Null broadcast detection method and device, equipment and storage medium

Abstract

The invention discloses an idle broadcast detection method and device, equipment and a storage medium, and belongs to the field of network live broadcast. The method comprises the steps of obtaining live broadcast data of a first live broadcast room; determining at least one classification result of the live broadcast data, wherein the at least one classification result is used for indicating one or more of whether a voiceprint of the first live broadcast room is matched, whether a picture contains a human body, whether a human body posture is static, and whether an anchor executes interaction operation; and determining an identification result based on the at least one classification result, wherein the identification result is used for indicating whether the first live broadcast room is idle. The method can improve the idle broadcast detection efficiency.

Description

Technical Field

[0001] The present application relates to the field of network live broadcasting, and in particular to a method, device, equipment and storage medium for detecting empty broadcasts. Background Art

[0002] Empty broadcast means that in the live broadcast room, the host does not conduct real-time live broadcast activities, or the host is not in front of the camera for a long time and does not interact with the audience.

[0003] In the related art, in order to reduce the empty broadcast rate of the live broadcast room, the live broadcast room will be manually reviewed. The reviewer watches the live broadcast video of the live broadcast room. When the anchor is not in the live broadcast room for a long time, the reviewer will determine that the live broadcast room is empty. For the empty live broadcast room, the system will punish it to ensure the live broadcast standard.

[0004] Due to the large number of live broadcast rooms, the workload of manual review is extremely large, and the review cost is high and the efficiency is low. Summary of the invention

[0005] The present application provides a method, device, equipment and storage medium for detecting empty broadcasts, which can improve the efficiency of empty broadcast detection. The technical solution is as follows:

[0006] According to one aspect of the present application, a method for detecting empty broadcast is provided, the method comprising:

[0007] Get the live broadcast data of the first live broadcast room;

[0008] Determine at least one classification result of the live broadcast data, where the at least one classification result is used to indicate one or more of whether the voiceprint of the first live broadcast room matches, whether the picture contains a human body, whether the human body posture is still, and whether the anchor performs an interactive operation;

[0009] An identification result is determined based on the at least one classification result, and the identification result is used to indicate whether the first live broadcast room is empty.

[0010] According to another aspect of the present application, a device for detecting empty broadcast is provided, the device comprising:

[0011] An acquisition module, used to acquire live broadcast data of the first live broadcast room;

[0012] a classification module, used to determine at least one classification result of the live broadcast data, wherein the at least one classification result is used to indicate one or more of whether the voiceprint of the first live broadcast room matches, whether the picture contains a human body, whether the human body posture is still, and whether the anchor performs an interactive operation;

[0013] An identification module is used to determine an identification result based on the at least one classification result, and the identification result is used to indicate whether the first live broadcast room is empty.

[0014] According to another aspect of the present application, a computer device is provided, comprising a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to implement the air broadcast detection method as described above.

[0015] According to another aspect of the present application, a computer-readable storage medium is provided, in which at least one instruction, at least one program, a code set or an instruction set is stored, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by a processor to implement the empty broadcast detection method as described above.

[0016] According to another aspect of the present application, a computer program product or a computer program is provided, the computer program product or the computer program including computer instructions, the computer instructions being stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the empty broadcast detection method provided in various optional implementations of the above aspects.

[0017] The beneficial effects of the technical solution provided by this application include at least:

[0018] Identify empty broadcasts based on the live broadcast data of the live broadcast room, for example, identify the host's voiceprint based on the live broadcast audio, identify the host's body or host's actions based on the live broadcast screen, and identify the host's interactive operations based on the host's operation data. Based on the data of at least one mode in the live broadcast room, screen the live broadcast room that may be empty from at least one dimension. This method can quickly screen the empty broadcast rooms from a large number of live broadcast rooms, reduce the cost of empty broadcast detection, and improve the efficiency of empty broadcast detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0020] Figure 1 is a schematic diagram of the structure of a computer system provided by an exemplary embodiment of the present application;

[0021] Figure 2 is a flow chart of a method for detecting empty broadcasts provided by an exemplary embodiment of the present application;

[0022] Figure 3 is a schematic diagram of a method for detecting empty broadcasts provided by an exemplary embodiment of the present application;

[0023] Figure 4 is a schematic diagram of a method for detecting empty broadcasts provided by an exemplary embodiment of the present application;

[0024] Figure 5 is a structural schematic diagram of an empty broadcast detection device provided by an exemplary embodiment of the present application;

[0025] Figure 6 It is a structural diagram of a computer device provided by an exemplary embodiment of the present application.

[0026] 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. DETAILED DESCRIPTION

[0027] In order to make the objectives, technical solutions and advantages of the present application clearer, the implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.

[0028] Figure 1 A schematic diagram of a computer system provided by an exemplary embodiment of the present application is shown, and the computer system may include a terminal device 101 and a server 103 .

[0029] Exemplarily, the empty broadcast detection method shown in the embodiment of the present application can be applied in a terminal device or a server. The terminal device 101 runs an application 102 with a live broadcast function. The terminal device may include a mobile phone, a tablet computer, a notebook computer, a laptop computer, a desktop computer, an all-in-one computer, an Internet of Things device, an intelligent robot workstation, a television, a set-top box, smart glasses, a smart watch, a digital camera, an MP4 player, an MP5 player, a learning machine, a point-reading machine, an electronic paper book, an electronic dictionary, a vehicle-mounted device, a virtual reality (VR) player or an augmented reality (AR) player, etc.

[0030] Exemplarily, the empty broadcast detection method provided in the present application can be executed by a client in a terminal device. The client is a client of an application that supports a live broadcast function. For example, the application may include at least one of the following: an audio application, a video application, a short video application, a social application, a shopping application, a local service application, a live broadcast application, an information application, a browser, and a game application.

[0031] Exemplarily, the client running in the terminal device may be an anchor client, and the anchor account of the first live broadcast room is logged in to the anchor client. The anchor client may use the method provided in the embodiment of the present application to locally detect whether the first live broadcast room is in an empty broadcast state. If the first live broadcast room is in an empty broadcast state, the anchor client reports to the server that the first live broadcast room is empty.

[0032] The terminal device 101 includes a first memory and a first processor. An empty broadcast detection program is stored in the first memory; the above-mentioned empty broadcast detection program is called and executed by the first processor to implement the empty broadcast detection method provided in the present application. The first memory may include but is not limited to the following: Random Access Memory (RAM), Read Only Memory (ROM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), and Electrically Erasable Read-Only Memory (EEPROM).

[0033] The first processor may be composed of one or more integrated circuit chips. Optionally, the first processor may be a general-purpose processor, such as a central processing unit (CPU) or a network processor (NP). Optionally, the first processor may implement the empty broadcast detection method provided in the present application by running a program or code.

[0034] In an optional embodiment, the terminal device 101 and the server 103 may be connected to each other via a wired or wireless network.

[0035] Exemplarily, the empty broadcast detection method provided in the present application can be executed by a server in the server. The server 103 is used to provide background services for the application of the terminal device 101. Optionally, the server 103 undertakes the main computing work, and the terminal device 101 undertakes the secondary computing work; or, the server 103 undertakes the secondary computing work, and the terminal device 101 undertakes the main computing work; or, the server 103 and the terminal device 101 adopt a distributed computing architecture for collaborative computing.

[0036] The server 103 may be a single server, or a server cluster consisting of multiple servers, or a cloud computing service center.

[0037] Optionally, the server 103 includes a second memory and a second processor. The second memory stores an empty broadcast detection program; the empty broadcast detection program is called by the second processor to implement the empty broadcast detection method provided in the present application. Optionally, the second memory may include but is not limited to the following: RAM, ROM, PROM, EPROM, EEPROM. Optionally, the second processor may be a general-purpose processor, such as a CPU or NP.

