Live broadcast stream pushing method, device and system, electronic equipment and storage medium

Abstract

The invention discloses a live broadcast stream pushing method, which comprises the steps of obtaining a task undertaking instruction used for indicating to undertake a task for pushing a first live broadcast data stream to a client, the first live broadcast data stream being a live broadcast data stream corresponding to a target live broadcast room, and the target live broadcast room corresponding to a plurality of live broadcast data streams; sending a live broadcast material data request message used for requesting to obtain live broadcast material data used for generating the first live broadcast data stream; obtaining live broadcast material data, wherein the live broadcast material data comprises user portrait data, anchor image data, anchor timbre data and first service item data which can be provided for a user by an anchor party; generating the first live broadcast data stream according to the live broadcast material data; and pushing the first live broadcast data stream to the first user side. By adopting the scheme of the invention, different users can see different live broadcast room contents when entering the same live broadcast room, so that the live broadcast room contents and live broadcast styles watched by the users better conform to user preferences.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a live streaming method, device, electronic device and storage medium. Background Art

[0002] Live streaming is a technology that transmits video and audio content in real time over the Internet. It is highly real-time, interactive, and rich in content, so it is widely used in different fields and scenarios. In the existing technology, every user who enters the same live broadcast room sees the same content. However, different users have different needs and preferences for live streaming. Therefore, the existing live streaming method cannot meet the different needs of different users for the same live broadcast room, which will further affect the viewing rate. If the anchor is a merchant, it may also affect the number of visits to the merchant.

[0003] Therefore, how to meet users' needs for live broadcast rooms and improve the matching degree between the live broadcast content watched by users and their preferences has become a problem that needs to be solved at present. Summary of the invention

[0004] The present application provides a live streaming method for improving the matching degree between users and live content and making the recommended live streaming room more in line with user preferences.

[0005] In a first aspect, the present application provides a live streaming method, comprising: Obtaining a task acceptance instruction issued by a live broadcast server for instructing to accept a task of pushing a first live broadcast data stream to a client, wherein the first live broadcast data stream is a live broadcast data stream corresponding to a target live broadcast room, the target live broadcast room corresponds to multiple live broadcast data streams, and the first live broadcast data stream is different from other live broadcast data streams corresponding to the target live broadcast room; In response to the task acceptance instruction, a live broadcast material data request message is sent to the live broadcast server to request acquisition of live broadcast material data used to generate the first live broadcast data stream; Obtaining the live broadcast material data returned by the live broadcast service end in response to the live broadcast material data request message, wherein the live broadcast material data includes user portrait data, anchor image data, anchor timbre data, and first service item data that the anchor can provide to the user; Generate the first live broadcast data stream according to the live broadcast material data; Obtaining a streaming instruction sent by the live streaming service end for instructing to push the first live streaming data stream to the first user end; In response to the streaming instruction, the first live data stream is pushed to the first user terminal.

[0006] Optionally, the task acceptance instruction includes identification data pre-allocated by the live broadcast server for the first live broadcast data stream, the live broadcast material data request message includes identification data pre-allocated by the live broadcast server for the first live broadcast data stream, and the streaming instruction includes identification data pre-allocated by the live broadcast server for the first live broadcast data stream.

[0007] Optionally, generating the first live broadcast data stream according to the live broadcast material data includes: generating text data corresponding to the first live broadcast data stream according to the first service item data and the user portrait data; generating host audio data corresponding to the first live broadcast data stream according to the host timbre data and the text data corresponding to the first live broadcast data stream; generating host lip drive data corresponding to the first live broadcast data stream according to the host image data and the host audio data corresponding to the first live broadcast data stream; generating the first live broadcast data stream according to the host audio data corresponding to the first live broadcast data stream and the host lip drive data corresponding to the first live broadcast data stream.

[0008] Optionally, it also includes: obtaining a live data stream update task sent by the live broadcast server, the live data stream update task is used to indicate that an updated live data stream is generated for the first user terminal according to the behavioral characteristic data of the first user terminal for the target live broadcast room, the live content characteristic data corresponding to the updated live data stream generated for the first user terminal matches the behavioral characteristic data of the first user terminal for the target live broadcast room, the live data stream update task includes identification data pre-allocated by the live broadcast server for updating the live data stream, the behavioral characteristic data of the first user terminal for the target live broadcast room, and the live material data for generating the updated live data stream; according to the behavioral characteristic data of the first user terminal for the target live broadcast room and the live material data for generating the updated live data stream The method comprises: generating an updated live data stream for the first user terminal by using live material data of the new live data stream; sending a response message to the live server terminal to indicate that the updated live data stream has been generated, wherein the response message includes identification data pre-allocated for the updated live data stream; obtaining a live data stream switching indication message sent by the live server terminal to indicate that the first live data stream that has been pushed to the first user terminal is switched to the updated live data stream, wherein the live data stream switching indication message includes identification data allocated by the live server terminal to the first live data stream and identification data pre-allocated for the updated live data stream; and in response to the live data stream switching indication message, switching the first live data stream that has been pushed to the first user terminal to the updated live data stream.

[0009] Optionally, the live broadcast data stream update task includes using the host image data corresponding to the first live broadcast data stream as indication information of the host image data corresponding to the updated live broadcast data stream and using the host timbre data corresponding to the first live broadcast data stream as indication information of the host timbre data corresponding to the updated live broadcast data stream; the live broadcast material data used to generate the updated live broadcast data stream includes other service item data that the host can provide to the user that is different from the first service item data; the live broadcast material data used to generate the updated live broadcast data stream for the first user terminal based on the behavioral feature data of the first user terminal for the target live broadcast room and the live broadcast material data used to generate the updated live broadcast data stream Updating the live broadcast data stream includes: generating text data corresponding to the updated live broadcast data stream according to other service item data that the anchor can provide to users and the behavioral characteristic data of the first user terminal for the target live broadcast room; generating anchor audio data corresponding to the updated live broadcast data stream according to the anchor timbre data and the text data corresponding to the updated live broadcast data stream; generating anchor lip drive data corresponding to the updated live broadcast data stream according to the anchor image data and the anchor audio data corresponding to the updated live broadcast data stream; generating the updated live broadcast data stream according to the anchor audio data corresponding to the updated live broadcast data stream and the anchor lip drive data corresponding to the updated live broadcast data stream.

[0010] Optionally, the live broadcast material data used to generate the updated live broadcast data stream includes host image data corresponding to the updated live broadcast data stream, host timbre data corresponding to the updated live broadcast data stream, and other service item data that the host can provide to users and is different from the first service item data; generating the updated live broadcast data stream for the first user terminal based on the behavioral characteristic data of the first user terminal for the target live broadcast room and the live broadcast material data used to generate the updated live broadcast data stream includes: generating text data corresponding to the updated live broadcast data stream based on other service item data that the host can provide to users and the behavioral characteristic data of the first user terminal for the target live broadcast room; generating host audio data corresponding to the updated live broadcast data stream based on the host timbre data corresponding to the updated live broadcast data stream and the text data corresponding to the updated live broadcast data stream; generating host lip drive data corresponding to the updated live broadcast data stream based on the host image data corresponding to the updated live broadcast data stream and the host audio data corresponding to the updated live broadcast data stream; generating the updated live broadcast data stream based on the host audio data corresponding to the updated live broadcast data stream and the host lip drive data corresponding to the updated live broadcast data stream.

[0011] In a second aspect, a live streaming method includes: Obtaining a live broadcast room access request message sent by the first user terminal for accessing a target live broadcast room; Selecting a target cloud device from the cloud device list for pushing the first live data stream to the client; Sending a task acceptance instruction to the target cloud device for instructing the target cloud device to accept the task of pushing the first live data stream to the client; Obtaining a live broadcast material data request message issued by the target cloud device in response to the task acceptance instruction for requesting to obtain live broadcast material data, wherein the live broadcast material data is material data used to generate the first live broadcast data stream; In response to the live broadcast material data request message, the live broadcast material data is sent to the target cloud device, wherein the live broadcast material data includes first user portrait data, anchor image data, anchor timbre data, and first service item data that the anchor can provide to the user; Obtaining identification data of the target live broadcast room, and obtaining a live broadcast stream distribution strategy corresponding to the identification data of the target live broadcast room; Determine, according to the first user portrait data and the live stream distribution strategy, a first live data stream corresponding to the target live broadcast room to be pushed to the first user terminal, wherein the first live data stream is different from a second live data stream pushed to a second user terminal accessing the target live broadcast room; In response to the live room access request message, the first live data stream is pushed to the first user terminal.

[0012] Optionally, also include: After determining the first live broadcast data stream corresponding to the target live broadcast room to be pushed to the first user terminal, determining the target cloud device in the cloud device list that has undertaken the task of pushing the first live broadcast data stream to the client; The pushing of the first live data stream to the first user terminal in response to the live room access request message includes: in response to the live room access request message, issuing a push instruction to the target cloud device to instruct the pushing of the first live data stream to the first user terminal.

[0013] Optionally, also include: Obtaining portrait data of multiple user groups corresponding to the target live broadcast room; Obtaining live broadcast content feature data corresponding to the target live broadcast room; Based on the portrait data of the multiple user groups and the live content feature data, multiple live data streams corresponding to the target live room are determined.

[0014] Optionally, obtaining the portrait data of multiple user groups corresponding to the target live broadcast room includes: Obtaining portrait data of historical users who visited the target live broadcast room, performing clustering processing on the portrait data of historical users who visited the target live broadcast room, and obtaining portrait data of multiple user groups corresponding to the target live broadcast room; Alternatively, portrait data of historical users who have visited the anchor data page corresponding to the target live broadcast room is obtained, and the portrait data of historical users who have visited the anchor data page corresponding to the target live broadcast room are clustered to obtain portrait data of multiple user groups corresponding to the target live broadcast room; Alternatively, portrait data of historical users who have accessed a data service system is obtained, where the data service system is the data service system in which the target live broadcast room is located, and the portrait data of historical users who have accessed the data service system are clustered to obtain portrait data of multiple user groups corresponding to the target live broadcast room.

[0015] Optionally, determining multiple live broadcast data streams corresponding to the target live broadcast room according to the portrait data of the multiple user groups and the live broadcast content feature data includes: Determine matching live content feature data for the portrait data of each user group; Set identification data of the live data stream for the portrait data of each user group; A live data stream list of the target live broadcast room is established, in which the corresponding relationship between the portrait data of each user group, the matching live content feature data and the matching live data stream identification data is recorded.

[0016] Optionally, determining, according to the first user portrait data, a first live broadcast data stream corresponding to the target live broadcast room to be pushed to the first user terminal includes: In the live data stream list, searching for portrait data of a user group matching the first user portrait data; In the live data stream list, search for identification data of a live data stream that matches the portrait data of the user group as the identification data of the first live data stream.

[0017] Optionally, also include: If the portrait data of the user group matching the first user portrait data is not found in the live data stream list, determining whether there is live content feature data matching the first user portrait data in the live data stream list; If there is live content feature data matching the first user portrait data in the live data stream list, then the identification data of the live data stream corresponding to the live content feature data matching the first user portrait data is searched in the live data stream list as the identification data of the first live data stream, and the correspondence between the first user portrait data, the live content feature data matching the first user portrait data, and the identification data of the live data stream corresponding to the live content feature data matching the first user portrait data is recorded in the live data stream list.

[0018] Optionally, also include: If there is no live content feature data matching the first user portrait data in the live data stream list, determining whether updated live content feature data corresponding to the target live room can be obtained, the updated live content feature data being live content feature data corresponding to the target live room but not recorded in the live data stream list; If the updated live content feature data can be obtained, determining whether there is live content feature data matching the first user portrait data in the updated live content feature data; If the updated live content characteristic data contains live content characteristic data that matches the first user portrait data, the identification data of the live data stream is set for the first user portrait data as the identification data of the first live data stream. In the live data stream list, the correspondence between the first user portrait data, the live content characteristic data in the updated live content characteristic data that matches the first user portrait data, and the identification data of the first live data stream is recorded.

[0019] In a third aspect, the present application provides a live streaming device, comprising: a task acceptance instruction receiving unit, configured to obtain a task acceptance instruction issued by a live broadcast server for instructing to accept a task of pushing a first live broadcast data stream to a client, wherein the first live broadcast data stream is a live broadcast data stream corresponding to a target live broadcast room, the target live broadcast room corresponds to multiple live broadcast data streams, and the first live broadcast data stream is different from other live broadcast data streams corresponding to the target live broadcast room; a live broadcast material data request message sending unit, configured to send, in response to the task acceptance instruction, to the live broadcast server a live broadcast material data request message for requesting to obtain the live broadcast material data for generating the first live broadcast data stream; A live broadcast material data acquisition unit, used to obtain the live broadcast material data returned by the live broadcast service end in response to the live broadcast material data request message, wherein the live broadcast material data includes user portrait data, anchor image data, anchor timbre data, and first service item data that the anchor can provide to the user; A first live data stream generating unit, configured to generate the first live data stream according to the live material data; A streaming instruction acquisition unit, used to obtain a streaming instruction sent by the live streaming service end to instruct to push the first live streaming data stream to the first user end; A pushing unit is used to push the first live data stream to the first user terminal according to the streaming instruction.

