Display device and voice interaction method

Abstract

This application discloses a display device and a voice interaction method, relating to the field of display devices. The method can identify the identity information matched with each voice message through voiceprint recognition, integrate the identity information with the text information corresponding to the voice message to generate first information, and update it to the conversation history corresponding to the current wake-up. When the first information requires the AI ​​agent to process business, the method transmits the multi-turn conversation history carrying the complete user identity identifier to the AI ​​agent, enabling the AI ​​agent to accurately identify different speakers in multi-turn dialogues, adapt to the multi-user collaborative interaction scenario of the display device, ensure the contextual coherence of multi-user turn-by-turn dialogues, and optimize the overall user experience of multi-user voice interaction.

Description

Technical Field

[0001] This application relates to the field of display devices, and more particularly to a display device and a voice interaction method. Background Technology

[0002] Currently, smart TVs and other display devices primarily deploy artificial intelligence (AI) agents for natural language dialogue in the cloud. These agents receive the user's natural language after being triggered by a preset wake word and then perform operations such as program retrieval, system function settings, media playback control, and message replies. However, this interaction method only supports one-on-one dialogue. In multi-user collaborative interaction scenarios such as multiple users taking turns speaking and discussing programs, there is a noticeable disconnect and fragmentation in human-computer dialogue, reducing the user experience. Summary of the Invention

[0003] This application provides a display device and a voice interaction method that can ensure the contextual coherence of multi-user turn-by-turn dialogue and AI intelligent agent in multi-user collaborative interaction scenarios, thereby improving the human-computer interaction experience.

[0004] In a first aspect, a display device is provided, including a display and a controller. The controller is configured to: receive a first voice when a preset wake-up word is detected; perform voiceprint recognition on the first voice to obtain user identity information corresponding to the first voice; generate first information based on the user identity information and the text information corresponding to the first voice; write the first information into a first session history to obtain an updated second session history, wherein both the first session history and the second session history are used to store all interaction records generated after the current wake-up trigger; when the first information contains an interaction intent to request an AI agent to handle business, send the second session history to the AI ​​agent; receive second information sent by the AI ​​agent, and perform corresponding interactive operations based on the second information.

[0005] Using the above method, the display device can identify the identity information matched with each voice message through voiceprint recognition, integrate the identity information with the text information corresponding to the voice message to generate the first information, and update it to the conversation history corresponding to this wake-up; and when the first information needs the AI ​​agent to process business, the multi-turn conversation history carrying the complete user identity identifier is transmitted to the AI ​​agent, so that the AI ​​agent can accurately identify different speakers in multi-turn dialogues, adapt to the multi-user collaborative interaction scenario of the display device, ensure the contextual coherence of multi-user turn-by-turn dialogue, and optimize the overall user experience of multi-user voice interaction.

[0006] The first voice is any voice spoken by the user after the display device is woken up by a preset wake-up word. It may include the first voice after being woken up, as well as other voices used in subsequent interactions.

[0007] The display device integrates and encapsulates the user identity information corresponding to the first voice message and the corresponding text information according to a preset data format into the first information. For example, the preset data format could be {user:text}.

[0008] In one possible implementation, after the controller executes the process of writing the first information into the first session history and obtaining the updated second session history, and before sending the second session history to the AI ​​agent when the first information contains an interactive intent to request the AI ​​agent to process the business, it is further configured to: input the second session history into a preset first model to obtain a first confidence level output by the first model; the first model is used to determine the interactive intent of the first information based on the second session history; if the first confidence level is greater than or equal to a preset first threshold, it is determined that the first information contains an interactive intent to request the AI ​​agent to process the business.

[0009] Using the above method, the display device determines whether a cloud-based AI agent needs to process business based on the first model. When there is a business need, it uploads a complete multi-turn conversation history with user identification, which helps the AI ​​agent distinguish different speakers in the dialogue, improves the accuracy of intent recognition and the stability of interaction, adapts to multi-user interaction scenarios on the display device, maintains the contextual coherence of multi-user turn-based dialogue, and optimizes the effect of multi-user voice interaction.

[0010] In one possible implementation, the controller generates first information based on the user identity information and the text information corresponding to the first speech conversion, and is configured to: determine the correlation between the first speech and the first session history; and, if there is a correlation between the first speech and the first session history, integrate the user identity information and the text information to generate the first information.

[0011] By determining the correlation between the first voice and the first conversation history using the above method, and then integrating identity and text generation information, we can accurately identify user intent and distinguish user data by combining context, while avoiding redundancy of irrelevant historical information and improving processing efficiency.

[0012] In one possible implementation, the controller is configured to determine the correlation between the first speech and the conversation history by: inputting the text information corresponding to the first speech and the first conversation history into a preset second model to obtain a second confidence level output by the second model; if the second confidence level is greater than or equal to a preset second threshold, it is determined that the first speech and the first conversation history are correlated; if the second confidence level is less than the preset second threshold, it is determined that the first speech and the first conversation history are not correlated.

[0013] By using the above method, the second model outputs a confidence metric to measure the semantic relevance, accurately distinguishing whether there is contextual association, reducing misjudgments, and improving the relevance and user experience of multi-user collaborative interaction.

[0014] In one possible implementation, when the second information is a response message used to reply to the first information, the controller performs a corresponding interactive operation based on the second information, configured to: write the second information into the second session history to obtain an updated third session history; convert the second information into a second voice and play it, or display the second information on a display.

[0015] Using the methods described above, In one possible implementation, the first session history includes a first temporary session history and a local session history; the controller executes writing the first information into the first session history to obtain an updated second session history, which is configured to: write the first information into the first temporary session history to obtain an updated second temporary session history, the second session history including the second temporary session history and the local session history.

[0016] By using the above method, the session history is divided into two categories: temporary and local. This distinguishes between short-term interaction caches and long-term retention records, enabling hierarchical data management and improving the accuracy of interactions.

[0017] In one possible implementation, after the controller executes the action of writing the first information into the first temporary session history and generating the updated second temporary session history, it is further configured to: if the data storage capacity of the second temporary session history is greater than a preset capacity threshold, compress the second temporary session history and write the compressed data into the local session history.