[0038] Exemplarily, the server may use the method provided in the embodiment of the present application to detect whether each live broadcast room in the application is in an idle broadcast state.

[0039] Figure 2 FIG. 1 is a flow chart of a method for detecting empty broadcasts provided by an exemplary embodiment of the present application. The method can be used for Figure 1 The terminal device or server shown in the figure. The method comprises the following steps.

[0040] Step 210: Obtain live broadcast data of the first live broadcast room.

[0041] Exemplarily, the method may be executed by a computer device, which may be a terminal device or a server. For example, the method may be executed by a server of a live broadcast application running in a server. And / or, the method may be executed by a host client of a live broadcast application running in a terminal device, and a host account is logged in to the host client. And / or, the method may be executed by a viewer client of a live broadcast application running in a terminal device, and an account is logged in to the viewer client.

[0042] In an optional embodiment, the method is executed by a server of a live broadcast application, and the server uses the empty broadcast detection method to detect the live broadcast status of each live broadcast room in the live broadcast application, and performs empty broadcast detection quickly and efficiently.

[0043] Alternatively, the method is executed by an anchor client of a live broadcast application program, and the anchor client uses the empty broadcast detection method to detect the live broadcast status of the live broadcast room corresponding to the anchor client, and reports to the server when the live broadcast room is empty.

[0044] Alternatively, the method is executed by a viewer client of a live broadcast application, and the viewer client uses the empty broadcast detection method to detect the live broadcast status of the live broadcast room currently being watched, and reports to the server when the live broadcast room is empty. Alternatively, the viewer client uses the empty broadcast detection method to detect the live broadcast status of the live broadcast room indicated by the server, and feeds back the empty broadcast detection result of the live broadcast room to the server when the live broadcast room is empty.

[0045] Taking the first live broadcast room as an example, after the first live broadcast room starts broadcasting, the computer device obtains the real-time live broadcast data of the first live broadcast room, and determines whether the first live broadcast room is currently in an idle broadcast state based on the real-time live broadcast data of the first live broadcast room.

[0046] Exemplarily, the live broadcast data is real-time data after the first live broadcast room starts broadcasting. Based on the real-time data, it is determined whether the first live broadcast room is currently in a state of no broadcasting. Alternatively, the live broadcast data may be historical live broadcast data of the first live broadcast room. Based on the historical live broadcast data, it is determined whether the first live broadcast room has no broadcasting behavior in historical live broadcast sessions.

[0047] The live broadcast data may be data within a period of time. The computer device may periodically obtain the live broadcast data of the first live broadcast room, and determine whether the first live broadcast room is in an idle broadcast state based on the live broadcast data in the current period. Alternatively, the computer device may perform an idle broadcast detection at a preset time interval, and each idle broadcast detection obtains the live broadcast data of the first live broadcast room within a period of time in the past according to a preset duration, and determines whether the first live broadcast room is in an idle broadcast state based on the live broadcast data.

[0048] Live broadcast data includes at least one of the following types of data: live broadcast audio, live broadcast picture, live broadcast video, and anchor-side operation data.

[0049] The anchor-side operation data is used to record the user operations received by the anchor client. For example, the anchor-side operation data can be used to record at least one of the following operations: live broadcast room management operations, product setting operations, activity editing operations, picture setting operations, sound setting operations, etc. received by the anchor client. The anchor-side operation data can also include live broadcast room data of the first live broadcast room, such as the chat text in the chat area of ​​the first live broadcast room, the audience information of the first live broadcast room, and the audience interaction information of the first live broadcast room (for example, interaction information of likes, favorites, attention, purchases, comments, microphones, etc.).

[0050] When the method is executed by the server, the live broadcast data can be obtained by the server reading locally. The live broadcast data can also be reported by the server receiving the anchor client; the live broadcast data can also be reported by the server receiving the audience client. When the method is executed by the anchor client, the live broadcast data can be obtained locally by the anchor client or from the server. When the method is executed by the audience client, the live broadcast data can be obtained locally by the audience client or from the server. The embodiment of the present application only takes the method executed by the server as an example for explanation, and the live broadcast data is data stored locally on the server, or the live broadcast data is reported in real time by the anchor client.

[0051] In an optional embodiment, when the live broadcast data includes at least two types of data, the computer device can simultaneously acquire the at least two types of data and perform empty broadcast detection based on the at least two types of data.

[0052] The computer device may also sequentially obtain at least two categories of data, and when the classification result of the previous category of data corresponds to a blank broadcast, obtain the next category of data, identify the classification result of the next category of data, until all classification results correspond to blank broadcasts, and determine the identification result of the first live broadcast room as a blank broadcast; when the classification result of the previous category of data corresponds to a normal live broadcast, stop obtaining the next category of data and determine the identification result of the first live broadcast room as a normal live broadcast.

[0053] Among them, the order of obtaining at least two types of data can be determined according to the size of the data, and the live broadcast data type with a smaller data volume is obtained first, and the live broadcast data type with a larger data volume is obtained last, so as to quickly identify the empty broadcast room based on a small amount of data and improve the efficiency of empty broadcast detection. For example, the anchor end operation data is obtained first, and when it is determined based on the anchor segment operation data that the anchor has not performed an interactive operation; the live broadcast audio is obtained, and based on the live broadcast audio, it is determined whether the real-time voiceprint matches the anchor voiceprint; if it does not match, the live broadcast picture is continued to be obtained, and the human body posture is identified based on the live broadcast picture; if there is no human body posture, it is determined that the first live broadcast room is in an empty broadcast state; if there is a human body posture, the live broadcast video is continued to be obtained to identify whether the human body posture in at least two frames of the live broadcast picture has performed an action. If no action is performed, it is determined that the first live broadcast room is in an empty broadcast state; if an action is performed, it is determined that the first live broadcast room is broadcasting normally.

[0054] Step 220: Determine at least one classification result of the live broadcast data, where the at least one classification result is used to indicate one or more of whether the voiceprint of the first live broadcast room matches, whether the picture contains a human body, whether the human body posture is still, and whether the anchor performs an interactive operation.

[0055] Exemplarily, the computer device identifies the type of the live broadcast data and obtains the classification result of the live broadcast data in at least one evaluation dimension. For example, the evaluation dimension may include at least one of voiceprint, human body, action, and interaction. The computer device determines the live broadcast status of the first live broadcast room based on the classification result of the at least one evaluation dimension. The at least one evaluation dimension is used to identify whether the first live broadcast room is empty.

[0056] For example, in the voiceprint dimension, it is possible to identify whether the voiceprint in the live broadcast data matches the voiceprint of the host in the first live broadcast room. If they match, it means that the host is conducting voice interaction or explanation in the first live broadcast room, and the first live broadcast room is broadcasting normally; if they do not match, it means that the host is not speaking in the first live broadcast room, and the first live broadcast room may be in an empty broadcast state.

[0057] For example, in the human body dimension, it can be identified whether the live broadcast screen in the live broadcast data contains a human body image. If it contains a human body image, it means that there is a real person in the live broadcast screen of the first live broadcast room, and the first live broadcast room may be broadcasting normally; if it does not match, it means that there is no real person in the first live broadcast room, and the first live broadcast room may be in an empty broadcast state.

[0058] For example, in the motion dimension, it is possible to identify whether the human body in the live video is moving. If there is movement, it means that the first live broadcast room may be a live broadcast of real people, and the first live broadcast room may be broadcasting normally; if there is no movement, it means that the human body in the first live broadcast room may be a static image, and there is no live broadcast of real people, so the first live broadcast room may be in an empty broadcast state.

