Voice control method and system of earphone case, earphone case and storage medium

Abstract

The application provides a voice control method and system of an earphone case, the earphone case and a storage medium. The method comprises: controlling the earphone case to collect user voice data; performing local preprocessing on the user voice data by the earphone case, wherein the local preprocessing comprises at least one of voice activity detection, keyword trigger detection and intent pre-classification; selecting earphone case local processing or remote AI calling processing according to the classification result of the local preprocessing; if the remote AI calling processing is selected, initiating a call to an AI application program of a mobile terminal to trigger an AI request of the AI application program to process the user voice data; and receiving a processing result returned by the AI application program and feeding back the processing result to a user. The voice control method of the earphone case can preprocess and classify user voice data, flexibly select local response or cloud processing according to the classification result, and improve the flexibility and response efficiency of AI interaction.

Description

Technical Field

[0001] This invention relates to the field of AI interaction technology, and in particular to a voice control method, system, earphone case, and storage medium for an earphone case. Background Technology

[0002] With the rapid maturation of Large Language Models (LLM) and voice AI technologies, smart headphones are gradually evolving from "audio playback devices" to "AI interaction gateways." Currently, most mainstream headphone voice AI interaction solutions in the industry are based on a smartphone-centric architecture, using a mobile application (APP) as the primary AI invocation platform. In this existing solution, the headphones are only responsible for voice capture and playback; all AI requests are transmitted to the mobile application via a transmission channel, and then the mobile application calls cloud models for processing. Because all user AI requests need to be routed from the mobile application to the cloud for processing, the response time for AI requests is relatively slow, thus affecting the user experience. Therefore, existing technologies still need improvement and development. Summary of the Invention

[0003] This invention provides a voice control method, system, earphone compartment, and storage medium for an earphone compartment. The main objective of this invention is to solve the technical problems mentioned in the background section of the prior art.

[0004] The first aspect of this invention provides a voice control method for an earphone charging case, comprising: Control the earphone compartment to collect user voice data; The user's voice data is preprocessed locally through the earphone compartment. The local preprocessing includes at least one of voice activity detection, keyword trigger detection, and intent pre-classification. Based on the classification results of the local preprocessing, the earphone compartment can be processed locally or remotely by AI. If it is a remote AI call, then control the earphone case to initiate a call to the AI ​​application of the mobile terminal to trigger the AI ​​application to process the user's voice data. The earphone compartment receives the processing results returned by the AI ​​application and provides feedback to the user.

[0005] In some optional embodiments of the first aspect of the present invention, the local preprocessing of the user's voice data through the earphone compartment includes: Perform voice activity detection on the user voice data to determine whether there is valid user voice in the user voice data; If it is determined that the valid user voice exists in the user voice data, then keyword trigger detection is further performed on the user voice data to detect whether there is a preset wake word or preset instruction prefix in the user voice data; And / or perform intent pre-classification on the user voice data to determine the instruction type corresponding to the user voice data; if the instruction type corresponding to the user voice data is a local processing instruction, then output the classification result of the user voice data in the earphone compartment; if the instruction type corresponding to the user voice data is an AI query instruction, then output the classification result of the user voice data processed by the AI ​​application of the mobile terminal.

[0006] In some optional embodiments of the first aspect of the present invention, the earphone compartment and the mobile terminal communicate via a standardized remote protocol, the standardized remote protocol including one or more of the following: request identifier, serial number, intent type, data type, user voice data or its encoded data, context reference identifier, authentication information, and return result format information.

[0007] In some optional embodiments of the first aspect of the present invention, if the process involves remote AI invocation, controlling the earphone case to initiate an invocation to the AI ​​application of the mobile terminal to trigger the AI ​​application to process the user's voice data includes: The running status of the AI ​​application is detected, including whether it is active in the foreground, dormant in the background, not started, or the process has been killed by the system. Based on the running status of the AI ​​application, the earphone case is controlled to adopt a corresponding calling strategy, which includes requesting to send the AI ​​application directly, sending the AI ​​application after warm wake-up, sending the AI ​​application after cold wake-up, or sending the AI ​​application after restarting.

[0008] In some optional embodiments of the first aspect of the present invention, the voice control method further includes: The earphone compartment is configured with a context manager for storing dialogue context information related to user AI interaction; Based on the context manager and through a standardized context synchronization protocol, the earphone case and the AI ​​application are synchronized bidirectionally to achieve continuous dialogue across devices. The standardized context synchronization protocol includes: after each AI dialogue is completed, the AI ​​application synchronizes the complete context of this round to the earphone case; when the AI ​​application generates recommended content in the background, the AI ​​application actively pushes context updates to the earphone case; and when the earphone case reconnects to the mobile terminal, the earphone case automatically pulls the full context from the AI ​​application.