[0020] In a fourth aspect, the present application provides a live streaming device, comprising: A live broadcast room access request message acquisition unit, used to obtain a live broadcast room access request message sent by the first user terminal for accessing a target live broadcast room; A target cloud device selection unit, used to select a target cloud device from a cloud device list for pushing the first live data stream to the client; A task acceptance instruction sending unit, used for sending a task acceptance instruction to the target cloud device for instructing the target cloud device to accept the task of pushing the first live data stream to the client; a live broadcast material data request message acquisition unit, configured to obtain a live broadcast material data request message issued by the target cloud device in response to the task acceptance instruction for requesting to obtain live broadcast material data, wherein the live broadcast material data is material data used to generate the first live broadcast data stream; A live broadcast material data sending unit, configured to send the live broadcast material data to the target cloud device in response to the live broadcast material data request message, wherein the live broadcast material data includes first user portrait data, anchor image data, anchor timbre data, and first service item data that the anchor can provide to the user; A live stream distribution strategy acquisition unit, used to obtain the identification data of the target live broadcast room, and obtain the live stream distribution strategy corresponding to the identification data of the target live broadcast room; A first live data stream determining unit is used to determine, according to the first user portrait data and the live stream distribution strategy, a first live data stream corresponding to the target live broadcast room to be pushed to the first user terminal, wherein the first live data stream is different from a second live data stream pushed to a second user terminal accessing the target live broadcast room; The first live data stream pushing unit is used to push the first live data stream to the first user terminal in response to the live room access request message.

[0021] In a fifth aspect, the present application provides a live streaming system, which is characterized by comprising a cloud end and a server end: The cloud end is used to obtain a task acceptance instruction issued by the live broadcast service end for instructing to accept the task of pushing a first live broadcast data stream to a client, the first live broadcast data stream is a live broadcast data stream corresponding to a target live broadcast room, the target live broadcast room corresponds to multiple live broadcast data streams, and the first live broadcast data stream is different from other live broadcast data streams corresponding to the target live broadcast room; in response to the task acceptance instruction, a live broadcast material data request message is issued to the live broadcast service end for requesting to obtain live broadcast material data for generating the first live broadcast data stream; the live broadcast material data returned by the live broadcast service end in response to the live broadcast material data request message is obtained, the live broadcast material data includes user portrait data, anchor image data, anchor timbre data and first service item data that the anchor can provide to the user; the first live broadcast data stream is generated according to the live broadcast material data; the push stream instruction sent by the live broadcast service end for instructing to push the first live broadcast data stream to a first user end; in response to the push stream instruction, the first live broadcast data stream is pushed to the first user end; The server is used to obtain a live broadcast room access request message for accessing a target live broadcast room issued by a first user end; select a target cloud device for pushing a first live broadcast data stream to a client from a cloud device list; issue a task acceptance instruction to the target cloud device for instructing the target cloud device to undertake the task of pushing the first live broadcast data stream to the client; obtain a live broadcast material data request message for requesting to obtain live broadcast material data issued by the target cloud device in response to the task acceptance instruction, wherein the live broadcast material data is material data for generating the first live broadcast data stream; and send the live broadcast material data to the target cloud device in response to the live broadcast material data request message. According to the data, the live broadcast material data includes first user portrait data, anchor image data, anchor voice data and first service item data that the anchor can provide to the user; obtain the identification data of the target live broadcast room, and obtain the live broadcast stream distribution strategy corresponding to the identification data of the target live broadcast room; according to the first user portrait data and the live broadcast stream distribution strategy, determine the first live broadcast data stream corresponding to the target live broadcast room to be pushed to the first user terminal, wherein the first live broadcast data stream is different from the second live broadcast data stream pushed to the second user terminal accessing the target live broadcast room; in response to the live broadcast room access request message, push the first live broadcast data stream to the first user terminal.

[0022] In a sixth aspect, the present application provides an electronic device comprising: a processor, a memory, and computer program instructions stored in the memory and executable on the processor; when the processor executes the computer program instructions, the method described in the first aspect is implemented.

[0023] In a seventh aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are executed by a processor, they are used to implement the method described in the first aspect.

[0024] Compared with the prior art, this application has the following advantages: The live streaming method provided in the embodiment of the present application generates a first live streaming data stream based on acquired live streaming material data, wherein the live streaming material data includes user portrait data, and the first live streaming data stream pushed to the user based on the user portrait data conforms to the user portrait.

[0025] The live streaming method provided in the embodiment of the present application pushes different live data streams based on the portrait data of different users. Even if the user clicks on the same target live room, different live data streams will be obtained due to different user portraits, thereby seeing different live content. When obtaining the task acceptance instruction issued by the live broadcast server, the cloud will return a request message for requesting live broadcast material data to the live broadcast server to obtain the live broadcast material data generated by the cloud, and generate a first live broadcast data stream corresponding to the user portrait data based on the live broadcast material data. This method can meet the user's demand for the live broadcast room, thereby increasing the viewing time of the live broadcast room, and can enable different users to see different live broadcast room contents when entering the same live broadcast room, so that the live broadcast room content and live broadcast style watched by users are more in line with user preferences. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 A schematic diagram of an application scenario provided for a scenario implementation example.

[0027] Figure 2 A schematic diagram of a live streaming push scenario provided for a scenario implementation example.

[0028] Figure 3 A flowchart of a live streaming method provided for a scenario embodiment.

[0029] Figure 4 This is a flowchart of the live streaming method provided in Example 1.

[0030] Figure 5 A flow chart of a method for determining multiple live data streams provided in Example 1.

[0031] Figure 6 A schematic diagram of a live streaming device provided in Example 2.

[0032] Figure 7 This is a flow chart of the cloud-based live streaming method provided in Example 3.

[0033] Figure 8Schematic diagram of the live streaming system provided in Example 4.

[0034] Fig. 9 This is a schematic diagram of an electronic device provided in Example 5. DETAILED DESCRIPTION

[0035] In order to enable those skilled in the art to better understand the technical solution of the present application, the present application is described clearly and completely below in conjunction with the drawings in the embodiments of the present application. However, the present application can be implemented in many other ways different from the following description. Therefore, based on the embodiments provided by the present application, all other embodiments obtained by those of ordinary skill in the art without creative work should fall within the scope of protection of the present application.

[0036] It should be noted that the terms "first", "source domain", "third", etc. in the claims, description and drawings of the present application are used to distinguish similar objects and are not used to describe a specific order or sequence. The data used in this way are interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including", "having" and their variants are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0037] Live streaming is a technology that transmits video and audio content in real time over the Internet. It has the characteristics of strong real-time, strong interactivity, and diverse platforms. The real-time nature of live streaming allows viewers to watch or listen to content instantly during playback without having to wait for the recorded file to be completed; the interactivity is reflected in the fact that viewers can interact with the anchor by sending bullet screens, liking, following, sending gifts, etc.; and the platform diversity lies in the fact that live streaming can be carried out through social media, live streaming platforms, e-commerce platforms, and enterprise-built live streaming systems. Therefore, whether it is news reports, entertainment programs, sports events, or educational courses, product launches, product recommendations, and even personal sharing, talent shows, and voice radio stations, they are all conducted in the form of live streaming.

[0038] Live streaming refers to the process of transmitting video or audio content from a local device or cloud to a live streaming platform, which is one of the important links in live streaming technology. Live streaming platforms refer to online platforms that provide live streaming media services, through which users can push video content live. Streaming platforms generally provide streaming software, streaming protocol support, and real-time video processing and distribution functions. At present, live streaming platforms will recommend and display live broadcast rooms on the user-side live streaming platform interface in the form of preview windows, cards, lists, etc. according to user preferences. Users can select preview windows, cards, and other entrances of their interest through the live streaming platform to enter and watch live content. This may cause too many people in the live broadcast room, and the anchor cannot explain and reply to a certain user in time, resulting in poor user experience. In order to reply to users' interactive information in time, the existing technology proposes digital human live broadcast technology. Digital humans include virtual characters or virtual anchors. Replacing real anchors with digital humans can improve interaction efficiency and reduce labor costs. However, the live content obtained by different users by clicking on the same live broadcast room entrance is still the same. Therefore, the live broadcast content and live broadcast style provided by the above two live broadcast methods cannot meet the preferences of each user, and the preferences of different users are different, which leads to a decrease in users' interest in the live broadcast room and a shorter viewing time.

[0039] In order to solve the above technical problems, this application proposes a live streaming method to recommend live content that is more in line with user preferences. When different user terminals send access requests to the same target live broadcast room, the live streaming method provided by this application determines the live data stream corresponding to the user terminal according to different user portrait data, so that each user receives different recommended content, which is closer to the user's needs and preferences, and helps to increase the user's interest and duration in watching live broadcasts.

[0040] To facilitate understanding of the live streaming method of this application, its application scenarios are introduced. Figure 1 , Figure 1 The application scenario diagram provided for the scenario embodiment. The application scenario mainly includes a server 101, a cloud 102, a live broadcast terminal 103 and a user terminal 104. Figure 2 and Figure 3 , Figure 2 A schematic diagram of a live streaming push scenario provided for a scenario embodiment, Figure 3 A flowchart of a live streaming method provided for a scenario embodiment. Figure 1 , Figure 2 and Figure 3, each end cooperates with each other to complete the live streaming. Among them, the server 101 includes a live streaming server 1011, a search and push server 1012, a machine scheduling server 1013 and a user server 1014; the cloud 102 includes a cloud host cluster 1021 and a digital human algorithm engineering 1022, and the cloud host cluster 1021 includes one or more cloud machines; the live streaming end 103 includes an anchor end 1031 and a central console 1032. The live streaming method provided in this embodiment includes the following steps: S301: Create a live broadcast. The anchor adjusts the live broadcast related data through the configuration interface on the anchor side and initiates a broadcast start request. The anchor side synchronizes the live broadcast related data to the central console, and the central console generates configuration information based on the live broadcast related data. The central console sends the live broadcast creation message to the live broadcast server to create a live broadcast for the anchor. At the same time, the live broadcast server returns the result of creating the live broadcast to the central console.

[0041] S302: Allocate cloud machines. The live broadcast server receives a message for creating a live broadcast, and calls the machine scheduling server to allocate a cloud machine for the live broadcast so as to provide live broadcast services on the cloud. The cloud machine information is pre-stored in the machine scheduling server, and the storage method can be a list, table, or other data structure that is easy for the machine to read and store. The allocation of cloud machines can be achieved in the following way: the machine scheduling server allocates the corresponding cloud machine for the live broadcast based on the received message for creating a live broadcast and the cloud machine information, and can also create an association between the cloud machine and the live broadcast. The allocated cloud machine information and the association between the cloud machine and the live broadcast room can be included in the streaming information, and the machine scheduling server sends the streaming information to the cloud and the live broadcast server.

[0042] Step S302 can complete the task of allocating a preset number of live data streams of preset types corresponding to the target live broadcast, wherein each live data stream can render different live content and pictures.

[0043] S303: Start the cloud machine. After receiving the push stream information sent by the machine scheduling server, the cloud machine parses the assigned cloud machine information carried in the push stream information, and transparently transmits the push stream information to the assigned cloud machine. After receiving the corresponding push stream information, the cloud machine parses the push stream information and starts the corresponding service program, preparing to provide live broadcast service for the corresponding live broadcast.

[0044] S304: Pulling materials. After receiving the streaming information, each cloud machine pulls the digital human materials related to the live broadcast from the live broadcast server. Specifically, this step can be implemented in any of the following two ways. Method 1: The cloud uniformly initiates a material pull request to the live broadcast server, and the live broadcast server responds to the material pull request and returns the corresponding digital human materials to the cloud; Method 2: Different cloud machines initiate material pull requests to the live broadcast server, and the live broadcast server responds to the material pull request and returns the corresponding digital human materials to each cloud machine. In comparison, method 1 is more efficient in initiating requests and obtaining return results, and although method 2 requests and obtains materials separately, since each cloud machine may render different digital humans and pictures in the subsequent streaming and rendering process, method 2 is conducive to improving the execution efficiency of subsequent tasks.

[0045] Among them, the digital human materials include the digital human's image information, the digital human's timbre information, etc. In addition to the digital human materials, the requested and obtained materials may also include other materials related to the live broadcast room, such as backgrounds, decorative patches, interactive controls, etc.

[0046] S305: Digital human data inference. After the cloud obtains the digital human material, the digital human algorithm engineering is called to infer the digital human data based on the digital human material. The digital human data includes the PCM audio data and lip inference data of the digital human. PCM audio data is used to convert analog audio signals into digital signals and store or transmit audio information in an original, uncompressed manner. Lip inference data can be used to drive the lip shape and lip movements of the digital human.

[0047] Specifically, the digital human image ID, timbre ID, etc. are extracted from the digital human material, combined with the corresponding copy information, to obtain the analog audio signal of the digital human, which is further converted into a digital signal and stored in the form of PCM (Pulse Code Modulation) audio data of the digital human. At the same time, the movement and morphological changes of the digital human's lips are obtained according to the digital human image ID and timbre ID, and then the lip reasoning data is obtained. For example, the corresponding lip shape, body movement and other data when the digital human makes a certain sound. Among them, the copy can be generated according to the user's preferences through the large language model deployed in the cloud, and it can be input into the digital human data.

[0048] After the digital human data is obtained by the digital human algorithm engineering reasoning, the digital human data is passed to the cloud host cluster through inter-process communication for subsequent rendering of the digital human. If step S304 adopts the second method, in this step, the digital human data is passed to the corresponding cloud host through inter-process communication. The inter-process communication methods include pipes, named pipes, message queues, etc., among which pipes are suitable for communication between parent and child processes or brother processes, named pipes are similar to unnamed pipes, but can communicate between unrelated processes, and message queues allow different processes to send and receive data asynchronously. In this step, the pipe method is preferably used to make data reading fast and easy to use.

[0049] S306: Rendering of digital human screen in the cloud. Based on the digital human data obtained in the above steps, the cloud renders the digital human screen, preliminarily constructs the prototype of the digital human, and transmits the digital human screen to the digital human cloud communication SDK (Software Development Kit). In addition, the cloud also sends the digital human data obtained by reasoning in the above steps to the digital human cloud communication SDK. The digital human cloud communication SDK can be understood as the communication interface of the digital human in the cloud, through which the digital human data can be transmitted to the user end.