[0059] For example, in the interactive dimension, it is possible to identify whether the host in the first live broadcast room has performed an interactive operation, such as the host client receiving a user operation, and / or the host client receiving an operation in which the host interacts with the audience (for example, commenting, starting a live broadcast room activity, putting a product on the shelf, etc.). If there is an interactive operation, it means that there may be real-person control in the first live broadcast room, and the first live broadcast room may be broadcasting normally; if there is no action, it means that there may be no real-person operation in the first live broadcast room, and the first live broadcast room may be in an empty broadcast state.

[0060] In an optional embodiment, the live broadcast data is multimodal data. The method provided in the embodiment of the present application obtains classification results from at least two dimensions based on the multimodal data, and determines whether the first live broadcast room is in an empty broadcast state based on the classification results in at least two dimensions.

[0061] Step 230: Determine an identification result based on at least one classification result, where the identification result is used to indicate whether the first live broadcast room is empty.

[0062] Exemplarily, an empty broadcast means that there is no host in the live broadcast room, and / or the host is not displayed in the live broadcast screen, and / or the live broadcast audio does not include the host's voice, and / or the host does not interact with the audience in the live broadcast room, and / or the host does not perform any live broadcast behavior in the live broadcast room, etc.

[0063] Empty broadcast rooms lack interaction and real-time feedback with viewers, and viewers cannot watch valuable content from the rooms. The room loses viewers and becomes vacant (no host broadcasts and no viewers watch the live broadcast). Empty broadcast rooms occupy server load and waste server resources. Therefore, live broadcast applications need to review the live broadcast status of the rooms, warn or punish empty broadcast rooms, reduce the empty broadcast rate of the rooms, and improve server resource utilization.

[0064] Exemplarily, at least one classification result may include at least one of the following: a first classification result, a second classification result, a third classification result, and a fourth classification result. The first classification result is used to indicate whether the voiceprint of the first live broadcast room matches. The second classification result is used to indicate whether the live broadcast screen contains a human body. The third classification result is used to indicate whether at least two frames of the live broadcast screen contain human body movements. The fourth classification result is used to indicate the frequency of interactive operations.

[0065] The identification result may include one of an empty broadcast and a normal live broadcast. Alternatively, the identification result may include one of an empty broadcast and a non-empty broadcast.

[0066] Exemplarily, when one of the at least one classification result corresponds to a blank broadcast, the recognition result may be determined as a blank broadcast. Alternatively, when all of the at least one classification result correspond to a blank broadcast, the recognition result may be determined as a blank broadcast.

[0067] For example, the first classification result is that voiceprint matching corresponds to normal live broadcast, and the first classification result is that voiceprint mismatch corresponds to empty broadcast; the second classification result is that containing human body corresponds to normal live broadcast, and the second classification result is that not containing human body corresponds to empty broadcast; the third classification result is that containing human body movement corresponds to normal live broadcast, and the third classification result is that not containing human body movement corresponds to empty broadcast; the fourth classification result is that high frequency of interactive operation corresponds to normal live broadcast, and the fourth classification result is that low frequency of interactive operation corresponds to empty broadcast.

[0068] Exemplarily, when the identification result of the first live broadcast room is that the broadcast is not broadcast, the computer device can control the anchor client of the first live broadcast room to display a pop-up warning to remind the anchor to return to the live broadcast room as soon as possible to broadcast live; and / or control the anchor client to emit a warning tone to remind the anchor to return to the live broadcast room as soon as possible to broadcast live. Alternatively, when the identification result of the first live broadcast room is that the broadcast is not broadcast, the computer device can add the first live broadcast room to the pending review list, and the live broadcast rooms in the pending review list are manually reviewed to confirm whether they are not broadcast, so as to reduce the error rate of the empty broadcast review, reduce the workload of manual review of the empty broadcast of the live broadcast room, and improve the efficiency and accuracy of the empty broadcast review.

[0069] In summary, the method provided in this embodiment identifies empty broadcasts based on the live broadcast data of the live broadcast room, for example, identifying the host's voiceprint based on the live broadcast audio, identifying the host's body or host's actions based on the live broadcast screen, and identifying the host's interactive operations based on the host's operation data. Based on the data of at least one modality in the live broadcast room, the live broadcast room that may be empty is screened from at least one dimension. This method can quickly screen empty broadcast rooms from a large number of live broadcast rooms, reduce the cost of empty broadcast detection, and improve the efficiency of empty broadcast detection.

[0070] Figure 3 FIG. 1 is a flow chart of a method for detecting empty broadcasts provided by an exemplary embodiment of the present application. The method can be used for Figure 1 The terminal device or server shown. Figure 2 In the illustrated embodiment, step 230 includes steps 231 to 233 .

[0071] Step 210: Obtain live broadcast data of the first live broadcast room.

[0072] For example, after the first live broadcast room starts broadcasting, the server periodically obtains the live broadcast data of the first live broadcast room within the past preset time period, and the live broadcast data may include: live broadcast audio, live broadcast screen, live broadcast video, and anchor terminal operation data. The server obtains the recognition result based on the live broadcast data.

[0073] In an optional embodiment, the server can determine whether the first live broadcast room is idle based on the identification result of a period. When the identification result of the first live broadcast room in the current period is idle, it is determined that the first live broadcast room is idle; when the identification result of the first live broadcast room in the current period is normal live broadcast, it is determined that the first live broadcast room is normal live broadcast.

[0074] In another optional embodiment, the server may determine whether the first live broadcast room is empty based on the identification results of x consecutive cycles, where x is a preset positive integer. If the identification results of x consecutive cycles are all empty, the first live broadcast room is determined to be empty; if the identification result of one cycle in the x consecutive cycles is normal live broadcast, the first live broadcast room is determined to be normal live broadcast.

[0075] Optionally, the value of x can be determined based on the historical empty broadcast situation of the first live broadcast room. For a live broadcast room without a history of empty broadcasts, x can be an initial value (e.g., 10); when there is empty broadcast behavior in the history of the live broadcast room, x is inversely proportional to the number of empty broadcasts in the live broadcast room.

[0076] When it is determined that the first live broadcast room is empty, a pop-up warning can be displayed on the anchor client of the first live broadcast room to prompt the anchor to return to the live broadcast room. If an operation to close the pop-up warning is received, the live broadcast data is continued to be periodically obtained to detect whether the first live broadcast room is empty. If an operation to close the pop-up warning is not received within the first time period after the pop-up warning is displayed, the server closes the first live broadcast room, releases the resources occupied by the first live broadcast room, and improves the resource utilization of the server.

[0077] Step 220: Determine at least one classification result of the live broadcast data, where the at least one classification result is used to indicate one or more of whether the voiceprint of the first live broadcast room matches, whether the picture contains a human body, whether the human body posture is still, and whether the anchor performs an interactive operation.

[0078] Exemplarily, the live broadcast data may include at least one type of data, for example, including live broadcast audio, live broadcast screen, live broadcast video, anchor operation data, etc. One type of live broadcast data may obtain at least one classification result, and the classification result is used to indicate whether the feature information of empty broadcast is recognized in the live broadcast data. For example, if the voiceprint in the live broadcast audio does not match the anchor voiceprint pre-stored in the first live broadcast room, the first live broadcast room may be empty; or, if the live broadcast screen does not contain a human body, the first live broadcast room may be empty; or, if the human body in the live broadcast screen is still, the first live broadcast room may be empty; or, if there is no anchor operation in the first live broadcast room, the first live broadcast room may be empty.

[0079] For example, the computer device can call a neural network model to classify the live broadcast data, obtain classification results in different dimensions, and judge whether the first live broadcast room is empty from multiple dimensions. The following is an example using four dimensions: voiceprint, human body, action, and interaction.