[0009] In some optional embodiments of the first aspect of the present invention, the voice control method further includes: A first dual-mode communication link is pre-established between the earphone compartment and the mobile terminal, and a second dual-mode communication link is established between the AI ​​application and the cloud AI model. The first dual-mode communication link includes a Bluetooth communication link and a Wi-Fi communication link, and the second dual-mode communication link includes an HTTP communication link and a WebSocket / SSE communication link. The Bluetooth communication link is used for short command issuance, the Wi-Fi communication link is used for file / voice transmission, the HTTP communication link is used for requesting uplink and responding downlink, and the WebSocket / SSE communication link is used for streaming response.

[0010] In some optional embodiments of the first aspect of the present invention, the voice control method further includes: A two-way keep-alive mechanism is established between the earphone case and the AI ​​application on the mobile terminal. The two-way keep-alive mechanism includes: periodically sending heartbeat detection requests to the AI ​​application through the earphone case and receiving responses from the AI ​​application; and guiding the user to add the AI ​​application to the system whitelist or wake up the AI ​​application using a manufacturer-level push activation channel when the earphone case and the mobile terminal are paired for the first time.

[0011] A second aspect of this invention provides a voice control system for an earphone charging case, the voice control system comprising: The voice receiving module is used to control the earphone compartment to collect user voice data; The voice preprocessing module is used to perform local preprocessing on the user's voice data through the earphone compartment. The local preprocessing includes at least one of voice activity detection, keyword trigger detection, and intent preclassification. The processing decision module is used to select whether to process the earphone compartment locally or remotely via AI based on the classification results of the local preprocessing. The remote call module is used to control the earphone case to initiate a call to the AI ​​application on the mobile terminal if the remote AI call is to be processed, so as to trigger the AI ​​application to process the user's voice data. The result feedback module is used to control the earphone compartment to receive the processing results returned by the AI ​​application and provide feedback to the user.

[0012] A third aspect of the present invention provides an earphone case, the earphone case comprising: a memory and at least one processor, the memory storing instructions, and the memory and the at least one processor being interconnected via a circuit; The at least one processor invokes the instructions in the memory to cause the earphone compartment to perform the voice control method for the earphone compartment as described in any one of the first aspects of the present invention.

[0013] A fourth aspect of the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the voice control method for an earphone compartment as described in any one of the first aspects of the present invention.

[0014] Beneficial Effects: This invention provides a voice control method, system, earphone compartment, and storage medium for an earphone charging case. The method includes controlling the earphone compartment to collect user voice data; performing local preprocessing on the user voice data through the earphone compartment, the local preprocessing including at least one of voice activity detection, keyword trigger detection, and intent pre-classification; selecting local processing or remote AI invocation processing based on the classification result of the local preprocessing; if remote AI invocation processing is selected, initiating a call to an AI application on a mobile terminal to trigger an AI request from the AI ​​application to process the user voice data; receiving the processing result returned by the AI ​​application and feeding it back to the user. This invention's voice control method for an earphone compartment allows the earphone compartment to preprocess and classify user voice data, flexibly selecting between local response and cloud processing based on the classification result, thus improving the flexibility and response efficiency of AI interaction. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of an embodiment of a voice control method for an earphone compartment according to the present invention; Figure 2 This is a schematic diagram of an embodiment of the voice control system for an earphone compartment according to the present invention; Figure 3 This is a schematic diagram of one embodiment of an earphone case according to the present invention. Detailed Implementation

[0016] The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of the embodiments of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0017] The first aspect of this invention provides a voice control method for an earphone charging case, which can be used in an AI interaction system including smart earphones, a mobile terminal, and cloud AI services. The architecture of the AI ​​interaction system mainly consists of five layers: a perception layer (earphone body) → an edge scheduling layer (earphone charging case) → a protocol adaptation layer (the mobile terminal's APP (i.e., the AI ​​application)) → a cloud capability layer (cloud AI model). The layers are connected through a standardized bidirectional communication channel. For example, the communication links used between the layers in this embodiment are shown in Table 1 below: Table 1. Communication Links Between Layers of the AI ​​Interaction System in Embodiments of the Present Invention

[0018] Based on Table 1 above, in an optional embodiment of the first aspect of the present invention, the voice control method further includes: pre-establishing a first dual-mode communication link between the earphone compartment and the mobile terminal, and establishing a second dual-mode communication link between the AI ​​application and the cloud AI model. The first dual-mode communication link includes a Bluetooth communication link and a Wi-Fi communication link, and the second dual-mode communication link includes an HTTP communication link and a WebSocket / SSE communication link. The Bluetooth communication link is used for short command issuance, the Wi-Fi communication link is used for file / voice transmission, the HTTP communication link is used for requesting uplink and responding downlink, and the WebSocket / SSE communication link is used for streaming response.

