A double-microphone beamforming anti-z-shaped ear clip structure

Abstract

The application belongs to the technical field of beam forming, and particularly relates to a double-microphone beam forming anti-Z-shaped ear clip structure which is composed of the following modules: an adaptive structure support module, a double-microphone sound pickup and beam forming module, a multi-modal sensing module, a core control module, a power supply and energy management module, and a sealing and protection module. The fixed form structure is adopted to adapt to ear types by relying on a preset radian, and the special ear type has poor adaptability. The scheme realizes dynamic adjustment of the ear clip angle and the microphone angle by adopting a three-section anti-Z-shaped framework+15 DEG elastic hinge+0.1 N pressure sensor, so that the adaptability is improved, the wearing pressure can be controlled in real time to be less than or equal to 5 N, and compression pain is avoided. Meanwhile, the rotating base can fine-tune the microphone sound pickup angle according to the user's head posture, solves the problem that the head rotation leads to sound pickup deviation in the prior art, and improves the voice intelligibility.

Description

Technical Field

[0001] This invention belongs to the field of beamforming technology, specifically a dual-microphone beamforming inverted Z-shaped ear clip structure. Background Technology

[0002] With the rapid development of consumer electronics and smart wearable technology, clip-on audio devices are increasingly widely used in scenarios such as call noise reduction, sports audio, and health monitoring due to their characteristics of "no pressure in the ear and convenient wearing". Among them, dual-microphone beamforming technology, as a key means of directional sound pickup, can suppress environmental noise through signal phase difference and has become a standard technology in mid-to-high-end clip-on products.

[0003] Currently, most ear clip-on devices on the market adopt a fixed-shape structure + basic beamforming technology: structurally, they are mostly single-segment or two-segment fixed frames, mainly made of medical-grade titanium alloy or shape memory alloy, and adapt to the ear shape of most adults through preset curvature; at the same time, users' demand for devices that "adapt to different ear shapes and integrate health monitoring" is driving the industry to upgrade from "single audio function" to "multimodal smart wearables".

[0004] In the current environment, clip-on devices still have shortcomings. Among them, the problems of narrow ear shape adaptation and weak scene sound pickup adaptation still exist, affecting daily use. In addition, the weak scene sound pickup adaptation, the beamforming mode needs to be manually switched, the wind noise suppression in motion scenes is insufficient, and the sound pickup angle is easily offset when the head is turned during stationary calls.

[0005] Therefore, the present invention provides a reverse Z-shaped ear clip structure with dual-microphone beamforming. Summary of the Invention

[0006] In order to overcome the shortcomings of the prior art, at least one technical problem raised in the background art is solved.

[0007] The technical solution adopted by this invention to solve its technical problem is: the dual-microphone beamforming inverted Z-shaped ear clip structure of this invention is composed of the following modules: Adaptive structural support module, dual-microphone pickup and beamforming module, multimodal sensing module, core control module, power supply and energy management module, sealing and protection module; The adaptive structural support module is implemented in three steps: Physical support and shape adaptation: It adopts a three-section reverse Z-shaped titanium alloy frame and achieves a close fit to the ear through elastic hinges; Real-time monitoring of wearing pressure: A pressure sensor with a sensitivity of 0.1N is embedded in the bottom of the microphone rotating base to detect the contact pressure between the base and the auricle in real time; Microphone directional assistance: The rotating base allows for fine-tuning of the microphone's pickup angle; The dual-microphone pickup and beamforming module is implemented in three steps: High-quality voice signal acquisition: Employs a high-sensitivity MEMS microphone that covers the full frequency range of human speech; the microphone surface is fitted with a wind noise reduction mesh. Scenario-based beamforming processing: Integrates a dedicated beamforming algorithm chip, supporting three core modes: 1. Narrow beam; 2. Wide beam; 3. Wind noise resistant wide beam; Real-time signal output and feedback: Transmits the processed clear voice signal to the core control module, and simultaneously receives the "mode switching command" from the core control module; The multimodal sensing module is implemented in three steps: Scene state recognition: Equipped with a three-axis accelerometer, it identifies scenes by analyzing the magnitude of acceleration changes; 1. Stationary; 2. Moving; 3. Multi-person scenes; Health data monitoring: Integrating a miniature PPG sensor, it monitors heart rate and blood oxygen saturation in real time by detecting changes in capillary blood flow under the skin of the concha. Abnormal status warning: When a heart rate > 120 bpm or blood oxygen < 93% is detected, an "abnormal warning signal" is immediately sent to the core control module to trigger subsequent reminder functions; The core control module is implemented in four steps: Data reception and parsing: Data from each module is received in parallel via multiple interfaces; 1. Raw / processed voice signals are received from the dual-microphone module; 2. Acceleration / PPG data are received from the multimodal module; 3. Pressure data is received from the adaptive structure module; Decision instruction generation: Instructions are generated based on the "scenario-parameter" linkage logic; Multi-module collaborative scheduling: through time-slice scheduling algorithms; System iteration and maintenance: Integrated OTA upgrade module, enabling remote updates of beamforming algorithms and health data analysis models; The power supply and energy management module is implemented in three steps: Multi-voltage power supply: Two core voltages are output through a voltage divider circuit; 1. 3.3V: supplies the core control module and multimodal sensing module; 2. 5V: supplies the dual microphone module. Charging and power management: Equipped with a charging management chip; integrates a power detection circuit to monitor the remaining power in real time; Energy consumption optimization control: "On-demand power supply" is achieved through low-power logic circuits; 1. Static scenario: Microphone wind noise reduction module is turned off; 2. Sleep scenario: Only the core control module retains the lowest power consumption. The sealing and protection module is implemented in three steps: Waterproof and sweat- and corrosion-resistant: The module interface uses a nitrile rubber sealing ring, and the outer shell is sealed by laser welding; Wear and impact resistance: The surface of the titanium alloy frame is nitrided; the core control module is surrounded by a 0.5mm thick silicone buffer layer; Dustproof and biocompatibility: The outer shell gaps are made of dustproof mesh (0.02mm aperture); the silicone pads that come into contact with the skin are made of medical-grade silicone rubber; Preferably, the adaptive structure support module, dual-microphone pickup and beamforming module, multimodal sensing module, core control module, power supply and energy management module, and sealing and protection module form a closed-loop system with the core control module as the central hub through "hardware physical connection + software data interaction". The connection relationship can be divided into three categories: power supply link, data transmission link, and control command link. Each type of link has a clear interface standard, transmission content, and collaborative logic. The power supply and energy management module is connected to each module via miniature copper foil wires, and provides an adaptive voltage according to the power consumption requirements of each module. The dual-microphone pickup and beamforming module transmits the original voice signal and beamformed audio data to the core control module through the I2S audio interface. The multimodal sensing module transmits preprocessed acceleration data, PPG heart rate / blood oxygen data, and abnormal warning signals to the core control module via the I2C communication interface. The adaptive structure support module transmits the following data to the core control module via the PWM control interface: Forward: real-time data from the pressure sensor; Reverse: "pressure threshold setting" data from the core control module. The core control module transmits the following to the dual-microphone pickup and beamforming module via the SPI control interface: forward: beam mode switching command, filter parameter adjustment command; reverse: module execution result feedback. The core control module transmits the following commands to the adaptive structural support module via a UART serial interface: forward: rotation base angle adjustment command, hinge elastic preload adjustment command; reverse: structural adjustment result feedback. The core control module transmits low-power mode switching commands and module power-off commands to the power supply and energy management module through the GPIO general interface.

