Intelligent sports earphone system based on electromyographic intention recognition and scene-based sound field reconstruction

Abstract

The application belongs to the field of intelligent sports equipment and relates to an intelligent sports earphone system based on electromyographic intention recognition and scene-based sound field reconstruction, which comprises: an electromyographic sensing module including an electromyographic sensing unit and a sports intention sensing unit; the electromyographic sensing unit is arranged on both sides of an ear hook and a head beam of the sports earphone and is used for collecting surface electromyographic signals of a wearer; the sports intention sensing unit is used for recognizing real-time sports intention of the wearer according to the surface electromyographic signals; a multi-source sensing fusion module is arranged at an ear cover of the sports earphone and is used for obtaining sound information, sports scene information and orientation information of an environment in which the wearer is located; and a scene-based sound field reconstruction module is used for adjusting sound field parameters according to the real-time sports intention, the sound information of the environment in which the wearer is located, the sports scene information and the orientation information to generate sound field reconstruction information corresponding to a scene. The application realizes accurate recognition of sports intention, scene-based intelligent reconstruction of a sound field and sports safety closed-loop early warning in the use process of the intelligent earphone.

Description

Technical Field

[0001] This invention belongs to the field of intelligent sports equipment, specifically relating to an intelligent sports headphone system based on electromyography intention recognition and scene-based sound field reconstruction. Background Technology

[0002] With the popularization of national fitness activities and the rapid development of the outdoor sports industry, sports headphones, as core wearable devices connecting athletes with the outside world, have undergone continuous technological evolution from simple audio playback to intelligent and scenario-based applications. Currently, sports headphone products mainly focus on three dimensions: basic audio playback, physical protection, and simple exercise data monitoring. However, in key areas such as accurate perception of exercise intentions, sound field adaptation for complex scenarios, and closed-loop protection for exercise safety, existing technologies still have significant structural deficiencies.

[0003] At the level of motion intention perception, existing sports headphones generally rely on inertial sensors or photoelectric sensors to collect macroscopic physiological parameters such as cadence and heart rate. While these parameters can reflect exercise intensity to some extent, they cannot establish a direct mapping relationship between the exerciser's real-time motion intention and audio control. Due to the lack of motion intention recognition, existing headphone systems struggle to synchronize audio rhythm during dynamic processes such as vigorous acceleration, turning, or sudden stops, resulting in a significant time lag between exercise audio and body movements, and a significant reduction in the sense of immersion and rhythm matching.

[0004] In terms of sound field adaptation and audio reconstruction, the acoustic environment of sports scenarios exhibits high spatiotemporal heterogeneity. In outdoor running scenarios, wind noise and road noise exhibit a wide-band steady-state distribution; in road cycling scenarios, transient safety signals such as vehicle horns have sudden and directional characteristics; in gym training scenarios, equipment collision noise and background music overlap; and in mountain hiking scenarios, both the perception of natural environmental sounds and the clarity of team communication are required. Existing sports headphones generally employ a fixed noise reduction depth or a universal equalization curve sound field processing strategy, failing to dynamically reconstruct the sound field according to the safety requirements and acoustic characteristics of different sports scenarios. This creates a structural contradiction between noise reduction performance and safety assurance—excessive noise reduction blocks environmental safety sounds, while insufficient noise reduction degrades the signal-to-noise ratio of sports audio. Existing products generally have a signal-to-noise ratio below 18 dB in typical sports scenarios, making it difficult to meet the listening needs and safety requirements of complex sports environments.

[0005] In terms of sports safety early warning, existing solutions mostly rely on a single heart rate threshold to trigger the warning mechanism. The warning logic lacks multi-dimensional fusion analysis of exercise posture, environmental sound source characteristics, and muscle fatigue state. As an indirect indicator of exercise intensity, heart rate changes lag behind the actual occurrence of muscle fatigue and cannot reflect the mechanical mechanisms of sports injuries. In outdoor scenarios, spatially oriented hazards such as vehicles approaching from behind are difficult to predict based on heart rate changes, and existing early warning mechanisms have significant deficiencies in both timeliness and spatial targeting.