[0018] Using the above methods, short-term interaction data relies on the real-time caching of temporary session history to ensure the continuity of interaction; after the data accumulates beyond the preset capacity threshold, it is compressed and key data is written to the local session history for long-term storage, while redundant temporary cache is cleared to improve the overall accuracy of interaction.

[0019] Secondly, a voice interaction method is provided, applicable to any of the display devices mentioned in the first aspect. The method, upon detecting a preset wake-up word, receives a first voice message; performs voiceprint recognition on the first voice message to obtain user identity information corresponding to the first voice message; generates first information based on the user identity information and the text information corresponding to the first voice message; writes the first information into a first session history to obtain an updated second session history, wherein both the first and second session histories are used to store all interaction records generated after this wake-up trigger; when the first information contains an interaction intent to request an AI agent to handle business, the second session history is sent to the AI ​​agent; receives the second information sent by the AI ​​agent and performs corresponding interactive operations based on the second information.

[0020] Thirdly, a voice interaction device is provided, further configured as a unit for performing any of the methods in the second aspect. This device may be a terminal device or a chip within a terminal device.

[0021] Fourthly, a computer-readable storage medium is provided, which stores a computer program that, when executed by a voice interaction device, causes the voice interaction device to perform any of the resource control methods in the second aspect.

[0022] Fifthly, a computer program product is provided, comprising: a computer program that, when run by a voice interaction device, causes the voice interaction device to execute any of the resource control methods described in the second aspect.

[0023] It is understood that the beneficial effects of the second and fifth aspects mentioned above can be found in the relevant descriptions in the first aspect mentioned above, and will not be repeated here. Attached Figure Description

[0024] Figure 1 This is a schematic diagram illustrating an operational scenario between a display device and a control device provided in some embodiments of this application; Figure 2 This application provides configuration block diagrams of display devices according to some embodiments; Figure 3 System block diagrams of display devices provided in some embodiments of this application; Figure 4 System architecture block diagram of a voice interaction system provided in some embodiments of this application; Figure 5 A flowchart illustrating a voice interaction method provided in some embodiments of this application is shown; Figure 6 A flowchart illustrating a voice interaction method provided in some embodiments of this application is shown; Figure 7 A flowchart illustrating a voice interaction method provided in some embodiments of this application is shown; Figure 8 The following is a timing interaction diagram illustrating the voice interaction method provided in some embodiments of this application; Figure 9 The following is a timing interaction diagram illustrating the voice interaction method provided in some embodiments of this application; Figure 10 A schematic diagram of the first temporary session history provided in some embodiments of this application is shown. Detailed Implementation

[0025] The technical solutions in the embodiments of this application will now be described with reference to the accompanying drawings.

[0026] It should be understood that "multiple" as mentioned in this application refers to two or more. In the description of this application, unless otherwise stated, " / " indicates "or," for example, A / B can mean A or B; "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist, for example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Furthermore, to facilitate a clear description of the technical solutions of this application, the terms "first," "second," etc., are used to distinguish identical or similar items with essentially the same function and effect. Those skilled in the art will understand that the terms "first," "second," etc., do not limit the quantity or execution order, and that "first," "second," etc., do not necessarily imply differences.

[0027] References to "one embodiment" or "some embodiments" in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. Furthermore, the terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0028] Before providing a detailed explanation of the display device and voice interaction method provided in the embodiments of this application, the application scenarios and related technologies involved in the embodiments of this application will be described first.

[0029] Currently, AI agents for natural language dialogue on smart TVs and other display devices are mainly deployed in the cloud or on local terminals. After the user speaks a preset wake-up word to trigger the device, the display device receives the user's voice and uses the AI ​​agent to complete operations such as program retrieval, system function settings, media playback control, and message replies.

[0030] AI agents refer to autonomous artificial intelligence entities that are based on large models, possess the ability to perceive, remember, plan autonomously, invoke tools, and reflect on themselves, and can autonomously complete complex goals.

[0031] The AI ​​agent lacks speaker recognition and identity differentiation capabilities, supporting only one-to-one dialogues between single users and failing to adapt to collaborative interaction scenarios such as multiple users taking turns speaking or group discussions. The AI ​​agent employs a single wake-up mechanism and independent session IDs for session isolation and management. Each user generates an independent session cache upon waking up the device, with data not interconnected, preventing the system from integrating discussion information from multiple users in the same scenario across session sets.

[0032] Therefore, in multi-user collaborative interaction scenarios where multiple users take turns communicating and discussing program content, the AI ​​agent cannot fully grasp the entire dialogue, which can easily lead to problems such as speaker confusion, loss of contextual information, and misplacement of the respondent. This directly causes interruptions in the multi-user dialogue process, logical breaks, and frequent awkward and disjointed interactions, thus reducing the overall user experience in multi-user movie-watching scenarios.

[0033] For example, if A and B are watching TV together in the living room, A says, "Let's find a science fiction movie." After the device pushes a list of movies, B immediately adds, "Not foreign ones." Because the AI ​​agent cannot distinguish between the two speakers, it will lose the progressive semantic meaning of "filtering domestic science fiction movies" and will still only display overseas science fiction movies.

[0034] To this end, embodiments of this application provide a display device and a voice interaction method. This method can identify the identity information matched with each voice message through voiceprint recognition, integrate the identity information with the text information corresponding to the voice message to generate first information, and update it to the conversation history corresponding to the current wake-up. When the first information requires the AI ​​agent to process business, the method transmits the multi-turn conversation history carrying the complete user identity identifier to the AI ​​agent, so that the AI ​​agent can accurately identify different speakers in multi-turn dialogues, adapt to the multi-user collaborative interaction scenario of the display device, ensure the contextual coherence of multi-user turn-by-turn dialogue, and optimize the overall user experience of multi-user voice interaction.

[0035] The display device provided in this application can have various implementation forms, such as a television, a smart television, a laser projection device, a mobile phone, a tablet computer, a laptop computer, a monitor, an electronic bulletin board, an electronic table, etc.

[0036] Figure 1 This is a schematic diagram illustrating an operational scenario between a display device and a control device provided in some embodiments of this application. For example... Figure 1 As shown, the user can operate the display device 200 through the control device 110 or the smart device 120.

[0037] The control device 110 can be a remote control. Communication between the remote control and the display device includes infrared protocol communication, Bluetooth protocol communication, and other short-range communication methods, controlling the display device 200 wirelessly or via wired means. Users can control the display device 200 by inputting user commands through buttons on the remote control, voice input, control panel input, etc.