[0019] In this embodiment of the invention, the core processing node of the AI ​​interaction system is the earphone compartment. The earphone compartment contains a main control MCU (such as ARM Cortex-M4+ or higher) running an AI scheduling module. The earphone compartment establishes a bidirectional communication channel with the earphone itself and the mobile terminal (with a compatible app installed) through a communication module (supporting BLE 5.3 + Wi-Fi 6 dual-mode). A detailed description of each functional module of the AI ​​interaction system in this embodiment of the invention is shown in Table 2 below: Table 2. Introduction to the Functional Modules of the AI ​​Interaction System in the Embodiments of the Invention

[0020] See Figure 1 The voice control method for the earphone compartment includes: S100. Control the earphone compartment to collect user voice data. In this embodiment of the invention, the user voice data can be collected using the microphone array built into the earphone compartment or by collecting it through the microphone array in the earphone body and then sending it to the earphone compartment. If it is collected through the microphone array in the earphone body, after the microphone array in the earphone body collects the user voice data, it will send it to the earphone compartment through a pre-established bidirectional communication channel between itself and the earphone compartment.

[0021] S200: The user's voice data is preprocessed locally via the earphone compartment. This local preprocessing includes at least one of voice activity detection, keyword trigger detection, and intent pre-classification. In this step of this embodiment, the AI ​​scheduling module within the earphone compartment performs local preprocessing on the user's voice data sent from the earphone itself.

[0022] Specifically, in some optional embodiments of the first aspect of the present invention, the local preprocessing of the user's voice data through the earphone compartment includes: S201. Perform voice activity detection on the user voice data to determine whether there is valid user voice in the user voice data; specifically, the main purpose of voice activity detection (VAD) in this embodiment of the invention is to determine whether the user is actually speaking (distinguishing background noise), avoid invalid requests triggering subsequent AI application (APP) wake-up, reduce the power consumption and communication overhead of the earphone compartment. In this embodiment of the invention, the algorithm used for voice activity detection can be the WebRTCVAD algorithm or a similar lightweight model voice detection algorithm, and ensure that the computational cost is <5 MIPS.

[0023] S202. If it is determined that the valid user voice exists in the user voice data, then keyword trigger detection is further performed on the user voice data to detect whether there is a preset wake word or preset command prefix in the user voice data. In this embodiment of the invention, keyword trigger detection can identify the preset wake word or preset command prefix through the TinyNLP model (approximately 50KB) running locally in the earphone case. Only when a wake word (e.g., "Hey Siri") or a specific command prefix (e.g., "Query", "Ask me") is detected is it determined that there is a valid AI request in the user voice data. Similarly, in this embodiment of the invention, the subsequent AI application is only triggered when a match is found, avoiding the AI ​​application being woken up every time a voice is used, thus saving mobile phone battery power.

[0024] S203, and / or perform intent pre-classification on the user voice data to determine the instruction type corresponding to the user voice data; if the instruction type corresponding to the user voice data is a local processing instruction, then output the classification result of the user voice data being responded to locally in the earphone compartment; if the instruction type corresponding to the user voice data is an AI query instruction, then output the classification result of the user voice data being processed by the AI ​​application of the mobile terminal. Specifically, in this embodiment of the invention, the requests in the user voice data are divided into "simple instructions (i.e., local processing instructions)" (which can be processed locally, such as "play music") and "AI query instructions" (which require a cloud-based LLM model). For simple instructions, the complete AI application → cloud link is not followed, but the response is directly performed on the earphone compartment side.

[0025] S300. Based on the classification result of the local preprocessing, select either local processing in the earphone compartment or remote AI invocation processing. In this embodiment of the invention, after the intent classification in step S200, if the output is a classification result of processing the user voice data locally in the earphone compartment, then the user voice data is directly processed by the local lightweight model of the earphone compartment and the result is fed back. If the output is a classification result of processing the user voice data by the AI ​​application of the mobile terminal, then the user voice data needs to be constructed into an AI request based on a standardized remote invocation protocol and sent to the AI ​​application of the mobile terminal for processing.

[0026] S400. If it is a remote AI call processing, then control the earphone case to initiate a call to the AI ​​application of the mobile terminal to trigger the AI ​​application to process the user's voice data. Specifically, the core idea of ​​this embodiment is to use the earphone case as a trigger + scheduler for the AI ​​application. The technical innovation is to transform the earphone case from a passive charging device into an active AI request initiator and APP scheduler (APP Deferred Invocation mechanism). In order to standardize the call between the earphone case and the AI ​​application, in some optional embodiments of the first aspect of this embodiment, the earphone case and the mobile terminal communicate through a standardized remote protocol. The standardized remote protocol includes one or more of the following: request identifier, sequence number, intent type, data type, user voice data or its encoded data, context reference identifier, authentication information, and return result format information.

