Distributed audio system construction method and system based on graphical intention

Abstract

The invention provides a method and a system for constructing a distributed audio system based on a graphical intention. The method comprises the following steps: acquiring the graphical intention; generating a local execution state corresponding to each agent body according to the graphical intention and the node capability and end point information corresponding to each agent body, and sending first adjustment information indicating the local execution state to each agent body; receiving local verification information returned by each agent, wherein the local verification information is used for indicating whether the corresponding local execution state can be executed or not; after it is determined that all the agents corresponding to the graphical intention pass local verification, link establishment information is sent to all the agents, the link establishment information is used for indicating all the agents to complete link establishment based on the current local execution states of all the agents, and a target audio system is obtained; and receiving state information uploaded by each agent, and judging whether to update the graphical intention or the graphical intention version according to the state information. The problems that a multi-node audio system lacks unified expression, media constraints are implied, and the reconfiguration capacity is insufficient are solved.

Description

Technical Field

[0001] This invention relates to the field of distributed audio system technology, specifically to a method and system for constructing a distributed audio system based on graphical intents. Background Technology

[0002] Currently, networked audio technologies have been widely applied in professional sound reinforcement, conferencing systems, educational settings, broadcast transmission, immersive performances, and multimedia control. Existing solutions typically enable audio acquisition and playback, device discovery, signal routing, channel mapping, and basic network transmission. Some mature systems further incorporate network diagnostics, clock synchronization, remote configuration, and status monitoring capabilities. However, the core organizational structure of most existing technologies still revolves around single devices, single links, or fixed platforms. Their system abstraction layers are mostly at the device, port, or channel routing levels, making it difficult to provide a unified expression and consistent control for multi-node audio systems that span hosts, platforms, and network environments.

[0003] The shortcomings of existing technologies are mainly reflected in the following aspects. First, there is a lack of unified modeling capabilities for the overall structure of distributed audio systems. Although many existing systems can achieve audio transmission between devices, their control methods usually rely on the item-by-item configuration of device parameters, channel connections, or static routing relationships. The overall system topology is not expressed in a unified, explicit, and computable way. Under such an architecture, once the number of nodes increases, the device types become heterogeneous, or the host platforms differ, the system structure easily relies on manual understanding, experience-based maintenance, and local conventions, leading to complex deployment, difficult maintenance, and hindering automated verification and large-scale expansion. Second, the handling of media constraints is too implicit and difficult to adapt to heterogeneous devices and complex deployment environments. Many existing systems default to all nodes sharing a unified sampling rate, buffer length, or link parameters, or treat these parameters as external preset conditions rather than internal, describable, negotiable, and verifiable components. This approach may work when the number of devices is small and the platform is relatively simple, but under conditions of multiple hosts, multiple interfaces, multiple network paths, and phased deployment, it is prone to compatibility issues, link establishment failures, instability, or difficulty in locating faults.

[0004] Secondly, existing technologies lack a unified collaborative mechanism for bidirectional interaction. Specifically, the relationship between the control end and the execution end is usually one-way downward transmission or static configuration. The relationship between nodes is often understood as a fixed-direction audio transmission relationship, making it difficult to support bidirectional control and state interaction between the control end and nodes, and also difficult to support interaction between nodes based on system objectives. Therefore, when the system requires continuous collaboration, state feedback, operational adjustments, and multi-node joint control, existing solutions often lack a unified semantics and mechanism. Furthermore, the coupling between control logic, execution logic, platform adaptation logic, and testing and verification logic is high. While many existing solutions can run stably on a fixed platform, when the control end, execution end, driver layer, and audio runtime evolve simultaneously, problems such as decreased compatibility, increased maintenance costs, and insufficient cross-platform capabilities easily arise. Simultaneously, the lack of a unified technical syntax and state expression method between the control end and nodes leads to inconsistent node capability descriptions, difficulty in reproducing configuration results, and unpredictable system behavior.

[0005] Furthermore, existing technologies have limited support for runtime state changes and system reconfiguration. Traditional solutions often enter static operation after system setup. When faced with topology changes, network quality fluctuations, configuration version updates, device offline / outdoor status, or fault recovery, manual intervention or reliance on decentralized diagnostic mechanisms is usually required, making it difficult to achieve unified state tracking, anomaly classification, and dynamic reconfiguration. As a result, although the system can connect during the initial setup phase, it lacks sufficient observability, recoverability, and maintainability in a continuous operating environment. Finally, existing technologies lack a systematic mechanism for pre-deployment verification. Many network audio systems still rely on joint debugging tests after actual hardware connection for verification. The stability of system behavior, the effectiveness of fault switching, and the compatibility between different nodes often need to be confirmed through repeated testing after the devices are connected. This approach not only leads to high R&D and deployment costs but also hinders problem reproduction and regression control, making it difficult to support the engineering delivery of complex audio systems.

[0006] Therefore, the lack of unified expression, implicit media constraints, insufficient two-way collaboration mechanism, high coupling between control and execution, insufficient runtime reconfiguration capability, and difficulty in pre-deployment verification are the technical problems that need to be solved in multi-node audio systems.