[0050] S307: Determine the distribution strategy. The distribution strategy of the live stream is determined by the live server. The live server stores the live room information, which includes the association between the live room and the live data stream and the live data stream identifier. The live data stream identifier can represent the live room type, live room partition and other features used for distribution. The distribution strategy can include the correspondence between the live data stream identifier and the live room and the correspondence between the live data stream identifier and the user information. For example, the live data stream identifier A corresponds to live broadcast 1 and corresponds to user type A. Then, through the distribution strategy, it can be known that the live data stream identifier A should be recommended to user a whose user information contains the user type A identifier, and user a can see the corresponding live broadcast 1 on the user end.

[0051] The live broadcast server determines the distribution strategy and sends it to the search and push server. The search and push server will perform the distribution task for the live broadcast room according to the distribution strategy and other related information.

[0052] At the end of this step, the data reasoning and rendering of the digital human on the cloud have been completed, but the digital human image has not been completed. Therefore, it is necessary to continue processing the digital human data and rendering the digital human image on the client in the subsequent process.

[0053] S308: Live broadcast distribution. The search and push server recommends the live broadcast room content to matching users according to the distribution strategy and live broadcast room information.

[0054] It should be noted that steps S307 and S308 are not necessarily executed after S306. In fact, if the live broadcast server has obtained the cloud machine allocation information and the live broadcast room information, it can generate a distribution strategy; and when the user clicks to enter the live broadcast room, the user end can obtain the user information and user portrait, and the live broadcast room can be distributed at this time. The user portrait can be represented by the user information, and the user information can contain multiple user identifiers, such as user type, the type of live broadcasts that the user often watches, the type of goods that the user often buys, and the type of time the user watches live broadcasts.

[0055] S309: Synchronize digital human data. The digital human cloud communication SDK sends the received digital human related information to the user end so that the user end can render the live screen according to the digital human related information. The user end is also provided with a digital human end-side rendering SDK as the communication interface of the digital human on the user end. The digital human cloud communication SDK sends the digital human's PCM audio data stream and lip drive data and other information to the digital human end-side rendering SDK through the HTTP communication protocol, and the user end can receive the digital human data transmitted from the cloud through the digital human end-side rendering SDK. Specifically, the digital human's PCM audio data and lip data are generated in the cloud. It is necessary to pull the PCM audio data and lip data of the live data stream corresponding to the live data stream identifier from the cloud according to the live data stream identifier, and then transmit it to the user end through the digital human cloud communication SDK and the digital human end-side rendering SDK, so as to assemble the digital human's lip shape and character image, and enhance the face and adjust the naturalness of the synthesis on the user end.

[0056] S310: User enters the live broadcast room. The user clicks to enter the live broadcast room, and the user end sends the user entry event to the user server. The user server sends the user information to the search and push server. The search and push server allocates the live broadcast data stream to the user according to the distribution strategy determined in S307, and the user automatically enters the recommended live broadcast data stream. This step can occur in step S308. Take the e-commerce platform as an example to illustrate the practical application of this step. The e-commerce platform recommendation page recommends the target live broadcast room to the user in the form of a card. User A clicks card A, and the user end sends the event of user A entering the live broadcast room to the user server. The user server sends the event and user information to the search and push server. The search and push server allocates the live broadcast data stream to user A according to the preset distribution strategy. Assuming that when analyzing the user portrait, it is found that user A is interested in mobile phones, then the live broadcast data stream A corresponding to the channel of live mobile phone content will be allocated to user A. On the user side, after the user clicks the card, it will automatically jump to the live broadcast data stream A to see the live broadcast content related to the mobile phone. Correspondingly, if user B is interested in the laptop, when user B clicks card A, the search and push server will distribute the live data stream B containing the laptop-related content to user B according to the preset distribution strategy, and user B can directly jump to the live content corresponding to the live data stream B on the user side. Therefore, even if user A and user B click on the entrance of the same live room, they will see different live content through different live data streams. Please refer to Figure 2 , Figure 2 Schematic diagram of the live streaming push scenario provided for the scenario implementation example. Taking the live streaming of an electronic goods merchant in an e-commerce platform as an example, the application of the distribution strategy of the live streaming room in the streaming process is explained. When different users click on the same live streaming room entrance on different user terminals, they may be recommended to the default live streaming data stream 201, or they may be recommended to the live streaming data stream 202 corresponding to each category. For example, if a user does not have a clear goal and just browses around on the live streaming platform, by clicking on the live streaming room entrance, he will be pushed to the default live streaming data stream 201 and see the default live streaming content; if the user portrait data shows that the user is interested in products in the mobile phone category, the mobile phone category live streaming data stream will be automatically recommended, and the content seen by the user includes the introduction and display of the mobile phone category.

[0057] S311: The user renders the digital human screen. The digital human end-side rendering SDK infers and assembles the relevant data on the client side, and finally generates the digital human screen and renders it to the mobile phone screen. It can be seen that after the digital human data inference and preliminary screen rendering are performed on the cloud, the digital human end-side rendering SDK is deployed on the client to provide digital human algorithm inference capabilities. Therefore, the client can further process the digital human, assemble the digital human's corresponding lip shape and digital human image into one, integrate the lip features into the digital human face, and appropriately adjust the naturalness and clarity of the facial synthesis, so that the live broadcast picture seen by the client is natural and smooth, and the audio and picture are synchronized. This does not need to occupy cloud resources and reduce the pressure of cloud data processing, and also allows users to quickly see the rendered picture on the client.

[0058] S312: Processing bullet screen information. The user end receives the bullet screen information sent by the user and transmits it to the cloud. The cloud generates live broadcast room reply information based on the bullet screen information and sends it to the user end to provide feedback on the bullet screen information of the user end. At the same time, the user end also synchronizes the user information to the cloud, which can be synchronized to the cloud through the user server or the digital human SDK. Based on the user information, the cloud uses RLHF (Reinforcement Learning with Human Feedback) preference learning to train the digital human's human interaction and other capabilities to train a personalized digital human anchor for the user, that is, the feedback information of the digital human in the live broadcast room seen by the user is obtained based on the user information training, which is more in line with the user's preferences. In the above steps, user A and user B who click on the same live broadcast room entrance are respectively assigned to different live broadcast data streams according to the user portrait. In this step, according to other information in the user portrait, the digital human in different live broadcast data streams can also be gradually optimized. For example, the host in live data stream A is digital human A, while the host in live data stream B is digital human B. If user A prefers to watch cute pet images, while user B prefers to watch anthropomorphic images, then the digital human in live data stream A uses cute pet images as the host, while the digital human in live data stream B uses anthropomorphic images as the host. For another example, if user A likes a humorous live broadcast style, while user B likes a straightforward and efficient communication method, then in live data stream A, the explanation style, body movements, and language style of the bullet screen of digital human A will be adjusted accordingly to match the humorous live broadcast style that user A likes; similarly, in live data stream B, the explanation style and language style of the bullet screen of digital human B will be adjusted to a straightforward and capable style.

[0059] The above steps split the live broadcast digital human rendering, with the cloud and the client processing part of the digital human rendering task respectively, and finally displaying the digital human rendering picture on the client, reducing the occupation of cloud resources, alleviating the pressure of cloud data processing, and improving the efficiency of live broadcast push and loading. Since the distribution strategy is determined according to information such as user preferences in the above steps, and the live broadcast data stream is recommended to the user according to the distribution strategy, and different live broadcast data streams generate different live broadcast content and live broadcast formats, the live broadcast content and live broadcast format seen by different users entering the same live broadcast room are not the same. This embodiment realizes the recommendation of different live broadcasts based on different users, and improves the differentiation and personalization of live broadcast recommendations.

[0060] Embodiment 1 provides a live streaming method, and the execution subject of the method may be a server. Part of the content in embodiment 1 has been mentioned in the above content, and the relevant content may refer to the above embodiment.

[0061] Please see Figure 4 , Figure 4 This is a flowchart of a live streaming method provided in Example 1, and the live streaming method includes the following steps: S401: Obtain a live broadcast room access request message sent by a first user terminal for accessing a target live broadcast room.

[0062] When the first user terminal detects that the user clicks on the live broadcast room entrance, it sends a live broadcast room access request message for accessing the target live broadcast room, and the server receives the live broadcast room access request message sent by the first user terminal. On a dedicated live broadcast platform, the recommendation page often displays preview images or covers of multiple live broadcasts as live broadcast entrances. On a comprehensive platform, in addition to displaying the live broadcast entrance, the recommendation page may also display text, pictures, mini-program jump links and other entrances. Users can enter the live broadcast room by clicking on the live broadcast entrance on any page. The target live broadcast room can be understood as a basic live broadcast room or a main live broadcast room. Taking an e-commerce platform as an example, the target live broadcast room can represent a full-category live broadcast room of a certain merchant, with rich live broadcast content and diverse live broadcast styles, which is suitable for users to watch when their purpose is not strong or their preferences are not obvious.

[0063] S402: Obtain portrait data of the first user corresponding to the first user terminal as first user portrait data.

[0064] User portrait data usually refers to a comprehensive user profile created by collecting and analyzing user behavior, interests, needs, preferences and other information. These data help companies or platforms better understand user groups and make personalized recommendations, product designs, and recommendation strategy formulation. The collection and analysis of user portraits can mainly start from the following aspects: basic personal information, including browsing history, purchase behavior and usage habits, search behavior data, interests and preferences, social data, device and platform information, active time periods, etc. Specific user portrait data is not limited to the scope listed above and can be collected according to specific needs during use.

[0065] S403: Determine, based on the first user portrait data, a first live broadcast data stream corresponding to the target live broadcast room to be pushed to the first user terminal, wherein the first live broadcast data stream is different from a second live broadcast data stream pushed to a second user terminal accessing the target live broadcast room.

[0066] According to the first user portrait data obtained in the above steps, the first live broadcast data stream to be pushed to the first user terminal can be preliminarily determined, thereby determining the corresponding live broadcast content displayed to the first user terminal. Similarly, a live broadcast room access request message issued by the second user terminal for accessing the target live broadcast room is obtained, and the portrait data of the second user corresponding to the second user terminal is obtained as the second user portrait data. According to the second user portrait data, the second live broadcast data stream corresponding to the target live broadcast room to be pushed to the second user terminal is determined. Since the second live broadcast data stream is determined based on the second user portrait data and is recommended to the second user terminal, the target live broadcast room content seen by the second user on the second user terminal may be different from that seen by the first user on the first user terminal. Among them, the second user terminal can be one or more user terminals different from the first user terminal.

[0067] S404: Pushing the first live broadcast data stream to the first user terminal in response to the live broadcast room access request message.

[0068] The above steps determine the first user portrait data, and further determine the first live data stream corresponding to the target live room to be pushed to the first user terminal. In this step, the server pushes the first live data stream to the first user terminal, and the first user can obtain the live content corresponding to the first live data stream through the first user terminal.

[0069] Optionally, the live streaming method provided in Embodiment 1 further includes: obtaining identification data of the target live room, and obtaining a live streaming distribution strategy corresponding to the identification data of the target live room.

[0070] The target live broadcast room is associated with one or more live broadcast data streams, and different live broadcast data streams can provide different live broadcast content for the user end. The live broadcast data streams associated with the same live broadcast room can be recommended to different user ends according to the live broadcast stream distribution strategy. The identification data of the target live broadcast room includes data that can represent the type of live broadcast room, the characteristics of the live broadcast room, the content of the live broadcast room, etc. The characteristic information of the target live broadcast room can be obtained through the identification data of the target live broadcast room. Combined with the identification representation data of the target live broadcast room, the technical effect of recommending the live broadcast data stream to the user end is achieved through the live broadcast stream distribution strategy.

[0071] The method of determining the first live broadcast data stream corresponding to the target live broadcast room to be pushed to the first user terminal based on the first user portrait data includes: determining the first live broadcast data stream to be pushed to the first user terminal based on the first user portrait data and the live broadcast stream distribution strategy.

[0072] By obtaining the identification data of the target live broadcast room, the characteristic information of the target live broadcast room can be obtained. At the same time, by obtaining the first user portrait data, the preference information of the first user can be obtained. Combined with the live broadcast stream distribution strategy, the live broadcast stream distribution strategy that can be obtained is more accurate, and it is determined that the first live broadcast data stream is more in line with the user's preferences, which improves the matching degree and matching efficiency between the live broadcast room and the user.

[0073] Before determining the first live broadcast data stream to be pushed to the first user terminal, the following steps may also be performed, including: selecting a target cloud device for pushing the first live broadcast data stream to the client from a cloud device list; issuing a task acceptance instruction to the target cloud device for instructing the target cloud device to undertake the task of pushing the first live broadcast data stream to the client; obtaining a live broadcast material data request message issued by the target cloud device for requesting to obtain live broadcast material data in response to the task acceptance instruction, wherein the live broadcast material data is material data used to generate the first live broadcast data stream; in response to the live broadcast material data request message, sending the live broadcast material data to the target cloud device; after determining the first live broadcast data stream corresponding to the target live broadcast room to be pushed to the first user terminal, determining the target cloud device in the cloud device list that has undertaken the task of pushing the first live broadcast data stream to the client; in response to the live broadcast room access request message, pushing the first live broadcast data stream to the first user terminal includes: in response to the live broadcast room access request message, issuing a push stream instruction to the target cloud device for instructing to push the first live broadcast data stream to the first user terminal.