[0080] 1) Voiceprint

[0081] The live broadcast data includes live broadcast audio; the computer device calls the voiceprint recognition model to identify the real-time voiceprint of the live broadcast audio; based on the voiceprint similarity between the real-time voiceprint and the host's voiceprint, a first classification result is determined; the first classification result is used to indicate whether the voiceprint of the first live broadcast room matches; the host's voiceprint is the preset voiceprint of the first live broadcast room.

[0082] The voiceprint recognition model may be a neural network model. For example, the voiceprint recognition model may adopt ECAPA-TDNN (Emphasized Channel Attention Propagation and Aggregation in Time Delay Neural Network) or Deep Speaker (deep speech recognition system).

[0083] Before the first live broadcast room starts broadcasting, the live broadcast application can guide the anchor to record the anchor audio, input the anchor audio into the voiceprint recognition model to obtain the anchor voiceprint and store it. Alternatively, the live broadcast application can guide the anchor to record the anchor audio before each live broadcast room starts broadcasting. The first live broadcast room can correspond to at least one anchor, and each anchor has its own anchor voiceprint. Before the first live broadcast room starts broadcasting, the anchors participating in this live broadcast can be set, and the live broadcast application can perform the above-mentioned voiceprint matching process based on the voiceprint of the anchor participating in this live broadcast.

[0084] Exemplarily, the computer device analyzes the continuous audio stream in the live broadcast room to detect whether the host's voice exists. For example, the frequency, intensity, and duration of the live broadcast audio are detected to ensure that the sound characteristics of the live broadcast audio are captured, and the similarity is calculated with the preset voiceprint of the host in the first live broadcast room to obtain the voiceprint similarity.

[0085] Optionally, when the voiceprint similarity is higher than a similarity threshold, the first classification result may be determined as a voiceprint match; when the voiceprint similarity is not higher than the similarity threshold, the first classification result may be determined as a voiceprint mismatch.

[0086] 2) Human body

[0087] The live broadcast data includes live broadcast images; the computer device calls a posture recognition model to recognize the live broadcast images and obtains posture recognition results; based on whether the posture recognition results contain human postures, a second classification result is determined; the second classification result is used to indicate whether the live broadcast images contain human bodies.

[0088] Optionally, the computer device can identify whether a frame of live broadcast image contains a human gesture based on it. For example, a frame of live broadcast image is input into a gesture recognition model, and a gesture recognition result is output. When the live broadcast image contains a human gesture, the gesture recognition result includes key point information identified from the live broadcast image, and the key points include key points corresponding to characteristic parts such as joints, head, hands, etc. of the human body, key point coordinates, key point types, and other information. When the live broadcast image does not contain a human gesture, the gesture recognition result is empty.

[0089] Exemplarily, when the posture recognition result includes human posture (key point information), the second classification result is determined to include a human body; when the posture recognition result does not include human posture (the posture recognition result is empty), the second classification result is determined to not include a human body.

[0090] Optionally, the computer device may determine the second classification result based on the gesture recognition results of at least two frames of live images. If one of the at least two gesture recognition results contains a human gesture (key point information), the second classification result is determined to contain a human body; if neither of the at least two gesture recognition results contains a human gesture (the gesture recognition result is empty), the second classification result is determined to not contain a human body.

[0091] Exemplarily, the posture recognition model may adopt an OpenPose model, a MoveNet model, a PoseNet model, or a BlazePose model.

[0092] 3) Action

[0093] The number of live broadcast images is at least two frames; when the posture recognition result includes human posture, the computer device calls the action recognition model based on the human postures corresponding to at least two frames of live broadcast images, and outputs a third classification result; the third classification result is used to indicate whether at least two frames of live broadcast images contain human motion.

[0094] Exemplarily, if at least two frames of live broadcast images contain human body gestures, the human body gestures (key point information) corresponding to at least two frames of live broadcast images are input into the action recognition model, and the third classification result is output. The action recognition model is trained to output whether it is a coherent human body action based on the input of at least two frames of human body gestures. If at least two frames of human body gestures are coherent human body gestures, the third classification result is that human body gestures are included; if at least two frames of human body gestures are not coherent human body gestures, or at least two frames of human body gestures are exactly the same (human body gestures are stationary), the third classification result is that human body gestures are not included.

[0095] Exemplarily, the computer device monitors the live screen in the live video stream to identify whether the host's body is moving, thereby ensuring the host's physical presence and activities during the live broadcast.

[0096] 4) Interaction

[0097] The live broadcast data includes the host-side operation data; the computer device calls the interactive recognition model based on the host-side operation data and outputs the fourth classification result; the fourth classification result is used to indicate the frequency of the interactive operation.

[0098] The host-side operation data may include user operations received by the host, such as the host's input of comments, the host's setting of the host room, the host's listing of products, the host's starting of live broadcast room activities, and the host's response to audience interactions.

[0099] The computer device inputs the host-side operation data into the interactive recognition model and outputs a fourth classification result. The interactive recognition model is trained to score the host-side operation data and output an interactive score based on the input host-side operation data. The interactive score is inversely proportional to the frequency of the interactive operation. For example, if the host-side operation data does not contain user operations, the interactive score is 10 points. If the host-side operation data includes frequent high-weight user operations, the interactive score is 0 points.

[0100] Exemplarily, the computer device analyzes the host's behavior pattern, such as the frequency of interaction, to determine whether the host is actively participating in the live broadcast.

[0101] Step 231: Determine scores corresponding to at least two classification results.

[0102] Exemplarily, when the first classification result is a voiceprint match, the first classification result corresponds to a first score (eg, 0 points); when the first classification result is a voiceprint mismatch, the first classification result corresponds to a second score (eg, 10 points).

[0103] Exemplarily, when the second classification result contains a human body, the second classification result corresponds to a third score (e.g., 0 points); when the second classification result does not contain a human body, the second classification result corresponds to a fourth score (e.g., 10 points).

[0104] Exemplarily, when the third classification result includes human body movements, the third classification result corresponds to a fifth score (e.g., 0 points); when the third classification result does not include human body movements, the third classification result corresponds to a sixth score (e.g., 10 points).

[0105] Exemplarily, the fourth classification result may be directly determined as the score corresponding to the fourth classification result.

[0106] Step 232: Based on the weights corresponding to the at least two classification results, a weighted result of the scores corresponding to the at least two classification results is calculated.

[0107] Exemplarily, different classification results correspond to respective weights, for example, the first classification result corresponds to a first weight, the second classification result corresponds to a second weight, the third classification result corresponds to a third weight, and the fourth classification result corresponds to a fourth weight. The live broadcast application can preset the weights of the various classification results.

[0108] The score corresponding to the first classification result is multiplied by the first weight to obtain a first weighted result; the score corresponding to the second classification result is multiplied by the second weight to obtain a second weighted result; the score corresponding to the third classification result is multiplied by the third weight to obtain a third weighted result; the score corresponding to the fourth classification result is multiplied by the fourth weight to obtain a fourth weighted result. The sum of the first weighted result, the second weighted result, the third weighted result, and the fourth weighted result is determined as the weighted result.

[0109] Optionally, the weight of each classification result can be determined according to the type of the first live broadcast room. Different types of live broadcast rooms have their own corresponding weight configurations.

[0110] For example, when the first live broadcast room is a live broadcast of real people, the first weight, the second weight, the third weight and the fourth weight are all set to a higher value (for example, 1). When the first live broadcast room is an audio live broadcast, the first weight and the fourth weight are set to a higher value, and the second weight and the third weight are set to a lower value (for example, 0.1). When the first live broadcast room is a game live broadcast, the first weight is set to a higher value, the second weight and the third weight are set to a lower value, and the fourth weight is set to 0. When the first live broadcast room is a shopping live broadcast, the fourth weight is set to a higher value, and the first weight, the second weight, and the third weight are set to a lower value.