[0038] The control device 110 can also use a smart device 120 (such as a mobile terminal, tablet computer, computer, laptop computer, etc.) to control the display device 200. For example, the display device 200 can be controlled using an application running on the smart device.

[0039] The display device 200 can also be controlled in ways other than the control device 110 and the smart device 120. For example, it can be controlled by receiving user voice commands directly through a module configured inside the display device 200 for acquiring voice commands, or it can be controlled by receiving user voice commands through a voice control device set outside the display device 200.

[0040] Display device 200 can also communicate with server 300. Display device 200 can communicate via local area network (LAN), wireless local area network (WLAN), and other networks. Server 300 can provide various content and interactive features to display device 200. Server 300 can be a cluster or multiple clusters, and may include one or more types of servers.

[0041] For example, server 300 can deploy an AI agent (also known as an AI conversational agent, which is an AI agent that uses natural language dialogue as the interaction medium) to manage media resources based on received conversation history, parse natural language interaction commands, and coordinate terminal scheduling and backend business linkage. This enables display device 200 to complete operations such as program retrieval, system function settings, media playback control, and message reply based on the AI ​​agent, achieving continuous memory of interaction context in multi-user collaborative interaction scenarios, making multi-user collaborative interaction smooth and stable.

[0042] In some embodiments, the AI ​​agent may be deployed on the display device 200.

[0043] Figure 2 This is a configuration block diagram of a display device provided in some embodiments of this application. For example... Figure 2 As shown, the display device 200 includes at least one of the following: a tuner 210, a communicator 220, a detector 230, an external device interface 240, a controller 250, a display 260, an audio output interface 270, a memory, a power supply, and a user interface.

[0044] In some embodiments, the controller includes a processor, a video processor, an audio processor, a graphics processor, RAM, ROM, and a first interface to an nth interface for input / output.

[0045] The display 260 includes a display screen assembly for presenting images, a driving assembly for driving image display, a component for receiving image signals from the controller output, and a user control UI interface for displaying video content, image content, menu control interface, and user control UI interface.

[0046] The display 260 can be an LCD display, an OLED display, or a projection display, and can also be a projection device and a projection screen.

[0047] The communicator 220 is a component used to communicate with external devices or servers according to various communication protocol types. For example, the communicator may include at least one of the following: a Wi-Fi module, a Bluetooth module, a wired Ethernet module, other network communication protocol chips or near-field communication protocol chips, and an infrared receiver. The display device 200 can establish the transmission and reception of control signals and data signals with the control device 110 or the server 300 through the communicator 220.

[0048] The user interface can be used to receive control signals from the control device 110 (such as an infrared remote control).

[0049] Detector 230 is used to collect signals from the external environment or to interact with the outside world. For example, detector 230 includes a light receiver, a sensor for collecting ambient light intensity; or, detector 230 includes an image collector, such as a camera, which can be used to collect external environmental scenes, user attributes, or user interaction gestures; or, detector 230 includes a sound collector, such as a microphone, array microphone, etc., for receiving external sounds, such as a preset wake word or a first voice.

[0050] The external device interface 240 may include, but is not limited to, one or more of the following: High Definition Multimedia Interface (HDMI), analog or high-definition component input interface (component), composite video input interface (CVBS), USB input interface (USB), RGB port, etc. It may also be a composite input / output interface formed by multiple interfaces mentioned above.

[0051] The tuner / demodulator 210 receives broadcast television signals via wired or wireless means, and demodulates audio and video signals, such as EPG data signals, from multiple wireless or wired broadcast television signals.

[0052] In some embodiments, the controller 250 and the tuner 210 may be located in different separate devices, that is, the tuner 210 may also be located in an external device of the main device where the controller 250 is located, such as an external set-top box.

[0053] The controller 250 controls the operation of the display device and responds to user operations through various software control programs stored in the memory. The controller 250 controls the overall operation of the display device 200. For example, in response to received voice, it performs voiceprint recognition to determine the user's identity information corresponding to the voice; it also integrates the user's identity information with the text information corresponding to the voice into first information and writes the first information into the session history; and it uploads the session history to a cloud server, etc.

[0054] In some embodiments, the controller includes at least one of a central processing unit (CPU), a video processor, an audio processor, a graphics processing unit (GPU), random access memory (RAM), read-only memory (ROM), a first to an nth interface for input / output, a communication bus, etc.

[0055] Users can input commands by entering specific sounds or gestures. The user input interface then uses sensors to recognize the sounds or gestures and receives the user input commands.

[0056] The audio output interface 270 includes a speaker for playing audio data, such as a second voice.

[0057] Figure 3 System block diagrams of display devices provided in some embodiments of this application, such as Figure 3 As shown, the system is divided into four layers, from top to bottom: the Applications layer (referred to as the "Application Layer"), the Application Framework layer (referred to as the "Framework Layer"), the Android runtime and system library layer (referred to as the "System Runtime Library Layer"), and the kernel layer.

[0058] The application layer runs at least one application, which can be a built-in Windows program, system settings program, or clock program of the operating system; or it can be an application developed by a third-party developer. In specific implementations, the application packages in the application layer are not limited to the examples above.

[0059] The framework layer provides application programming interfaces (APIs) and a programming framework for applications. The application framework layer includes predefined functions. It acts as a central processing unit, determining the actions taken by applications within the application layer. Through the API, applications can access system resources and obtain system services during execution.

[0060] like Figure 3 As shown, the application framework layer in this embodiment includes managers, content providers, etc., wherein the managers include at least one of the following modules: ActivityManager, which interacts with all activities running in the system; LocationManager, which provides access to system location services for system services or applications; PackageManager, which retrieves various information related to application packages currently installed on the device; NotificationManager, which controls the display and clearing of notification messages; and WindowManager, which manages icons, windows, toolbars, wallpapers, and desktop widgets on the user interface.

[0061] The system runtime library layer provides support for the upper layer, namely the framework layer. When the framework layer is used, the Android operating system will run the C / C++ libraries contained in the system runtime library layer to implement the functions that the framework layer needs to achieve.