[0074] Among them, the relevant information of the cloud machines in the cloud host cluster is stored in the cloud device list in the form of a list. One or more instances can be run on each cloud machine, and the cloud machine is the cloud device. When the live broadcast is created, a start message is received, and the live broadcast server needs to allocate a cloud machine to provide live broadcast service. This step can be understood in conjunction with the content in the aforementioned step S302. After determining the target cloud device for pushing the first live broadcast data stream, the live broadcast server sends a task acceptance instruction to the target cloud device, and the task acceptance instruction is used to instruct the target cloud device to push the first live broadcast data stream to the client. In other words, determining the target cloud device means that the target cloud device needs to be responsible for pushing the first live broadcast data stream to the client during the subsequent streaming process.

[0075] When the target cloud device receives the task acceptance instruction, it sends a live material data request message to the live broadcast server for requesting to obtain live material data. The live material data includes material data for generating the first live data stream, and may also include material data for generating the second live data stream, etc., or includes material data that can generate all associated live data streams. The live broadcast server receives the live material data request message sent by the target cloud device, and returns the corresponding live material data to the target cloud device. The live material data request message may include requirements such as the type information and feature information of the live material. The live broadcast server will send the corresponding live material data to each target cloud device according to the requirement information, or the live broadcast server will return all live material data used to generate the live data stream to the cloud.

[0076] The live material data obtained above can be determined as the first live data stream corresponding to the target live room pushed by the first user terminal. The target cloud device is found in the cloud device list, and a push stream instruction is sent to the target cloud device to instruct the target cloud device to push the first live data stream to the first user terminal.

[0077] The above content introduces the process of determining the first live broadcast data stream based on the first user portrait data, so that the live broadcast content seen by the first user can be consistent with the user portrait and have a higher match with the user. Similarly, a target live broadcast room may be recommended to multiple different users, and multiple different users may have the same attributes. Therefore, the live broadcast streaming method also includes: obtaining portrait data of multiple user groups corresponding to the target live broadcast room; obtaining live content feature data corresponding to the target live broadcast room; and determining multiple live broadcast data streams corresponding to the target live broadcast room based on the portrait data of the multiple user groups and the live content feature data.

[0078] Users with the same attributes or portraits can be divided into a user group, and each user will be divided into at least one user group. The specific division criteria can be determined according to actual needs. Different users here can be users who have watched the live broadcast room in the past. Specifically, by obtaining the portrait data of historical viewers and extracting the corresponding user features, the users are divided. By dividing different users, multiple user groups can be obtained. Each user group has corresponding portrait data, and the portrait data can reflect his preferences and habits for live broadcasts. According to the portrait data of multiple user groups and the live content feature data, different multiple live data streams corresponding to the target live broadcast room can be determined. The division of historical audiences can realize the pre-division of the audience of the live broadcast room even when no one is watching the live broadcast.

[0079] The portrait data of multiple user groups corresponding to the target live broadcast room may be obtained in a variety of ways, including: Method 1: obtaining the portrait data of historical users who visited the target live broadcast room, performing clustering processing on the portrait data of historical users who visited the target live broadcast room, and obtaining the portrait data of multiple user groups corresponding to the target live broadcast room; Alternatively, method 2: obtaining portrait data of historical users who have visited the anchor data page corresponding to the target live broadcast room, performing clustering processing on the portrait data of historical users who have visited the anchor data page corresponding to the target live broadcast room, and obtaining portrait data of multiple user groups corresponding to the target live broadcast room; Alternatively, method three: obtain portrait data of historical users who have accessed a data service system, where the target live broadcast room is located, and cluster the portrait data of historical users who have accessed the data service system to obtain portrait data of multiple user groups corresponding to the target live broadcast room.

[0080] Method 1 means obtaining the portrait data of multiple user groups corresponding to the target live broadcast room through historical users who have visited the target live broadcast room. Among them, historical users can be all users who have entered the live broadcast room, or users who have entered the live broadcast room within a certain period of time, or users whose single or cumulative duration of stay in the live broadcast room exceeds a certain threshold. This is to reflect the portrait data related to the user and the target live broadcast room to a certain extent. Therefore, method 1 can accurately analyze historical users, and the obtained portrait data has a certain regularity. Method 2 means obtaining the portrait data of multiple user groups corresponding to the target live broadcast room through users who have visited the page data of the live broadcast initiator (such as a merchant). Among them, the page data of the live broadcast initiator (such as a merchant) can be the anchor data page, for example, the anchor will correspond to the dynamic information of the merchant, the store page, etc. Therefore, method 2 can accurately measure the user's preference for the product and the anchor. Method 3 means obtaining the portrait data of multiple user groups corresponding to the target live broadcast room through users who have visited the e-commerce platform (live broadcast platform). Among them, the portrait data of historical users can include browsing records, purchase information, attention information, viewing history, etc. generated after the user settles in the e-commerce platform. Similarly, historical users in methods 2 and 3 can be all users who have visited the anchor data page or data service system, or users who have visited the anchor data page or data service system within a certain period of time, or users whose single or cumulative duration of visits to the anchor data page or data service system exceeds a certain threshold. The above three methods include users obtained from different channels, and users obtained in each method may enter the live broadcast room. Therefore, further obtaining the portrait data of these users to obtain the portrait data of the user group used by the target live broadcast room will help improve the accuracy of the results.

[0081] The user portrait data obtained above can represent the user's preferences, habits and other characteristics. The following content will explain how to obtain the characteristics of the live broadcast room.

[0082] The live broadcast content feature data corresponding to the target live broadcast room may be obtained in the following two ways, including: Method 1: obtaining historical live broadcast material data of the target live broadcast room, and obtaining live broadcast content feature data corresponding to the target live broadcast room according to the historical live broadcast material data of the target live broadcast room; Method 2: Alternatively, characteristic data of service items that the anchor corresponding to the target live broadcast room can provide to users are obtained as characteristic data of live broadcast content corresponding to the target live broadcast room.

[0083] Among them, the historical live broadcast materials of the target live broadcast room in the first method include the background materials of the live broadcast room, the anchor materials, the copywriting materials, the live broadcast data, etc. The historical live broadcast materials can represent the content and theme of the historical live broadcast, so that the live broadcast content feature data corresponding to the target live broadcast room can be obtained from them. The second method is to use the service items that the anchor can provide as a guide to obtain the characteristics of the service items to determine the live broadcast content feature data. For example, when the anchor is a merchant on an e-commerce platform, the service items are the goods provided by the merchant; when the anchor is a merchant on a takeaway platform, the service items are the goods and delivery services provided by the merchant; when the anchor is an online education platform, the service items are online courses, teaching materials, online tutoring and other services provided by the anchor. These service items represent the characteristics of the anchor and reflect the content characteristics of the live broadcast room. Therefore, the feature data of the service items that the anchor corresponding to the target live broadcast room can provide to users can be used as the live broadcast content feature data corresponding to the target live broadcast room. Furthermore, the anchor is a target merchant, and the live broadcast content feature data corresponding to the target live broadcast room includes the category data of the goods that the anchor can provide to users. Taking the e-commerce platform as an example, when the anchor is the target merchant, the live content feature data can also include the category data of the goods. For example, the target merchant mainly provides sports goods, and its domestic goods include hiking shoes, raincoats, sun-proof clothes, river tracing shoes, jackets, swimsuits, swimming caps, trekking poles, etc., then the categories of the target merchant's goods can include mountaineering sports, river tracing sports, swimming sports, and ball sports, and each category contains different product data.

[0084] After determining the portrait data and live content feature data of multiple user groups corresponding to the target live broadcast room, you can determine the multiple live broadcast data streams corresponding to the target live broadcast room. Figure 5 , Figure 5 A flow chart of a method for determining multiple live data streams provided in Example 1.

[0085] The determining, according to the portrait data of the multiple user groups and the live content feature data, multiple live data streams corresponding to the target live broadcast room comprises: S501: Determine matching live content feature data for the portrait data of each user group.

[0086] S502: Setting identification data of the live data stream for the portrait data of each user group.

[0087] S503: Establish a live data stream list of the target live broadcast room, in which the corresponding relationship between the portrait data of each user group, the matching live content feature data and the matching live data stream identification data is recorded.

[0088] Even when the live broadcast room has not yet started broadcasting or the live broadcast data stream has not yet been generated, the portrait data of each user group can be pre-matched with the live broadcast content feature data, and the type of live broadcast that should be recommended to each user group can be pre-determined, and the identification data of the live broadcast data stream can be pre-set for each user's portrait data, so as to pre-match the corresponding live broadcast data stream for each user group. Of course, after the live broadcast room is started or the live broadcast data stream is generated, the matching live broadcast content feature data and the identification data of the live broadcast data stream can be determined. Based on the above process, a live broadcast data stream list of the target live broadcast room is established, and the live broadcast data stream list records the correspondence between the portrait data of each user group, the matching live broadcast content feature data and the matching live broadcast data stream identification data.

[0089] After establishing the live broadcast data stream list of the target live broadcast room, optionally, determining the first live broadcast data stream corresponding to the target live broadcast room to be pushed to the first user terminal based on the first user portrait data, including: searching the live broadcast data stream list for portrait data of a user group that matches the first user portrait data; searching the live broadcast data stream list for identification data of a live broadcast data stream that matches the portrait data of the user group as identification data of the first live broadcast data stream.

[0090] For example, the live content feature data includes A, B, and C; and the user group portrait data includes user group 1, user group 2, and user group 3. The identification data of the live data stream corresponding to user groups 1, 2, and 3 are identification 1, identification 2, and identification 3, respectively; identification 1 corresponds to live content feature data A, indicating that identification 1 represents a live data stream whose live content feature data includes A; identification 2 corresponds to live content feature data B, and identification 3 corresponds to live content feature data C. Then, user group 1 matching the first user can be found in the live data stream list as the portrait data of the user group matching the first user portrait data. Further, based on the matching relationship, the identification data of the live data stream corresponding to user group 1, i.e. identification 1, can be found. The live data stream corresponding to identification 1 can be used to generate a live broadcast including live content feature data A.

[0091] Although users with the same attributes can be divided into a group, and each user will be divided into at least one user group, due to different methods of generating the live data stream list, the first user portrait data may not necessarily exist in the user group portrait data recorded in the live data stream list, so it is not necessarily possible to find the portrait data of the user group that matches the first user's speech data from the live data stream list according to the above method. For example, when the portrait data of historical users is used to obtain the portrait data of multiple user groups corresponding to the target live broadcast room, the first user portrait data may not have appeared in the historical user portrait data. Therefore, the first user portrait data may belong to a new user group. At this time, the first user portrait data does not belong to the portrait data of any user group. In this case, the live streaming method provided in Example 1 also includes: if the portrait data of the user group matching the first user portrait data is not found in the live data stream list, then judging whether there is live content feature data matching the first user portrait data in the live data stream list; if there is live content feature data matching the first user portrait data in the live data stream list, then searching the live data stream list for identification data of the live data stream corresponding to the live content feature data matching the first user portrait data as the identification data of the first live data stream, and recording the correspondence between the first user portrait data, the live content feature data matching the first user portrait data, and the identification data of the live data stream corresponding to the live content feature data matching the first user portrait data in the live data stream list. The above method determines the live data stream identification data corresponding to the live content feature data matching the first user portrait data by judging whether there is live content feature data matching the first user portrait data in the live data stream list.

[0092] That is to say, the portrait data of the user group that was originally matched by the first user portrait data matched the identification data of the corresponding live data stream in the live data stream list, but because the portrait data of the user group that matched the first user portrait data was not found in the live data stream list, and each live content feature data corresponds to the portrait data of one or more user groups. Therefore, the live data stream identification data can be determined by the live content feature data that matches the first user portrait data. For example, the first user portrait data belongs to user group 4, and the portrait data of this user group cannot be found in the live data stream list, but the user group 4 described in the first user portrait data matches the live content feature data A, that is, the live content feature data A is not only applicable to user group 1, but also to user group 4, then the corresponding live data stream identification data 1 can be determined by the live content feature data A, and it can be used as the identification data of the first live data stream that matches the first user portrait data.

[0093] Thus, a method for determining the identification data of the first live data stream is provided when the portrait data of the user group matching the first user portrait data cannot be found in the live data stream list. In addition, since the first user portrait data is new user portrait data, the user group 4 to which it belongs is also the portrait data of a new user group. Therefore, the correspondence between the first user portrait data, the live content feature data matching the first user portrait data, and the identification data of the live data stream corresponding to the live content feature data matching the first user portrait data is recorded in the live data stream list, so as to match the identification data of the live data stream for other users next time.

[0094] Taking into account that there may be no live content feature data matching the first user portrait data in the live data stream list, optionally, a live streaming method is provided, including: if there is no live content feature data matching the first user portrait data in the live data stream list, then judging whether updated live content feature data corresponding to the target live room can be obtained, the updated live content feature data being live content feature data corresponding to the target live room but not recorded in the live data stream list; if the updated live content feature data can be obtained, then judging whether there is live content feature data matching the first user portrait data in the updated live content feature data; if there is live content feature data matching the first user portrait data in the updated live content feature data, setting identification data of the live data stream for the first user portrait data as the identification data of the first live data stream, and recording in the live data stream list the correspondence between the first user portrait data, the live content feature data matching the first user portrait data in the updated live content feature data, and the identification data of the first live data stream.

[0095] Since the live content feature data may change every time the anchor performs a live broadcast, therefore, when there is no live content feature data matching the first user portrait data in the live data stream list, it can be considered whether the live content feature data of the target live broadcast room can be updated, and the updated live content feature data is more in line with the recent live broadcast characteristics. For example, if the live data stream list is created on January 1, 2024, and the anchor has performed several live broadcasts as of August 1, 2024, the live content feature data can be updated. Compare the updated live content feature data with the live content feature data matching the first user portrait data to determine whether there is live content feature data matching the first user portrait data in the updated live content feature data.