[0111] Step 233: When the weighted result is higher than the empty broadcast threshold, the recognition result is determined as empty broadcast.

[0112] For example, if the score of the weighted result is high, it means that there may be empty broadcasting in the first live broadcast room, and the identification result is determined as empty broadcasting; if the score of the weighted result is low, it means that the first live broadcast room may be broadcasting normally, and the identification result is determined as normal live broadcasting.

[0113] The idle broadcast threshold can be a preset value. The idle broadcast threshold can also be set based on the server load. The idle broadcast threshold is inversely proportional to the server load rate. When the server load is high, the idle broadcast threshold is set to a lower value, and the idle broadcast room is closed in time to release the server load; when the server load is low, the idle broadcast threshold is set to a higher value to allow some live broadcast rooms with the possibility of idle broadcast to continue live broadcasting.

[0114] Optionally, when the live broadcast application determines that there is a live broadcast in the live broadcast room through the above analysis, the relevant information of the live broadcast room can be automatically added to the pending review list. The pending review list will be used for subsequent manual review to confirm the accuracy of the judgment and impose corresponding penalties on the anchor according to the regulations of the live broadcast platform, such as warning, suspension of live broadcast rights or banning of accounts.

[0115] In summary, the method provided in this embodiment determines whether the live broadcast room is empty from multiple dimensions based on the live broadcast data of the live broadcast room. Recognize the real-time voiceprint based on the live broadcast audio to identify whether the real-time voiceprint matches the host's voiceprint; recognize the human posture based on the live broadcast screen to identify whether the human posture is included in the live broadcast screen; recognize human body movements based on the live video to identify whether the human body in the live video is still; recognize the host's interactive operation based on the host's operation data to identify whether the host has performed the operation in the live broadcast room. The judgment results of multiple dimensions are combined to finally determine whether there is an empty broadcast behavior in the live broadcast room, thereby improving the accuracy and reliability of empty broadcast identification. This method can quickly screen out empty broadcast rooms from massive live broadcast rooms, reduce the cost of empty broadcast detection, and improve the efficiency of empty broadcast detection.

[0116] The method provided in this embodiment can calculate weighted results based on the classification results of live broadcast data in multiple dimensions and the weight coefficients of each dimension, and judge whether the live broadcast room is empty based on the weighted results. The live broadcast application can make the method applicable to multiple types of live broadcast rooms by adjusting the weights corresponding to each classification result. Different types of live broadcast rooms use different weight coefficients to perform empty broadcast identification, thereby improving the accuracy of empty broadcast identification.

[0117] The method provided in this embodiment comprehensively analyzes the continuous audio and video stream segments in the live broadcast room to detect the host's behavioral characteristics, including but not limited to the host's voice, body presence, and behavioral movements, so as to accurately determine whether the host in the live broadcast room is in a state of idle broadcast. Once the idle broadcast behavior is detected, the system will automatically mark the live broadcast room and push it to the list for review, so as to carry out subsequent manual review and necessary punishment measures.

[0118] Exemplarily, a method for processing an empty live broadcast room is provided.

[0119] Figure 4 FIG. 1 is a flow chart of a method for detecting empty broadcasts provided by an exemplary embodiment of the present application. The method can be used for Figure 1 The terminal device or server shown. Figure 2 In the illustrated embodiment, step 230 also includes steps 241 to 244 .

[0120] Step 210: Obtain live broadcast data of the first live broadcast room.

[0121] Step 220: Determine at least one classification result of the live broadcast data, where the at least one classification result is used to indicate one or more of whether the voiceprint of the first live broadcast room matches, whether the picture contains a human body, whether the human body posture is still, and whether the anchor performs an interactive operation.

[0122] Step 230: Determine an identification result based on at least one classification result, where the identification result is used to indicate whether the first live broadcast room is empty.

[0123] Illustratively, steps 210 to 230 may refer to the relevant descriptions in the above embodiments.

[0124] Step 241: when the identification result is empty broadcast, the historical empty broadcast data of the first live broadcast room is obtained, and the historical empty broadcast data is used to indicate the historical empty broadcast situation of the first live broadcast room.

[0125] For live broadcast rooms that are identified as empty broadcasts, the processing method can be decided based on historical empty broadcast data.

[0126] Historical blank broadcast data may include at least one of the following information: the number of blank broadcasts in the first live broadcast room that were judged to be blank broadcasts in history, the start time of each blank broadcast, the duration of the blank broadcast, the time when normal live broadcast is resumed, the change in the number of viewers during the blank broadcast, the frequency of blank broadcasts, the blank broadcast threshold used when judging blank broadcasts, the weight of each classification result, etc.

[0127] Step 242: Call the multimodal model to output a comprehensive score of the first live broadcast room based on the historical idle broadcast data and live broadcast data. The comprehensive score is used to evaluate the activity level of the first live broadcast room.

[0128] The multimodal model is used to score the first live broadcast room according to at least two modal data of the first live broadcast room to obtain a comprehensive score. The comprehensive score is used to evaluate the activity of the live broadcast room based on the historical empty broadcast situation and the current live broadcast situation.

[0129] The data of at least two modes include historical idle broadcast data and live broadcast data. The live broadcast data may be the same as the live broadcast data used in step 210. Alternatively, the live broadcast data may be the anchor end operation data.

[0130] Optionally, the live broadcast data may include status data of the live broadcast room. For example, the status data includes at least one of the following: the number of viewers in the current live broadcast room, the sales volume of the live broadcast room, the activity of the live broadcast room, the broadcast time of the live broadcast room, the popularity value of the live broadcast room, and the level of the live broadcast room.

[0131] In an optional embodiment, the anchor's historical empty broadcast data and recent live broadcast room behavior data (including but not limited to human voice, action, operation events on the live broadcast room system and anchor level, etc.) are comprehensively scored through LLM (Large Language Model). Based on these scores, a pop-up warning is implemented for the identified low-scoring anchors, and the anchor is required to confirm within the specified time. If the anchor fails to respond in time, the system will automatically execute the penalty of interruption and suspension of broadcast to ensure the standardization of the live broadcast content. For high-quality high-scoring anchors, unnecessary crackdowns on them are exempted, further optimizing the live broadcast experience of high-quality anchors. It can not only improve the efficiency of supervision of live broadcast content, but also protect the legitimate rights and interests of anchors and maintain the order and quality of live broadcast platforms.

[0132] Step 243: When the comprehensive score is higher than the score threshold, the first live broadcast room is added to the list to be reviewed.

[0133] For highly active live broadcast rooms, such as high-value live broadcast rooms with a large number of views, in order to avoid pop-up warnings affecting the audience's viewing of the live broadcast, they can be put into the pending review list, and the live broadcast status of the live broadcast room can be manually confirmed, and the penalty method for empty live broadcast rooms can be manually determined to avoid the impact of misjudgment on high-value live broadcast rooms as much as possible.

[0134] Step 244: When the comprehensive score is lower than the score threshold, the first live broadcast room is controlled to display a pop-up warning; the pop-up warning is used to prompt that the first live broadcast room is in an idle broadcast state, and guide the first live broadcast room to resume normal live broadcast.

[0135] For low-activity live broadcast rooms, a pop-up warning can be displayed to guide the anchor to resume normal live broadcast and avoid empty live broadcast rooms occupying server resources.

[0136] Exemplarily, in response to receiving an operation to close the pop-up warning, the live broadcast status of the first live broadcast room is determined to be normal live broadcast; in response to not receiving an operation to close the pop-up warning within a first time period, the first live broadcast room is closed.