[0008] Preferably, it includes the following steps: S1, Adaptive anti-Z-shaped frame and pressure closed-loop regulation: It adopts a three-section inverted Z-shaped titanium alloy frame, with each section connected by ±15° elastic hinges; a 0.1N sensitivity pressure sensor is embedded in the bottom of the microphone rotating base; S2, Dual-microphone scene-specific beamforming and wind noise suppression: It uses a MEMS microphone with a frequency response of 20Hz-20kHz and a signal-to-noise ratio of ≥62dB, and is equipped with a 0.1mm aperture wind noise reduction mesh cover; it integrates a dedicated beamforming algorithm chip; it automatically activates a 40°-50° narrow beam when making calls while stationary to enhance the voice directly in front; it switches to a 70°-80° wide beam to expand the sound pickup range during multi-person conferences; and it triggers a wind noise reduction wide beam in motion scenarios. S3, Multimodal Scene Recognition and Health Anomaly Early Warning: A triaxial accelerometer with a range of ±8g and a sampling rate of 50Hz is integrated in the lower part of the ear clip. The scene is identified by acceleration fluctuations: fluctuations ≤0.5g indicate stationary conditions, 0.5g-3g indicate movement, and irregular fluctuations indicate multi-person scenes. Simultaneously, a PPG sensor with a sampling rate of 100Hz and an accuracy of ±2bpm is embedded to monitor blood flow in the concha and acquire heart rate and blood oxygen data. The data is then transmitted after filtering and baseline drift removal preprocessing. S4. Core module multi-interface collaborative scheduling and OTA upgrade: It employs an ARM Cortex-M55 processor and receives microphone audio, multimodal sensing, and pressure data in parallel via I2S, I2C, and PWM interfaces, with data cached in 16KBSRAM to prevent data loss. Instructions are generated based on "scene-parameter" logic, and time-slice scheduling is used to avoid conflicts. S5. Multi-voltage power supply and dynamic energy consumption control: It outputs 3.3V and 5V voltages via a DC-DC voltage divider circuit, is equipped with a 5V / 1A fast charging chip, disables the microphone and wind noise reduction module in static scenes, and reduces the PPG sampling rate to 30Hz; it enters sleep mode after 30 minutes of inactivity, retaining only the core module 10. Minimum power consumption, other modules powered off; S6. Multi-layer sealing protection and biocompatibility treatment: The module interface uses a nitrile rubber sealing ring with a compression of 0.2mm, and the outer shell is laser welded; the titanium alloy skeleton is nitrided; the core module is wrapped with a 0.5mm silicone buffer layer; the gaps in the outer shell are fitted with a 0.02mm diameter dustproof mesh, and the skin contact parts are made of FDA-certified medical silicone rubber.

[0009] Preferably, the adaptive anti-Z-shaped frame and the pressure closed-loop adjustment are configured using the following formula: The adaptive pressure and angle closed-loop control algorithm is calculated according to the following formula: in, Microphone base angle adjustment; Pressure deviation; , , 0.1: PID control parameters, obtained through ANSYS simulation optimization.

[0010] Preferably, the dual-microphone scene-specific beamforming and wind noise suppression are configured using the following formula: The scenario-based beamforming weight allocation algorithm follows the following formula: in, : No. The microphone's first Scenario weights; , , ; Beamforming output signal; : No. Microphone delay The input signal after; This is the final weight.

[0011] Preferably, the multimodal scene recognition and health anomaly early warning are constructed using the following formula: The multimodal scene classification and health warning algorithm follows the following formula in its calculation method: in, Scene classification results; Variance of triaxial acceleration data; Health warning signals; For real-time heart rate, This refers to blood oxygen saturation.