[0027] In some optional embodiments of the first aspect of the present invention, if the remote AI call processing is involved, controlling the earphone case to initiate a call to the AI ​​application of the mobile terminal to trigger the AI ​​application to process the user's voice data includes: detecting the running status of the AI ​​application, the running status including foreground active, background dormant, not started, or the process being killed by the system; and controlling the earphone case to adopt a corresponding calling strategy according to the running status of the AI ​​application, the calling strategy including requesting to send the request directly to the AI ​​application, warmly waking up the AI ​​application and then sending it, cold waking up the AI ​​application and then sending it, or restarting the AI ​​application and then sending it.

[0028] Specifically, to achieve efficient response, this embodiment of the invention employs a differentiated scheduling strategy based on the running status of the AI ​​application: For example, if the AI ​​application is active in the foreground: the earphone case directly sends an AI request via the standardized remote call protocol (ERCP), and the AI ​​application responds and processes it within 50ms, achieving real-time interaction; if the AI ​​application is dormant in the background: the earphone case first sends a wake-up command via the standardized remote call protocol, and the AI ​​application hot-starts from the background within approximately 200ms, restoring its processing capabilities; if the AI ​​application process is killed or cold-started: the earphone case first sends a cold wake-up command via the standardized remote call protocol, triggering the restart of the AI ​​application process, which takes approximately 2-5 seconds. An exemplary intelligent scheduling table for the AI ​​application in this embodiment of the invention is shown in Table 3 below: Table 3. Intelligent Scheduling Table of AI Application in Embodiments of the Invention

[0029] In this embodiment of the invention, the internal processing flow of the AI ​​application (i.e., APP) of the mobile terminal after receiving an AI request sent from the earphone case based on the standardized Remote Procedure Call Protocol (ERCP) can be as follows: (1) Processing based on request reception and security verification rules: The AI ​​application (i.e., APP) of the mobile terminal in this embodiment of the invention has a built-in daemon process. This process runs continuously in the background and is responsible for listening to and receiving ERCP data packets (i.e. AI requests) sent by the pairing earphone compartment via Bluetooth or Wi-Fi. When the ERCP packet is received, the APP does not immediately process the business logic, but first performs strict security verification to ensure the security and stability of the communication channel. The security verification includes: PSK signature verification (anti-forgery): The APP uses the pre-shared key (PSK) with the earphone compartment to digitally verify the authentication order in the ERCP packet. Only requests that pass the signature verification are considered to come from legitimate and trusted devices to prevent malicious forgery attacks by third-party devices; seq_id verification (anti-replay): The APP will verify and record the sequence number (seq_id) in the request packet. The APP will maintain a seq_id window of processed requests. If the received seq_id is a duplicate or expired value, the request packet will be discarded directly to prevent network delays or malicious attackers from intercepting and replaying historical requests.

[0030] (2) Processing based on context management rules: After passing the security check, the APP uses the context reference identifier (ctx_ref) in the ERCP packet to process the continuity of the dialogue and query the historical context: The APP uses ctx_ref as an index to query the corresponding historical dialogue context in the local database or cache; Validity judgment: If ctx_ref is valid and the corresponding context is found, it indicates that the current request is a continuation of a multi-turn dialogue, and the APP will load the context so that the AI ​​model can understand the context in the future; If ctx_ref is empty or invalid (cannot be found), the APP regards this request as the beginning of a new dialogue and builds a brand new dialogue session.

[0031] (3) Processing based on intent routing and processing rules: After completing context preparation, the APP enters the intent routing stage. The APP parses the intent field in the ERCP packet and distributes the request to different processing modules according to its defined type. In this embodiment of the invention, the ERCP packet includes at least three intent types: Intent type 1: ai.query (AI query): This indicates that the headphone compartment has initiated a complete AI query request. The APP will start a complete AI processing flow. This flow first determines the type of input data. If the data type is VOICE_PCM (voice stream), ASR speech recognition is enabled. The cloud or local automatic speech recognition (ASR) engine is called to convert the original PCM audio stream into a text string. Then, the text recognized by ASR and the dialogue context loaded in the aforementioned step (2) are submitted to the cloud large language model (LLM). After understanding the complete context, the LLM generates the corresponding text response. Then, the text-to-speech (TTS) engine is called to synthesize the text response generated by the LLM into natural and fluent speech data (such as MP3 or Opus format). If the data type is TEXT (text), the text input and dialogue context will be directly submitted to the LLM to generate a text response. In some other optional embodiments, the generated text response can also be selectively synthesized into TTS speech according to the request parameters or application settings.

[0032] Intent type 2: ai.context_sync (context synchronization): This means that the headphone charging case only requests or pushes context information. The APP will perform a lightweight context synchronization operation, such as synchronizing the latest context generated by the APP (such as an operation performed by the user on the phone) to the headphone charging case, or responding to the headphone charging case's context retrieval request. This process does not involve complex AI calculations.