[0137] In summary, the method provided in this embodiment can call a multimodal model to comprehensively score the first live broadcast room after it is identified as an empty broadcast room, and determine the subsequent processing method based on the scoring result. For live broadcast rooms with higher comprehensive scores, they can be added to the pending review list, and their live broadcast status can be manually reviewed; for live broadcast rooms with lower comprehensive scores, pop-up warnings can be directly displayed. This can prevent pop-up warnings from affecting the live broadcast of highly active live broadcast rooms, and timely remind low-activity live broadcast rooms that are empty, while reducing the workload of manual review of empty live broadcast rooms, optimizing the audience's experience of watching live broadcasts, and improving the efficiency of empty broadcast review.

[0138] The method provided in this embodiment effectively filters out non-empty live broadcast rooms by deeply analyzing the real-time audio and video stream content of the anchor in the list to be reviewed, combining the anchor's behavior data and the anchor's rating data, improving review efficiency and ensuring that review resources can be more concentratedly applied to real empty broadcast situations.

[0139] The method provided in this embodiment comprehensively scores the anchor by comprehensively analyzing the anchor's historical empty broadcast data and recent live broadcast room behavior data (including but not limited to human voice, action, operation events on the live broadcast room system and anchor level, etc.). Based on these comprehensive scores, the system implements pop-up warnings for identified low-scoring anchors and requires the anchor to confirm within the specified time. If the anchor fails to respond in time, the system will automatically execute the penalty of interruption and suspension of broadcast to ensure the standardization of the live broadcast content. For high-quality anchors who receive high scores, the system will exempt them from unnecessary audits, thereby optimizing their live broadcast experience and encouraging the continuous provision of high-quality live broadcast content. It not only improves the accuracy of the audit, but also reflects the differentiated management of different anchor behaviors, aiming to create a healthier and more positive live broadcast environment.

[0140] The method provided in this embodiment determines whether there is an anchor in the live broadcast room by detecting the anchor's behavior (including but not limited to the anchor's voice, anchor's body, anchor's movements, etc.) of the continuous audio / video stream segments in the live broadcast room. The non-empty live broadcast rooms are filtered out by detecting the behavior data of the anchor's real-time audio and video stream content in the list to be reviewed, thereby reducing the review volume of the anchors who have returned.

[0141] The method provided in this embodiment uses LLM to comprehensively score the anchor's historical empty broadcast data and recent live broadcast behavior data (including but not limited to human voice, action, operation events on the live broadcast room system, and the anchor's level, etc.). Based on these scores, the system will issue a pop-up warning to the identified low-scoring anchors, and the anchor is required to click to confirm within a specified time, otherwise the anchor will be punished by interrupting the stream and stopping the broadcast. For high-quality high-scoring anchors, unnecessary attacks on them are avoided, and the live broadcast experience of high-quality anchors is optimized.

[0142] It should be noted that before collecting relevant data of users and during the process of collecting relevant data of users, this application can display a prompt interface, pop-up window or output voice prompt information, and the prompt interface, pop-up window or voice prompt information is used to prompt the user that its relevant data is currently being collected, so that this application only starts to execute the relevant steps of obtaining relevant data of users after obtaining the confirmation operation issued by the user to the prompt interface or pop-up window, otherwise (that is, when the confirmation operation issued by the user to the prompt interface or pop-up window is not obtained), the relevant steps of obtaining relevant data of users are terminated, that is, the relevant data of users are not obtained. In other words, all user data collected by this application are collected with the consent and authorization of the user, and the collection, use and processing of relevant user data need to comply with the relevant laws, regulations and standards of relevant countries and regions.

[0143] It should be noted that the sequence of the method steps provided in the embodiments of the present application can be appropriately adjusted, and the steps can be increased or decreased accordingly according to the circumstances. Any technician familiar with the technical field can easily think of the changed methods within the technical scope disclosed in the present application, and they should be included in the protection scope of the present application, so they will not be repeated here.

[0144] Figure 5 : is a schematic diagram of the structure of a device for detecting empty broadcast provided by an exemplary embodiment of the present application. The device comprises:

[0145] An acquisition module 601 is used to acquire live broadcast data of a first live broadcast room;

[0146] A classification module 602 is used to determine at least one classification result of the live broadcast data, wherein the at least one classification result is used to indicate one or more of whether the voiceprint of the first live broadcast room matches, whether the picture contains a human body, whether the human body posture is still, and whether the anchor performs an interactive operation;

[0147] The identification module 603 is used to determine an identification result based on the at least one classification result, and the identification result is used to indicate whether the first live broadcast room is empty.

[0148] In an optional embodiment, the number of the classification results is at least two; the identification module 603 is used to determine the scores corresponding to the at least two classification results respectively;

[0149] The identification module 603 is used to calculate a weighted result of the scores corresponding to the at least two classification results based on the weights corresponding to the at least two classification results;

[0150] The identification module 603 is configured to determine the identification result as empty broadcast if the weighted result is higher than an empty broadcast threshold.

[0151] In an optional embodiment, the live broadcast data includes live broadcast audio;

[0152] The classification module 602 is used to call the voiceprint recognition model to identify the real-time voiceprint of the live audio;

[0153] The classification module 602 is used to determine a first classification result based on the voiceprint similarity between the real-time voiceprint and the host voiceprint; the first classification result is used to indicate whether the voiceprint of the first live broadcast room matches; the host voiceprint is a preset voiceprint for the first live broadcast room.

[0154] In an optional embodiment, the live broadcast data includes the live broadcast picture;

[0155] The classification module 602 is used to call the posture recognition model to recognize the live broadcast picture and obtain the posture recognition result;

[0156] The classification module 602 is used to determine a second classification result based on whether the posture recognition result contains a human posture; the second classification result is used to indicate whether the live broadcast screen contains a human body.

[0157] In an optional embodiment, the number of the live broadcast pictures is at least two frames;

[0158] The classification module 602 is used to call the action recognition model based on the human postures corresponding to the at least two frames of live broadcast images when the posture recognition result includes human posture, and output a third classification result; the third classification result is used to indicate whether the at least two frames of live broadcast images contain human body movements.

[0159] In an optional embodiment, the live broadcast data includes the anchor terminal operation data;

[0160] The classification module 602 is used to call the interactive recognition model based on the anchor end operation data to output a fourth classification result; the fourth classification result is used to indicate the frequency of the interactive operation.

[0161] In an optional embodiment, the device further includes:

[0162] The acquisition module 601 is used to acquire historical empty broadcast data of the first live broadcast room when the identification result is empty broadcast, and the historical empty broadcast data is used to indicate the historical empty broadcast situation of the first live broadcast room;

[0163] Processing module 604, configured to call a multimodal model to output a comprehensive score of the first live broadcast room based on the historical idle broadcast data and the live broadcast data, wherein the comprehensive score is used to evaluate the activity level of the first live broadcast room;

[0164] The processing module 604 is used to add the first live broadcast room to the list to be reviewed when the comprehensive score is higher than the score threshold.

[0165] In an optional embodiment, the device further includes:

[0166] The acquisition module 601 is used to acquire historical empty broadcast data of the first live broadcast room when the identification result is empty broadcast, and the historical empty broadcast data is used to indicate the historical empty broadcast situation of the first live broadcast room;

[0167] Processing module 604, configured to call a multimodal model to output a comprehensive score of the first live broadcast room based on the historical idle broadcast data and the live broadcast data, wherein the comprehensive score is used to evaluate the activity level of the first live broadcast room;

[0168] The processing module 604 is used to control the first live broadcast room to display a pop-up warning when the comprehensive score is lower than the score threshold; the pop-up warning is used to prompt that the first live broadcast room is in an idle broadcast state and guide the first live broadcast room to resume normal live broadcast.