[0012] Preferably, the core module's multi-interface collaborative scheduling and OTA upgrade are configured using the following formula: The multi-module time-slice scheduling algorithm is calculated according to the following formula: in, : No. Time slice allocation for each module; Module priority; The scheduling period; Data integrity verification results; To receive the amount of data, For standard data volume, This is the error threshold.

[0013] Preferably, the multi-voltage adaptive power supply and dynamic energy consumption control are configured using the following formula: The dynamic energy consumption regulation algorithm is calculated according to the following formula: in, Power consumption after module adjustment; Standard power consumption for the module; : Scenario power consumption coefficient; Power consumption coefficient.

[0014] Preferably, the multi-layer sealing protection and biocompatibility treatment are configured using the following formula: The protection status monitoring algorithm follows the following formula in its calculation method: in, Waterproof and safe status; For ambient humidity; Time spent in contact with water; Remaining wear resistance cycles; 5000 is the maximum number of wear cycles; for The number of times of wear and tear.

[0015] The beneficial effects of this invention are as follows: 1. The dual-microphone beamforming inverted Z-shaped ear clip structure of this invention, by adopting a fixed shape structure and relying on a preset curvature to adapt to ear shape, has poor adaptability to special ear shapes. This solution uses a three-segment inverted Z-shaped frame + ±15° elastic hinge + 0.1N pressure sensor to achieve dynamic adjustment of the ear clip angle and microphone angle, thereby improving the fit rate and controlling the wearing pressure ≤5N in real time to avoid pressure pain. At the same time, the rotating base can finely adjust the microphone pickup angle according to the user's head posture, solving the problem of pickup offset caused by head rotation in the prior art, thus improving voice clarity.

[0016] 2. The dual-microphone beamforming inverted Z-shaped ear clip structure of the present invention integrates a triaxial accelerometer and a PPG sensor, and achieves functional coordination through the scene-parameter linkage algorithm of the core control module: automatically switching to "anti-wind noise wide beam" in motion scenarios, while reducing the PPG sampling rate to save power consumption; switching to "narrow beam pickup" in static call scenarios, while simultaneously improving PPG monitoring accuracy; triggering "omnidirectional pickup" in multi-person conference scenarios to meet the needs of multi-sound source acquisition, and realizing "function scheduling on demand and dynamic energy consumption optimization". Attached Figure Description

[0017] The invention will now be further described with reference to the accompanying drawings.

[0018] Figure 1 This is a flowchart of the adaptive pressure regulation system in this invention; Figure 2 This is a flowchart of the scene adaptive beamforming system in this invention; Figure 3 This is a flowchart of the multimodal sensing and health monitoring system in this invention; Figure 4 This is a flowchart of the multi-core collaborative scheduling and OTA upgrade system in this invention; Figure 5 This is a flowchart of the dynamic power management system in this invention; Figure 6 This is a flowchart of the multi-layer sealing and protection system in this invention. Detailed Implementation