[0062] The kernel layer is the layer between hardware and software. For example... Figure 3 As shown, the kernel layer includes at least one of the following drivers: audio driver, display driver, Bluetooth driver, camera driver, Wi-Fi driver, USB driver, HDMI driver, sensor driver (such as fingerprint sensor, temperature sensor, pressure sensor, etc.), and power driver.

[0063] Figure 4 This is a system architecture block diagram of a voice interaction system provided in some embodiments of this application. For example... Figure 4 As shown, taking the deployment of an AI agent on a server as an example, the voice interaction system includes a display device and a server, and the display device and the server are connected in communication.

[0064] The display device includes a communication module 410, a session management module 420, a voiceprint recognition module 430, a speech-to-text (STT) module 440, and a text-to-speech (TTS) module 450.

[0065] It should be understood that the communication module 410, session management module 420, voiceprint recognition module 430, STT module 440 and TTS module 450 encapsulate underlying capabilities such as communication, session management, voiceprint recognition and voice interaction, and isolate the application layer from the system's underlying interface. They belong to the voice service middleware category.

[0066] The communication module 410 is used for data communication and interaction between the display device and the server, and the AI ​​agent within the server.

[0067] The conversation management module 420 is used to complete all lifecycle conversation management operations. It works with the neural network model to realize conversation topic relevance determination, AI wake-up recognition, conversation content compression and other processing flows, and has management and control capabilities such as writing and deleting conversation history records.

[0068] It should be understood that the neural network model mounted on the display device can perform at least one of the following tasks: voice and conversation history correlation analysis (such as the correlation between the first voice and the first conversation history), AI agent wake-up determination (such as whether the first message contains an interactive intent to request the AI ​​agent to handle business), and conversation content compression. The deployment of this neural network model can adopt an integrated large-parameter model (such as a 7B parameter model), with a single model uniformly performing all the above tasks; or it can adopt a multi-model distributed architecture, with each sub-model independently undertaking its corresponding business task.

[0069] In some embodiments, the neural network model may include a preset first model, a second model, and a third model.

[0070] The first model is used to identify, based on the updated conversation history (second conversation history), whether the most recently written voice in the conversation history contains an interactive intent to request the AI ​​agent to handle business, thereby determining whether to trigger an AI agent response. The input to the first model is: information parsed from the received voice and the updated conversation history; the output is a value in the range of 0 to 1 (first confidence level), representing the probability of waking up the AI ​​agent to respond.

[0071] The second model is used to analyze the correlation between the received speech and the previous conversation history (first conversation history); the input of the second model is the information after speech parsing and the conversation history stored before the speech (first conversation history); the output is a value between 0 (unrelated) and 1 (related) (second confidence level), which represents the correlation between speech and conversation history.

[0072] The third model is used when the data storage capacity of temporary session history exceeds a preset capacity threshold. It compresses the content of the temporary session history, such as extracting the central arguments of the session. The third model can combine local session history and temporary session history to compress the content of all sessions; it can also combine local session history and temporary session history to compress the content of temporary sessions.

[0073] The local session history is a file used to store key session information. The temporary session history is used to store recently received interactions, and the number of interactions that can be stored can be configured to be 20-40. The data storage capacity of the temporary session history can be determined by the AI ​​agent's ability to process context.

[0074] When the data storage capacity of temporary session history exceeds the preset capacity threshold, session compression is triggered. The compressed key information is stored in the local session history, and the temporary session history is cleared.

[0075] It should be understood that either the first session history or the second session history may include both temporary session history and local session history; for example, the first session history may include the first temporary session history and local session history, and the second session history may include the second temporary session history and local session history.

[0076] In some other embodiments, the first model, the second model, and the third model support integrated deployment, and are uniformly integrated into a single large-parameter model.

[0077] The voiceprint recognition module 430 is used to identify the voiceprint features of the received speech (such as the first speech) and match the voiceprint features with the voiceprint feature templates stored in the user's voiceprint database; if the match is successful, the user identity information corresponding to the speech is determined; if the match fails, temporary identity information is generated based on the voiceprint features.

[0078] Voiceprint features are unique biological characteristic parameters in speech, such as spectrum, timbre, speech rate, and prosody.

[0079] In some embodiments, if no voice corresponding to the temporary identity information is received within a preset first time period, the temporary identity information can be deleted and cleared to reduce the local cache usage of the display device and the redundancy of invalid identity data.

[0080] It should be understood that the user voiceprint database is used to centrally store the voiceprint feature templates of all registered users, with each template bound to a unique user identity. The database can encrypt and store the voiceprint templates and manage them uniformly. During the verification and authentication phase, the module retrieves all voiceprint feature templates in the database and compares them one by one with the real-time extracted voiceprint features to determine the user's identity. After the verification and matching are completed, the corresponding user identity information is output, such as the user identity (user ID): userA, user1001, which serves as the identity recognition for various text information in the subsequent conversation history.

[0081] It should also be understood that the voiceprint recognition module 430 can also be used to collect users' voice samples, extract corresponding voiceprint features, generate voiceprint feature templates, and store the voiceprint feature templates in the user's voiceprint database to complete the user's voiceprint registration.

[0082] The STT module 440 is used for speech-to-text conversion.

[0083] The TTS module 450 is used for text-to-speech playback conversion.

[0084] To facilitate a further understanding of the technical solutions in some embodiments of this application, the technical solutions of the display device and display control method, and how these solutions solve the aforementioned technical problems, are described in detail below with reference to specific embodiments and accompanying drawings. The embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. Obviously, the described embodiments are only some, not all, of the embodiments of this application.

[0085] Figure 5 A flowchart illustrating a voice interaction method provided in some embodiments of this application is shown. This voice interaction method is applied to a display device or a controller of a display device. The following description uses a display device and an AI agent deployed on a server as an example. The display device is configured to perform the following steps S510-S570: S510: When a preset wake-up word is detected, the display device receives the first voice message.

[0086] Among them, the preset wake word is a dedicated voice trigger command that is pre-configured and effective in the display device. It is a prerequisite for enabling the device's voice interaction function, and voice reception will only be initiated after the preset wake word is matched.

[0087] The first voice is any voice spoken by the user after the display device is woken up by a preset wake word. It may include the first voice after being woken up, as well as other voices used in subsequent interactions.

[0088] S520: The display device performs voiceprint recognition on the first voice to obtain the user identity information corresponding to the first voice.