[0169] It should be noted that the empty broadcast detection device provided in the above embodiment is only illustrated by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the empty broadcast detection device provided in the above embodiment and the empty broadcast detection method embodiment belong to the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0170] The embodiment of the present application further provides a computer device, which includes: a processor and a memory, wherein at least one instruction, at least one program, code set or instruction set is stored in the memory, and the at least one instruction, at least one program, code set or instruction set is loaded and executed by the processor to implement the empty broadcast detection method provided by the above-mentioned method embodiments. The computer device can be implemented as a terminal.

[0171] For example, Figure 6 It is a structural diagram of a computer device provided by an exemplary embodiment of the present application.

[0172] Typically, the computer device 1700 includes a processor 1701 and a memory 1702 .

[0173] The processor 1701 may include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor 1701 may be implemented in at least one hardware form of DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). The processor 1701 may also include a main processor and a coprocessor. The main processor is a processor for processing data in the awake state, also known as a CPU (Central Processing Unit); the coprocessor is a low-power processor for processing data in the standby state. In some embodiments, the processor 1701 may be integrated with a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content to be displayed on the display screen. In some embodiments, the processor 1701 may also include an AI (Artificial Intelligence) processor, which is used to process computing operations related to machine learning.

[0174] The memory 1702 may include one or more computer-readable storage media, which may be non-transitory. The memory 1702 may also include a high-speed random access memory, and a non-volatile memory, such as one or more disk storage devices, flash memory storage devices. In some embodiments, the non-transitory computer-readable storage medium in the memory 1702 is used to store at least one instruction, which is used to be executed by the processor 1701 to implement the empty broadcast detection method provided in the method embodiment of the present application.

[0175] In some embodiments, the computer device 1700 may also optionally include: a peripheral device interface 1703 and at least one peripheral device. The processor 1701, the memory 1702 and the peripheral device interface 1703 may be connected via a bus or a signal line. Each peripheral device may be connected to the peripheral device interface 1703 via a bus, a signal line or a circuit board. Specifically, the peripheral device includes: at least one of a radio frequency circuit 1704, a display screen 1705, a camera assembly 1706, an audio circuit 1707 and a power supply 1708.

[0176] The peripheral device interface 1703 may be used to connect at least one peripheral device related to I / O (Input / Output) to the processor 1701 and the memory 1702. In some embodiments, the processor 1701, the memory 1702, and the peripheral device interface 1703 are integrated on the same chip or circuit board; in some other embodiments, any one or two of the processor 1701, the memory 1702, and the peripheral device interface 1703 may be implemented on a separate chip or circuit board, which is not limited in the embodiments of the present application.

[0177] The radio frequency circuit 1704 is used to receive and transmit RF (Radio Frequency) signals, also known as electromagnetic signals. The radio frequency circuit 1704 communicates with the communication network and other communication devices through electromagnetic signals. The radio frequency circuit 1704 converts the electrical signal into an electromagnetic signal for transmission, or converts the received electromagnetic signal into an electrical signal. Optionally, the radio frequency circuit 1704 includes: an antenna system, an RF transceiver, one or more amplifiers, a tuner, an oscillator, a digital signal processor, a codec chipset, a user identity module card, and the like. The radio frequency circuit 1704 can communicate with other computer devices through at least one wireless communication protocol. The wireless communication protocol includes, but is not limited to: the World Wide Web, a metropolitan area network, an intranet, various generations of mobile communication networks (2G, 3G, 4G and 5G), a wireless local area network and / or a WiFi (Wireless Fidelity) network. In some embodiments, the radio frequency circuit 1704 may also include circuits related to NFC (Near Field Communication), which is not limited in this application.

[0178] The display screen 1705 is used to display a UI (User Interface). The UI may include graphics, text, icons, videos, and any combination thereof. When the display screen 1705 is a touch display screen, the display screen 1705 also has the ability to collect touch signals on the surface or above the surface of the display screen 1705. The touch signal can be input to the processor 1701 as a control signal for processing. At this time, the display screen 1705 can also be used to provide virtual buttons and / or virtual keyboards, also known as soft buttons and / or soft keyboards. In some embodiments, the display screen 1705 can be one, and the front panel of the computer device 1700 is set; in other embodiments, the display screen 1705 can be at least two, which are respectively set on different surfaces of the computer device 1700 or are folded; in some other embodiments, the display screen 1705 can be a flexible display screen, which is set on the curved surface or folded surface of the computer device 1700. Even, the display screen 1705 can also be set to a non-rectangular irregular shape, that is, a special-shaped screen. The display screen 1705 can be made of materials such as LCD (Liquid Crystal Display) and OLED (Organic Light-Emitting Diode).

[0179] The camera assembly 1706 is used to capture images or videos. Optionally, the camera assembly 1706 includes a front camera and a rear camera. Typically, the front camera is arranged on the front panel of the computer device 1700, and the rear camera is arranged on the back of the computer device. In some embodiments, there are at least two rear cameras, which are any one of a main camera, a depth of field camera, a wide-angle camera, and a telephoto camera, so as to realize the fusion of the main camera and the depth of field camera to realize the background blur function, the fusion of the main camera and the wide-angle camera to realize the panoramic shooting and VR (Virtual Reality) shooting function or other fusion shooting functions. In some embodiments, the camera assembly 1706 may also include a flash. The flash can be a monochrome temperature flash or a dual-color temperature flash. A dual-color temperature flash refers to a combination of a warm light flash and a cold light flash, which can be used for light compensation at different color temperatures.

[0180] The audio circuit 1707 may include a microphone and a speaker. The microphone is used to collect sound waves from the user and the environment, and convert the sound waves into electrical signals and input them into the processor 1701 for processing, or input them into the radio frequency circuit 1704 to achieve voice communication. For the purpose of stereo acquisition or noise reduction, there may be multiple microphones, which are respectively arranged at different parts of the computer device 1700. The microphone may also be an array microphone or an omnidirectional acquisition microphone. The speaker is used to convert the electrical signal from the processor 1701 or the radio frequency circuit 1704 into sound waves. The speaker may be a traditional film speaker or a piezoelectric ceramic speaker. When the speaker is a piezoelectric ceramic speaker, it can not only convert the electrical signal into sound waves audible to humans, but also convert the electrical signal into sound waves inaudible to humans for purposes such as ranging. In some embodiments, the audio circuit 1707 may also include a headphone jack.

[0181] The power supply 1708 is used to power various components in the computer device 1700. The power supply 1708 can be an alternating current, a direct current, a disposable battery, or a rechargeable battery. When the power supply 1708 includes a rechargeable battery, the rechargeable battery can be a wired rechargeable battery or a wireless rechargeable battery. A wired rechargeable battery is a battery that is charged through a wired line, and a wireless rechargeable battery is a battery that is charged through a wireless coil. The rechargeable battery can also be used to support fast charging technology.

[0182] In some embodiments, the computer device 1700 further includes one or more sensors 1709 , including but not limited to: an acceleration sensor 1710 , a gyroscope sensor 1711 , a pressure sensor 1712 , an optical sensor 1713 , and a proximity sensor 1714 .

[0183] The acceleration sensor 1710 can detect the magnitude of acceleration on the three coordinate axes of the coordinate system established by the computer device 1700. For example, the acceleration sensor 1710 can be used to detect the components of gravity acceleration on the three coordinate axes. The processor 1701 can control the touch display screen 1705 to display the user interface in a horizontal view or a vertical view according to the gravity acceleration signal collected by the acceleration sensor 1710. The acceleration sensor 1710 can also be used to collect game or user motion data.