[0006] In summary, existing sports headphone technology has not yet achieved effective technological breakthroughs in areas such as closed-loop control of "sports intention-audio response", dynamic balance of "noise reduction depth-safety perception", and integrated decision-making of "multi-dimensional risk-graded early warning". The sports audio experience and sports safety protection are still in a state of separation. There is an urgent need to build a collaborative intelligent sports headphone system that can achieve accurate recognition of sports intention, intelligent reconstruction of scene-based sound field, and closed-loop early warning of sports safety. Summary of the Invention

[0007] In view of the shortcomings of the prior art described above, the purpose of this invention is to provide an intelligent sports headphone system based on electromyography intention recognition and scene-based sound field reconstruction, which can realize accurate recognition of movement intention, intelligent reconstruction of scene-based sound field, and closed-loop warning of movement safety during the use of intelligent headphones.

[0008] This invention provides an intelligent sports headphone system based on electromyography intention recognition and contextualized sound field reconstruction, comprising: The electromyography (EMG) sensing module includes an EMG sensing unit and a motor intention sensing unit. The electromyography (EMG) sensing unit is located on both sides of the ear hooks and headband of the sports headphones and is used to collect the wearer's surface EMG signals. The motion intention sensing unit is used to identify the wearer's real-time motion intention based on the surface electromyography signal; A multi-source sensing fusion module is installed in the earcups of the sports headphones to acquire sound information of the wearer's environment, sports scene information, and location information; The scene-based sound field reconstruction module is used to adjust the sound field parameters according to the real-time motion intention, the sound information of the surrounding environment, the motion scene information and the orientation information, so as to generate sound field reconstruction information corresponding to the scene.

[0009] In one embodiment of the present invention, the motion intention sensing unit is used to extract integral electromyography (EMG) values, root mean square (RMS) and zero-crossing number features based on the surface electromyography (SMG) signal, and to identify the wearer's motion intention based on the integral EMG values, the RMS and the zero-crossing number features using a long short-term memory (LSTM) network algorithm.

[0010] In one embodiment of the present invention, the electromyography sensing unit integrates a flexible electromyography sensor and uses a multi-channel flexible dry electrode electromyography array to collect the surface electromyography signals of the wearer during exercise. The electromyography (EMG) sensing module also includes a signal preprocessing unit, which performs adaptive notch filtering on the surface EMG signal to eliminate motion artifacts and power frequency interference, and uses wavelet packet decomposition to extract EMG signal features to obtain the preprocessed surface EMG signal.

[0011] In one embodiment of the present invention, the electromyography sensing module further includes a fatigue recognition unit, used to assess the wearer's muscle fatigue state in real time based on the surface electromyography signal; The fatigue recognition unit calculates the muscle fatigue index based on the rate of decrease of the median frequency of the surface electromyography signal over time, in order to characterize the wearer's muscle fatigue state.

[0012] In one embodiment of the present invention, a motion safety warning module is further included. The motion safety warning module is connected to the scene-based sound field reconstruction module and is used to send sound field adjustment commands to the scene-based sound field reconstruction module to drive the scene-based sound field reconstruction module to perform corresponding sound field parameter adjustments.

[0013] In one embodiment of the present invention, the motion safety warning module includes an intention-based dynamic audio control unit, a fatigue and safety-based dual warning unit, and a personalized sound field iteration unit. The intention-based dynamic audio control unit is used to generate dynamic sound field control parameters according to the real-time motion intention, so as to adjust the audio rhythm, frequency band gain or spatial sound field distribution in the sound field reconstruction information. The dual early warning unit based on fatigue and safety is used to provide graded sound field early warning based on the sound information of the surrounding environment and the muscle fatigue state. The sound field control command includes the dynamic sound field control parameters and the graded sound field warning command. The personalized sound field iteration unit is used to establish a sound field parameter model based on the wearer's initial test data, and to optimize the sound field parameter model at a preset period by combining the wearer's feedback data on sound field reconstruction information based on different scenarios and the data detected by the smart sports headphone system during the wearer's exercise.