[0007] An existing distributed control system includes a first server and multiple device nodes. The first server establishes communication connections with each of the device nodes. The first server obtains the registration information of the first online device node. After successful registration, the server retrieves the device interface of the first device node based on its registration information and calls the device interface to control the device node. The first server includes a first executor that acts as a script executor. The script can directly control the device, obtain device status and data, and trigger corresponding event handling scripts according to the defined time. All hardware capabilities of the device are aggregated into the script, which can implement all hardware control processes required by external systems. The first communication terminal of the device control system provides control and data soft links. Data interaction via data interface calls is achieved through data soft links. However, this distributed system is designed for rapid device node access and is not intended for distributed audio systems. Specifically, the server still calls device interfaces sequentially via scripts. When used in a multi-node audio system, the aforementioned technical problems persist. Summary of the Invention

[0008] The primary objective of this invention is to provide a method for constructing a distributed audio system based on graph intents, addressing issues such as the lack of unified expression, implicit media constraints, insufficient bidirectional collaboration mechanisms, coupling between control and execution, and inadequate runtime reconfiguration capabilities in multi-node audio systems.

[0009] A second objective of this invention is to provide a system for implementing the above-described method for constructing a distributed audio system based on schematic intent.

[0010] To achieve the aforementioned first objective, this invention provides a method for constructing a distributed audio system based on graph intents, comprising the following steps: establishing communication relationships with multiple agents and obtaining the node capabilities and endpoint information corresponding to each agent; obtaining graph intents, wherein the graph intents include a node set, an endpoint set, a connection set, and a media constraint set, the node set representing nodes participating in the construction of the target audio system, the endpoint set representing endpoints provided by each node, the connection set representing media stream interaction relationships between endpoints, and the media constraint set representing media conditions corresponding to each media stream interaction relationship; and obtaining the graph intents based on the graph intents. The system generates a local execution state for each agent based on the graph and the corresponding node capabilities and endpoint information. It then sends first adjustment information to each agent to indicate its local execution state. The system receives local verification information from each agent, which indicates whether the corresponding local execution state is executable. After confirming that all agents corresponding to the graph intent have passed local verification, the system sends link establishment information to each agent, instructing each agent to establish a link based on its current local execution state to obtain the target audio system. Finally, the system receives status information uploaded by each agent and determines whether to update the graph intent or its version based on the status information.

[0011] As can be seen from the above scheme, the present invention provides a unified system expression method for distributed audio scenarios. It uses graph intents to uniformly describe the nodes, endpoints, connection relationships and related media constraints in the target audio system, enabling multi-node audio systems to be expressed and processed in a holistic manner. Furthermore, based on graph intents, a technical mechanism maps the global target to the local execution states of each agent, allowing each agent to verify according to its own capabilities and constraints. The verification is achieved by executing the local execution states to realize media stream interaction between agents. In addition, after the target audio system is built, it can monitor the system status through state information and determine whether to incorporate the current state change into the current target audio system. Thus, when the system status changes, it is not necessarily updated (for example, the addition of a new agent or the disconnection of an agent does not affect the normal operation of the current target audio system). The graph intents or graph diagram versions are only updated when necessary. As can be seen, this invention constructs an audio system construction method for complex networked digital audio scenarios through technical features such as graphical intent, distributed mapping, explicit constraints, bidirectional collaboration, state awareness, and verification support. It has significant advantages in terms of system expressiveness, heterogeneous adaptability, runtime controllability, engineering feasibility, and deployment reliability.

[0012] A further approach includes modifying the first adjustment information based on local verification information to obtain the second adjustment information; and sending the second adjustment information to the agent.

[0013] Therefore, it can be seen that the present invention can change the local execution state of the relevant agent by sending the second adjustment information if the local verification of the agent corresponding to the schematic diagram is not passed. This adjustment mechanism ensures that the chain establishment information is sent only when the local verification of the agent corresponding to the schematic diagram is passed.

[0014] A further approach is to obtain the graph intent based on the node capabilities and endpoint information corresponding to the agent.

[0015] As can be seen, the present invention obtains and stores the node capabilities and endpoint information corresponding to each agent to determine the schematic intent.

[0016] A further approach involves establishing communication relationships with the agent by: discovering each node participating in the system through a network discovery mechanism, a preset address mechanism, or a node active registration mechanism, and establishing communication relationships with the agent corresponding to the node.

[0017] Therefore, the present invention can communicate with the nodes involved in the construction of the target audio system and construct a graph intent based on the node capabilities and endpoint information.

[0018] A further approach is to send a second adjustment message to the corresponding agent based on the diagram intent, local verification information, node capabilities, and endpoint information.

[0019] Therefore, by using diagrammatic intent, local verification information, node capabilities, and endpoint information, the present invention can better address the reasons why local verification fails and readjust the local execution state of the relevant agent by changing the second adjustment information.

[0020] A further approach is to include graph version information in the graph intent.

[0021] Therefore, this invention can manage graphical intents using graphical version information, thereby facilitating the construction of different target audio systems and allowing for rollback to the previous graphical intent version based on the graphical version information when an anomaly occurs.

[0022] A further approach involves sending connection establishment information to each agent, and also includes: if it is determined from the status information that the diagram intent or diagram intent version needs to be updated, sending third adjustment information to the agent based on the updated diagram intent or diagram intent version to adjust the target audio system.