[0096] When the live content feature data matching the first user portrait data has not been found through the above-mentioned multiple methods, it is determined that the default live data stream corresponding to the target live room is to be pushed to the first user terminal. Specifically, it also includes: if the updated live content feature data is not obtained or the updated live content feature data does not contain the live content feature data matching the first user portrait data, it is determined that the default live data stream corresponding to the target live room is to be pushed to the first user terminal, and the correspondence between the first user portrait data and the identification data of the default live data stream corresponding to the target live room is recorded in the live data stream list, and the default live data stream corresponding to the target live room is a data stream that can be pushed to all user terminals that access the target live room; in response to the live room access request message, the default live data stream corresponding to the target live room is pushed to the first user terminal.

[0097] The first live data stream corresponding to the target live broadcast room to be pushed to the first user terminal can be indirectly determined based on the user terminal of the determined live data stream. The first live data stream corresponding to the target live broadcast room to be pushed to the first user terminal based on the first user portrait data includes: judging whether there is a user terminal matching the first user portrait data among the user terminals of the determined live data streams, and determining whether there is a user terminal matching the first user portrait data among the user terminals of the determined live data streams as a portrait data matching user terminal; if there is a portrait data matching user terminal among the user terminals of the determined live data streams, then determining the live data stream determined for the portrait data matching user terminal as the first live data stream.

[0098] The method obtains a user terminal that matches the first user portrait data from the user terminals of the determined live data stream, and further determines the first live data stream. It can be seen that the first live data stream that matches the first user portrait data can be quickly obtained. However, if there is no user terminal matching the portrait data in the user terminals of the determined live data stream, the identification data of the live data stream is set for the first user portrait data as the identification data of the first live data stream, that is, the identification data of the new live data stream is allocated to the first user; wherein the identification data of the first live data stream is different from the identification data of the live data stream determined for other user terminals. If the above steps cannot find the first live data stream that matches the first user portrait data, the following method for preparing to push the stream can also be used to push the default live data stream to the first user terminal. Specifically, it also includes: if the first user portrait data is not obtained, it is determined that the default live data stream corresponding to the target live room is to be pushed to the first user terminal, and the default live data stream corresponding to the target live room is a data stream that can be pushed to all user terminals that access the target live room; for the live room access request message, the default live data stream corresponding to the target live room is pushed to the first user terminal.

[0099] The above steps push the default live data stream of the target live broadcast room to the first user terminal, so that the user can watch the default live content. At the same time, through the feedback of the first user after entering the default live stream, other live data streams that are more suitable for the first user can be switched or recommended, and a new live data stream can be generated for the first user; the feedback may include the behavioral feature data of the first user in the target live broadcast room for the live content of the default live data stream, which can be specifically obtained through user terminal information combined with barrage and comments, likes and votes, viewing time and interaction frequency, click behavior, text analysis and sentiment analysis, user leaving and returning data, etc.

[0100] Furthermore, the live streaming method also includes: after pushing the default live data stream corresponding to the target live room to the first user terminal, obtaining the behavior characteristic data of the first user terminal for the target live room; determining the updated live data stream corresponding to the target live room to be pushed to the first user terminal based on the behavior characteristic data of the first user terminal for the target live room; switching the default live data stream corresponding to the target live room that has been pushed to the first user terminal to the updated live data stream corresponding to the target live room.

[0101] Among them, the first user terminal can determine the specific acquisition method of the behavioral characteristic data for the target live broadcast room by judging whether there is a user terminal among the user terminals of the determined live broadcast data stream whose behavioral characteristic data for the target live broadcast room matches the behavioral characteristic data for the target live broadcast room of the first user terminal, that is, judging whether the user behavior characteristics of different user terminals in the same live broadcast room are the same.

[0102] The method of determining, based on the behavior characteristic data of the first user terminal for the target live broadcast room, an updated live broadcast data stream corresponding to the target live broadcast room to be pushed to the first user terminal includes: judging whether there is a user terminal whose behavior characteristic data for the target live broadcast room matches the behavior characteristic data of the first user terminal for the target live broadcast room among the user terminals of the determined live broadcast data stream, and taking the user terminal whose behavior characteristic data for the target live broadcast room matches the behavior characteristic data of the first user terminal for the target live broadcast room among the user terminals of the determined live broadcast data stream as a behavior characteristic matching user terminal; if the behavior characteristic matching user terminal exists among the user terminals of the determined live broadcast data stream, determining the live broadcast data stream determined for the behavior characteristic matching user terminal as the updated live broadcast data stream corresponding to the target live broadcast room.

[0103] The above steps match the behavior feature data of the first user terminal by referring to the behavior features of the user terminal of the determined live data stream, so as to determine the behavior feature matching user terminal, and then determine the updated live data stream corresponding to the target live room pushed by the first user terminal. However, if the behavior feature matching user terminal does not exist in the user terminals of the determined live data stream, it is necessary to judge the live content feature data corresponding to the default live data stream corresponding to the target live room to determine it as the updated live data stream corresponding to the target live room.

[0104] If the user terminal with the behavior characteristic matching does not exist among the user terminals of the determined live broadcast data stream, it is determined whether the behavior characteristic data of the first user terminal for the target live broadcast room matches the live content characteristic data corresponding to the default live broadcast data stream corresponding to the target live broadcast room; if the behavior characteristic data of the first user terminal for the target live broadcast room does not match the live content characteristic data corresponding to the default live broadcast data stream corresponding to the target live broadcast room, a live broadcast data stream update task is generated, and the live broadcast data stream update task is used to indicate that an updated live broadcast data stream is generated for the first user terminal according to the behavior characteristic data of the first user terminal for the target live broadcast room, and the live content characteristic data corresponding to the updated live broadcast data stream generated for the first user terminal matches the behavior characteristic data of the first user terminal for the target live broadcast room; the updated live broadcast data stream generated for the first user terminal is determined as the updated live broadcast data stream corresponding to the target live broadcast room.

[0105] Optionally, the live streaming method also includes: if there is no user terminal with the behavior characteristic matching among the user terminals of the determined live data stream, then determining whether the behavior characteristic data of the first user terminal for the target live room matches the live content characteristic data corresponding to the default live data stream corresponding to the target live room; if the behavior characteristic data of the first user terminal for the target live room does not match the live content characteristic data corresponding to the default live data stream corresponding to the target live room, then generating a live data stream update task, the live data stream update task is used to indicate that an updated live data stream is generated for the first user terminal according to the behavior characteristic data of the first user terminal for the target live room, the live content characteristic data corresponding to the updated live data stream generated for the first user terminal matches the behavior characteristic data of the first user terminal for the target live room; and the updated live data stream generated for the first user terminal is determined as the updated live data stream corresponding to the target live room.

[0106] In the above steps, the live broadcast server can update the live broadcast data stream to obtain the updated live broadcast data stream, and the live broadcast server can also instruct the cloud to update the live broadcast data stream to generate updated live broadcast stream data.

[0107] The first live data stream is pushed to the first user terminal. After the first user enters the live broadcast room and sees the live broadcast, it can be determined whether the user's behavioral characteristics match the live content characteristics. For example, the live content characteristics in the live broadcast room are mobile phone categories, but the first user is asking questions about wearable devices in the live broadcast room. At this time, it can be determined that the first user's behavioral characteristics do not match the live content characteristics corresponding to the current live data stream. At this time, it can be found whether there is a live data stream of wearable devices in other live data streams corresponding to the current live broadcast room. If so, the live stream of the mobile phone category recommended on the first user terminal is switched to the live data stream corresponding to the wearable device category, so as to recommend the live broadcast service of the wearable device that the first user is more interested in. The live streaming method can also determine the updated live data stream recommended for the first user terminal based on whether there is a live data stream in other live data streams corresponding to the target live broadcast room that matches the behavioral characteristic data of the first user terminal for the target live broadcast room. Specifically, it also includes: after pushing the first live broadcast data stream to the first user terminal, judging whether the behavior characteristic data of the first user terminal for the target live broadcast room matches the live content characteristic data corresponding to the first live broadcast data stream; if the behavior characteristic data of the first user terminal for the target live broadcast room does not match the live content characteristic data corresponding to the first live broadcast data stream, judging whether there is a live broadcast data stream matching the behavior characteristic data of the first user terminal for the target live broadcast room in other live broadcast data streams corresponding to the target live broadcast room, and using the live broadcast data stream matching the behavior characteristic data of the first user terminal for the target live broadcast room in other live broadcast data streams corresponding to the target live broadcast room as the updated live broadcast data stream of the first user terminal; if the updated live broadcast data stream of the first user terminal exists in other live broadcast data streams corresponding to the target live broadcast room, switching the first live broadcast data stream pushed to the first user terminal to the updated live broadcast data stream of the first user terminal.

[0108] Among them, whether the behavioral characteristic data of the first user terminal for the target live broadcast room matches the live content characteristic data corresponding to the first live broadcast data stream can be determined by inputting the behavioral characteristic data of the first user and the live content characteristic data corresponding to the first live broadcast data stream into the matching model, and the matching model will output the corresponding matching result.

[0109] If the updated live data stream of the first user terminal does not exist in the other live data streams corresponding to the target live room, it indicates that the behavioral characteristic data of the first user terminal for the target live room may have never been seen in the target live room and the corresponding other data live streams. Therefore, even from the other live data streams corresponding to the target live room, it is impossible to find a live data stream that matches the behavioral characteristic data of the first user for the target live room. At this time, an updated live data stream can be provided by generating a new live data stream for the first user terminal. Specifically, it also includes: if the updated live data stream of the first user terminal does not exist in the other live data streams corresponding to the target live room, a live data stream update task is generated, and the live data stream update task is used to indicate that an updated live data stream is generated for the first user terminal according to the behavioral characteristic data of the first user terminal for the target live room, and the live content characteristic data corresponding to the updated live data stream generated for the first user terminal matches the behavioral characteristic data of the first user terminal for the target live room; the first live data stream that has been pushed to the first user terminal is switched to the updated live data stream generated for the first user terminal.

[0110] After determining the updated live data stream of the first user terminal, the method further includes: executing the live data stream update task to generate an updated live data stream for the first user terminal.

[0111] Continuing with the example given above, considering that the identification data of the live data stream is to be allocated to the user end, a live stream identifier is pre-assigned to the value data stream of the wearable device to be generated. After determining that it is the updated live data stream corresponding to the target live room, a response message sent by the cloud to indicate that the updated live data stream has been generated is obtained to execute the switching of the live data stream.

[0112] Corresponding to the aforementioned determination of the updated live data stream generated for the first user terminal as the updated live data stream corresponding to the target live broadcast room, the live broadcast streaming method also includes: pre-allocating identification data for the updated live data stream; sending the live data stream update task to the cloud, the live data stream update task including the identification data pre-allocated for the updated live data stream; obtaining a response message sent by the cloud to indicate that the updated live data stream has been generated, the response message including the identification data pre-allocated for the updated live data stream; the switching of the default live data stream corresponding to the target live broadcast room that has been pushed to the first user terminal to the updated live data stream corresponding to the target live broadcast room includes: sending a first live data stream switching indication message to the cloud for instructing the cloud to switch the default live data stream corresponding to the target live broadcast room that has been pushed to the first user terminal to the updated live data stream generated for the first user terminal, the first live data stream switching indication message including the identification data pre-allocated for the updated live data stream.

[0113] In the above process, the identification data of the corresponding live data stream is pre-assigned to the live stream of the wearable device, and then the task of generating the live stream of the wearable device is sent to the cloud. The server receives the message returned by the cloud that the live stream of the wearable device has been generated. The server will instruct the cloud to switch the live stream of the first user end: the default live stream originally pushed to the first user end is switched to the wearable live stream.

[0114] Corresponding to the aforementioned switching of the first live data stream that has been pushed to the first user terminal to the updated live data stream generated for the first user terminal, the live streaming method also includes: pre-allocating identification data for the updated live data stream; sending the live data stream update task to the cloud, the live data stream update task including the identification data pre-allocated for the updated live data stream; obtaining a response message sent by the cloud to indicate that the updated live data stream has been generated, the response message including the identification data pre-allocated for the updated live data stream; the switching of the first live data stream that has been pushed to the first user terminal to the updated live data stream generated for the first user terminal includes: sending a second live data stream switching indication message to the cloud for instructing the cloud to switch the first live data stream that has been pushed to the first user terminal to the updated live data stream generated for the first user terminal, the second live data stream switching indication message including the identification data pre-allocated for the updated live data stream.

[0115] In the above steps, the server instructs the cloud to switch the original first live data stream of the first user end to the live stream of the wearable device.

[0116] Among them, the updating live data stream is generated for the first user terminal according to the behavior characteristic data of the first user terminal for the target live room, including: obtaining the live content characteristic data matching the behavior characteristic data of the first user terminal for the target live room according to the behavior characteristic data of the first user terminal for the target live room; generating the updating live data stream for the first user terminal according to the behavior characteristic data of the first user terminal for the target live room and the live content characteristic data matching the behavior characteristic data of the first user terminal for the target live room. That is to say, according to the behavior of the first user terminal asking questions about wearable devices in the live room, find the live room that matches the behavior of the first user. Among them, the question of the user asking about wearable devices in the live room is a kind of behavior characteristic, and the live stream is generated according to these behavior characteristics and the live content characteristic of the wearable device.