[0019] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0020] like Figures 1 to 6 As shown in the embodiment of the present invention, a dual-microphone beamforming inverted Z-shaped ear clip structure is composed of the following modules: Adaptive structural support module, dual-microphone pickup and beamforming module, multimodal sensing module, core control module, power supply and energy management module, sealing and protection module; The aforementioned adaptive structural support module is implemented in three steps: Physical support and shape adaptation: It adopts a three-section inverted Z-shaped titanium alloy frame (the upper section fits the upper edge of the auricle, the middle section is suspended and bears the weight, and the lower section fits the concha cavity). The elastic hinge (±15° adjustable) achieves a close fit to the ear. For example, for protruding ears, the hinge can bend outward by 10° to make the lower section fit the concha cavity better; for small ears, the hinge bends inward by 5° to improve clamping stability. Real-time monitoring of wearing pressure: A pressure sensor with a sensitivity of 0.1N is embedded at the bottom of the microphone rotating base to detect the contact pressure between the base and the auricle in real time (threshold setting ≤5N, corresponding to the physiological feedback of "no obvious pressure" in the human body). If the pressure exceeds the standard (such as wearing too tightly), an "overpressure signal" is immediately sent to the core control module to trigger the adjustment command. Microphone directional assistance: The rotating base (0°-30° adjustable) can drive the microphone to fine-tune the pickup angle. For example, when the user lowers their head, the base will automatically rotate downwards by 5° to ensure that the microphone is always facing the mouth, avoiding a decrease in speech clarity caused by the offset of the pickup angle. The aforementioned dual-microphone pickup and beamforming module is implemented in three steps: High-quality voice signal acquisition: Employs a high-sensitivity MEMS microphone (frequency response 20Hz-20kHz, signal-to-noise ratio ≥62dB), covering the full frequency range of human speech (300Hz-3.4kHz as the core speech range); the microphone surface is equipped with a wind noise reduction mesh (0.1mm aperture ultra-fine metal mesh), which can reduce wind noise and airflow impact by 80%, avoiding interference from the "whooshing" sound when running; Scenario-based beamforming processing: Integrates a dedicated beamforming algorithm chip, supporting three core modes: 1. Narrow beam (40°-50°): Suitable for one-on-one calls, enhancing the voice directly in front and suppressing noise from both sides (such as colleagues talking in an office); 2. Wide beam (70°-80°): Suitable for multi-person conferences, expanding the sound pickup range and avoiding missing any speakers; 3. Anti-wind noise wide beam: Suitable for motion scenarios, using dynamic filtering to eliminate low-frequency wind noise (below 100Hz). Real-time signal output and feedback: The processed clear voice signal (sampling rate 48kHz, bit depth 16bit) is transmitted to the core control module, and the "mode switching command" of the core control module is received at the same time. For example, when switching from "stationary" to "running", the switching from narrow beam to wind noise resistant wide beam is completed within 10ms. The aforementioned multimodal sensing module is implemented in three steps: Scene state recognition: Equipped with a three-axis accelerometer (range ±8g, sampling rate 50Hz), the scene is identified by analyzing the magnitude of acceleration changes; 1. Stationary: Acceleration fluctuation ≤0.5g (e.g., office work, phone call); 2. Motion: Acceleration fluctuation 0.5g-3g (e.g., running, walking); 3. Multi-person scene: Irregular acceleration fluctuation (e.g., frequent head turning during a meeting), providing data support for beamforming mode switching; Health data monitoring: Integrated miniature PPG sensor (sampling rate 100Hz, accuracy ±2bpm) monitors heart rate (normal range 60-100bpm) and blood oxygen saturation (normal ≥95%) in real time by detecting changes in capillary blood flow under the skin of the concha. The data is preprocessed (filtered, baseline drift removed) before being transmitted to the core control module to avoid misjudgment caused by noise in the raw data. Abnormal status warning: When a heart rate > 120 bpm (outside of exercise) or blood oxygen < 93% is detected, an "abnormal warning signal" is immediately sent to the core control module, triggering subsequent reminder functions (such as pop-up windows through connected devices), extending the practical value of the module; The aforementioned core control module is implemented in four steps: Data reception and parsing: Data from each module is received in parallel through multiple interfaces; 1. Raw / processed voice signals are received from the dual-microphone module; 2. Acceleration / PPG data are received from the multimodal module; 3. Pressure data is received from the adaptive structure module and temporarily stored through a data buffer unit (16KBSRAM) to avoid concurrent data loss. Decision command generation: Commands are generated based on the "scene-parameter" linkage logic; for example: acceleration data shows "motion" (fluctuation 1.2g) → generate "switch to anti-wind noise wide beam" command; pressure data shows "5.2N" → generate "rotate base angle down by 2°" command. The commands are converted into signals that can be recognized by each module (such as SPI protocol commands) by the parsing circuit. Multi-module collaborative scheduling: Through time-slice scheduling algorithm (priority: voice > health > structure adjustment), command conflicts are avoided. For example, in motion scenarios, microphone beam switching is prioritized, and then the PPG data sampling rate is reduced (from 100Hz to 50Hz) to ensure that core functions are not affected. System Iteration and Maintenance: The system integrates an OTA upgrade module (supporting BLE Bluetooth transmission), which can remotely update beamforming algorithms and health data analysis models. For example, the system can upgrade the functionality of adding a "sleep scene beam pattern" without hardware modifications, thus extending the product lifecycle. The above power supply and energy management module is implemented in three steps: Multi-voltage power supply: Two core voltages are output through a voltage divider circuit (DC-DC converter); 1. 3.3V: supplies the core control module and multimodal sensing module (low power consumption requirement); 2. 5V: supplies the dual microphone module (high sensitivity sound pickup requires stable high voltage), and the voltage fluctuation is controlled within ±5% to avoid voltage instability leading to a decrease in microphone signal-to-noise ratio or sensor data distortion. Charging and power management: Equipped with a charging management chip (supports 5V / 1A fast charging), the 200mAh lithium battery can be fully charged in 1.5 hours; integrated power detection circuit (accuracy ±2%) monitors the remaining power in real time, and when the power is <10%, it sends a "low power warning" to the core control module to trigger the low power mode; Energy consumption optimization control: "On-demand power supply" is achieved through low-power logic circuits; I. Static scenario: Microphone wind noise reduction module is turned off, PPG sampling rate is reduced to 30Hz; II. Sleep scenario (30 minutes of inactivity): Only the minimum power consumption of the core control module is retained (10... When other modules are powered off, the battery life will be extended from 12 hours to 15 hours; The above-mentioned sealing and protection module is implemented in three steps: Waterproof and sweat- and corrosion-resistant: The module interface uses a nitrile rubber sealing ring (compression 0.2mm), and the outer shell is sealed by laser welding (weld width 0.3mm), achieving an IPX5 waterproof rating. It can withstand 30 minutes of low-pressure water jets (such as sweat and rain) to prevent microphone short circuits or PPG sensor failure due to moisture. Wear and impact resistance: The surface of the titanium alloy skeleton is nitrided (film thickness 5). The hardness has been increased to HV800, and the wear resistance is ≥5000 times (avoiding scratches caused by long-term wear); the core control module is surrounded by a 0.5mm thick silicone buffer layer, which can absorb 70% of the impact force when dropped from 1.8 meters, protecting the chip from damage; Dustproof and biocompatibility: The shell gaps are made of dustproof mesh (0.02mm aperture) to prevent dust from entering and affecting hinge adjustment; the silicone patch that comes into contact with the skin is made of medical-grade silicone rubber (compliant with FDA biocompatibility standards) to avoid skin allergies caused by long-term wear (allergy rate <0.1%).