[0014] In one embodiment of the present invention, when the intention-based dynamic audio control unit executes the generation of dynamic sound field control parameters according to the real-time motion intention, it includes: When an acceleration intention is detected, the dynamic sound field modulation parameters are used to enhance the rhythm parameters and high-frequency gain of the motion audio. When a deceleration intention is detected, the dynamic sound field modulation parameters are used to gradually reduce the rhythm parameters and low-frequency gain of the motion audio. When a turning intention is detected, the dynamic sound field control parameters are used to spatially focus the safety sound channel in the turning direction and reduce the audio volume in the opposite direction. When an emergency stop intention is detected, the dynamic sound field control parameters are used to pause the motion audio output within a preset response time and amplify the omnidirectional environmental safety sound signal. When the intention to exert force is detected, the dynamic sound field control parameters are used to synchronize the accent nodes of the motion audio with the timing of the force exertion and to enhance the low-frequency sound field feedback.

[0015] In one embodiment of the present invention, when the fatigue- and safety-based dual early warning unit performs the graded sound field early warning based on the sound information of the surrounding environment and the muscle fatigue state, it includes: When the muscle fatigue state is mild, a mild warning is triggered, and a prompt sound is superimposed on the sound field reconstruction signal; When the muscle fatigue state is severe fatigue, a severe warning is triggered, the volume of the exercise audio is automatically reduced, and a fatigue reminder voice is played. When the intensity of the detected transient safety noise exceeds the preset threshold, a safety warning is triggered. The motion audio output is paused within the preset response time, and the location information of the ambient sound source where the wearer is located is focused and a location prompt sound is output.

[0016] In one embodiment of the present invention, the multi-source sensing fusion module integrates several hardware units, including an environmental noise microphone, a motion scene recognition sensor, and a GPS positioning unit. The ambient noise microphone is used to collect ambient acoustic data, the motion scene recognition sensor is used to collect the wearer's motion posture data, and the GPS positioning unit is used to collect spatial positioning data.

[0017] In one embodiment of the present invention, the multi-source sensing fusion module includes a noise classification unit, a scene recognition unit, and a location positioning unit; The noise classification unit is used to identify the sound information of the wearer's environment based on the data collected by the hardware unit. The sound information of the environment includes steady-state motion noise and transient safety noise. The scene recognition unit is used to identify motion scene information based on the data collected by the hardware unit; The orientation positioning unit is used to determine the wearer's direction of movement and the orientation information of the ambient sound source based on the data collected by the hardware unit.

[0018] The beneficial effects of this invention are as follows: In terms of motion intention perception, this invention integrates an eight-channel flexible dry electrode electromyography array at the ear hook and headband, combined with the temporal modeling capability of long short-term memory network, to achieve accurate recognition of various motion intentions, enabling audio control to follow the changes in limb movements in real time.

[0019] At the sound field reconstruction level, this invention dynamically adjusts the audio rhythm, frequency band gain and spatial sound field distribution based on real-time motion intention, motion scene type, environmental noise category and external sound source location information, realizing a differentiated noise reduction strategy. It effectively suppresses steady-state interference such as wind noise and equipment noise, while selectively amplifying and spatially focusing transient safety sounds such as vehicle horns and teammates' shouts.

[0020] At the level of safety early warning and closed-loop intervention, this invention establishes a graded early warning mechanism that simultaneously monitors internal fatigue and external threats from two dimensions. The internal dimension calculates the muscle fatigue index using the rate of decline of the median frequency of electromyographic signals; the external dimension uses a multi-source perception fusion module to locate external sound sources. Upon triggering an early warning, the motion safety early warning module directly drives the scenario-based sound field reconstruction module to enforce intervention on audio parameters. Simultaneously, the personalized sound field iteration unit automatically optimizes the sound field parameter model by combining the wearer's subjective evaluation with objective motion data, enabling the system to continuously evolve and adapt to the user's movement habits. Attached Figure Description