[0023] Therefore, this invention has strong runtime observability and dynamic adjustment capabilities, and can send third adjustment information to make corresponding adjustments according to the actual needs of the system in the event of topology changes, configuration changes, etc.

[0024] A further solution is that each node includes a unique identifier, and the corresponding attributes of the node include host information, platform type, running status, and functional role; the endpoints provided by each node include at least one of the following: input endpoint, output endpoint, acquisition endpoint, playback endpoint, processing endpoint, and virtual endpoint; media conditions include at least one of the following: sampling rate, bit depth, number of channels, buffer length, encoding format, clock requirements, maximum allowable latency, link priority, and fault tolerance strategy.

[0025] Therefore, this invention is applicable to continuous collaboration and operational adjustments in complex scenarios.

[0026] To achieve the second objective mentioned above, the present invention provides a distributed audio system based on graph intents, comprising: a control terminal and multiple agent entities, wherein the control terminal is connected to the agent entities; the control terminal is used to execute the aforementioned distributed audio system construction method based on graph intents; the agent entities are used to upload node capability and endpoint information, local verification information, and status information to the control terminal, and to establish links between agent entities based on local execution status.

[0027] As can be seen from the above scheme, the present invention enables the control end to uniformly organize the target audio system based on the diagram intent, and then maps the global target to the local execution state of each agent, so that each agent can complete local verification and execution according to its own capabilities and constraints. This transforms the distributed audio system from the traditional local configuration method to a systematic construction method oriented towards the overall goal, thereby improving the expressiveness, constructability, collaborativeness and maintainability of complex networked audio systems.

[0028] A further approach is that, before establishing links between agents based on their local execution states, the control end performs pre-verification based on graph intents, node capabilities, and endpoint information.

[0029] Therefore, for complex distributed audio systems, relying solely on integration testing after actual hardware access is not only inefficient but also detrimental to problem reproduction and version regression. This invention, in its architecture design, supports system verification before deployment. It verifies the target system structure, link compatibility, node capability adaptation, and operational behavior before actual hardware deployment, allowing for early checks of system behavior, compatibility, and anomalies. This helps reduce deployment risks and improve testing efficiency. Attached Figure Description

[0030] Figure 1 This is a general architecture diagram of an embodiment of the distributed audio system based on schematic intent of the present invention.

[0031] Figure 2 This is a schematic diagram of the functional units of the control terminal and the agent in an embodiment of the distributed audio system based on schematic intent of the present invention.

[0032] Figure 3 This is a mapping diagram from graphical intents to local execution states in an embodiment of the distributed audio system based on graphical intents of the present invention.

[0033] Figure 4 This is a flowchart of an embodiment of the distributed audio method based on graphical intent of the present invention.

[0034] The present invention will be further described below with reference to the accompanying drawings and embodiments. Detailed Implementation

[0035] This invention presents a distributed audio system construction method based on graph intents, addressing the challenges of system construction, link establishment, operational coordination, and dynamic adjustment in multi-node, heterogeneous, and networked audio scenarios. It uses graph intents to provide a unified description of the target audio system, and the control terminal maps the global target to the local execution states of each agent based on these graph intents. Here, "graph intent" refers to a system-level description that uses a graph structure to uniformly express the target audio system; and "distributed audio system" refers to a collaborative system composed of a control terminal, multiple agents, and a shared network.

[0036] The distributed audio system based on graph intents provided by this invention is not built around a single device, a single audio link, or a single control software. Instead, it takes a target audio system with multiple nodes participating as the object, uses graph intents to uniformly express the target system, and then the control terminal maps the graph intents to the local execution states of each agent. This allows each agent to complete verification, link establishment, execution, monitoring, and adjustment based on its own capabilities, resource conditions, and connection constraints, thereby forming a constructible, negotiable, observable, and reconfigurable target audio system.

[0037] For ease of description, the main terms used in this invention are explained below: Controller: Refers to the global control entity in the system, which is used to complete graph intent management, node discovery, capability acquisition, global mapping, local execution state generation, state summary and system adjustment.

[0038] Agent: refers to the running entity deployed on each node, used to complete local capability exposure, local verification, link establishment, execution, and status reporting.

[0039] Node: refers to the physical or virtual operating unit that carries the agent, which can be a general host, a special device, an embedded unit, or a virtual instance.

[0040] Endpoint: refers to the functional or logical interface within a node that can participate in audio acquisition, input, output, processing, or forwarding.

[0041] Shared network: refers to the network infrastructure that enables interaction between the control terminal and each agent, as well as among the agents themselves.

[0042] Media stream interaction (stream): refers to the streaming interaction relationship between nodes that transmit audio data.

[0043] Control interaction (control): refers to the interaction between the control terminal and the node regarding information such as capabilities, status, link establishment, and adjustments.

[0044] Local execution state: refers to the state description generated by the controller for a certain agent based on the global schema intent, which can be verified and executed locally by the agent.