[0117] Before switching the default live data stream corresponding to the target live room pushed to the first user terminal to the updated live data stream corresponding to the target live room, the method may further include: generating a first switching live data stream generation task, the first switching live data stream generation task being used to indicate generating a first switching live data stream, the first switching live data stream being a live data stream pushed to the first user terminal after stopping pushing the default live data stream corresponding to the target live room to the first user terminal and before starting pushing the updated live data stream corresponding to the target live room to the first user terminal; the method of switching the default live data stream corresponding to the target live room pushed to the first user terminal to the updated live data stream corresponding to the target live room includes: after stopping pushing the default live data stream corresponding to the target live room to the first user terminal and before starting pushing the updated live data stream corresponding to the target live room to the first user terminal, executing the operation of pushing the first switching live data stream to the first user terminal. The operation of pushing the first switching live data stream may be executed by the server terminal or may be executed by the server terminal instructing the cloud terminal to instruct the cloud terminal to execute.

[0118] When the live content displayed by the user end switches from one live stream to another, the switching screen change seen by the user is not natural enough, which may make the user feel a little sudden. In addition, the two live data streams have different characteristics and the content played is also different. Therefore, the above steps propose a transition method for pushing the first switching live data stream to the first user end, so as to provide a seamless experience when the two live data streams are switched. In addition, since the aforementioned content provides multiple different live data streams before switching and live data streams after switching, according to different situations, embodiment 1 also provides a second switching live data stream and a third switching live data stream, which are similar to the functions and generation processes of the first switching live data stream. For specific instructions, please refer to the part after the introduction of the first switching live data stream.

[0119] In order to provide a first switching live broadcast data stream, the live broadcast streaming method also includes: according to the first switching live broadcast data stream generation task, obtaining the live broadcast data corresponding to the default live broadcast data stream corresponding to the target live broadcast room at the time point when the push to the first user terminal stops, as the first end live broadcast data; obtaining live broadcast data matching the first end live broadcast data from the updated live broadcast data stream corresponding to the target live broadcast room, as the first start live broadcast data; generating the first switching live broadcast data stream according to the first end live broadcast data, the live content feature data corresponding to the default live broadcast data stream corresponding to the target live broadcast room, the first start live broadcast data and the live content feature data corresponding to the updated live broadcast data stream corresponding to the target live broadcast room.

[0120] Among them, the generating of the first switching live data stream according to the first live broadcast end data, the live content feature data corresponding to the default live data stream corresponding to the target live broadcast room, the first live broadcast start data and the live content feature data corresponding to the updated live data stream corresponding to the target live broadcast room includes: inputting the first live broadcast end data, the live content feature data corresponding to the default live data stream corresponding to the target live broadcast room, the first live broadcast start data and the live content feature data corresponding to the updated live data stream corresponding to the target live broadcast room into a live content feature data acquisition model to obtain the live content feature data corresponding to the first switching live data stream; generating the first switching live data stream according to the live content feature data corresponding to the first switching live data stream.

[0121] The above steps are used to generate the first switching live data stream. Specifically, the actual or expected stop progress of the original default live data stream can be obtained, and the live data that matches the stop progress of the default live data stream can be found from the updated live data stream, without having to play from the beginning of the updated live stream. For example, when the default live stream stops playing, the host introduces the geographic location of electronic products. At this time, it is possible to find out whether there is content matching the geographic location in the live data stream of the wearable device. If so, the content of the live data stream of the wearable device is played from the matching content part. Furthermore, based on the data obtained above, the first switching live data stream is generated as a transition part connecting the default live data stream and the updated live data stream, so as to avoid the problem of poor connection when the live stream is switched.

[0122] The server can also push the transition live stream to the user by instructing the cloud, and the cloud generates the transition live stream. Specifically, it also includes: sending the first switching live data stream generation task to the cloud; obtaining a response message returned by the cloud indicating that the first switching live data stream has been generated; the execution of the operation of pushing the first switching live data stream to the first user terminal includes: sending an indication message to the cloud for instructing the cloud to push the first switching live data stream to the first user terminal.

[0123] Similar to the aforementioned first switching live data stream task, when the first live data stream switches to the updated live data stream, it also includes: generating a second switching live data stream generation task, the second switching live data stream generation task is used to represent the generation of a second switching live data stream, the second switching live data stream is the live data stream pushed to the first user end after stopping pushing the first live data stream to the first user end and before starting to push the updated live data stream to the first user end; the switching of the first live data stream that has been pushed to the first user end to the updated live data stream of the first user end includes: after stopping pushing the first live data stream to the first user end and before starting to push the updated live data stream to the first user end, executing the operation of pushing the second switching live data stream to the first user end.

[0124] Similarly, generating the second switching live broadcast data stream also includes: according to the second switching live broadcast data stream generation task, obtaining the live broadcast data corresponding to the time point when the first live broadcast data stream stops pushing to the first user terminal as the second end live broadcast data; obtaining the live broadcast data matching the second end live broadcast data from the updated live broadcast data stream as the second start live broadcast data; generating the second switching live broadcast data stream according to the second end live broadcast data, the live broadcast content feature data corresponding to the first live broadcast data stream, the second start live broadcast data and the live broadcast content feature data corresponding to the updated live broadcast data stream.

[0125] The method of generating the second switching live data stream according to the second live broadcast end data, the live content feature data corresponding to the first live broadcast data stream, the second live broadcast start data and the live content feature data corresponding to the updated live broadcast data stream comprises: inputting the second live broadcast end data, the live content feature data corresponding to the first live broadcast data stream, the second live broadcast start data and the live content feature data corresponding to the updated live broadcast data stream into a live content feature data acquisition model to obtain the live content feature data corresponding to the second switching live data stream; generating the second switching live data stream according to the live content feature data corresponding to the second switching live data stream.

[0126] Similarly, the server can instruct the cloud to complete the task of generating and pushing the second switching live data stream. Therefore, the live streaming method also includes: sending the second switching live data stream generation task to the cloud; obtaining a response message returned by the cloud indicating that the second switching live data stream has been generated; the operation of pushing the second switching live data stream to the first user terminal includes: sending an instruction message to the cloud to instruct the cloud to push the second switching live data stream to the first user terminal.

[0127] The process of switching from the first live data stream to the updated live data stream generated by the first user terminal also includes: generating a third switching live data stream generation task, the third switching live data stream generation task is used to indicate the generation of a third switching live data stream, the third switching live data stream is the live data stream pushed to the first user terminal after stopping pushing the first live data stream to the first user terminal and before starting to push the updated live data stream generated for the first user terminal to the first user terminal; switching the first live data stream pushed to the first user terminal to the updated live data stream generated for the first user terminal includes: after stopping pushing the first live data stream to the first user terminal and before starting to push the updated live data stream generated for the first user terminal to the first user terminal, executing the operation of pushing the third switching live data stream to the first user terminal. Similarly, the operation of pushing the third switching live data stream to the first user terminal can be performed by the server, or the server can transmit information to indicate that the cloud is completed.

[0128] Regarding the generation of the third switching live broadcast data stream, it also includes: according to the third switching live broadcast data stream generation task, obtaining the live broadcast data corresponding to the time point when the first live broadcast data stream stops pushing to the first user end, as the third end live broadcast data; obtaining live broadcast data matching the third end live broadcast data from the updated live broadcast data stream generated for the first user end, as the third start live broadcast data; generating the third switching live broadcast data stream according to the third end live broadcast data, the live content feature data corresponding to the first live broadcast data stream, the third start live broadcast data and the live content feature data corresponding to the updated live broadcast data stream generated for the first user end.

[0129] Among them, the generating of the third switching live data stream according to the third live broadcast end data, the live content feature data corresponding to the first live broadcast data stream, the third live broadcast start data and the live content feature data corresponding to the updated live broadcast data stream generated for the first user terminal includes: inputting the third live broadcast end data, the live content feature data corresponding to the first live broadcast data stream, the third live broadcast start data and the live content feature data corresponding to the updated live broadcast data stream generated for the first user terminal into a live content feature data acquisition model to obtain the live content feature data corresponding to the third switching live data stream; generating the third switching live data stream according to the live content feature data corresponding to the third switching live data stream.

[0130] The cloud can also generate and push a third switching live data stream, which also includes: sending the third switching live data stream generation task to the cloud; obtaining a response message returned by the cloud to indicate that the third switching live data stream has been generated; the execution of the operation of pushing the third switching live data stream to the first user terminal includes: sending an indication message to the cloud to instruct the cloud to push the third switching live data stream to the first user terminal.

[0131] Embodiment 1 also provides a more specific description of the distribution strategy. The distribution strategy may include the correspondence between user portrait data and live data stream, so that if you know the information of a certain user and obtain the user portrait data from it, you can find the live data stream that matches the user's portrait from the distribution strategy. The live stream distribution strategy includes the correspondence between user portrait data and live stream identification data; the determination of the first live data stream to be pushed to the first user terminal based on the first user portrait data and the live stream distribution strategy includes: finding the live stream identification data that matches the first user portrait data from the live stream distribution strategy as the identification data of the first live data stream.

[0132] The aforementioned content mentioned that the server will allocate cloud machines through the scheduling machine scheduling server. In fact, the server can store a cloud device list, and the cloud device list can record a live stream corresponding to a cloud device, that is, the cloud device is responsible for pushing the corresponding live stream. If it is determined that a live stream is to be pushed to a certain user, the corresponding cloud device can be found from the cloud device list to push the live stream. Specifically, it also includes: in the cloud device list, recording the correspondence between the identification data of the target cloud device and the identification data of the first live data stream; the target cloud device that has undertaken the task of pushing the first live data stream to the client is determined in the cloud device list, including: according to the identification data of the first live data stream, searching the cloud device list for the identification data of the target cloud device corresponding to the identification data of the first live data stream.

[0133] On the basis of the above, the server can also store a mapping table of live data streams and live materials. If it is known that the cloud pushes live data stream information, the live material data required by the cloud can be found from the mapping table. Specifically, it also includes: establishing a mapping table of live data streams and live material data for recording the correspondence between the identification data of the live data stream and the identification data of the live material data, and the mapping table of live data streams and live material data is a mapping table corresponding to the target live broadcast room; after obtaining the live material data request message, obtaining the identification data of the first live data stream from the live material data request message; according to the identification data of the first live data stream, from the mapping table of live data streams and live material data corresponding to the target live broadcast room, find the identification data of the first live material data corresponding to the first live data stream; according to the identification data of the first live material data, obtain the first live material data as the material data for generating the first live data stream. Among them, the identification data is a mark made to reduce storage space and search costs.

[0134] The above steps search for the corresponding live broadcast material according to the live broadcast data stream information, and the corresponding live broadcast material can also be obtained by acquiring the live broadcast data stream corresponding to the cloud device information according to the cloud device information. Specifically, it also includes: in the cloud device list, recording the correspondence between the identification data of the target cloud device and the identification data of the first live data stream; establishing a mapping table between live data stream and live material data for recording the correspondence between the identification data of the live data stream and the identification data of the live material data, wherein the mapping table between live data stream and live material data is a mapping table corresponding to the target live room; after obtaining the live material data request message, obtaining the identification data of the target cloud device that sends the live material data request message; according to the identification data of the target cloud device, searching the cloud device list for the identification data of the first live data stream corresponding to the identification data of the target cloud device; according to the identification data of the first live data stream, searching the mapping table between live data stream and live material data for the identification data of the first live material data corresponding to the first live data stream, wherein the mapping table between live data stream and live material data records the correspondence between the identification data of the live data stream and the identification data of the live material data; according to the identification data of the first live material data, obtaining the first live material data as the material data for generating the first live data stream.

[0135] The mapping table of live data stream and live data stream material is used in the above steps, and the establishment of the mapping table is explained below. In this step, the live data stream and the live material data are connected by using historical user portraits, so that there is also an indirect correspondence between the live data stream and the live material data. Specifically, it also includes: generating live material data corresponding to each user portrait data according to each user portrait data obtained in advance; assigning identification data to the live material data corresponding to each user portrait data; determining the corresponding live data stream for each user portrait data; obtaining the identification data of the corresponding live data stream determined for each user portrait data; the live data stream and live material data mapping table for recording the correspondence between the identification data of the live data stream and the identification data of the live material data includes: establishing the mapping table of live data stream and live material data according to the identification data assigned to the live material data corresponding to each user portrait data and the identification data of the corresponding live data stream determined for each user portrait data. It can be understood that clustering a large number of historical user portraits can make the determined live data stream more accurate.

[0136] For interactive information and actions such as barrage sent by users in the live broadcast room, the live broadcast streaming method also includes: after pushing the first live broadcast data stream to the first user terminal, obtaining the interactive information sent by the first user terminal to the target live broadcast room as user interactive information; for the user interactive information, generating anchor interactive information for replying to the user interactive information in the target live broadcast room; executing the operation of sending the anchor interactive information to the first user terminal in the target live broadcast room. For detailed description of this part, please refer to the introduction of step S312 in the previous text. In this step, a large language model can be used to generate reply copy for user interactive information. The large language model can be trained based on the copy of massive live broadcast platform data, reference live broadcast type information, etc.

[0137] The live streaming method provided in Example 1 splits the live digital human rendering, and the cloud and the client process part of the digital human rendering task respectively, and finally displays the digital human rendering picture on the client, reducing the occupation of cloud resources, alleviating the pressure of cloud data processing, and improving the efficiency of live push and loading. Since Example 1 recommends live data streams to users according to user preferences, different live data streams generate different live content and live formats, so different users entering the same live room see different live content and live formats, realizing live streaming with thousands of faces, and improving the differentiation and personalization of live recommendations. Example 1 also obtains the characteristic information of the target live room by obtaining the identification data of the target live room, and obtains the preference information of the first user by obtaining the first user portrait data, and then combines the live stream distribution strategy to improve the accuracy of the live stream distribution strategy, and determine that the first live data stream is more in line with the user's preference, so that the matching degree and matching efficiency of the live room and the user are improved.