[0021] like Figures 1 to 6 As shown, the aforementioned adaptive structure support module, dual-microphone pickup and beamforming module, multimodal sensing module, core control module, power supply and energy management module, and sealing and protection module; with the core control module as the central hub, a closed-loop system is formed through "hardware physical connection + software data interaction". The connection relationship can be divided into three categories: power supply link, data transmission link, and control command link. Each type of link has a clear interface standard, transmission content, and collaborative logic. The aforementioned power supply and energy management module is connected to each module via miniature copper foil wires (0.1mm diameter, low resistance), providing an adaptive voltage according to the power consumption requirements of each module; The aforementioned dual-microphone pickup and beamforming module transmits the original voice signal (16bit / 48kHz) and beamformed audio data to the core control module via the I2S audio interface. The aforementioned multimodal sensing module transmits preprocessed acceleration data (X / Y / Z axis values), PPG heart rate / blood oxygen data, and abnormal warning signals to the core control module via the I2C communication interface; The aforementioned adaptive structure support module transmits the following data to the core control module via a PWM control interface (which supports analog signal transmission): Forward: Real-time data from the pressure sensor (0-10N value); Reverse: "Pressure threshold setting" data from the core control module (e.g., set to ≤5N). The aforementioned core control module transmits the following commands to the dual-microphone pickup and beamforming module via the SPI control interface: Forward: Beam mode switching commands (such as "switch to anti-wind noise wide beam"), filter parameter adjustment commands (such as "wind noise threshold -45dB"); Reverse: Module execution result feedback (such as "mode switching completed"). The aforementioned core control module transmits the following commands to the adaptive structural support module via a UART serial interface: Forward: Rotation base angle adjustment commands (e.g., "lower by 2°"), hinge elastic preload adjustment commands (e.g., "increase clamping force by 1N"); Reverse: Structural adjustment result feedback (e.g., "angle adjusted to 13°, pressure 4.8N"). The aforementioned core control module transmits low-power mode switching instructions (such as "battery < 10%, enter low power") and module power-off instructions (such as "turn off microphone module when in sleep mode") to the power supply and energy management module through the GPIO general interface.

[0022] like Figures 1 to 6 As shown, it includes the following steps: S1, Adaptive anti-Z-shaped frame and pressure closed-loop regulation: It adopts a three-section inverted Z-shaped titanium alloy frame (upper section fits the upper edge of the auricle, middle section bears weight, and lower section fits the concha), and the sections are connected by ±15° elastic hinges; a 0.1N sensitivity pressure sensor is embedded in the bottom of the microphone rotating base (0°-30° adjustable); for protruding ears, the hinge is controlled to bend outward by 10° to make the lower section fit the concha; for small ears, the hinge is bent inward by 5° to increase the clamping force; the wearing pressure is detected in real time, and if it is >5N, an overpressure signal is sent to the core module to trigger the base angle to fine-tune until the pressure is ≤5N, solving the problems of poor fit and pressure pain of existing fixation structures; S2, Dual-microphone scene-specific beamforming and wind noise suppression: It employs a MEMS microphone with a frequency response of 20Hz-20kHz and a signal-to-noise ratio of ≥62dB, and is equipped with a 0.1mm aperture wind noise reduction mesh (reducing wind noise impact by 80%). It integrates a dedicated beamforming algorithm chip; during stationary calls, it automatically activates a 40°-50° narrow beam to enhance the voice directly in front; in multi-person conferences, it switches to a 70°-80° wide beam to expand the pickup range; in motion scenarios (acceleration fluctuations of 0.5g-3g), it triggers an anti-wind noise wide beam, dynamically filtering out low-frequency wind noise below 100Hz. The processed 48kHz / 16bit audio signal is transmitted to the core module with a mode switching delay of ≤10ms, overcoming the limitations of existing manual switching. S3, Multimodal Scene Recognition and Health Anomaly Early Warning: A triaxial accelerometer with a range of ±8g and a sampling rate of 50Hz is integrated into the lower part of the ear clip. The scene is identified by acceleration fluctuations: fluctuations ≤0.5g indicate stillness, 0.5g-3g indicate movement, and irregular fluctuations indicate a multi-person scene. Simultaneously, a PPG sensor with a sampling rate of 100Hz and an accuracy of ±2bpm is embedded to monitor blood flow in the concha and obtain heart rate (60-100bpm) and blood oxygen (≥95%) data. After filtering and baseline drift removal preprocessing, the data is transmitted. If the heart rate is >120bpm or blood oxygen is <93% in an external scene, an abnormal warning signal is immediately sent to the core module, realizing scene and health collaborative monitoring, which is different from the existing independent functional solutions. S4. Core module multi-interface collaborative scheduling and OTA upgrade: Employing an ARM Cortex-M55 processor, it receives microphone audio, multimodal sensing, and pressure data in parallel via I2S, I2C, and PWM interfaces, buffering them in 16KBSRAM to prevent data loss. Instructions are generated based on a "scene-parameter" logic, and time-slice scheduling (priority: voice > health > structural adjustment) avoids conflicts. For example, during exercise, beam switching is prioritized, and the PPG sampling rate is reduced to 50Hz. An integrated BLEOTA module allows for remote updates to the beam algorithm (such as adding a sleep mode) and health analysis model without hardware modifications, solving the problem of needing to replace hardware for existing system iterations. S5. Multi-voltage power supply and dynamic energy consumption control: The system outputs 3.3V (supplying the core and sensing modules) and 5V (supplying the microphone module) voltages via a DC-DC voltage divider circuit, with fluctuations controlled within ±5%. It features a 5V / 1A fast charging chip, fully charging a 200mAh lithium battery in 1.5 hours, with a power detection accuracy of ±2%. In static scenes, the microphone wind noise reduction module is disabled, and the PPG sampling rate is reduced to 30Hz. After 30 minutes of inactivity, the system enters sleep mode, retaining only the core module 10. With the lowest power consumption and other modules powered off, the battery life is extended from 12 hours to 15 hours, optimizing the existing fixed power consumption design; S6. Multi-layer sealing protection and biocompatibility treatment: The module interface uses a nitrile rubber sealing ring with a compression of 0.2mm, and the outer shell is laser welded (weld width 0.3mm) to achieve IPX5 waterproofing; the titanium alloy skeleton is nitrided (film thickness 5μm, hardness HV800), with wear resistance ≥5000 times; the core module is wrapped with a 0.5mm silicone buffer layer, absorbing 70% of the energy from a 1.8-meter drop; the gaps in the outer shell are fitted with a 0.02mm aperture dustproof mesh, and the skin contact parts use FDA-certified medical silicone rubber with an allergy rate of <0.1%, solving the problems of existing single protection and easy allergies.