[0089] For example, user identity information may include a user identifier.

[0090] In some embodiments, user identity information may also include structured information such as user-specific session configuration and user identity tags (administrator, ordinary family member, child, etc.) to provide a basis for subsequent personalized interactions.

[0091] The display device extracts the voiceprint features of the first speech and compares these features one by one with the voiceprint feature templates stored in the user's voiceprint database stored locally on the device to calculate the similarity. When the similarity score reaches the identity recognition threshold, the corresponding user profile is matched, and standardized user identity information is output; if the match fails, the user who spoke the first speech is determined to be a temporary user, and temporary identity information, such as tourist identity information, is generated based on the voiceprint features.

[0092] S530: The display device generates first information based on the user's identity information and the text information corresponding to the first voice.

[0093] It should be understood that the STT module in the display device can convert the first speech into corresponding text information.

[0094] For example, the display device can use a local speech-to-text algorithm to transcribe, correct, and remove redundancy from the first speech to obtain the corresponding text (i.e., text information), thus restoring the semantics of the first speech.

[0095] The display device integrates and encapsulates the user identity information corresponding to the first voice message and the corresponding text information according to a preset data format into first information. The preset data format can be {user:text}.

[0096] For example, the display device transcribes the first voice message into text: "Check the weather for tomorrow for me"; it retrieves the user's identity information as user1001. After integrating the two types of information, the device generates the first message as "user1001: Check the weather for tomorrow for me".

[0097] In some embodiments, the first information may also include auxiliary information such as the timestamp of the current session interaction and the identifier of the display device.

[0098] S540, the display device writes the first information into the first session history to obtain the updated second session history.

[0099] The first session history is a collection of session data that exists after this wake-up is triggered and before the first information is written, either blank or containing records of previous interactions. After this wake-up and before the first interaction, the first session history is blank; if there are multiple rounds of interaction, it contains session data updated in the previous round.

[0100] The second session history is the latest set of session data obtained by updating and iterating after the first information is completely written into the first session history.

[0101] The session history (including the first session history and the second session history) is used to store all interaction records generated after this wake-up is triggered. All voice, data, and operation records within a single wake-up cycle are summarized here and are only effective during this wake-up. After the device goes to sleep or wakes up again, it will be reset to a new session history.

[0102] S550: When the first information contains an interactive intent to request the AI ​​agent to process business, the display device sends the second session history to the server.

[0103] AI agents refer to intelligent processing modules deployed on servers, which have capabilities such as semantic understanding, information retrieval, business processing, content generation, and data response, and can handle complex interactive business that cannot be completed locally on display devices.

[0104] The display device parses the intent of the first information. When the first information contains an interactive intent to request the AI ​​agent to handle business, the second session history is sent to the AI ​​agent on the server so that the AI ​​agent can perform relevant business processing.

[0105] S560, the display device receives the second information sent by the server.

[0106] After receiving the second session history, the server uses an AI agent to analyze the user's needs, complete the corresponding business processing, generate second information that matches the user's needs, and send it back to the display device.

[0107] It should be understood that the second information can be a response message to the first information, or it can be operation information corresponding to the operation performed based on the first information, which can be used to provide data support for various subsequent interactive operations.

[0108] S570. Perform the corresponding interactive operation based on the second information.

[0109] When the second message is a response to the first message, it is either converted into a second voice message and played, or displayed on a screen. At this point, the second message can be written to the second session history, resulting in an updated third session history.

[0110] When the second information is operation information corresponding to the operation performed based on the first information, the second information is converted into an operation instruction and the instruction is executed.

[0111] The voice interaction method provided in this application can identify the identity information matched with each voice message through voiceprint recognition, integrate the identity information with the text information corresponding to the voice message to generate first information, and update it to the conversation history corresponding to this wake-up; and when the first information needs the AI ​​agent to process business, the multi-turn conversation history carrying the complete user identity identifier is transmitted to the AI ​​agent, so that the AI ​​agent can accurately identify different speakers in multi-turn dialogues, adapt to the multi-user collaborative interaction scenario of the display device, ensure the contextual coherence of multi-user turn-by-turn dialogue, and optimize the overall user experience of multi-user voice interaction.

[0112] Based on the voice interaction scheme described above, in some embodiments, after executing step S540 to write the first information into the first session history and obtaining the updated second session history, and before sending the second session history to the server in step S550 when the first information has an interactive intent to request the AI ​​agent to process business, it can also be determined whether the first information has an interactive intent to request the AI ​​agent to process business. This will be further explained below.

[0113] Figure 6 The following is a flowchart illustrating a voice interaction method provided in some embodiments of this application. This voice interaction method is applied to a display device or a controller of a display device. Taking a display device and an AI agent deployed on a server as an example, the process of determining whether the first information contains an interaction intent to request the AI ​​agent to process business includes the following steps S610-S620: S610, The display device inputs the second session history into a preset first model and obtains the first confidence level output by the first model.

[0114] It should be understood that the first model is a model deployed locally on the display device, used to determine the interaction intent of the first message based on the second session history; the first confidence level is the output value of the first model, which can identify the degree of credibility of the interaction intent of the first message as requiring AI intelligent agent to process business.

[0115] The display device reads the complete history data of the generated second session, inputs the data into the first model for calculation, and the model outputs the corresponding first confidence value.

[0116] For example, the second conversation history contains three rounds of dialogue: User A asks, "Want to watch TV series XX later?", User B replies, "I want to watch it, but I don't know which episode it's on.", and User C responds, "Could you check the complete synopsis of this show for me?", and the model outputs a first confidence level of 0.98.

[0117] For example, the second conversation history contains 5 rounds of dialogue: User X asks "What movie should I watch tonight?", User H replies "A suspense movie", User M says "Suspense movies are depressing", User H then says "How about a comedy instead?", and User X responds "Okay"; the model outputs a first confidence level of 0.45.

[0118] It should also be understood that the model deployed locally on the display device can be an integrated large-parameter model, and the functions corresponding to the first model can be implemented by the large-parameter model. This application does not limit this.

[0119] S620. If the first confidence level is greater than or equal to the preset first threshold, the display device determines that the first information contains an interactive intent to request the AI ​​agent to process the business.

[0120] The display device determines the presence of an interactive intent to request an AI agent to process business by judging a threshold value (preset first threshold) based on a pre-set confidence level.