[0138] Embodiment 2 provides a live streaming device, which includes the live streaming method in the scenario embodiment and embodiment 1. The following description of the live streaming device embodiment is only illustrative, and the relevant implementation details can be understood by referring to the scenario embodiment and embodiment 1.

[0139] Please see Figure 6 , Figure 6 This is a schematic diagram of a live streaming device provided in Embodiment 2. The live streaming device provided in this embodiment includes: A live broadcast room access request message acquisition unit 601 is used to obtain a live broadcast room access request message sent by a first user terminal for accessing a target live broadcast room; A target cloud device selection unit 602 is used to select a target cloud device from a cloud device list for pushing the first live data stream to the client; The task acceptance instruction sending unit 603 is used to send a task acceptance instruction to the target cloud device to instruct the target cloud device to accept the task of pushing the first live data stream to the client; A live broadcast material data request message acquisition unit 604 is used to obtain a live broadcast material data request message issued by the target cloud device in response to the task acceptance instruction for requesting to obtain live broadcast material data, wherein the live broadcast material data is material data used to generate the first live broadcast data stream; The live broadcast material data sending unit 605 is used to send the live broadcast material data to the target cloud device in response to the live broadcast material data request message, wherein the live broadcast material data includes first user portrait data, anchor image data, anchor timbre data, and first service item data that the anchor can provide to the user; A live stream distribution strategy acquisition unit 606 is used to obtain the identification data of the target live broadcast room and obtain the live stream distribution strategy corresponding to the identification data of the target live broadcast room; A first live broadcast data stream determining unit 607 is configured to determine, based on the first user portrait data, a first live broadcast data stream corresponding to the target live broadcast room and to be pushed to the first user terminal, wherein the first live broadcast data stream is different from a second live broadcast data stream pushed to a second user terminal accessing the target live broadcast room; The first live data stream pushing unit 608 is used to push the first live data stream to the first user terminal in response to the live room access request message.

[0140] Corresponding to the aforementioned embodiment 1, embodiment 3 further provides a live streaming method, and the execution subject of the method can be the cloud. Some steps in embodiment 1 can also be executed by the cloud. Part of the content of this embodiment has been described in the aforementioned content, and the relevant content can be referred to the aforementioned embodiment. Figure 7 , Figure 7 This is a flow chart of a live streaming method in the cloud provided in Example 3. The live streaming method includes: S701: Obtaining a task acceptance instruction issued by the live broadcast server for instructing to accept a task of pushing a first live broadcast data stream to a client, where the first live broadcast data stream is a live broadcast data stream corresponding to a target live broadcast room, the target live broadcast room corresponds to multiple live broadcast data streams, and the first live broadcast data stream is different from other live broadcast data streams corresponding to the target live broadcast room; S702: in response to the task acceptance instruction, sending a live broadcast material data request message to the live broadcast server for requesting to obtain live broadcast material data for generating the first live broadcast data stream; S703: Obtain the live broadcast material data returned by the live broadcast service end in response to the live broadcast material data request message; S704: Generate the first live broadcast data stream according to the live broadcast material data; S705: Obtaining a streaming instruction sent by the live streaming service end to instruct to push the first live streaming data stream to the first user end; S706: In response to the streaming instruction, push the first live data stream to the first user terminal.

[0141] The task acceptance instruction includes identification data pre-allocated by the live broadcast server for the first live broadcast data stream, the live broadcast material data request message includes identification data pre-allocated by the live broadcast server for the first live broadcast data stream, and the streaming instruction includes identification data pre-allocated by the live broadcast server for the first live broadcast data stream.

[0142] Among them, the live broadcast material data used to generate the first live broadcast data stream is different from the live broadcast material data used to generate the other live broadcast data streams. The live broadcast material data includes anchor image data, anchor timbre data and first service item data that the anchor can provide to users; the generating the first live broadcast data stream according to the live broadcast material data includes: generating text data corresponding to the first live broadcast data stream according to the first service item data that the anchor can provide to users; generating anchor audio data corresponding to the first live broadcast data stream according to the anchor timbre data and the text data corresponding to the first live broadcast data stream; generating anchor lip drive data corresponding to the first live broadcast data stream according to the anchor image data and the anchor audio data corresponding to the first live broadcast data stream; generating the first live broadcast data stream according to the anchor audio data corresponding to the first live broadcast data stream and the anchor lip drive data corresponding to the first live broadcast data stream.

[0143] The live broadcast material data also includes user portrait data; the copywriting data corresponding to the first live broadcast data stream is generated based on the first service item data that the anchor can provide to the user, including: the copywriting data corresponding to the first live broadcast data stream is generated based on the first service item data that the anchor can provide to the user and the user portrait data.

[0144] Optionally, it also includes: obtaining a live broadcast data stream update task sent by the live broadcast server, the live broadcast data stream update task is used to indicate that an updated live broadcast data stream is generated for the first user terminal according to the behavioral characteristic data of the first user terminal for the target live broadcast room, the live content characteristic data corresponding to the updated live broadcast data stream generated for the first user terminal matches the behavioral characteristic data of the first user terminal for the target live broadcast room, the live broadcast data stream update task includes identification data pre-allocated by the live broadcast server for updating the live broadcast data stream, the behavioral characteristic data of the first user terminal for the target live broadcast room and the live broadcast material data used to generate the updated live broadcast data stream; according to the behavioral characteristic data of the first user terminal for the target live broadcast room and the live broadcast material data used to generate the updated live broadcast data stream The method comprises: generating an updated live data stream for the first user terminal by using live material data of the new live data stream; sending a response message to the live server terminal to indicate that the updated live data stream has been generated, wherein the response message includes identification data pre-allocated for the updated live data stream; obtaining a live data stream switching indication message sent by the live server terminal to indicate that the first live data stream that has been pushed to the first user terminal is switched to the updated live data stream, wherein the live data stream switching indication message includes identification data allocated by the live server terminal to the first live data stream and identification data pre-allocated for the updated live data stream; and in response to the live data stream switching indication message, switching the first live data stream that has been pushed to the first user terminal to the updated live data stream.

[0145] The live broadcast data stream update task includes using the host image data corresponding to the first live broadcast data stream as indication information of the host image data corresponding to the updated live broadcast data stream and using the host timbre data corresponding to the first live broadcast data stream as indication information of the host timbre data corresponding to the updated live broadcast data stream. The live broadcast material data used to generate the updated live broadcast data stream includes other service item data that the host can provide to the user that is different from the first service item data; the update task generates an update for the first user terminal based on the behavioral feature data of the first user terminal for the target live broadcast room and the live broadcast material data used to generate the updated live broadcast data stream. The live broadcast data stream includes: generating text data corresponding to the updated live broadcast data stream according to other service item data that the anchor can provide to users and the behavioral characteristic data of the first user terminal for the target live broadcast room; generating anchor audio data corresponding to the updated live broadcast data stream according to the anchor timbre data and the text data corresponding to the updated live broadcast data stream; generating anchor lip drive data corresponding to the updated live broadcast data stream according to the anchor image data and the anchor audio data corresponding to the updated live broadcast data stream; generating the updated live broadcast data stream according to the anchor audio data corresponding to the updated live broadcast data stream and the anchor lip drive data corresponding to the updated live broadcast data stream.

[0146] The live broadcast material data used to generate the updated live broadcast data stream includes the host image data corresponding to the updated live broadcast data stream, the host timbre data corresponding to the updated live broadcast data stream, and other service item data that the host can provide to the user that is different from the first service item data; the updated live broadcast data stream is generated for the first user terminal according to the behavioral characteristic data of the first user terminal for the target live broadcast room and the live broadcast material data used to generate the updated live broadcast data stream, including: generating the text data corresponding to the updated live broadcast data stream according to the other service item data that the host can provide to the user and the behavioral characteristic data of the first user terminal for the target live broadcast room; generating the host audio data corresponding to the updated live broadcast data stream according to the host timbre data corresponding to the updated live broadcast data stream and the text data corresponding to the updated live broadcast data stream; generating the host lip drive data corresponding to the updated live broadcast data stream according to the host image data corresponding to the updated live broadcast data stream and the host audio data corresponding to the updated live broadcast data stream; generating the updated live broadcast data stream according to the host audio data corresponding to the updated live broadcast data stream and the host lip drive data corresponding to the updated live broadcast data stream.

[0147] Optionally, it also includes: obtaining a switching live broadcast data stream generation task sent by the live broadcast server; the switching live broadcast data stream generation task is used to indicate the generation of a switching live broadcast data stream, the switching live broadcast data stream is a live broadcast data stream pushed to the first user end after stopping pushing the first live broadcast data stream to the first user end and before starting to push the updated live broadcast data stream to the first user end, the switching live broadcast data stream generation task includes the behavior characteristic data of the first user end for the target live broadcast room; generating the switching live broadcast data stream according to the behavior characteristic data of the first user end for the target live broadcast room; and sending a response to the live broadcast server to indicate that the switching live broadcast data stream has been generated. a response message; obtaining a stream change instruction issued by the live broadcast server, wherein the stream change instruction is used to instruct to stop pushing the first live data stream to the first user terminal, push the switching live data stream to the first user terminal, and push the updated live data stream to the first user terminal after the live content corresponding to the target live broadcast room of the switching live data stream is completed; in response to the stream change instruction, stop pushing the first live data stream to the first user terminal, and push the switching live data stream to the first user terminal; after the live content corresponding to the target live broadcast room of the switching live data stream is completed, push the updated live data stream to the first user terminal.

[0148] Embodiment 4 provides a live streaming system. The following description of the live streaming system embodiment is only illustrative, and the relevant implementation details can be understood by referring to the above embodiments. Figure 8 , Figure 8 The live streaming system 800 provided in Embodiment 4 includes a server 101, a cloud 102, a live streaming terminal 103 and a user terminal 104. The figure only shows part of the capabilities of each terminal, and other capabilities are not shown in the figure. For relevant parts, please refer to the above embodiments.

[0149] The live streaming system provided in Example 4 can improve the accuracy of live room recommendations, and can provide different live content for different users when they enter the same live room. At the same time, it also splits the live digital human rendering, with the cloud and the client processing part of the digital human rendering task respectively, and finally displays the digital human rendered picture on the client, reducing the occupation of cloud resources, alleviating the cloud data processing pressure, and improving the efficiency of live push and loading.

[0150] Embodiment 5 provides an electronic device embodiment corresponding to the live streaming method provided in the first embodiment, and the live streaming method of the aforementioned embodiment can be executed on the electronic device. The following description of the electronic device embodiment is merely illustrative, and the relevant content can refer to the aforementioned embodiment.

[0151] Please see Fig. 9 Understand the above electronic devices, Fig. 9 Schematic diagram of an electronic device provided in Embodiment 5. The electronic device provided in Embodiment 5 includes: a processor 901, a memory 902, a communication bus 903, and a communication interface 904; The memory 902 is used to store computer instructions for data processing. When the computer instructions are read and executed by the processor 901, the steps of the first embodiment and the scenario embodiment are executed.

[0152] Embodiment 6 also provides a computer-readable storage medium for implementing the method described in the scenario embodiment and embodiment 1. The computer-readable storage medium embodiment provided in embodiment 6 is described in a relatively simple manner. For the relevant parts, please refer to the corresponding description of the above method embodiment. The embodiment described below is only exemplary. The computer-readable storage medium stores computer instructions, and when the computer-executable instructions are executed by the processor, they are used to implement the method described in any one of the scenario embodiment and embodiment 1.

[0153] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0154] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0155] 1. Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be used to store information by any method or technology. Information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include non-transitory media such as modulated data signals and carrier waves.

[0156] 2. Those skilled in the art should understand that the embodiments of the present application can be provided as methods, systems or computer program products. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0157] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0158] Although the present application is disclosed as above in the form of a preferred embodiment, it is not intended to limit the present application. Any technical personnel in this field may make possible changes and modifications without departing from the spirit and scope of the present application. Therefore, the scope of protection of the present application shall be based on the scope defined by the claims of the present application.

Claims

1. A live streaming method, characterized in that: include: Obtaining a task acceptance instruction issued by a live broadcast server for instructing to accept a task of pushing a first live broadcast data stream to a client, wherein the first live broadcast data stream is a live broadcast data stream corresponding to a target live broadcast room, the target live broadcast room corresponds to multiple live broadcast data streams, and the first live broadcast data stream is different from other live broadcast data streams corresponding to the target live broadcast room; In response to the task acceptance instruction, a live broadcast material data request message is sent to the live broadcast server to request acquisition of live broadcast material data used to generate the first live broadcast data stream; Obtaining the live broadcast material data returned by the live broadcast service end in response to the live broadcast material data request message, wherein the live broadcast material data includes user portrait data, anchor image data, anchor timbre data, and first service item data that the anchor can provide to the user; Generate the first live broadcast data stream according to the live broadcast material data; Obtaining a streaming instruction sent by the live streaming service end for instructing to push the first live streaming data stream to the first user end; In response to the streaming instruction, the first live data stream is pushed to the first user terminal.

2. The live streaming method according to claim 1, characterized in that: The task acceptance instruction includes identification data pre-allocated by the live broadcast server for the first live broadcast data stream, the live broadcast material data request message includes identification data pre-allocated by the live broadcast server for the first live broadcast data stream, and the streaming instruction includes identification data pre-allocated by the live broadcast server for the first live broadcast data stream.

3. The live streaming method according to claim 1, characterized in that: The generating the first live broadcast data stream according to the live broadcast material data includes: Generate text data corresponding to the first live broadcast data stream according to the first service item data and the user portrait data; Generate host audio data corresponding to the first live data stream according to the host timbre data and the text data corresponding to the first live data stream; Generate host lip drive data corresponding to the first live broadcast data stream according to the host image data and the host audio data corresponding to the first live broadcast data stream; The first live broadcast data stream is generated according to the host audio data corresponding to the first live broadcast data stream and the host lip drive data corresponding to the first live broadcast data stream.