[0021] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application. It is obvious that the drawings described below are merely some embodiments of this application, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0022] Figure 1 This is a functional block diagram of an intelligent sports headphone system based on electromyography intention recognition and scene-based sound field reconstruction provided in one embodiment of the present invention; Figure 2 This is a functional block diagram of an electromyography sensing module provided in one embodiment of the present invention; Figure 3 This is a functional block diagram of a multi-source sensing fusion module provided in one embodiment of the present invention; Figure 4 This is a functional block diagram of a motion safety early warning module provided in one embodiment of the present invention; Among them, the electromyography (EMG) sensing module is 101, the EMG sensing unit is 1011, the motion intention sensing unit is 1012, the fatigue recognition unit is 1013, the signal preprocessing unit is 1014, the multi-source sensing fusion module is 102, the noise classification unit is 1021, the scene recognition unit is 1022, the orientation positioning unit is 1023, the scene-based sound field reconstruction module is 103, the motion safety warning module is 104, the intention-based dynamic audio control unit is 1041, the fatigue and safety-based dual warning unit is 1042, the personalized sound field iteration unit is 1043, and the audio output module is 105. Detailed Implementation

[0023] To facilitate understanding of this application, a more complete description will be provided below with reference to the accompanying drawings, which illustrate embodiments of the present application. However, the present application can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that the disclosure of this application will be thorough and complete.

[0024] This invention provides an intelligent sports headphone system based on electromyography (EMG) intention recognition and scene-based sound field reconstruction, which can be applied to various sports scenarios. This system organically integrates surface EMG sensing technology, multi-source environmental sensing technology, scene-based sound field reconstruction technology, and sports safety early warning technology, upgrading the headphones from a simple audio playback device into an intelligent sports assistant with the capabilities of motion intention perception, scene-adaptive audio control, and proactive safety risk warning.

[0025] See Figure 1 As shown, an intelligent sports headphone system based on electromyography intention recognition and scene-based sound field reconstruction provided by an embodiment of the present invention includes: an electromyography sensing module 101, a multi-source sensing fusion module 102, a scene-based sound field reconstruction module 103, and a sports safety warning module 104.

[0026] The intelligent sports headphone system based on electromyography (EMG) intention recognition and contextualized sound field reconstruction provided by this invention has a three-layer progressive architecture. The bottom layer is the perception layer, consisting of an EMG perception module 101 and a multi-source perception fusion module 102, responsible for collecting the wearer's internal physiological signals and external environmental information, respectively. The middle layer is the decision layer, consisting of a contextualized sound field reconstruction module 103 and a sports safety warning module 104, responsible for the fusion processing of multi-source information and the generation of sound field control commands. The top layer is the execution layer, namely the headphone's own audio output module 105, responsible for converting the reconstructed sound field signal into audible sound wave output, that is, converting the sound field reconstruction information generated by the contextualized sound field reconstruction module into sound wave signal output. The modules communicate in real time via a data bus, forming a closed-loop link of perception, decision-making, and execution.

[0027] Further, see Figure 2 As shown, the electromyography sensing module 101 includes an electromyography sensing unit 1011, a motion intention sensing unit 1012, a fatigue recognition unit 1013, and a signal preprocessing unit 1014.

[0028] The electromyography (EMG) sensing unit 1011 is located on both sides of the ear hooks and headband of the sports headphones. It integrates flexible EMG sensors and uses a multi-channel flexible dry electrode EMG array (such as an 8-channel flexible dry electrode EMG array) to collect the wearer's surface EMG signals during exercise. After wearing the sports headphones, the electrode array fits closely to the wearer's temporalis and upper trapezius muscles. During exercise, the bioelectrical signals generated by muscle contraction are picked up by the electrodes and converted into digital signals at a sampling rate of 512Hz, with a common-mode rejection ratio of not less than 110dB.