[0121] The display device compares a first confidence level with a preset first threshold. When the first confidence level is greater than or equal to the first threshold, it determines that the first information contains an interactive intent to request an AI agent to handle business. For example, if the preset first threshold is 0.8, the above example yields a first confidence level of 0.98. Since 0.98 ≥ 0.8, the display device determines that the message "Please help me find the complete synopsis of this drama" contains an interactive intent to request an AI agent to handle business.

[0122] It should be understood that when the first confidence level is less than the first threshold, it is determined that the first information does not contain an interactive intent to request the AI ​​agent to process the business, and the subsequent steps will not be executed.

[0123] In some embodiments, keyword matching, intent retrieval, regular expression matching, and other methods can be used to determine whether the first information contains an interactive intent to request the AI ​​agent to process business.

[0124] For example, the display device extracts words from the text information corresponding to the first information and matches them with a preset business terminology (such as querying the weather, information, or plot retrieval) or wake word terminology (such as Xiaoxin, Xiaoxin Xiaoxin). If the match is successful, it is determined that the first information contains an interactive intent to request the AI ​​agent to process the business.

[0125] The display device can preset regular expression templates for consultation and query sentence patterns, such as sentence patterns like "help me check XXX", "introduce XX", "what is XX", etc. If the text information corresponding to the first information matches the preset regular expression template, it is determined that the first information has an interactive intent to request the AI ​​intelligent agent to process business.

[0126] The voice interaction method provided in this application utilizes voiceprint recognition to obtain the user identity corresponding to each voice message, combines the voice and text to generate first information, and updates it to the conversation history after the current wake-up. A first model is used to determine whether a cloud-based AI agent needs to process business. When there is a business requirement, a complete multi-turn conversation history with user identification is uploaded, facilitating the AI ​​agent to distinguish different speakers in the dialogue. This also improves the accuracy of intent recognition and the stability of interaction, adapts to multi-user interaction scenarios on display devices, maintains the contextual coherence of multi-user turn-based dialogue, and optimizes the multi-user voice interaction effect.

[0127] Based on the voice interaction scheme introduced above, during the execution step S530, when generating the first information according to the user's identity information and the text information corresponding to the first voice conversion, the correlation between the first voice and the first conversation history can also be determined. This will be further explained below.

[0128] Figure 7 The following is a flowchart illustrating a voice interaction method provided in some embodiments of this application. This voice interaction method is applied to a display device or a controller of a display device. Taking a display device and an AI agent deployed on a server as an example, the process of determining the correlation between a first voice and a first conversation history includes the following steps S710~S720: S710, The display device determines the correlation between the first voice and the first session history.

[0129] It should be understood that there are two types of speech: one is speech that is related to the current conversation, and the other is speech generated in the environment that is not related to the current conversation.

[0130] For example, the living room and dining room are connected. Users A and B are watching TV in the living room, while users C and D are dining in the dining room. The audio received by the TV is the voices of users C and D chatting during their meal. This audio data is background noise and has no relation to the interactive session triggered by this wake-up.

[0131] In some embodiments, the process by which the display device determines the relevance of the first voice to the first conversation history includes the following steps (1) to (2): (1) The display device inputs the text information corresponding to the first voice and the second model preset by the first conversation history to obtain the second confidence level output by the second model.

[0132] It should be understood that the second model is deployed locally on the display device and is used to perform semantic matching, scene association, and intent coherence calculation on the text information corresponding to the first speech and the first conversation history, and output a second confidence score; the second confidence score can identify the degree of correlation between the text information corresponding to the first speech and the first conversation history.

[0133] For example, the dialogue record of the first conversation history is as follows: User R speaks "Open the children's animation channel", and the TV replies "The children's animation section has been played for you". The first voice is user S saying "Play cartoon XX". The text information corresponding to the first voice, along with the first conversation history, is input into the second model, and the model calculates and outputs a second confidence level of 0.92.

[0134] It should also be understood that the model deployed locally on the display device can be an integrated large-parameter model, and the functions corresponding to the second model can be implemented by the large-parameter model, or the functions corresponding to the first model and the second model can both be implemented by the large-parameter model. This application does not limit this.

[0135] (2) If the second confidence level is greater than or equal to the preset second threshold, the display device determines that the first voice and the first conversation history are related.

[0136] The display device determines the critical value for determining the correlation between the first voice and the first conversation history by using a pre-set confidence threshold value (a preset second threshold).

[0137] The display device compares the output second confidence level with a preset second threshold to determine the correlation: if the second confidence level is greater than or equal to the preset second threshold, then the first speech is determined to be correlated with the first conversation history. For example, if the preset second threshold is 0.6, the above example yields a second confidence level of 0.92. Since 0.92 ≥ 0.6, the display device determines that "playing animation XX" is correlated with the first conversation history.

[0138] It should be understood that if the second confidence level is less than the preset second threshold, the display device determines that the first voice and the first conversation history are not related.

[0139] In some other embodiments, the correlation between the first speech and the first conversation history can also be determined by methods such as contextual timing logic verification and semantic similarity matching.

[0140] For example, the display device can calculate the semantic vector distance between the first voice and the first conversation history, and if the value meets the standard, it is determined to be relevant; or, for another example, the display device can combine the business logic before and after the conversation to determine whether the content of the first voice carries over the requirements of the previous round of conversation, and if so, it is determined to be relevant.

[0141] S720, when there is a correlation between the first voice and the first conversation history, the display device integrates the user identity information and text information to generate the first information.

[0142] For details on the implementation of step S720, please refer to the detailed description of step S530 above, which will not be repeated here.

[0143] It should be understood that if the first voice is not related to the first conversation history, the display device will not record or store the first voice and will directly discard the voice data.

[0144] The voice interaction method provided in this application, when the received first voice is related to the conversation history, combines the voice text to generate first information and updates it to the conversation history after the current wake-up, thereby improving the contextual coherence of the conversation and the accuracy of voice interaction intent recognition; furthermore, when the first information requires the AI ​​agent to process business, the multi-turn conversation history carrying complete user identity identifiers is transmitted to the AI ​​agent, enabling the AI ​​agent to accurately identify different speakers in multi-turn dialogues, adapt to multi-user collaborative interaction scenarios of display devices, ensure the contextual coherence of multi-user turn-by-turn dialogues, and optimize the overall user experience of multi-user voice interaction.