4. The live streaming method according to claim 1, characterized in that: Also includes: Obtaining a live data stream update task sent by the live broadcast server, wherein the live data stream update task is used to indicate that an updated live data stream is generated for the first user terminal according to the behavior characteristic data of the first user terminal for the target live broadcast room, and the live content characteristic data corresponding to the updated live data stream generated for the first user terminal matches the behavior characteristic data of the first user terminal for the target live broadcast room, and the live data stream update task includes identification data pre-allocated by the live broadcast server for updating the live data stream, the behavior characteristic data of the first user terminal for the target live broadcast room, and live material data for generating the updated live data stream; Generate an updated live broadcast data stream for the first user terminal according to the behavior feature data of the first user terminal for the target live broadcast room and the live broadcast material data used to generate the updated live broadcast data stream; Sending a response message to the live broadcast server indicating that the updated live broadcast data stream has been generated, wherein the response message includes identification data pre-allocated for updating the live broadcast data stream; Obtaining a live data stream switching indication message sent by the live broadcast server for indicating that the first live data stream pushed to the first user terminal is switched to the updated live data stream, wherein the live data stream switching indication message includes identification data allocated by the live broadcast server to the first live data stream and identification data pre-allocated to the updated live data stream; In response to the live data stream switching indication message, the first live data stream that has been pushed to the first user terminal is switched to the updated live data stream.

5. The live streaming method according to claim 4, characterized in that: The live data stream update task includes using the host image data corresponding to the first live data stream as indication information of the host image data corresponding to the updated live data stream and using the host timbre data corresponding to the first live data stream as indication information of the host timbre data corresponding to the updated live data stream, and the live material data used to generate the updated live data stream includes other service item data that the host can provide to the user, which is different from the first service item data; The step of generating an updated live broadcast data stream for the first user terminal according to the behavior feature data of the first user terminal for the target live broadcast room and the live broadcast material data used to generate the updated live broadcast data stream comprises: Generate text data corresponding to the updated live broadcast data stream according to other service item data that the anchor can provide to the user and the behavioral feature data of the first user terminal for the target live broadcast room; Generate host audio data corresponding to the updated live data stream according to the host timbre data and the text data corresponding to the updated live data stream; Generate host lip drive data corresponding to the updated live data stream according to the host image data and the host audio data corresponding to the updated live data stream; The updated live data stream is generated according to the host audio data corresponding to the updated live data stream and the host lip drive data corresponding to the updated live data stream.

6. The live streaming method according to claim 4, characterized in that: The live broadcast material data used to generate the updated live broadcast data stream includes the anchor image data corresponding to the updated live broadcast data stream, the anchor timbre data corresponding to the updated live broadcast data stream, and other service item data that the anchor can provide to the user and is different from the first service item data; The step of generating an updated live broadcast data stream for the first user terminal according to the behavior feature data of the first user terminal for the target live broadcast room and the live broadcast material data used to generate the updated live broadcast data stream comprises: Generate text data corresponding to the updated live broadcast data stream according to other service item data that the anchor can provide to the user and the behavioral feature data of the first user terminal for the target live broadcast room; Generate host audio data corresponding to the updated live data stream according to the host timbre data corresponding to the updated live data stream and the text data corresponding to the updated live data stream; Generate host lip drive data corresponding to the updated live data stream according to the host image data corresponding to the updated live data stream and the host audio data corresponding to the updated live data stream; The updated live data stream is generated according to the host audio data corresponding to the updated live data stream and the host lip drive data corresponding to the updated live data stream.

7. A live streaming method, characterized in that: include: Obtaining a live broadcast room access request message sent by the first user terminal for accessing a target live broadcast room; Selecting a target cloud device from the cloud device list for pushing the first live data stream to the client; Sending a task acceptance instruction to the target cloud device for instructing the target cloud device to accept the task of pushing the first live data stream to the client; Obtaining a live broadcast material data request message issued by the target cloud device in response to the task acceptance instruction for requesting to obtain live broadcast material data, wherein the live broadcast material data is material data used to generate the first live broadcast data stream; In response to the live broadcast material data request message, the live broadcast material data is sent to the target cloud device, wherein the live broadcast material data includes first user portrait data, anchor image data, anchor timbre data, and first service item data that the anchor can provide to the user; Obtaining identification data of the target live broadcast room, and obtaining a live broadcast stream distribution strategy corresponding to the identification data of the target live broadcast room; Determine, according to the first user portrait data and the live stream distribution strategy, a first live data stream corresponding to the target live broadcast room to be pushed to the first user terminal, wherein the first live data stream is different from a second live data stream pushed to a second user terminal accessing the target live broadcast room; In response to the live room access request message, the first live data stream is pushed to the first user terminal.

8. The live streaming method according to claim 7, characterized in that: Also includes: After determining the first live broadcast data stream corresponding to the target live broadcast room to be pushed to the first user terminal, determining the target cloud device in the cloud device list that has undertaken the task of pushing the first live broadcast data stream to the client; The pushing of the first live data stream to the first user terminal in response to the live room access request message includes: in response to the live room access request message, issuing a push instruction to the target cloud device to instruct the pushing of the first live data stream to the first user terminal.

9. The live streaming method according to claim 7, characterized in that: Also includes: Obtaining portrait data of multiple user groups corresponding to the target live broadcast room; Obtaining live broadcast content feature data corresponding to the target live broadcast room; Based on the portrait data of the multiple user groups and the live content feature data, multiple live data streams corresponding to the target live room are determined.

10. The live streaming method according to claim 9, characterized in that: The obtaining of the portrait data of multiple user groups corresponding to the target live broadcast room includes: Obtaining portrait data of historical users who visited the target live broadcast room, performing clustering processing on the portrait data of historical users who visited the target live broadcast room, and obtaining portrait data of multiple user groups corresponding to the target live broadcast room; Alternatively, portrait data of historical users who have visited the anchor data page corresponding to the target live broadcast room is obtained, and the portrait data of historical users who have visited the anchor data page corresponding to the target live broadcast room are clustered to obtain portrait data of multiple user groups corresponding to the target live broadcast room; Alternatively, portrait data of historical users who have accessed a data service system is obtained, where the data service system is the data service system in which the target live broadcast room is located, and the portrait data of historical users who have accessed the data service system are clustered to obtain portrait data of multiple user groups corresponding to the target live broadcast room.

11. The live streaming method according to claim 9, characterized in that: The determining, according to the portrait data of the multiple user groups and the live content feature data, multiple live data streams corresponding to the target live broadcast room comprises: Determine matching live content feature data for the portrait data of each user group; Set identification data of the live data stream for the portrait data of each user group; A live data stream list of the target live broadcast room is established, in which the corresponding relationship between the portrait data of each user group, the matching live content feature data and the matching live data stream identification data is recorded.

12. The live streaming method according to claim 11, characterized in that: The determining, according to the first user portrait data, a first live broadcast data stream corresponding to the target live broadcast room to be pushed to the first user terminal includes: In the live data stream list, searching for portrait data of a user group matching the first user portrait data; In the live data stream list, search for identification data of a live data stream that matches the portrait data of the user group as the identification data of the first live data stream.

13. The live streaming method according to claim 12, characterized in that: Also includes: If the portrait data of the user group matching the first user portrait data is not found in the live data stream list, determining whether there is live content feature data matching the first user portrait data in the live data stream list; If there is live content feature data matching the first user portrait data in the live data stream list, then the identification data of the live data stream corresponding to the live content feature data matching the first user portrait data is searched in the live data stream list as the identification data of the first live data stream, and the correspondence between the first user portrait data, the live content feature data matching the first user portrait data, and the identification data of the live data stream corresponding to the live content feature data matching the first user portrait data is recorded in the live data stream list.

14. The live streaming method according to claim 11, characterized in that: Also includes: If there is no live content feature data matching the first user portrait data in the live data stream list, determining whether updated live content feature data corresponding to the target live room can be obtained, the updated live content feature data being live content feature data corresponding to the target live room but not recorded in the live data stream list; If the updated live content feature data can be obtained, determining whether there is live content feature data matching the first user portrait data in the updated live content feature data; If the updated live content characteristic data contains live content characteristic data that matches the first user portrait data, the identification data of the live data stream is set for the first user portrait data as the identification data of the first live data stream. In the live data stream list, the correspondence between the first user portrait data, the live content characteristic data in the updated live content characteristic data that matches the first user portrait data, and the identification data of the first live data stream is recorded.

15. A live streaming device, characterized in that: include: a task acceptance instruction receiving unit, configured to obtain a task acceptance instruction issued by a live broadcast server for instructing to accept a task of pushing a first live broadcast data stream to a client, wherein the first live broadcast data stream is a live broadcast data stream corresponding to a target live broadcast room, the target live broadcast room corresponds to multiple live broadcast data streams, and the first live broadcast data stream is different from other live broadcast data streams corresponding to the target live broadcast room; a live broadcast material data request message sending unit, configured to send, in response to the task acceptance instruction, to the live broadcast server a live broadcast material data request message for requesting to obtain the live broadcast material data for generating the first live broadcast data stream; A live broadcast material data acquisition unit, used to obtain the live broadcast material data returned by the live broadcast service end in response to the live broadcast material data request message, wherein the live broadcast material data includes user portrait data, anchor image data, anchor timbre data, and first service item data that the anchor can provide to the user; A first live data stream generating unit, configured to generate the first live data stream according to the live material data; A streaming instruction acquisition unit, used to obtain a streaming instruction sent by the live streaming service end to instruct to push the first live streaming data stream to the first user end; A pushing unit is used to push the first live data stream to the first user terminal according to the streaming instruction.

16. A live streaming device, characterized in that: include: A live broadcast room access request message acquisition unit, used to obtain a live broadcast room access request message sent by the first user terminal for accessing a target live broadcast room; A target cloud device selection unit, used to select a target cloud device from a cloud device list for pushing the first live data stream to the client; A task acceptance instruction sending unit, used for sending a task acceptance instruction to the target cloud device for instructing the target cloud device to accept the task of pushing the first live data stream to the client; a live broadcast material data request message acquisition unit, configured to obtain a live broadcast material data request message issued by the target cloud device in response to the task acceptance instruction for requesting to obtain live broadcast material data, wherein the live broadcast material data is material data used to generate the first live broadcast data stream; A live broadcast material data sending unit, configured to send the live broadcast material data to the target cloud device in response to the live broadcast material data request message, wherein the live broadcast material data includes first user portrait data, anchor image data, anchor timbre data, and first service item data that the anchor can provide to the user; A live stream distribution strategy acquisition unit, used to obtain the identification data of the target live broadcast room, and obtain the live stream distribution strategy corresponding to the identification data of the target live broadcast room; A first live data stream determining unit is used to determine, according to the first user portrait data and the live stream distribution strategy, a first live data stream corresponding to the target live broadcast room to be pushed to the first user terminal, wherein the first live data stream is different from a second live data stream pushed to a second user terminal accessing the target live broadcast room; The first live data stream pushing unit is used to push the first live data stream to the first user terminal in response to the live room access request message.

17. A live streaming system, characterized in that: Including cloud and server: The cloud end is used to obtain a task acceptance instruction issued by the live broadcast server end for instructing to accept a task of pushing a first live broadcast data stream to a client end, wherein the first live broadcast data stream is a live broadcast data stream corresponding to a target live broadcast room, the target live broadcast room corresponds to multiple live broadcast data streams, and the first live broadcast data stream is different from other live broadcast data streams corresponding to the target live broadcast room; In response to the task acceptance instruction, a live broadcast material data request message is sent to the live broadcast service end to request the live broadcast material data used to generate the first live broadcast data stream; the live broadcast material data returned by the live broadcast service end in response to the live broadcast material data request message is obtained, and the live broadcast material data includes user portrait data, anchor image data, anchor timbre data, and first service item data that the anchor can provide to the user; the first live broadcast data stream is generated according to the live broadcast material data; a push stream instruction sent by the live broadcast service end to instruct the first user end to push the first live broadcast data stream; in response to the push stream instruction, the first live broadcast data stream is pushed to the first user end; The server is used to obtain a live broadcast room access request message for accessing a target live broadcast room issued by a first user terminal; select a target cloud device for pushing a first live broadcast data stream to a client from a cloud device list; issue a task acceptance instruction to the target cloud device for instructing the target cloud device to undertake the task of pushing the first live broadcast data stream to the client; obtain a live broadcast material data request message for requesting to obtain live broadcast material data issued by the target cloud device in response to the task acceptance instruction, wherein the live broadcast material data is material data for generating the first live broadcast data stream; in response to the live broadcast material data request message, send the live broadcast material data to the target cloud device, wherein the live broadcast material data includes first user portrait data, anchor image data, anchor timbre data, and first service item data that the anchor can provide to the user; obtain identification data of the target live broadcast room, and obtain a live broadcast stream distribution strategy corresponding to the identification data of the target live broadcast room; Based on the first user portrait data and the live stream distribution strategy, determine the first live data stream corresponding to the target live room to be pushed to the first user terminal, wherein the first live data stream is different from the second live data stream pushed to the second user terminal accessing the target live room; in response to the live room access request message, push the first live data stream to the first user terminal.

18. An electronic device, characterized in that: include: A processor, a memory, and computer program instructions stored on the memory and executable on the processor; When the processor executes the computer program instructions, the method according to any one of claims 1 to 14 is implemented.

19. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, which are used to implement the method described in any one of claims 1 to 14 when executed by a processor.

Images