[0029] The signal preprocessing unit 1014 performs two-stage processing on the digitized surface electromyography (EMG) signal. First, it uses an adaptive notch filter to eliminate 50Hz power frequency interference and its harmonic components, while suppressing low-frequency motion artifacts caused by the relative sliding of the electrodes and skin. This achieves adaptive notch filtering of the EMG signal to eliminate motion artifacts and power frequency interference. Then, it uses a wavelet packet decomposition algorithm to decompose the signal into different frequency bands and extract the effective EMG components to obtain the preprocessed EMG signal.

[0030] The motion intention sensing unit 1012 is used to identify the wearer's real-time motion intention based on the surface electromyography (EMG) signals. The motion intentions include acceleration, deceleration, turning, sudden stop, and exertion of force. Specifically, a sliding time window (preferably 100 to 200 milliseconds) can be used to extract data segments from the continuous EMG signal stream. Within each window, the integrated EMG value, root mean square (RMS) value, and zero-crossing count feature are calculated. The integrated EMG value is obtained by accumulating the absolute values ​​of the signals within the window, reflecting the total intensity of muscle force exertion. The RMS value is obtained by taking the square root of the average of the squared amplitude values, characterizing the effective power of muscle contraction. The zero-crossing count feature counts the number of times the signal waveform crosses the zero-level line, reflecting the frequency of muscle discharge. These three features form a feature vector, which is input to a pre-trained long short-term memory (LSTM) network classifier. This classifier takes the feature sequences from multiple consecutive windows as input and outputs the probability distribution of the five motion intentions (acceleration, deceleration, turning, sudden stop, and exertion of force), selecting the highest probability as the current intention recognition result.

[0031] The fatigue recognition unit 1013 is used to assess the wearer's muscle fatigue state in real time based on the surface electromyography (EMG) signal; and to calculate a muscle fatigue index based on the rate of decrease of the median frequency of the EMG signal over time, to characterize the wearer's muscle fatigue state. Specifically, a fast Fourier transform is performed on the signal segment per second to obtain a power spectrum, from which the median frequency is located, which is the high-frequency cutoff frequency at which the total power of the spectrum is evenly divided. The system records the median frequency value per second and calculates its rate of decrease over time, normalizes and compares the current rate of decrease, and maps it to a muscle fatigue index between 0 and 1. A value greater than 0.8 indicates no fatigue, 0.6 to 0.8 indicates mild fatigue, and less than 0.5 indicates severe fatigue.

[0032] This invention utilizes a long short-term memory network recognition mechanism based on integral electromyography values, root mean square (RMS) values, and the number of zero-crossing points. This mechanism can predict core motion intentions such as acceleration, deceleration, steering, sudden stops, and force exertion in advance, providing a preliminary decision-making basis for sound field reconstruction and significantly improving the immersiveness and rhythm matching of motion.

[0033] In this embodiment of the invention, the multi-source sensing fusion module 102 is disposed on the earcups of the sports headphones and is used to acquire sound information, motion scene information and location information of the wearer's environment. It integrates several hardware units, including an environmental noise microphone, a motion scene recognition sensor and a GPS positioning unit.

[0034] The ambient noise microphone is used to collect ambient acoustic data, the motion scene recognition sensor is used to collect the wearer's motion posture data, and the GPS positioning unit is used to collect spatial positioning data.

[0035] For example, at the hardware level, each earcup is equipped with an ambient noise microphone to form a binaural array for collecting ambient acoustic data and establishing a spatial auditory baseline; the motion scene recognition sensor is a six-axis inertial measurement unit (including a three-axis accelerometer and a three-axis gyroscope) for collecting motion posture data; and the GPS positioning unit is used to collect spatial positioning data.

[0036] Specifically, see Figure 3 As shown, the multi-source perception fusion module 102 includes a noise classification unit 1021, a scene recognition unit 1022, and a location positioning unit 1023.

