish different user members in the family. This function not only provides a personalized service experience, but also enhances the security and privacy protection capabilities of the system. Because different user members may have different voice habits and preferences, voiceprint recognition technology can ensure that the system provides the most suitable silent time window setting for each user member. Then, the system uses advanced endpoint detection technology to detect the user's audio input and calculate the interval time between sentence breaks. This step is crucial for accurately identifying the user's voice instructions. By continuously collecting and analyzing the user's sentence break data, the system can deeply understand the user's voice habits and provide strong support for subsequent prediction models. Finally, using the weighted moving average algorithm prediction model, the system can dynamically adjust the silent time window setting for each user member based on historical data and current data. This prediction model can capture changes in user voice habits in real time and adjust the length of the silent time window accordingly, thereby optimizing the efficiency and accuracy of voice interaction.

[0127] That is, the optional embodiments of the present application can improve the interaction success rate: through voiceprint recognition and endpoint detection technology, the system can more accurately identify the user's voice commands, reduce misjudgments and missed judgments, and thus improve the interaction success rate of abnormal interaction groups. It can also shorten the interaction time: through the weighted moving average algorithm prediction model, the system can output the most suitable silent time window setting for the user, reduce unnecessary waiting time, and shorten the overall interaction time of the normal interaction group. It can also provide a personalized service experience: the system can provide a personalized service experience based on the voice habits and preferences of different user members to meet the needs and expectations of different users.

[0128] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods of each embodiment of the present application.

[0129] Figure 7 is a structural block diagram of a silent time window prediction device according to an embodiment of the present application; Figure 7 As shown, including:

[0130] An acquisition module 72, configured to acquire historical voice interaction data of a target object, wherein the historical voice interaction data is used to indicate the interactive voice between the target object and a voice device;

[0131] A first determination module 74, configured to determine a first time difference between a voice interaction time corresponding to the historical voice interaction data and a current time, and to determine a first sentence segmentation interval time corresponding to the historical voice interaction data;

[0132] A second determination module 76, configured to determine a target weight corresponding to the historical voice interaction data according to the first time difference;

[0133] The prediction module 78 is used to predict the target silence time window for the interaction between the target object and the voice device according to the target weight and the first sentence interval time.

[0134] Through the above-mentioned device, the historical voice interaction data between the target object and the voice device is obtained, the first time difference between the historical voice interaction data and the corresponding voice interaction time and the current time is determined, and the first time interval of the sentence segmentation time corresponding to the historical voice interaction data is determined, and the target weight corresponding to the historical voice interaction data is determined according to the first time difference; the target silent time window for the interaction between the target object and the voice device is predicted according to the target weight and the first sentence segmentation time interval. That is to say, the embodiment of the present application determines the target weight according to the first time difference between the voice interaction time corresponding to the historical voice interaction data and the current time, and then predicts the target silent time window according to the target weight and the first sentence segmentation time interval. According to the embodiment of the present application, the problem of increasing the overall voice interaction time caused by selecting a longer silent time window to adapt to the different sentence segmentation habits of different user groups when speaking in the related art can be solved. Then different silent time windows are determined to adapt to the different sentence segmentation habits of different user groups, reducing the overall voice interaction time.

[0135] In an exemplary embodiment, the second determination module 76 is also used to obtain multiple voice interaction data corresponding to the target object, and determine the second time difference between the voice interaction time corresponding to each voice interaction data and the current time; divide the multiple voice interaction data and the second time difference corresponding to each voice interaction data into training set data and test set data, and obtain an initial attenuation coefficient; train the target model according to the training set data and the initial attenuation coefficient to obtain a trained target model, wherein the target model is used to determine the weight corresponding to each voice interaction data; verify the trained target model according to the test set data; when it is determined that the trained target model has passed the verification, input the first time difference into the trained target model so that the trained training model outputs the target weight.

[0136] In an exemplary embodiment, the first determination module 74 is further used to determine whether there is a first voice interruption event in the historical voice interaction data, wherein the first voice interruption event includes: a voice disappearance event or a silence event; when it is determined that the first voice interruption event exists in the historical voice interaction data, determine a first time when the first voice interruption event starts, and determine whether there is a voice occurrence event after the first voice interruption event; when it is determined that there is a voice occurrence event after the first voice interruption event, determine a second time when the voice occurrence event starts, and determine the first sentence interval time according to the difference between the second time and the first time; when it is determined that there is no voice occurrence event after the first voice interruption event, determine a third time when the historical voice interaction data ends, and determine the first sentence interval time according to the difference between the third time and the first time.