[0023] like Figures 1 to 6 As shown, the above-mentioned adaptive inverse Z-shaped frame and pressure closed-loop regulation are constructed using the following formula: The adaptive pressure and angle closed-loop control algorithm is calculated according to the following formula: in, Microphone base rotation angle adjustment (unit: °, range 0°-30°); Pressure deviation ( Real-time wearing pressure, in N; (Pressure threshold); (proportion coefficient) (Integral coefficient) 0.1 (derivative coefficient): PID control parameter, obtained through ANSYS simulation optimization; A PID closed-loop control algorithm is used to calculate the deviation between the wearing pressure and the threshold in real time. :when (Overvoltage) For positive, The output can be adjusted in the positive direction (e.g., 2°), and the base rotates downwards to reduce contact pressure. Negative, The output reverse adjustment angle (e.g., 1°) improves clamping stability; compared with the existing fixed angle design, pressure stability control is achieved through dynamic deviation adjustment (error ≤ 0.2N).

[0024] like Figures 1 to 6 As shown, the above dual-microphone scene-specific beamforming and wind noise suppression are constructed using the following formula: The scenario-based beamforming weight allocation algorithm follows the following formula: in, : No. The microphone's first Scenario weights ( 2 is a dual microphone; still, sports, (Multiple people) (Scene weighting coefficient) (Scene base weights, static) ,sports Multiple people ), (Frequency weighting, for the speech frequency band of 300Hz-3.4kHz) ); Beamforming output signal; : No. Microphone delay (73-88) The input signal after ) For the final weight; Microphone weights are calculated by combining scene and frequency dimensions: static scene ( , When the microphone is in motion, the weights are tilted towards the microphone directly in front, forming a narrow beam of 40°-50°; When doing so, reduce the scene weight ratio, increase the frequency weight for mid-to-high frequency speech gain, and add a wind noise filtering factor (below 100Hz). This achieves a wide beamwidth with low wind noise. Compared to traditional fixed-weight algorithms, scene adaptability is improved by 30%.

[0025] like Figures 1 to 6 As shown, the above multimodal scene recognition and health anomaly early warning are constructed using the following formula: The multimodal scene classification and health warning algorithm follows the following formula in its calculation method: in, Scene classification results (1=static, 2=moving, 3=multiple people); : Variance of triaxial acceleration data (reflecting the amplitude of fluctuation); Health warning signal (1 = warning triggered, 0 = normal); For real-time heart rate, Blood oxygen saturation; By calculating the variance of the acceleration data Scene classification: Values ​​with variance ≤ 0.5g (small fluctuations) are classified as stationary. =1), 0.5g-3g (regular fluctuation) is motion ( =2), >3g (irregular fluctuations) are for multi-person scenarios ( =3); In the health warning logic, the warning is triggered only when the heart rate is >120 bpm or the blood oxygen level is <93% in a static scenario (excluding increased heart rate due to exercise). This avoids the problem of false alarms in motion scenes in existing technologies, and improves the accuracy of early warning to over 95%.

[0026] like Figures 1 to 6 As shown, the multi-interface collaborative scheduling and OTA upgrade of the above core modules are constructed using the following formula: The multi-module time-slice scheduling algorithm is calculated according to the following formula: in, : No. Time slice allocation for each module ( language, healthy, (structural adjustment) Module priority ( , , ); The scheduling period; Data integrity verification result (1 = qualified, 0 = retransmission); To receive the amount of data, For standard data volume, This is the error threshold; Priority-based time slice allocation: Voice module ( ) Allocate 60ms to prioritize beamforming commands; health module ( ) Allocate 30ms to process PPG data; structural adjustment module ( ) Allocate 10ms to perform angle adjustment; simultaneously verify data integrity ( Ensure that no data is lost during transmission via the I2S / I2C interface. This triggers retransmission, resolving the existing issues of "instruction conflict and data loss" in multi-module collaboration, with a scheduling delay of ≤5ms.

[0027] like Figures 1 to 6 As shown, the above-mentioned multi-voltage adaptive power supply and dynamic energy consumption control are constructed using the following formula: The dynamic energy consumption regulation algorithm is calculated according to the following formula: in, Power consumption after module adjustment (unit: ); Standard power consumption for the module (microphone) PPG ); Scene power consumption coefficient (static) 0.7, Exercise 1.0, Hibernation 0.1); Battery capacity coefficient (battery capacity ≥ 30%) 1.0, Battery level ≤ 30% 0.8, Battery level ≤10% 0.5); Combined with scene and power dynamic adjustment module power consumption: static scene ( At 0.7), the microphone power consumption drops to 10.5. The PPG sampling rate was reduced from 100Hz to 30Hz. ); Battery level <10% ( At 0.5), the microphone power consumption was further reduced to 7.5mW, retaining only the core functions; compared with the existing fixed power consumption design, the full-function battery life was extended from 8 hours to 15 hours, and the low-power battery life was improved by 50%.