[0045] Specifically, the control terminal of this invention organizes the distributed target audio system to be constructed based on a graph intent, and then maps the global target corresponding to the graph intent to the local execution state of each agent. This allows each agent to complete local verification and execution according to its own capabilities and constraints. The control terminal and each agent interact bidirectionally through a shared network for control and state interaction. Agents can also interact bidirectionally through the shared network for media stream interaction. The media stream interaction relationship between agents is dynamically determined by the graph intent, rather than being preset to a fixed topology, thus enabling the system to adapt to different node combinations and different audio task requirements. It should be noted that there is no control interaction relationship between agents, only a media stream interaction relationship. Therefore, the distributed audio system can be transformed from a traditional local configuration method to a systematic construction method oriented towards the overall goal, thereby improving the expressiveness, constructability, collaborativeness, and maintainability of complex networked audio systems.

[0046] This invention is applicable to scenarios requiring highly reliable networked audio collaboration, such as professional sound reinforcement, conferences, education, and cultural tourism performances. It is also applicable to complex digital audio systems that require multi-node joint operation, flexible expansion, and unified control. Its technical solution is not limited to a single device, a single platform, or a single network structure, but has good versatility and adaptability.

[0047] The terms "first," "second," and "third" in this invention are used to distinguish similar things and do not imply any order of precedence.

[0048] Example of a distributed audio system based on graph intent: The distributed audio system in this embodiment includes a control terminal and multiple agents. The control terminal is connected to each agent to achieve bidirectional data transmission.

[0049] The agent is used to upload node capability and endpoint information, local verification information, and status information to the control terminal. It also determines its own local execution state based on received adjustment information and executes its current local execution state according to received link establishment information to complete the link establishment. The control terminal instructs each agent on its own local execution state through adjustment information and controls each agent to execute its current local execution state through link establishment information, thereby establishing links between the agents.

[0050] See Figure 1 This embodiment includes a control terminal 10, a first agent 21, a second agent 22, a third agent 23, and a shared network 30. The control terminal 10 is connected to the first agent 21, the second agent 22, and the third agent 23 via the shared network 30. The first agent 21, the second agent 22, and the third agent 23 can selectively connect to each other via the shared network 30 (indicated by dashed lines). The media stream relationship between the agents is dynamically controlled by the control terminal based on graphical intent.

[0051] It should be noted that the present invention does not limit the number of proxy entities. In different embodiments, the number of proxy entities in the target audio system is determined according to the specific schematic intent. Furthermore, one proxy entity can correspond to one or more nodes.

[0052] The control terminal 10 can run on a physical unit, such as a central server. The control terminal 10 is used to discover nodes, obtain node capabilities and endpoint information, receive or generate graphical intents, determine the composition of the target audio system, generate adjustment information indicating the local execution state, instruct the agent to build the target audio system according to the local execution state, receive the running status of the nodes of the built target audio system, and trigger the dynamic adjustment or reconstruction of the system.

[0053] The first agent 21, the second agent 22, and the third agent 23 are deployed in different nodes to expose local identity, capabilities, endpoint information and current status, and to complete local verification, link establishment, audio data processing, status reporting and dynamic adjustment according to the local execution status issued by the control terminal.

[0054] The shared network 30 is used to carry control interactions between the control terminal 10 and the first agent 21, the second agent 22, and the third agent 23, and also to carry media stream interactions between the first agent 21, the second agent 22, and the third agent 23.

[0055] See Figure 2 The control unit 10 includes a node discovery and capability acquisition unit, a graph intent management unit, a local execution status generation unit, and a control information generation unit. Each agent includes a capability and endpoint exposure unit, a local verification unit, a link establishment and execution unit, and a status reporting unit.

[0056] The node discovery and capability acquisition unit is used to acquire and store the node capability and endpoint information sent by the agent's capability and endpoint exposure unit. The node capability and endpoint information provides the agent with computable capability descriptions of its corresponding nodes. Specifically, the node capability and endpoint information includes: node identity information, audio capability information, endpoint inventory information, network and link capability information, and local constraint information. This allows the control end to determine the executability of the graph intent based on the capability description information reported by each node. This method differs from existing methods that only perform static device discovery; the key point is that node capabilities are not merely for display but directly participate in subsequent graph mapping, link negotiation, and execution generation.

[0057] The Graph Intent Management Unit manages graph intents. Graph intents describe the nodes, endpoints, media stream interactions, and corresponding media conditions involved in the entire target audio system. Graph intents can be generated or received from the control terminal. Typically, after viewing the node discovery and capability acquisition units of each agent on the control terminal, the user selects the agents participating in the construction of the target audio system, the required endpoints, the media stream interactions between different agents, and the corresponding media conditions according to business needs. The control terminal then generates graph intents based on this. Graph intents can also be imported from preset templates, historical configurations, or external orchestration results. It should be noted that graph intents describe the relationships that the target audio system currently being built hopes to establish, not a simple complete set of all agents already connected to the system. Therefore, even if an agent has been discovered and its node capabilities and endpoint information has been stored, it will not participate in the audio path of the current target audio system unless it is included in a graph intent.

[0058] The control information generation unit generates adjustment information and link establishment information. Adjustment information indicates the local execution state of the agent. The agent can determine its own local execution state based on the adjustment information. It should be noted that the adjustment information received by the agent can be a global change to its current state or a local change to its current state. Link establishment information instructs the agent to execute the corresponding current local execution state to complete the link establishment. The current local execution state refers to the most recently received local execution state by the agent; the link is established after the agent executes the corresponding current local execution state.