[0137] In an exemplary embodiment, the first determination module 74 is further used to determine whether the voice occurrence event exists within a target time period after the first time when it is determined that the first voice interruption event exists in the historical voice interaction data; determine that the voice interruption event is the voice disappearance event when it is determined that the voice occurrence event exists within the target time period; and determine that the voice interruption event is the silence event when it is determined that the voice occurrence event does not exist within the target time period.

[0138] In an exemplary embodiment, the prediction module 78 is further used to determine whether there is a second voice interruption event in the first interactive voice when it is determined that the voice device receives the first interactive voice emitted by the target object; determine whether the second voice interruption event is a silent event when it is determined that the first interactive voice exists in the first interactive voice; determine the sum of the target silent time window and the preset extension time when it is determined that the second voice interruption event is the silent event, and determine a first size relationship between the second sentence interval time corresponding to the first interactive voice and the sum; when the first size relationship indicates that the second sentence interval time is greater than or equal to the sum, end the pickup of the first interactive voice, and determine a first weight corresponding to the first interactive voice after the pickup is ended; predict the first silent time window of the interaction between the target object and the voice device according to the first weight and the second sentence interval time, and update the target silent time window according to the first silent time window.

[0139] In an exemplary embodiment, the prediction module 78 is also used to perform endpoint detection on the first interactive speech to obtain a speech occurrence event after the second sentence interval when the first size relationship indicates that the second sentence interval time is less than the sum value; determine whether the silent event exists in the second interactive speech corresponding to the speech occurrence event; determine the second size relationship between the third sentence interval time corresponding to the second interactive speech and the sum value when it is determined that the silent event exists in the second interactive speech; end the pickup of the second interactive speech when the second size relationship indicates that the third sentence interval time is greater than or equal to the sum value, and merge the first interactive speech and the second interactive speech to update the first interactive speech; determine the second weight corresponding to the updated first interactive speech; predict the second silent time window for the interaction between the target object and the voice device based on the second weight, the second sentence interval time and the third sentence interval time, and update the target silent time window based on the second silent time window.

[0140] In an exemplary embodiment, the prediction module 78 is also used to predict a second silent time window for interaction between the target object and the voice device based on the second weight, the second sentence interval time and the third sentence interval time, including: determining a third size relationship between the second sentence interval time and the third sentence interval time; determining the maximum sentence interval time between the second sentence interval time and the third sentence interval time based on the third size relationship; and determining the second silent time window based on the second weight and the maximum sentence interval time.

[0141] In an exemplary embodiment, the prediction module 78 is also used to determine whether the sound object corresponding to the third interactive voice is the target object when it is determined that the voice device has received a third interactive voice; when it is determined that the sound object is not the target object, determine whether there is a third silent time window in the silent time window library for the interaction between the sound object and the voice device, wherein the silent time window library includes: a target silent time window corresponding to the target object; when it is determined that the third silent time window does not exist in the silent time window library, determine the third weight corresponding to the third interactive voice and the fourth sentence interval time corresponding to the third interactive voice; predict the third silent time window based on the third weight and the fourth sentence interval time, and store the third silent time window corresponding to the sound object in the silent time window library.

[0142] An embodiment of the present application further provides a storage medium, which includes a stored program, wherein the program executes any of the above methods when it is run.

[0143] Optionally, in this embodiment, the storage medium may be configured to store program codes for executing the following steps:

[0144] S1, obtaining historical voice interaction data of a target object, wherein the historical voice interaction data is used to indicate the interactive voice between the target object and a voice device;

[0145] S2, determining a first time difference between a voice interaction time corresponding to the historical voice interaction data and the current time, and determining a first sentence segmentation interval time corresponding to the historical voice interaction data;

[0146] S3, determining a target weight corresponding to the historical voice interaction data according to the first time difference;

[0147] S4: predicting a target silent time window for interaction between the target object and the voice device according to the target weight and the first sentence interval time.