[0145] Figure 8 The following is a timing interaction diagram of the voice interaction method provided in some embodiments of this application, such as... Figure 8 As shown, the display device includes a voiceprint recognition module, a speech-to-text module, a text-to-speech module, and a conversation management module. When a preset wake-up word is detected, the method includes the following steps S801~S817: S801, the voiceprint recognition module performs voiceprint recognition on the first voice and obtains the user identity information corresponding to the first voice.

[0146] The voiceprint recognition module identifies the voiceprint features of the first speech and matches these features with the voiceprint feature templates stored in the user's voiceprint database. If the match is successful, the user's identity information corresponding to the speech is determined. If the match fails, temporary identity information is generated based on the voiceprint features, thus determining the user's identity information corresponding to the first speech.

[0147] S802, the speech-to-text module converts the first speech into text information.

[0148] S803, the session management module can determine whether the current interactive session is a brand new session or an existing session.

[0149] The first voice is any voice spoken by the user after the display device is woken up by a preset wake word. It may include the first voice after being woken up, as well as other voices used in subsequent interactions.

[0150] For example, if the first voice is the first voice after wake-up, the current interaction session is a brand new session; if the first voice is another voice that is not the first voice in subsequent interactions after wake-up, the current interaction session is an existing session.

[0151] If the current interaction session is a brand new session, perform the following steps S804~S809: S804, The session management module clears the session history.

[0152] S805 The session management module generates the first message based on the user's identity information and text information.

[0153] S806, the session management module sends the first information to the server's AI agent.

[0154] S807, The conversation management module receives the response information returned by the AI ​​agent.

[0155] S808, the session management module writes the first message and the reply message into the session history.

[0156] S809 The text-to-speech module generates and plays a reply voice message based on the reply information.

[0157] If the current interactive session is an existing, ongoing session, perform the following steps: S810, the conversation management module compares the text information corresponding to the first voice with the second model preset by the first conversation history input to obtain the second confidence level output by the second model.

[0158] The session management module compares the second confidence level with a preset second threshold: if the second confidence level is greater than or equal to the preset second threshold, it is determined that the first voice is related to the first session history; if the second confidence level is less than the preset second threshold, the display device determines that the first voice is not related to the first session history.

[0159] If the first voice message is unrelated to the first session history, the session management module will not record or store the first voice message and will directly discard the voice data.

[0160] If the first voice correspondence is related to the first session history, perform the following steps S811~S813: S811, the session management module integrates user identity information with text information to generate the first message.

[0161] S812, The session management module writes the first information into the first session history to obtain the updated second session history.

[0162] S813. The session management module inputs the second session history into the preset first model and obtains the first confidence level output by the first model.

[0163] The display device compares a first confidence level with a preset first threshold. When the first confidence level is greater than or equal to the first threshold, it determines that the first information has an interactive intent to request an AI agent to process business; when the first confidence level is less than the first threshold, it determines that the first information does not have an interactive intent to request an AI agent to process business.

[0164] If the initial information does not contain an interactive intent to request the AI ​​agent to process business, the session management module will not execute subsequent steps.

[0165] If the first piece of information contains an interactive intent to request the AI ​​agent to process business, the following steps S814~S81 are executed: S814, The session management module sends the second session history to the AI ​​agent.

[0166] For example, the session management module may send a prompt word containing the second session history. For instance, the prompt word may include the second session history and the response format requirements (such as a concise and friendly reply).

[0167] S815, the session management module receives the second information returned by the AI ​​agent.

[0168] S816. The session management module writes the second information into the second session history to obtain the updated third session history.

[0169] S817 The text-to-speech module generates and plays the second speech based on the second information.

[0170] In some embodiments, if no interaction record with the AI ​​agent is generated within a preset second time period (e.g., 10 minutes), the display device will pop up a conversation confirmation interface, prompting the user to choose whether to continue the previous conversation; the user can also manually initiate a new conversation through voice commands or control devices; after the user confirms the start of a new conversation, the device will clear all conversation history records.

[0171] The voice interaction method provided in this application uses voiceprint recognition to obtain the user identity corresponding to each voice message. When the received first voice message is related to the conversation history, it combines the voice text to generate first information and updates it to the conversation history after the current wake-up. It uses a first model to determine whether the cloud AI agent needs to process the business. When the first information indicates that the AI ​​agent needs to process the business, it transmits the multi-turn conversation history carrying the complete user identity identifier to the AI ​​agent, so that the AI ​​agent can accurately identify different speakers in the multi-turn dialogue, adapt to the multi-user collaborative interaction scenario of the display device, ensure the contextual coherence of multi-user turn-by-turn dialogue, and optimize the overall user experience of multi-user voice interaction.

[0172] Based on the voice interaction scheme introduced above, in some embodiments, the conversation history can be divided into two types: temporary and local conversation history. This can distinguish between short-term interaction cache and long-term retention records, realize hierarchical data management, and improve the accuracy of interaction. The following is a supplementary explanation of this.

[0173] Figure 9The diagram illustrates a timing interaction diagram of a voice interaction method provided in some embodiments of this application. The first session history includes a first temporary session history and a local session history, and the second session history includes a second temporary session history and a local session history.

[0174] like Figure 9 As shown, the process of writing the first information into the first session history to obtain the updated second session history includes the following steps: S901, The session management module writes the first information into the first temporary session history.

[0175] The first temporary session history is a temporary cache branch of the first session history, storing short-term, real-time, and coherent interaction data within the current interactive session of the display device, such as... Figure 10 As shown, the data is only temporarily stored in the device's running memory to handle semantic recognition for multiple rounds of continuous voice interaction. It has no long-term retention attribute and can be dynamically updated.

[0176] It should be understood that the first piece of information is written in an append-only manner to ensure that the data from this round of interactions is consistent with the historical short-term interaction data.

[0177] S902, The session management module determines the updated second temporary session history.

[0178] The second temporary session history is a new short-term session cache data set that is dynamically updated after the first temporary session history is written with the first information. It belongs to the temporary branch of the second session history.

[0179] It should be understood that during the above process cycle, after updating the temporary session history, when the content stored in the temporary session history is large, the display device can compress the content as needed, clear the temporary session history, and extract the key information of the session to store in the local session history.