[0059] Optionally, the link information may include specified link execution parameters, which correspond to specific media conditions. Thus, the current local execution state refers to the local execution state that the agent has just received in time, and also the local execution state after modifying the media conditions according to the link execution parameters.

[0060] Optionally, the adjustment information can be generated directly from the diagram intent, generated from the diagram intent and local verification information, generated from the diagram intent, local verification information and node capabilities and endpoint information, or generated from the state information.

[0061] Local verification information refers to the verification result that the agent, upon receiving adjustment information, determines whether a local execution state is executable based on local endpoint capabilities, resource usage, platform limitations, and operational status, and then reports the result back to the control unit. The verification result can include success, failure, failure category, unmet parameters, and a suggested range of feasible parameters. Based on the verification result, the control unit decides whether to proceed to the formal chain-building phase, whether to recalculate, or whether to formulate a degradation plan.

[0062] Therefore, the control end can determine whether to send connection establishment information based on the verification results. If the local verification information of all agents corresponding to the graph intent is successful, that is, all agents participating in building the target audio system have passed local verification, then connection establishment information can be sent to the agents. If the local verification information of any agent in the graph intent is unsuccessful, the control end can regenerate adjustment information based on the reason for the failure, and the agent will continue to perform local verification until the local verification is successful. For example, if the verification result is unsatisfactory parameters and the suggested feasible parameter range, then the control end will read the target constraints (such as target media conditions) defined in the current graph intent, read the actual capability range of the relevant endpoints, form a requirement to relax the target constraints under the premise of allowing degradation, and regenerate adjustment information.

[0063] The Status Summary and System Adjustment Unit is used to obtain the status information reported by the status reporting units of each agent. Status information refers to the information uploaded by each agent describing its own operational status after adjusting itself to the required local execution state according to the link establishment information issued by the control terminal. Status information can be any combination of the following: whether the link was successfully established, current operational quality, resource usage, and any abnormal alarms.

[0064] The introduction of state information enables operational status monitoring and dynamic adjustment. This allows the target audio system to continuously maintain its state even after construction. Each agent can continuously report node online status, endpoint occupancy status, link establishment status, link interruption status, current latency, packet loss, resource usage, anomaly type, and local recovery results. The control unit maintains an operational view of the entire distributed audio system based on this state information. When events such as node online / offline status, endpoint addition / deletion, changes in network path quality, changes in media conditions, graph intent version updates, or local link establishment failures or interruptions occur, the control unit can re-execute graph resolution, local execution state generation, agent verification, and link establishment processes based on the state information, thereby completing partial adjustments or overall system reconstruction. If necessary, only the affected connections can be recalculated to reduce the scope of reconstruction.

[0065] The Capability and Endpoint Exposure Unit is used to upload the corresponding node's capability and endpoint exposure unit to the control terminal.

[0066] The local verification unit is used to determine whether execution is possible based on the local endpoint capabilities, resource usage, platform limitations, and running status of the received adjustment information, and then feeds back the verification result to the control unit. Specific verifications can be freely combined from the following options: endpoint existence verification, resource availability verification, constraint satisfiability verification, version consistency verification, and execution condition integrity verification.

[0067] The link establishment and execution unit is used to execute its own local execution state based on the link establishment information, so as to establish a link with other agents indicated by the diagram intent.

[0068] The status reporting unit is used to continuously report status information to the control terminal.

[0069] Optionally, the control unit can also be configured with a verification and simulation support unit. This unit performs verification and simulation before sending link establishment information. The verification and simulation process may include reading the target graph intent, reading node capability descriptions, constructing a simulated topology environment, performing consistency checks on graph mapping results, performing compatibility checks on constraint matching results, rehearsing the link establishment process, performing injection tests for abnormal situations, and outputting verification results, failure locations, and suggested adjustments. By performing verification before deployment, on-site debugging costs can be reduced, the system's success rate can be improved, and the ability to reproduce problems can be enhanced.

[0070] See Figure 3 , Figure 3This illustrates the mapping from graphical intents to local execution states. The graphical intents are used to describe the target audio system at a global level, including at least nodes, endpoints, and media stream interaction relationships, with corresponding media conditions set for these relationships. The control unit parses the target system, mapping and generating different agents that produce corresponding local execution states. Each agent then performs local verification and execution based on its own local execution state, thus achieving the transformation from a global target to distributed local execution.

[0071] The schematic intent is not a simple list of devices or a channel routing table, but a structured object used to express the system-level target state and to describe the distributed target audio system that needs to be built.

[0072] Graph intents include a set of nodes, a set of endpoints, a set of connections and a set of media constraints, and graph version information.

[0073] The node set represents all nodes participating in the target audio system. Each node has a unique identifier, and its attributes include host information, platform type, running status, and functional role.

[0074] The endpoint set represents the endpoints provided by each node. The endpoints provided by each node include at least one of the following: input endpoint, output endpoint, acquisition endpoint, playback endpoint, processing endpoint, and virtual endpoint. Endpoints may include corresponding attribute information, such as direction attributes, number of channels attributes, supported format attributes, clock attributes, and buffering capability attributes.