[0028] like Figures 1 to 6 As shown, the above-mentioned multi-layer sealing protection and biocompatibility treatment are constructed using the following formula: The protection status monitoring algorithm follows the following formula in its calculation method: in, Waterproof safety status (1 = safe, 0 = warning); For ambient humidity ( (for threshold) Time spent in contact with water; Remaining wear resistance cycles; 5000 is the maximum number of wear cycles; for The number of wear cycles at any given moment (calculated by acceleration fluctuations; one severe fluctuation counts as one wear cycle). Real-time monitoring of waterproof and abrasion-resistant status: when ambient humidity ≤ Furthermore, the contact time with water is ≤30 minutes (meets IPX5 standard), thus determining the waterproof safety. If humidity exceeds the standard or contact with water exceeds the specified time, a waterproof warning will be triggered; the wear resistance life is calculated based on the cumulative number of wear cycles. At this time, a replacement reminder is sent to the core module; compared with the existing stateless monitoring design, the equipment reliability is improved by 40%, and users can keep abreast of the protection status.

[0029] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention.

Claims

1. A dual-microphone beamforming inverted Z-shaped ear clip structure, characterized in that: It consists of the following modules: Adaptive structural support module, dual-microphone pickup and beamforming module, multimodal sensing module, core control module, power supply and energy management module, sealing and protection module; The adaptive structural support module is implemented in three steps: Physical support and shape adaptation: It adopts a three-section reverse Z-shaped titanium alloy frame and achieves a close fit to the ear through elastic hinges; Real-time monitoring of wearing pressure: A pressure sensor with a sensitivity of 0.1N is embedded in the bottom of the microphone rotating base to detect the contact pressure between the base and the auricle in real time; Microphone directional assistance: The rotating base allows for fine-tuning of the microphone's pickup angle; The dual-microphone pickup and beamforming module is implemented in three steps: High-quality voice signal acquisition: Employs a high-sensitivity MEMS microphone that covers the full frequency range of human speech; the microphone surface is fitted with a wind noise reduction mesh. Scenario-based beamforming processing: Integrates a dedicated beamforming algorithm chip, supporting three core modes:

1. Narrow beam; 2. Wide beam; 3. Wind noise resistant wide beam; Real-time signal output and feedback: Transmits the processed clear voice signal to the core control module, and simultaneously receives the "mode switching command" from the core control module; The multimodal sensing module is implemented in three steps: Scene state recognition: Equipped with a three-axis accelerometer, it identifies scenes by analyzing the magnitude of acceleration changes; 1. Stationary; 2. Moving; 3. Multi-person scenes; Health data monitoring: Integrating a miniature PPG sensor, it monitors heart rate and blood oxygen saturation in real time by detecting changes in capillary blood flow under the skin of the concha. Abnormal status warning: When a heart rate > 120 bpm or blood oxygen < 93% is detected, an "abnormal warning signal" is immediately sent to the core control module to trigger subsequent reminder functions; The core control module is implemented in four steps: Data reception and parsing: Data from each module is received in parallel via multiple interfaces; 1. Raw / processed voice signals are received from the dual-microphone module; 2. Acceleration / PPG data are received from the multimodal module; 3. Pressure data is received from the adaptive structure module; Decision instruction generation: Instructions are generated based on the "scenario-parameter" linkage logic; Multi-module collaborative scheduling: through time-slice scheduling algorithms; System iteration and maintenance: Integrated OTA upgrade module, enabling remote updates of beamforming algorithms and health data analysis models; The power supply and energy management module is implemented in three steps: Multi-voltage power supply: Two core voltages are output through a voltage divider circuit; 1. 3.3V: supplies the core control module and multimodal sensing module; 2. 5V: supplies the dual microphone module. Charging and power management: Equipped with a charging management chip; integrates a power detection circuit to monitor the remaining power in real time; Energy consumption optimization control: "On-demand power supply" is achieved through low-power logic circuits; 1. Static scenario: Microphone wind noise reduction module is turned off; 2. Sleep scenario: Only the core control module retains the lowest power consumption. The sealing and protection module is implemented in three steps: Waterproof and sweat- and corrosion-resistant: The module interface uses a nitrile rubber sealing ring, and the outer shell is sealed by laser welding; Wear and impact resistance: The surface of the titanium alloy frame is nitrided; the core control module is surrounded by a 0.5mm thick silicone buffer layer; Dustproof and biocompatibility: The outer shell gaps are made of dustproof mesh (0.02mm aperture); the silicone patches that come into contact with the skin are made of medical-grade silicone rubber.

2. The dual-microphone beamforming inverted Z-shaped ear clip structure according to claim 1, characterized in that: The adaptive structure support module, dual-microphone pickup and beamforming module, multimodal sensing module, core control module, power supply and energy management module, and sealing and protection module; with the core control module as the hub, a closed-loop system is formed through "hardware physical connection + software data interaction". The connection relationship can be divided into three categories: power supply link, data transmission link, and control command link. Each type of link has a clear interface standard, transmission content and collaborative logic. The power supply and energy management module is connected to each module via miniature copper foil wires, and provides an adaptive voltage according to the power consumption requirements of each module. The dual-microphone pickup and beamforming module transmits the original voice signal and beamformed audio data to the core control module through the I2S audio interface. The multimodal sensing module transmits preprocessed acceleration data, PPG heart rate / blood oxygen data, and abnormal warning signals to the core control module via the I2C communication interface. The adaptive structure support module transmits the following data to the core control module via the PWM control interface: Forward: real-time data from the pressure sensor; Reverse: "pressure threshold setting" data from the core control module. The core control module transmits the following to the dual-microphone pickup and beamforming module via the SPI control interface: positive: beam mode switching command and filter parameter adjustment command; Reverse: Feedback of module execution results; The core control module transmits the following commands to the adaptive structural support module via the UART serial interface: positive: rotation base angle adjustment command and hinge elastic preload adjustment command; Reverse: Feedback of structural adjustment results; The core control module transmits low-power mode switching commands and module power-off commands to the power supply and energy management module through the GPIO general interface.