[0033] Intent Type 3: ai.completion (Cloud Verification / Optimization): This indicates that the data after the lightweight model on the earphone case has completed preliminary processing (preprocessing) needs to be verified, completed, or optimized by more powerful AI capabilities in the cloud. After receiving this request, the APP will send the relevant data to a specific model in the cloud for processing and return the optimized result.

[0034] (4) Processing based on result encapsulation and feedback rules: After any of the above intentions are processed, the output results (such as synthesized speech, LLM-generated text, or synchronization status confirmation) will be uniformly encapsulated according to the ERCP protocol format. Finally, the APP selects the most efficient method to push the encapsulated ERCP result packet back to the headphone compartment based on the current network connection status with the headphone compartment. For example, BLE Notify is used in scenarios requiring low latency and small data volume, while Wi-Fi WebSocket channel is used when large data (such as high-quality audio) needs to be transmitted. After the process is completed, the APP continues to wait for the next ERCP request from the headphone compartment.

[0035] S500: Control the earphone case to receive the processing result returned by the AI ​​application and feed it back to the user. In this embodiment of the invention, the processing result can be fed back to the user through the display screen on the earphone case (which can be a 0.8~1.5-inch touch screen), through the speaker on the earphone case, or by first sending it to the earphone itself and then having the earphone itself play it to the user. In this embodiment of the invention, after the AI ​​application receives the AI ​​request (non-contextual synchronization intent), it converts the request into a cloud API call through the protocol adaptation layer, interacts with the cloud Large Language Model (LLM), and feeds back the processing result to the earphone case through a standardized remote procedure call protocol, which is then displayed to the user by the earphone case. Taking display screen interaction as an example, an exemplary display screen interaction design for an earphone case in this embodiment of the invention is shown in Table 4 below: Table 4. Display Screen Interaction Design of the Earphone Case in Embodiments of the Invention

[0036] In some optional embodiments of the first aspect of the present invention, the AI ​​invocation method further includes: configuring a context manager in the earphone case for storing dialogue context information related to user AI interaction; and, based on the context manager and through a standardized context synchronization protocol, bidirectionally synchronizing the dialogue context information between the earphone case and the AI ​​application to achieve continuous dialogue across devices. The standardized context synchronization protocol includes: after each AI dialogue is completed, the AI ​​application synchronizes the complete context of this round to the earphone case; when the AI ​​application generates recommended content in the background, the AI ​​application actively pushes context updates to the earphone case; and when the earphone case reconnects to the mobile terminal, the earphone case automatically retrieves the full context from the AI ​​application.

[0037] Specifically, in traditional solutions, the AI ​​dialogue context is stored locally on the mobile app, and the earphone charging case is unaware of it, causing each dialogue to start "from scratch." To solve the problem of context breakage in multi-turn dialogues, this embodiment of the invention establishes a distributed context manager on the earphone charging case side to achieve continuous dialogue across devices. Context synchronization: A standardized context synchronization protocol is used between the earphone charging case and the app. After each AI dialogue, the app synchronizes the complete context (including a summary) of this turn to the local cache of the earphone charging case via BLENotify or Wi-Fi WebSocket. Reconnection mechanism: When the earphone charging case disconnects and reconnects with the app, the earphone charging case automatically initiates a context retrieval request to synchronize the full context data from the app. An exemplary context synchronization protocol table of this embodiment of the invention is shown in Table 5 below: Table 5. Context Synchronization Protocol Table for the Earphone Case and AI Application (APP) in Embodiments of the Invention

[0038] In some optional embodiments of the first aspect of the present invention, the AI ​​invocation method further includes: establishing a two-way keep-alive mechanism between the earphone case and the AI ​​application on the mobile terminal, the two-way keep-alive mechanism including: periodically sending heartbeat detection requests to the AI ​​application through the earphone case and receiving a response from the AI ​​application; and when the earphone case and the mobile terminal are paired for the first time, guiding the user to add the AI ​​application to the system whitelist or wake up the AI ​​application using a manufacturer-level push activation channel.

[0039] Specifically, in this embodiment of the invention, to prevent the earphone case from being unable to use AI due to the system killing the APP in the background, this embodiment of the invention designs a two-way keep-alive mechanism between the earphone case and the APP: Heartbeat detection: The earphone case sends an ERCP Heartbeat request to the APP every 60 seconds, and the APP responds within 500ms. If there is no response for 3 consecutive times, the earphone case determines that the APP is unavailable and automatically switches to "offline mode" (only supporting simple local commands); System whitelist guidance: When the earphone case is paired for the first time, the APP guides the user to add the APP to the "high-priority background application" list in the phone's system settings to prevent the system from automatically killing the background application; Push activation: For Android devices, a manufacturer-grade push activation channel is used to ensure that the APP can be reawakened by the system even if it is cleared.