[0075] The connection set represents the media stream interaction relationships between endpoints. Each media stream interaction describes at least the source endpoint, the destination endpoint, and the connection identifier. Since there are only media stream interaction relationships and no control interaction relationships between endpoints, the risk of chaos in complex topologies can be reduced, and system stability and subsequent maintainability can be improved.

[0076] The media constraint set represents the media conditions corresponding to each media stream interaction relationship. These media conditions include at least one of the following: sampling rate, bit depth, number of channels, buffer length, encoding format, clock requirements, maximum allowable latency, link priority, backup path strategy, packet loss tolerance strategy, and recovery strategy. Therefore, this invention does not handle media constraints implicitly, but rather incorporates them into the diagrammatic intent and link establishment process. By explicitly handling media constraints, this invention can adapt to the link establishment needs between heterogeneous nodes, avoiding compatibility issues that only surface after runtime.

[0077] Graph version information is used to distinguish the target state of the system at different stages to support graph changes, version switching, rollback, and state consistency verification during operation.

[0078] During the link establishment process, the system preferably handles media constraints in the following manner: read the target constraints defined in the diagram intent; read the actual capability range of the relevant endpoints; perform constraint matching; generate execution parameters when the constraints can be satisfied; return the failure reason when the constraints cannot be satisfied, or generate alternative execution parameters under the premise of allowing degradation. For any media stream interaction relationship, its establishment process may at least include determining the relevant nodes and endpoints, determining the final execution media parameters, starting the sending and receiving capabilities in the relevant agents, establishing control associations and state associations, and reporting the link establishment results to the control end.

[0079] Once the link is established, each agent begins executing its audio tasks according to its local execution state, such as acquisition, reception, processing, output, or forwarding. Depending on the different requirements of the schema intent, agents can establish unidirectional or bidirectional media stream interactions. The control plane does not need to directly participate in the processing of each frame of the audio data itself, but is responsible for maintaining system-level state and control relationships. This mechanism helps avoid the mixing of the control plane and data plane, improving the overall scalability of the system.

[0080] Through the above diagrammatic intent, the control end can directly address the question of how the entire target system should be structured, without having to perform piecemeal configurations on each node separately, thereby enabling the behavior of the agent to be driven by the system-level target.

[0081] It should be noted that, to improve executability, this invention does not require each agent to understand the entire graph intent. Instead, the control end maps the global graph to local execution states. The graph mapping mechanism for parsing the graph intent and mapping it to agents may include node affiliation identification, local relationship extraction, local constraint extraction, execution task generation, and state association. Using this graph mapping method, each agent only needs to process local content relevant to itself, reducing node implementation complexity and improving maintainability during system expansion.

[0082] The process described in this embodiment, based on the construction of a graph intent, will be used as an example. In this graph intent, the first agent 21 corresponds to node A, the second agent 22 corresponds to node B, and the third agent 23 corresponds to node C. Node A provides the audio acquisition endpoint, node B provides the intermediate processing endpoint, and node C provides the playback endpoint. The control terminal receives a graph intent requesting that the acquisition output of node A be processed by node B and then sent to node C for playback. The graph intent also specifies media conditions for two media stream interaction relationships: one between node A and node B, and the other between node B and node C. Both media stream interaction relationships are media stream interactions, and the corresponding media conditions—namely, the sampling rate, the number of channels, and the maximum latency requirements—are different.

[0083] First, the control terminal discovers node A, node B, and node C, and reads their node capabilities and endpoint information.

[0084] Next, the control terminal parses the diagram intent and identifies two media stream interaction relationships: one from node A to node B, and the other from node B to node C. Based on the node affiliation, the control terminal sends adjustment information to the corresponding agents to indicate the corresponding local execution status.

[0085] Then, the local execution state of the first agent 21 corresponding to node A includes starting the acquisition endpoint and sending data to node B; the local execution state of the second agent 22 corresponding to node B includes receiving the link from node A, starting the local processing link, and sending data to node C; the local execution state of the third agent 23 corresponding to node C includes receiving the link from node B and starting the playback endpoint. Each agent performs local verification upon receiving its local execution state. If the second agent 22 corresponding to node B finds that its local buffer capacity is insufficient to meet the original constraints, it feeds back the corresponding local verification information to the control terminal. The control terminal regenerates the adjustment information according to the allowable range in the diagram intent and sends it again to the second agent 22 corresponding to node B. After each agent passes the local verification, the control terminal sends the link establishment information, and each agent sequentially executes its local execution state, establishing the relevant links, and the distributed target audio system enters the running state.

[0086] If the link from node B to node C fails during the operation of the target audio system, the relevant agent reports status information to the control terminal to indicate the interruption status. The control terminal can regenerate the local execution status corresponding to the affected agent based on the graph version information of the current graph intent and the status of each node, and then issue it to restore the system to operation.

[0087] The above embodiments demonstrate that the present invention can uniformly express system objectives as graphical intents, and then form a distributed audio system with engineering practicality through distributed mapping, explicit constraint processing, local verification, and dynamic adjustment mechanisms.