3. The dual-microphone beamforming inverted Z-shaped ear clip structure according to claim 1, characterized in that: Includes the following steps: S1, Adaptive anti-Z-shaped frame and pressure closed-loop regulation: It adopts a three-section inverted Z-shaped titanium alloy frame, with each section connected by ±15° elastic hinges; a 0.1N sensitivity pressure sensor is embedded in the bottom of the microphone rotating base; S2, Dual-microphone scene-specific beamforming and wind noise suppression: It uses a MEMS microphone with a frequency response of 20Hz-20kHz and a signal-to-noise ratio of ≥62dB, and is equipped with a 0.1mm aperture wind noise reduction mesh cover; it integrates a dedicated beamforming algorithm chip; it automatically activates a 40°-50° narrow beam when making calls while stationary to enhance the voice directly in front; it switches to a 70°-80° wide beam to expand the sound pickup range during multi-person conferences; and it triggers a wind noise reduction wide beam in motion scenarios. S3, Multimodal Scene Recognition and Health Anomaly Early Warning: A triaxial accelerometer with a range of ±8g and a sampling rate of 50Hz is integrated in the lower part of the ear clip. The scene is identified by acceleration fluctuations: fluctuations ≤0.5g indicate stationary conditions, 0.5g-3g indicate movement, and irregular fluctuations indicate multi-person scenes. Simultaneously, a PPG sensor with a sampling rate of 100Hz and an accuracy of ±2bpm is embedded to monitor blood flow in the concha and acquire heart rate and blood oxygen data. The data is then transmitted after filtering and baseline drift removal preprocessing. S4. Core module multi-interface collaborative scheduling and OTA upgrade: It uses an ARM Cortex-M55 processor and receives microphone audio, multimodal sensing, and pressure data in parallel through I2S, I2C, and PWM interfaces, with 16KBSRAM buffering to prevent data loss; it generates instructions based on "scene-parameter" logic and avoids conflicts through time-slice scheduling. S5. Multi-voltage power supply and dynamic energy consumption control: It outputs 3.3V and 5V voltages via a DC-DC voltage divider circuit, is equipped with a 5V / 1A fast charging chip, disables the microphone and wind noise reduction module in static scenes, and reduces the PPG sampling rate to 30Hz; it enters sleep mode after 30 minutes of inactivity, retaining only the core module 10. Minimum power consumption, other modules powered off; S6. Multi-layer sealing protection and biocompatibility treatment: The module interface uses a nitrile rubber sealing ring with a compression of 0.2mm, and the outer shell is laser welded; the titanium alloy skeleton is nitrided; the core module is wrapped with a 0.5mm silicone buffer layer; the gaps in the outer shell are fitted with a 0.02mm diameter dustproof mesh, and the skin contact parts are made of FDA-certified medical silicone rubber.

4. The dual-microphone beamforming inverted Z-shaped ear clip structure according to claim 3, characterized in that: The adaptive inverse Z-shaped frame and pressure closed-loop regulation are constructed using the following formula: The adaptive pressure and angle closed-loop control algorithm is calculated according to the following formula: in, Microphone base angle adjustment; Pressure deviation; , , 0.1: PID control parameters, obtained through ANSYS simulation optimization.

5. The dual-microphone beamforming inverted Z-shaped ear clip structure according to claim 3, characterized in that: The dual-microphone scene-specific beamforming and wind noise suppression are constructed using the following formula: The scenario-based beamforming weight allocation algorithm follows the following formula: in, : No. The microphone's first Scenario weights; 、 、 ； Beamforming output signal; : No. Microphone delay The input signal after; This is the final weight.

6. The dual-microphone beamforming inverted Z-shaped ear clip structure according to claim 3, characterized in that: The multimodal scene recognition and health anomaly early warning are constructed using the following formula: The multimodal scene classification and health warning algorithm follows the following formula in its calculation method: in, Scene classification results; Variance of triaxial acceleration data; Health warning signals; For real-time heart rate, This refers to blood oxygen saturation.

7. The dual-microphone beamforming inverted Z-shaped ear clip structure according to claim 3, characterized in that: The core module's multi-interface collaborative scheduling and OTA upgrade are constructed using the following formula: The multi-module time-slice scheduling algorithm is calculated according to the following formula: in, : No. Time slice allocation for each module; Module priority; The scheduling period; Data integrity verification results; To receive the amount of data, For standard data volume, This is the error threshold.

8. The dual-microphone beamforming inverted Z-shaped ear clip structure according to claim 3, characterized in that: The multi-voltage adaptive power supply and dynamic energy consumption control are constructed using the following formula: The dynamic energy consumption regulation algorithm is calculated according to the following formula: in, Power consumption after module adjustment; Standard power consumption for the module; : Scenario power consumption coefficient; Power consumption coefficient.

9. The dual-microphone beamforming inverted Z-shaped ear clip structure according to claim 3, characterized in that: The multi-layer sealing protection and biocompatibility treatment are constructed using the following formula: The protection status monitoring algorithm follows the following formula in its calculation method: in, Waterproof and safe status; For ambient humidity; Time spent in contact with water; Remaining wear resistance cycles; 5000 is the maximum number of wear cycles; for The number of times of wear and tear.

Images