[0040] In summary, the purpose of this invention is to construct an AI invocation and control system with the earphone charging case as the core node. This differs from existing AI invocation methods centered on mobile terminals, forming an AI request initiation and scheduling mechanism led by the earphone charging case. Specifically, it includes: First, by setting up an AI scheduling module in the earphone case, the earphone case is equipped with the ability to receive, preprocess, and recognize user voice data. Based on the recognition results, AI requests are classified and processed to realize a hierarchical decision-making mechanism of local execution and cloud call, thereby reducing unnecessary external calls and improving system response efficiency.

[0041] Secondly, a mechanism for the earphone case to actively invoke mobile terminal applications is proposed. By defining a standardized remote invocation protocol, the earphone case can wake up the APP on demand, transmit requests, and return results. This transforms the APP from a fixed running entity into a callable service unit, thereby realizing the modularization and scalable access of AI capabilities.

[0042] Furthermore, this embodiment of the invention uses a multi-communication link collaboration mechanism to dynamically select communication methods such as BLE or Wi-Fi between the earphone, earphone case, and mobile terminal to adapt to different data types and transmission requirements, thereby improving communication efficiency and system stability.

[0043] In addition, by establishing a dialogue context caching mechanism on the earphone case side, local storage and cross-device synchronization of multi-turn dialogue information can be achieved, enabling AI interaction to have continuity and contextual association capabilities.

[0044] Compared with the prior art, the embodiments of the present invention have at least the following core innovations: Innovation Point 1: AI Scheduling Hub Architecture for the Earphone Case: This invention redefines the smart earphone case from a "charging and storage device" to an "AI edge scheduling node," making it the central hub of the three-layer architecture of earphones-APP-cloud. This is the first time in the industry that complete AI request management capabilities have been achieved on the earphone case side, breaking the existing paradigm that earphone AI interaction must be centered on the mobile phone.

[0045] Innovation Point Two: Standardized Remote Procedure Call Protocol (ERCP) for Earphone Case to App: The ERCP protocol defined in this embodiment of the invention specifies a standardized bidirectional communication interface between the earphone case and the app, enabling the app to be woken up on demand, routed according to intent, and scheduled according to priority by the earphone case, achieving the effect of "earphone case-driven, plug-and-play AI capabilities". The ERCP protocol shields the underlying hardware differences, allowing different apps (corresponding to different AI service providers) to seamlessly access this system.

[0046] Innovation Point 3: Local Preprocessing and Intelligent Traffic Diversion Mechanism on the Earphone Case Side: This embodiment of the invention introduces a local preprocessing pipeline of VAD+TinyNLP on the earphone case side, realizing intelligent traffic diversion of "simple local commands do not wake up the APP" and "only cloud AI is required to trigger APP wake-up", which balances response speed and AI capability richness, while significantly reducing APP wake-up frequency and mobile phone power consumption.

[0047] Innovation Point 4: Distributed Cross-Device Multi-Turn Dialogue Context Management (DCM): This embodiment of the invention establishes a distributed context manager on the earphone case side. Through a standardized context synchronization protocol, the dialogue context that originally only existed in the mobile APP is extended to the earphone case side, realizing bidirectional context synchronization between the earphone case and the APP, and supporting users to continue the same round of dialogue on different devices.

[0048] Innovation Point 5: AI Result Preview and Instant Playback Interaction of the Case Display Unit: This invention provides a complete AI interactive display interface for the earphone case. Users can preview AI response summaries and click to play the complete voice reply on the earphone case screen without taking out their phone or wearing earphones, truly achieving "blind operation AI interaction".

[0049] See Figure 2 A second aspect of the present invention provides a voice control system for an earphone charging case, the voice control system for the earphone charging case comprising: The voice receiving module 10 is used to control the earphone compartment to collect user voice data; The voice preprocessing module 20 is used to perform local preprocessing on the user's voice data through the earphone compartment. The local preprocessing includes at least one of voice activity detection, keyword trigger detection, and intent preclassification. The processing decision module 30 is used to select whether the headphone compartment is processed locally or remotely by AI based on the classification results of the local preprocessing. The remote call module 40 is used to control the earphone compartment to initiate a call to the AI ​​application on the mobile terminal if it is the remote AI call processing, so as to trigger the AI ​​application to process the AI ​​request of the user's voice data.

[0050] The result feedback module 50 is used to control the earphone compartment to receive the processing results returned by the AI ​​application and provide feedback to the user.