[0088] In addition to the typical implementations described above, the present invention can also have the following variations: scalable node types, scalable sources of graph intents, scalable constraint handling strategies, scalable reconstruction scope, and scalable verification mechanisms. In summary, the present invention unifies the expression of the target system through graph intents, achieves system construction through node capability descriptions and local execution state generation mechanisms, realizes chain establishment and execution through explicit media constraints and local verification mechanisms, and achieves operation and maintenance through continuous monitoring and dynamic adjustment mechanisms, thereby forming a distributed audio system suitable for heterogeneous multi-node scenarios.

[0089] Example of a distributed audio method based on graph intent: See Figure 4This embodiment is implemented based on the execution of the computer program corresponding to the control terminal, and specifically includes the following steps: S11: Establish communication relationships with each agent.

[0090] S12: Obtain the corresponding node capabilities and endpoint information from each agent.

[0091] S13: Obtain the diagram intent.

[0092] S14: Send the corresponding first adjustment information to the corresponding agent according to the diagram intent.

[0093] S15: Determine whether each agent has passed local verification.

[0094] If the judgment result is yes, then continue to execute step S16; if the judgment result is no, then jump to step S19, and after executing step S19, return to step S15.

[0095] S16: Send chain establishment information to each agent.

[0096] S17: Obtain the status information uploaded by the proxy.

[0097] S18: Determine if there is any runtime abnormality.

[0098] If the judgment result is yes, then return to step S17; if the judgment result is no, then jump to step S20. After executing step S20, return to step S15.

[0099] S19: Send the second adjustment information to the agent.

[0100] S20: Send third adjustment information to the agent.

[0101] In step S11 above, the control terminal discovers each node that can participate in building the target audio system through a network discovery mechanism, a preset address mechanism, or a node active registration mechanism, and establishes a communication relationship with the agent corresponding to the node.

[0102] In step S12 above, the control terminal obtains the node capabilities and endpoint information corresponding to each agent, and the node capabilities and endpoint information can form a computable node resource view.

[0103] In step S13 above, the control terminal can generate a graphical intent corresponding to the target audio system to be constructed based on user input, system preset templates, external arrangement results, or historical configuration results.

[0104] In step S14 above, the control terminal parses the graph intent, identifies the node set, endpoint set, connection relationship set, media constraint set, and graph version information, thereby determining the nodes, endpoints, media stream interaction relationships involved in constructing the target audio system and their corresponding media conditions. Based on the capabilities of each node and the endpoint information, the graph intent is mapped to first adjustment information corresponding to each agent involved in constructing the target audio system and sent to the relevant agents. The first adjustment information indicates a first local execution state that includes at least the endpoints in which this node participates, the media stream interactions that this node needs to establish (such as input and output relationships), the media conditions corresponding to each media stream interaction relationship, and the local resources that this node needs to start, stop, or adjust.

[0105] In step S15 above, after the agent involved in the target audio system determines its corresponding local execution state, it determines whether the local execution state is executable based on the local endpoint capabilities, resource usage, platform limitations, and running status, and then feeds back the verification result to the control terminal.

[0106] In step S16 above, the link establishment information instructs each agent involved in the target audio system to execute its own local execution state to establish media stream interaction relationships. Furthermore, for some agents involved in the target audio system, the control terminal can determine the actual link establishment parameters corresponding to the media conditions based on the intersection of the capabilities of relevant endpoints and the media conditions in the schema intent. These actual link establishment parameters are then included in the link establishment information, thereby changing the local execution state of that agent. That agent then executes the changed local execution state based on the link establishment information. Thus, after the agents involved in building the target audio system execute their own local execution states, a runnable target audio system is obtained.

[0107] In step S17 above, each agent of the target audio system continuously (according to a preset period) reports status information to the control terminal. The status information includes whether the link has been successfully established, the current operating quality, resource usage, and abnormal alarms.

[0108] In step S18 above, "operational anomaly" means that the target audio system can no longer achieve its original design purpose, such as when a certain agent goes offline and affects the normal operation of the system. In this case, it is necessary to update the local execution state of the agent involved in the operation anomaly.

[0109] In step S19 above, the local execution state can be changed only for agents that fail local verification by regenerating the second adjustment information based on the diagram intent. Alternatively, the second adjustment information can be regenerated for all agents.

[0110] In step S20 above, the third adjustment information can be based on a new diagram intent. For example, if the current diagram intent needs a version update, it can be regenerated using the new version of the diagram intent. Alternatively, it can revert to the previous diagram intent version based on the diagram version information and send the third adjustment information based on the previous diagram intent version. It can also be generated after adjustments to the existing diagram intent, so that the third adjustment information only changes the local execution state of the agent body related to the runtime exception.

[0111] Therefore, this embodiment first completes node discovery and registration, then acquires capability and endpoint information, and generates or receives graph intents. Based on this, the control terminal generates the local execution states of each agent. After each agent completes local verification, it performs link establishment to obtain the target audio system. During the operation of the target audio system, continuous status reporting and operational monitoring are performed. When topology changes, constraint changes, node capability changes, or link anomalies are detected that cause operational abnormalities, the control terminal triggers dynamic adjustment or reconstruction, regenerating the local execution states through third-party adjustment information to maintain the continuous operation of the system. Through the above steps, an abstract target audio system is stably converted into concrete execution states on multiple nodes, supporting subsequent continuous operation and dynamic maintenance.