[0148] An embodiment of the present application further provides an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0149] Optionally, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0150] Optionally, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0151] S1, obtaining historical voice interaction data of a target object, wherein the historical voice interaction data is used to indicate the interactive voice between the target object and a voice device;

[0152] S2, determining a first time difference between a voice interaction time corresponding to the historical voice interaction data and the current time, and determining a first sentence segmentation interval time corresponding to the historical voice interaction data;

[0153] S3, determining a target weight corresponding to the historical voice interaction data according to the first time difference;

[0154] S4: predicting a target silent time window for interaction between the target object and the voice device according to the target weight and the first sentence interval time.

[0155] An embodiment of the present application further provides a computer program product, including a computer program, which is used by a processor to execute the steps in any of the above method embodiments.

[0156] Optionally, in this embodiment, the computer program product may be executed by a processor to perform the following steps:

[0157] S1, obtaining historical voice interaction data of a target object, wherein the historical voice interaction data is used to indicate the interactive voice between the target object and a voice device;

[0158] S2, determining a first time difference between a voice interaction time corresponding to the historical voice interaction data and the current time, and determining a first sentence segmentation interval time corresponding to the historical voice interaction data;

[0159] S3, determining a target weight corresponding to the historical voice interaction data according to the first time difference;

[0160] S4: predicting a target silent time window for interaction between the target object and the voice device according to the target weight and the first sentence interval time.

[0161] Optionally, in this embodiment, the above-mentioned storage medium may include but is not limited to: a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk, and other media that can store program codes.

[0162] Optionally, the specific examples in this embodiment may refer to the examples described in the above embodiments and optional implementation modes, and this embodiment will not be described in detail here.

[0163] Obviously, those skilled in the art should understand that the above modules or steps of the present application can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, and optionally, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order from that herein, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present application is not limited to any specific combination of hardware and software.

[0164] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A method for predicting a silent time window, characterized in that: include: Acquire historical voice interaction data of a target object, wherein the historical voice interaction data is used to indicate interactive voice between the target object and a voice device; Determine a first time difference between a voice interaction time corresponding to the historical voice interaction data and a current time, and determine a first sentence segmentation interval time corresponding to the historical voice interaction data; Determining a target weight corresponding to the historical voice interaction data according to the first time difference; A target silent time window for interaction between the target object and the voice device is predicted according to the target weight and the first sentence interval time.

2. The method for predicting a silent time window according to claim 1, characterized in that: Determining a target weight corresponding to the historical voice interaction data according to the first time difference includes: Acquire multiple voice interaction data corresponding to the target object, and determine a second time difference between the voice interaction time corresponding to each voice interaction data and the current time; Divide the plurality of voice interaction data and the second time difference corresponding to each voice interaction data into training set data and test set data, and obtain an initial attenuation coefficient; Training the target model according to the training set data and the initial attenuation coefficient to obtain a trained target model, wherein the target model is used to determine the weight corresponding to each voice interaction data; Verifying the trained target model according to the test set data; When it is determined that the trained target model passes the verification, the first time difference is input into the trained target model so that the trained training model outputs the target weight.

3. The method for predicting a silent time window according to claim 1, characterized in that: Determining a first sentence segmentation interval corresponding to the historical voice interaction data includes: Determine whether there is a first voice interruption event in the historical voice interaction data, wherein the first voice interruption event includes: a voice disappearance event or a silence event; In the case of determining that the first voice interruption event exists in the historical voice interaction data, determining a first time when the first voice interruption event starts, and determining whether a voice occurrence event occurs after the first voice interruption event; In the case where it is determined that the voice occurrence event exists after the first voice interruption event, determining a second time when the voice occurrence event starts, and determining the first sentence segmentation interval time according to a difference between the second time and the first time; When it is determined that there is no voice occurrence event after the first voice interruption event, a third time when the historical voice interaction data ends is determined, and the first sentence segmentation interval time is determined according to the difference between the third time and the first time.

4. The method for predicting a silent time window according to claim 3, characterized in that: After determining whether there is a first voice interruption event in the historical voice interaction data, the method further includes: In the case where it is determined that the first voice interruption event exists in the historical voice interaction data, determining whether the voice occurrence event exists within a target time period after the first time; In the case where it is determined that the voice occurrence event exists in the target time period, determining the voice interruption event as the voice disappearance event; When it is determined that the voice occurrence event does not exist within the target time period, the voice interruption event is determined to be the silence event.