[0180] S903. If the data storage capacity of the second temporary session history is greater than the preset capacity threshold, the session management module will compress the second temporary session history.

[0181] The temporary session history can be compressed using the third model. When the data storage capacity of the temporary session history exceeds the preset capacity threshold, session compression is triggered. The compressed key information is stored in the local session history, and the temporary session history is cleared.

[0182] It should be understood that the session management module's lightweight processing of temporary session data can retain core key information of the interaction (such as user identity, core semantics, and interaction scenario), remove redundant descriptions, duplicate fields, and invalid time-series notes, and reduce the data volume while retaining valid information.

[0183] S904, The session management module writes the compressed data into the local session history.

[0184] The voice interaction method provided in this application divides the conversation history into two types: temporary and local. Short-term interaction data relies on the real-time caching of temporary conversation history to ensure the continuity of interaction. After the data accumulates beyond a preset capacity threshold, it is compressed, and key data is written into the local conversation history for long-term storage. At the same time, redundant temporary cache is cleared to improve the accuracy of the overall interaction.

[0185] It should be understood that the sequence numbers of the processes in the above embodiments do not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention. The various embodiments described herein can be independent solutions or combinations based on internal logic, and all such solutions fall within the protection scope of this application.

[0186] It should also be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowchart may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0187] Based on the same concept, as an implementation of the above method, this application provides a voice interaction device. This device embodiment corresponds to the aforementioned method embodiment. For ease of reading, this device embodiment will not repeat the details of the aforementioned method embodiment one by one, but it should be clear that the device in this embodiment can implement all the contents of the aforementioned method embodiment.

[0188] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the voice interaction methods provided in the above embodiments.

[0189] This application also provides a chip, which includes a processor and a memory. The memory stores a computer program, which, when executed by the processor, implements the voice interaction methods provided in the above embodiments.

[0190] This application also provides a computer program product, which includes a computer program that, when executed by an execution device, causes the execution device to implement the voice interaction methods provided in the above embodiments.

[0191] It should be understood that the processor mentioned in the embodiments of this application can be a CPU, GPU, image signal processor (ISP), and / or neural network processing unit (NPU), or other general-purpose processors. A general-purpose processor can be a microprocessor or any conventional processor.

[0192] It should also be understood that the memory mentioned in the embodiments of this application may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory.

[0193] In the embodiments provided in this application, the division of various frameworks, modules, or units is merely a logical functional division. In actual implementation, there may be other division methods. For example, multiple frameworks, modules, or units may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the functional modules in the various embodiments of this application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0194] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0195] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A display device, characterized in that, include: monitor; The controller is configured as follows: Upon detecting a preset wake-up word, the first voice message is received; Perform voiceprint recognition on the first voice to obtain the user identity information corresponding to the first voice; First information is generated based on the user identity information and the text information corresponding to the first voice. Write the first information into the first session history to obtain the updated second session history, wherein both the first session history and the second session history are used to store all interaction records generated after this wake-up trigger; When the first information contains an interactive intent to request the AI ​​agent to process business, the second session history is sent to the AI ​​agent; The system receives the second information sent by the AI ​​agent and performs the corresponding interactive operation based on the second information.

2. The display device according to claim 1, characterized in that, After the controller executes the process of writing the first information into the first session history and obtaining the updated second session history, and before sending the second session history to the AI ​​agent when the first information contains an interaction intent to request the AI ​​agent to process business, it is further configured to: The second conversation history is input into a preset first model to obtain a first confidence level output by the first model; the first model is used to determine the interaction intent of the first information based on the second conversation history. If the first confidence level is greater than or equal to a preset first threshold, then it is determined that the first information contains an interactive intent to request the AI ​​agent to process business.

3. The display device according to claim 1, characterized in that, The controller generates first information based on the user identity information and the text information corresponding to the first speech conversion, and is configured as follows: Determine the correlation between the first voice and the first session history; If the first voice message is related to the first conversation history, the user identity information and the text information are integrated to generate the first information.

4. The display device according to claim 3, characterized in that, The controller is also configured to: If the first voice message is not related to the first session history, delete the first voice message.

5. The display device according to claim 3 or 4, characterized in that, The controller is configured to perform the determination of the relevance between the first voice and the session history. The text information corresponding to the first speech is compared with the second model preset by the first conversation history input to obtain the second confidence level output by the second model; If the second confidence level is greater than or equal to a preset second threshold, then it is determined that the first voice is related to the first conversation history; If the second confidence level is less than a preset second threshold, then it is determined that the first voice is not related to the first conversation history.

6. The display device according to any one of claims 1 to 4, characterized in that, When the second information is a response message used to reply to the first information, the controller performs a corresponding interactive operation based on the second information, configured as follows: Write the second information into the second session history to obtain the updated third session history; The second information is converted into a second voice and played, or the second information is displayed on the display.

7. The display device according to any one of claims 1 to 4, characterized in that, The first session history includes a first temporary session history and a local session history; the controller executes the process of writing the first information into the first session history to obtain an updated second session history, which is configured as follows: The first information is written into the first temporary session history to obtain the updated second temporary session history, which includes the second temporary session history and the local session history.

8. The display device according to claim 7, characterized in that, After the controller executes the process of writing the first information into the first temporary session history and generating the updated second temporary session history, it is further configured to: If the data storage capacity of the second temporary session history is greater than the preset capacity threshold, the second temporary session history is compressed, and the compressed data is written into the local session history.

9. The display device according to any one of claims 1 to 4, characterized in that, The AI ​​agent is deployed on a server that establishes a communication connection with the display device.

10. A voice interaction method, characterized in that, Applied to the display device according to any one of claims 1 to 9, comprising: Upon detecting a preset wake-up word, the first voice message is received; Perform voiceprint recognition on the first voice to obtain the user identity information corresponding to the first voice; First information is generated based on the user identity information and the text information corresponding to the first voice. Write the first information into the first session history to obtain the updated second session history, wherein both the first session history and the second session history are used to store all interaction records generated after this wake-up trigger; When the first information contains an interactive intent to request the AI ​​agent to process business, the second session history is sent to the AI ​​agent; The system receives the second information sent by the AI ​​agent and performs the corresponding interactive operation based on the second information.

Images