[0051] In some optional embodiments of the second aspect of the present invention, the speech preprocessing module includes: A voice activity detection unit is used to perform voice activity detection on the user voice data and determine whether there is valid user voice in the user voice data. The keyword detection unit is used to further perform keyword trigger detection on the user voice data if it is determined that there is valid user voice in the user voice data, and to detect whether there is a preset wake word or preset instruction prefix in the user voice data.

[0052] In some optional embodiments of the second aspect of the present invention, the speech preprocessing module further includes: An intent classification unit is used to pre-classify the intent of the user voice data and determine the instruction type corresponding to the user voice data. The classification result output unit is configured to output a classification result of the user voice data in the earphone compartment if the instruction type corresponding to the user voice data is a local processing instruction; and to output a classification result of the user voice data processed by the AI ​​application of the mobile terminal if the instruction type corresponding to the user voice data is an AI query instruction.

[0053] In some optional embodiments of the second aspect of the present invention, the remote invocation module includes: The program status detection unit is used to detect the running status of the AI ​​application, which includes foreground active, background dormant, not started, or the process being killed by the system. The strategy matching unit is used to control the earphone case to adopt a corresponding calling strategy according to the running state of the AI ​​application. The calling strategy includes requesting to send the AI ​​application directly, sending the AI ​​application after warm wake-up, sending the AI ​​application after cold wake-up, or sending the AI ​​application after restarting.

[0054] In some optional embodiments of the second aspect of the present invention, the voice control system further includes: The context management module is configured to configure a context manager in the earphone compartment for storing dialogue context information related to user AI interaction; and to perform bidirectional synchronization of the dialogue context information between the earphone compartment and the AI ​​application based on the context manager and through a standardized context synchronization protocol, so as to achieve continuous dialogue across devices.

[0055] The standardized context synchronization protocol includes: after each AI dialogue is completed, the AI ​​application synchronizes the complete context of this round to the earphone case; when the AI ​​application generates recommended content in the background, the AI ​​application actively pushes context updates to the earphone case; and when the earphone case reconnects to the mobile terminal, the earphone case automatically pulls the full context from the AI ​​application.

[0056] In some optional embodiments of the second aspect of the present invention, the voice control system further includes: a communication establishment module, configured to pre-establish a first dual-mode communication link between the earphone compartment and the mobile terminal, and to establish a second dual-mode communication link between the AI ​​application and the cloud AI model. The first dual-mode communication link includes a Bluetooth communication link and a Wi-Fi communication link, and the second dual-mode communication link includes an HTTP communication link and a WebSocket / SSE communication link. The Bluetooth communication link is used for short command issuance, the Wi-Fi communication link is used for file / voice transmission, the HTTP communication link is used for requesting uplink and responding to downlink, and the WebSocket / SSE communication link is used for streaming response.

[0057] In some optional embodiments of the second aspect of the present invention, the voice control system further includes: The keep-alive module is used to establish a two-way keep-alive mechanism between the earphone case and the AI ​​application on the mobile terminal. The two-way keep-alive mechanism includes: periodically sending heartbeat detection requests to the AI ​​application through the earphone case and receiving responses from the AI ​​application; and guiding the user to add the AI ​​application to the system whitelist or wake up the AI ​​application using a manufacturer-level push activation channel when the earphone case and the mobile terminal are paired for the first time.

[0058] Figure 3 This is a schematic diagram of an earphone charging case provided in an embodiment of the present invention. The earphone charging case can vary significantly depending on its configuration or performance, and may include one or more processors 60 (central processing units, CPUs) (e.g., one or more processors) and memory 70, and one or more storage media 80 (e.g., one or more mass storage devices) for storing applications or data. The memory and storage media can be temporary or persistent storage. The program stored in the storage media may include one or more modules (not shown in the diagram), each module including a series of instruction operations on the earphone charging case. Furthermore, the processor may be configured to communicate with the storage media and execute the series of instruction operations stored in the storage media on the earphone charging case.

[0059] The headphone case in this embodiment of the invention may further include one or more power supplies 90, one or more wired or wireless network interfaces 100, one or more input / output interfaces 110, and / or one or more operating systems, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art will understand that... Figure 3 The earphone case structure shown does not constitute a limitation on the earphone case, and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0060] This invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. The computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the steps of the voice control method for the headphone compartment.

[0061] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system or system / unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0062] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiments of the present invention, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0063] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention 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 the present invention.

Claims

1. A voice control method for an earphone charging case, characterized in that, include: Control the earphone compartment to collect user voice data; The user's voice data is preprocessed locally through the earphone compartment. The local preprocessing includes at least one of voice activity detection, keyword trigger detection, and intent pre-classification. Based on the classification results of the local preprocessing, the earphone compartment can be processed locally or remotely by AI. If it is a remote AI call, then control the earphone case to initiate a call to the AI ​​application on the mobile terminal to trigger the AI ​​application to process the user's voice data. The earphone compartment receives the processing results returned by the AI ​​application and provides feedback to the user.