5. The method for predicting a silent time window according to claim 1, characterized in that: After predicting the target silence time window for the interaction between the target object and the voice device according to the target weight and the first sentence interval time, the method further includes: In a case where it is determined that the voice device receives a first interactive voice issued by the target object, determining whether there is a second voice interruption event in the first interactive voice; In the case of determining that the second voice interruption event exists in the first interactive voice, determining whether the second voice interruption event is a silent event; In the case where it is determined that the second voice interruption event is the silence event, determining a sum of the target silence time window and the preset extension time, and determining a first magnitude relationship between a second sentence interval time corresponding to the first interactive voice and the sum; When the first size relationship indicates that the second segmentation interval is greater than or equal to the sum value, ending the sound pickup of the first interactive speech, and determining a first weight corresponding to the first interactive speech after the sound pickup is ended; A first silent time window for interaction between the target object and the voice device is predicted according to the first weight and the second sentence interval time, and the target silent time window is updated according to the first silent time window.

6. The method for predicting a silent time window according to claim 5, characterized in that: After determining a first magnitude relationship between a second sentence segmentation interval time corresponding to the first interactive speech and the sum value, the method further includes: When the first size relationship indicates that the second segmentation interval is less than the sum value, performing endpoint detection on the first interactive speech to obtain a speech occurrence event after the second segmentation interval; Determining whether the silence event exists in the second interactive speech corresponding to the speech occurrence event; In the case where it is determined that the silence event exists in the second interactive speech, determining a second magnitude relationship between a third sentence segmentation interval time corresponding to the second interactive speech and the sum value; When the second size relationship indicates that the third segmentation interval is greater than or equal to the sum value, ending the pickup of the second interactive voice, and merging the first interactive voice with the second interactive voice to update the first interactive voice; Determining a second weight corresponding to the updated first interactive voice; A second silent time window for interaction between the target object and the voice device is predicted according to the second weight, the second sentence interval time and the third sentence interval time, and the target silent time window is updated according to the second silent time window.

7. The method for predicting a silent time window according to claim 6, characterized in that: Predicting a second silent time window for interaction between the target object and the voice device according to the second weight, the second sentence interval time and the third sentence interval time includes: Determine a third magnitude relationship between the second segmentation interval time and the third segmentation interval time; Determine the maximum interval between the second interval and the third interval according to the third size relationship; The second silence time window is determined according to the second weight and the maximum sentence segmentation interval.

8. The method for predicting a silent time window according to claim 1, characterized in that: After predicting the target silence time window for the interaction between the target object and the voice device according to the target weight and the first sentence interval time, the method further includes: In a case where it is determined that the voice device receives a third interactive voice, determining whether a sounding object corresponding to the third interactive voice is the target object; In the case where it is determined that the sounding object is not the target object, determining whether there is a third silent time window in which the sounding object interacts with the voice device in the silent time window library, wherein the silent time window library includes: a target silent time window corresponding to the target object; In the case of determining that the third silent time window does not exist in the silent time window library, determining a third weight corresponding to the third interactive speech and a fourth sentence interval time corresponding to the third interactive speech; The third silent time window is predicted according to the third weight and the fourth sentence interval time, and the third silent time window corresponding to the sound object is stored in the silent time window library.

9. A prediction device for a silent time window, characterized in that: include: An acquisition module, used to acquire historical voice interaction data of a target object, wherein the historical voice interaction data is used to indicate the interactive voice between the target object and a voice device; A first determination module, used to determine a first time difference between a voice interaction time corresponding to the historical voice interaction data and a current time, and to determine a first sentence segmentation interval time corresponding to the historical voice interaction data; A second determination module, configured to determine a target weight corresponding to the historical voice interaction data according to the first time difference; A prediction module is used to predict a target silent time window for the interaction between the target object and the voice device according to the target weight and the first sentence interval time.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein the program executes the method described in any one of claims 1 to 8 when executed.

11. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to execute the method according to any one of claims 1 to 8 through the computer program.

Images