[0184] The gyro sensor 1711 can detect the body direction and rotation angle of the computer device 1700, and the gyro sensor 1711 can cooperate with the acceleration sensor 1710 to collect the user's 3D actions on the computer device 1700. The processor 1701 can implement the following functions based on the data collected by the gyro sensor 1711: motion sensing (such as changing the UI according to the user's tilt operation), image stabilization during shooting, game control, and inertial navigation.

[0185] The pressure sensor 1712 can be set on the side frame of the computer device 1700 and / or the lower layer of the touch display screen 1705. When the pressure sensor 1712 is set on the side frame of the computer device 1700, it can detect the user's grip signal of the computer device 1700, and the processor 1701 performs left and right hand recognition or shortcut operations according to the grip signal collected by the pressure sensor 1712. When the pressure sensor 1712 is set on the lower layer of the touch display screen 1705, the processor 1701 controls the operability controls on the UI interface according to the user's pressure operation on the touch display screen 1705. The operability controls include at least one of a button control, a scroll bar control, an icon control, and a menu control.

[0186] The optical sensor 1713 is used to collect the ambient light intensity. In one embodiment, the processor 1701 can control the display brightness of the touch display screen 1705 according to the ambient light intensity collected by the optical sensor 1713. Specifically, when the ambient light intensity is high, the display brightness of the touch display screen 1705 is increased; when the ambient light intensity is low, the display brightness of the touch display screen 1705 is reduced. In another embodiment, the processor 1701 can also dynamically adjust the shooting parameters of the camera assembly 1706 according to the ambient light intensity collected by the optical sensor 1713.

[0187] The proximity sensor 1714, also called a distance sensor, is usually arranged on the front panel of the computer device 1700. The proximity sensor 1714 is used to collect the distance between the user and the front of the computer device 1700. In one embodiment, when the proximity sensor 1714 detects that the distance between the user and the front of the computer device 1700 is gradually decreasing, the processor 1701 controls the touch display screen 1705 to switch from the screen-on state to the screen-off state; when the proximity sensor 1714 detects that the distance between the user and the front of the computer device 1700 is gradually increasing, the processor 1701 controls the touch display screen 1705 to switch from the screen-off state to the screen-on state.

[0188] Those skilled in the art will understand that Figure 6 The structure shown in the figure does not constitute a limitation on the computer device 1700, and the computer device 1700 may include more or less components than shown in the figure, or combine some components, or adopt a different arrangement of components.

[0189] A computer-readable storage medium is also provided in an embodiment of the present application, in which at least one instruction, at least one program, code set or instruction set is stored. When the at least one instruction, at least one program, code set or instruction set is loaded and executed by a processor of a computer device, the empty broadcast detection method provided by the above-mentioned method embodiments is implemented.

[0190] The present application also provides a computer program product or a computer program, which includes a computer instruction stored in a computer-readable storage medium. A processor of a computer device reads the computer instruction from the computer-readable storage medium, and the processor executes the computer instruction, so that the computer device executes the empty broadcast detection method provided by each of the above method embodiments.

[0191] A person skilled in the art will understand that all or part of the steps to implement the above embodiments may be accomplished by hardware or by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, and the above-mentioned readable storage medium may be a read-only memory, a disk or an optical disk, etc.

[0192] The above description is only an optional embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent switching, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for detecting empty broadcast, characterized in that: The method comprises: Get the live broadcast data of the first live broadcast room; Determine at least one classification result of the live broadcast data, where the at least one classification result is used to indicate one or more of whether the voiceprint of the first live broadcast room matches, whether the picture contains a human body, whether the human body posture is still, and whether the anchor performs an interactive operation; An identification result is determined based on the at least one classification result, and the identification result is used to indicate whether the first live broadcast room is empty.

2. The method according to claim 1, characterized in that The number of the classification results is at least two; and determining the recognition result based on the at least one classification result includes: Determine a score corresponding to each of the at least two classification results; Calculating a weighted result of the score corresponding to each of the at least two classification results based on the weights corresponding to each of the at least two classification results; When the weighted result is higher than the empty broadcast threshold, the recognition result is determined as empty broadcast.

3. The method according to claim 2, characterized in that The live broadcast data includes live broadcast audio; Determining at least one classification result of the live broadcast data includes: Calling a voiceprint recognition model to identify the real-time voiceprint of the live audio; Based on the voiceprint similarity between the real-time voiceprint and the host's voiceprint, a first classification result is determined; the first classification result is used to indicate whether the voiceprint of the first live broadcast room matches; the host's voiceprint is a preset voiceprint for the first live broadcast room.

4. The method according to claim 2, characterized in that: The live broadcast data includes the live broadcast picture; Determining at least one classification result of the live broadcast data includes: Calling a posture recognition model to recognize the live broadcast screen and obtain a posture recognition result; Based on whether the posture recognition result contains a human posture, a second classification result is determined; the second classification result is used to indicate whether the live broadcast screen contains a human body.

5. The method according to claim 4, characterized in that The number of live broadcast pictures is at least two frames; The determining of at least one classification result of the live broadcast data further includes: When the posture recognition result includes human posture, the action recognition model is called to output a third classification result based on the human postures corresponding to the at least two frames of live broadcast images; the third classification result is used to indicate whether the at least two frames of live broadcast images contain human motion.

6. The method according to claim 2, characterized in that The live broadcast data includes the anchor terminal operation data; The determining of at least one classification result of the live broadcast data further includes: Calling the interactive recognition model to output a fourth classification result based on the anchor terminal operation data; The fourth classification result is used to indicate the frequency of the interactive operation.

7. The method according to any one of claims 1 to 6, characterized in that: The method further comprises: When the identification result is that the broadcast is empty, obtaining historical empty broadcast data of the first live broadcast room, where the historical empty broadcast data is used to indicate the historical empty broadcast situation of the first live broadcast room; Calling a multimodal model to output a comprehensive score of the first live broadcast room based on the historical empty broadcast data and the live broadcast data, where the comprehensive score is used to evaluate the activity level of the first live broadcast room; When the comprehensive score is higher than the score threshold, the first live broadcast room is added to the list to be reviewed.

8. The method according to any one of claims 1 to 6, characterized in that: The method further comprises: When the identification result is that the broadcast is empty, obtaining historical empty broadcast data of the first live broadcast room, where the historical empty broadcast data is used to indicate the historical empty broadcast situation of the first live broadcast room; Calling a multimodal model to output a comprehensive score of the first live broadcast room based on the historical empty broadcast data and the live broadcast data, where the comprehensive score is used to evaluate the activity level of the first live broadcast room; When the comprehensive score is lower than the score threshold, the first live broadcast room is controlled to display a pop-up warning; the pop-up warning is used to prompt that the first live broadcast room is in an idle broadcast state, and guide the first live broadcast room to resume normal live broadcast.

9. A device for detecting empty broadcast, characterized in that: The device comprises: An acquisition module, used to acquire live broadcast data of the first live broadcast room; a classification module, used to determine at least one classification result of the live broadcast data, wherein the at least one classification result is used to indicate one or more of whether the voiceprint of the first live broadcast room matches, whether the picture contains a human body, whether the human body posture is still, and whether the anchor performs an interactive operation; An identification module is used to determine an identification result based on the at least one classification result, and the identification result is used to indicate whether the first live broadcast room is empty.

10. A computer device, characterized in that: The computer device includes a processor and a memory, wherein the memory stores at least one program, and the at least one program is loaded and executed by the processor to implement the empty broadcast detection method according to any one of claims 1 to 8.

11. A computer-readable storage medium, characterized in that: The readable storage medium stores at least one program, and the at least one program is loaded and executed by a processor to implement the empty broadcast detection method according to any one of claims 1 to 8.

12. A computer program product, characterized in that The computer program product includes computer instructions, which are stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the air broadcast detection method as described in any one of claims 1 to 8.

Images