[0112] In summary, this invention changes the organization, expression, and execution methods of distributed audio systems, enabling multi-node audio systems to be built, linked, run, and adjusted driven by a unified goal. It solves problems in existing multi-node audio systems such as lack of unified expression, implicit media constraints, insufficient bidirectional collaboration mechanisms, coupling between control and execution, insufficient runtime reconfiguration capabilities, and difficulties in pre-deployment verification. This invention addresses problems unique to audio system construction. First, instead of focusing on single-device control logic, it uses a graphical intent to express the entire target audio system, including nodes, endpoints, connection relationships, and media constraints, solving the problem of unifying the expression of multi-node audio topology. Second, the control end of this invention generates the local execution state corresponding to each agent based on the graphical intent and the node capabilities and endpoint information of the agent, solving the problem of how to implement the global audio goal in a heterogeneous multi-node system. Third, this invention elevates media stream relationships and media constraints to system-level objects. By organizing media constraints uniformly within the graphical intent, executableness can be determined before linking, avoiding problems being exposed after execution, thereby improving the determinism of linking, system consistency, and the organizeability of complex audio paths. Fourth, this invention clearly distinguishes between control interaction and media stream interaction. Agents only have media stream interaction relationships corresponding to media conditions; agents do not issue control commands to each other. This avoids the propagation of control logic between nodes, reduces the risk of chaos in complex topologies, and improves system stability and subsequent maintainability. Fifth, graph intents or graph versions are updated as needed when states change. Updates are not automatically triggered when a new agent is added or when an agent goes offline, reducing unnecessary reconstruction overhead and improving the continuous operation capability of the audio system.

[0113] Finally, it should be emphasized that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for constructing a distributed audio system based on graph intents, characterized in that, Includes the following steps: Establish communication relationships with multiple agents and obtain the node capabilities and endpoint information corresponding to each agent; Obtain a graph intent, wherein the graph intent includes a node set, an endpoint set, a connection set, and a media constraint set. The node set is used to represent nodes participating in the construction of the target audio system. The endpoint set is used to represent the endpoints provided by each node. The connection set is used to represent the media stream interaction relationship between the endpoints. The media constraint set is used to represent the media conditions corresponding to each media stream interaction relationship. Based on the diagram intent and the node capabilities and endpoint information corresponding to each of the agents, a local execution state corresponding to each of the agents is generated, and first adjustment information indicating the local execution state is sent to each of the agents. Receive local verification information returned by each of the agents, wherein the local verification information is used to indicate whether the corresponding local execution state is executable; After confirming that each of the agents corresponding to the schematic intent has passed local verification, a link establishment information is sent to each agent. The link establishment information is used to instruct each agent to complete the link establishment based on its current local execution state and obtain the target audio system. Receive status information uploaded by each of the agents, and determine whether to update the diagram intent or diagram intent version based on the status information.

2. The method for constructing a distributed audio system based on graph intents as described in claim 1, characterized in that, Also includes: The first adjustment information is modified based on the local verification information to obtain the second adjustment information; The second adjustment information is sent to the agent.

3. The method for constructing a distributed audio system based on graph intents as described in claim 1, characterized in that: The schematic intent is derived based on the node capabilities and endpoint information corresponding to the agent.

4. The method for constructing a distributed audio system based on graph intents as described in claim 1, characterized in that: Establishing the communication relationship with the agent includes: The system discovers each node participating in the system through network discovery mechanisms, preset address mechanisms, or node active registration mechanisms, and establishes communication relationships with the agent corresponding to the node.

5. The method for constructing a distributed audio system based on graph intents as described in claim 2, characterized in that: The second adjustment information is sent to the corresponding agent based on the diagram intent, the local verification information, the node capabilities, and the endpoint information.

6. The method for constructing a distributed audio system based on graph intents as described in claim 1, characterized in that: The diagram intent also includes diagram version information.

7. The method for constructing a distributed audio system based on graph intents as described in claim 1, characterized in that: If it is determined from the status information that the diagram intent or diagram intent version needs to be updated, a third adjustment message is sent to the agent to adjust the target audio system according to the updated diagram intent or diagram intent version.

8. The method for constructing a distributed audio system based on schematic intent as described in any one of claims 1 to 7, characterized in that: The node includes a unique identifier, and the attributes corresponding to the node include host information, platform type, running status, and functional role; The endpoints provided by each node include at least one of the following: input endpoint, output endpoint, acquisition endpoint, playback endpoint, processing endpoint, and virtual endpoint; The media conditions include at least one of the following: sampling rate, bit depth, number of channels, buffer length, encoding format, clock requirements, maximum allowable latency, link priority, and fault tolerance strategy.

9. A distributed audio system based on graph intent, characterized in that, include: A control terminal and multiple agent entities, wherein the control terminal is connected to the agent entities; The control terminal is used to execute the distributed audio system construction method based on graph intent as described in any one of claims 1 to 8; The agent is used to upload the node capability and endpoint information, local verification information, and status information to the control terminal, and to establish a link between the agents based on the local execution status.

10. The distributed audio system based on schematic intent as described in claim 9, characterized in that: Before the agent establishes a link between the agents based on the local execution state, the control terminal performs pre-verification.

Images