2. The voice control method for the earphone compartment according to claim 1, characterized in that, The local preprocessing of the user's voice data via the earphone compartment includes: Perform voice activity detection on the user voice data to determine whether there is valid user voice in the user voice data; If it is determined that the valid user voice exists in the user voice data, then keyword trigger detection is further performed on the user voice data to detect whether there is a preset wake word or preset instruction prefix in the user voice data; And / or perform intent pre-classification on the user voice data to determine the instruction type corresponding to the user voice data; if the instruction type corresponding to the user voice data is a local processing instruction, then output the classification result of the user voice data in the earphone compartment; if the instruction type corresponding to the user voice data is an AI query instruction, then output the classification result of the user voice data processed by the AI ​​application of the mobile terminal.

3. The voice control method for the earphone compartment according to claim 2, characterized in that, The earphone compartment and the mobile terminal communicate via a standardized remote protocol, which includes one or more of the following: request identifier, serial number, intent type, data type, user voice data or its encoded data, context reference identifier, authentication information, and return result format information.

4. The voice control method for the earphone compartment according to claim 1, characterized in that, If the process involves a remote AI call, then controlling the earphone case to initiate a call to the AI ​​application on the mobile terminal to trigger the AI ​​application to process the user's voice data includes: The running status of the AI ​​application is detected, including whether it is active in the foreground, dormant in the background, not started, or the process has been killed by the system. Based on the running status of the AI ​​application, the earphone case is controlled to adopt a corresponding calling strategy, which includes requesting to send the AI ​​application directly, sending the AI ​​application after warm wake-up, sending the AI ​​application after cold wake-up, or sending the AI ​​application after restarting.

5. The voice control method for the earphone compartment according to claim 1, characterized in that, The voice control method further includes: The earphone compartment is configured with a context manager for storing dialogue context information related to user AI interaction; Based on the context manager and through a standardized context synchronization protocol, the earphone case and the AI ​​application are synchronized bidirectionally to achieve continuous dialogue across devices. The standardized context synchronization protocol includes: after each AI dialogue is completed, the AI ​​application synchronizes the complete context of this round to the earphone case; when the AI ​​application generates recommended content in the background, the AI ​​application actively pushes context updates to the earphone case; and when the earphone case reconnects to the mobile terminal, the earphone case automatically pulls the full context from the AI ​​application.

6. The voice control method for the earphone compartment according to claim 1, characterized in that, The voice control method further includes: A first dual-mode communication link is pre-established between the earphone compartment and the mobile terminal, and a second dual-mode communication link is established between the AI ​​application and the cloud AI model. The first dual-mode communication link includes a Bluetooth communication link and a Wi-Fi communication link, and the second dual-mode communication link includes an HTTP communication link and a WebSocket / SSE communication link. The Bluetooth communication link is used for short command issuance, the Wi-Fi communication link is used for file / voice transmission, the HTTP communication link is used for requesting uplink and responding downlink, and the WebSocket / SSE communication link is used for streaming response.

7. The voice control method for the earphone compartment according to claim 1, characterized in that, The voice control method further includes: A two-way keep-alive mechanism is established between the earphone case and the AI ​​application on the mobile terminal. The two-way keep-alive mechanism includes: periodically sending heartbeat detection requests to the AI ​​application through the earphone case and receiving responses from the AI ​​application; and guiding the user to add the AI ​​application to the system whitelist or wake up the AI ​​application using a manufacturer-level push activation channel when the earphone case and the mobile terminal are paired for the first time.

8. A voice control system for an earphone charging case, characterized in that, The voice control system of the earphone compartment includes: The voice receiving module is used to control the earphone compartment to collect user voice data; The voice preprocessing module is used to perform local preprocessing on the user's voice data through the earphone compartment. The local preprocessing includes at least one of voice activity detection, keyword trigger detection, and intent preclassification. The processing decision module is used to select whether to process the earphone compartment locally or remotely via AI based on the classification results of the local preprocessing. The remote call module is used to control the earphone case to initiate a call to the AI ​​application on the mobile terminal if the remote AI call is to be processed, so as to trigger the AI ​​application to process the user's voice data. The result feedback module is used to control the earphone compartment to receive the processing results returned by the AI ​​application and provide feedback to the user.

9. An earphone case, characterized in that, The earphone compartment includes: a memory and at least one processor, wherein the memory stores instructions, and the memory and the at least one processor are interconnected via a circuit; The at least one processor invokes the instructions in the memory to cause the earphone compartment to perform the voice control method of the earphone compartment as described in any one of claims 1-7.

10. A computer-readable storage medium storing a computer program thereon, characterized in that, When the computer program is executed by the processor, it implements the voice control method for the headphone compartment as described in any one of claims 1-7.

Images