[0037] The noise classification unit 1021 is used to identify the sound information of the wearer's environment based on the data collected by the hardware unit. The sound information of the environment includes steady-state motion noise (such as wind noise and treadmill noise) and transient safety noise (such as vehicle horns and shouts). For example, the noise classification unit 1021 is used to analyze the acoustic waveforms collected by the ambient noise microphone in real time, extract acoustic features such as spectral envelope, onset time, duration, and peak factor, and uses a support vector machine classifier to distinguish ambient sounds into steady-state motion noise (continuous and stable, fixed spectrum) and transient safety noise (sudden pulses, spectral abrupt changes). Taking vehicle horns as an example, their onset time is extremely short (less than 20 milliseconds), their peak is prominent and decays rapidly, which contrasts sharply with the stable and wide-bandwidth characteristics of wind noise, and can be accurately identified.

[0038] The scene recognition unit 1022 is used to recognize sports scene information based on the data collected by the hardware unit; the sports scene information includes sports scenes such as road cycling, outdoor running, gym training, and mountain hiking.

[0039] Taking a road cycling scenario as an example: data from the GPS positioning unit shows a moving speed of 20 to 35 km / h; posture data from the motion scene recognition sensor shows a stable torso tilt and no vertical undulations in the gait (distinct from the obvious body undulations during running); and acoustic data collected by the environmental noise microphone shows continuous wind noise superimposed with intermittent vehicle traffic noise. These three sets of features are input into a decision tree classifier, which outputs a "road cycling" scene label.

[0040] Similarly, when the GPS speed is 6 to 15 km / h, the posture data shows regular vertical fluctuations (step frequency of about 160 to 190 steps / minute), and the acoustic data shows outdoor ambient noise, the "Outdoor Running" label is output; when the GPS signal is weak or lost, the posture data shows an explosive force-recovery alternation pattern, and the acoustic data shows indoor equipment noise, the "Gym Training" label is output; when the GPS speed is 3 to 6 km / h and the altitude data shows significant fluctuations, the posture data shows uneven stride length, and the acoustic data shows natural ambient noise, the "Mountain Hiking" label is output.

[0041] The orientation positioning unit 1023 is used to determine the wearer's direction of movement and the orientation information of the ambient sound source based on the data collected by the hardware unit. Specifically, it first calculates the relative azimuth angle of the sound source relative to the wearer's head by using the time difference and volume difference of the same sound signal received by the left and right ambient noise microphones; then it determines the wearer's absolute direction of movement in the geographic coordinate system by using the heading angle data provided by the GPS positioning unit; finally, it superimposes the relative azimuth angle with the absolute direction of movement to output the absolute azimuth angle of the external sound source relative to the wearer's direction of movement.

[0042] This invention constructs a multi-dimensional mapping relationship between motion intention, environmental sound type, motion scene, sound source location, and sound field parameters by classifying and identifying steady-state motion noise and transient safety noise, automatically identifying core sports scenarios such as outdoor running, road cycling, gym training, and mountain hiking, and accurately locating the location of environmental sound sources through the multi-source perception fusion module 102.

[0043] In this embodiment of the invention, the scene-based sound field reconstruction module 103 is used to adjust the sound field parameters according to the real-time motion intention, the sound information of the surrounding environment, the motion scene information and the orientation information, so as to generate sound field reconstruction information corresponding to the scene.

[0044] Optionally, the scene-based sound field reconstruction module 103 receives in real time the outputs of the motion intention perception unit 1012 (motion intention), the scene recognition unit 1022 (motion scene), the noise classification unit 1021 (noise type), and the orientation unit 1023 (sound source orientation), and generates suitable sound field parameters through a three-dimensional mapping model. The three-dimensional mapping model is stored in a built-in memory, and its core is a multi-dimensional mapping table. Each row in the table corresponds to a combination of "motion intention - motion scene," and each column corresponds to a type of sound field parameter (rhythm shift, gain shift in each frequency band, spatial phase shift, safety channel settings, etc.). When the system recognizes the current motion intention and motion scene, it obtains the sound field parameters by looking up the table.

[0045] In a practical application scenario of this invention, in outdoor running mode, the scenario-based sound field reconstruction module 103 boosts the gain of the mid-to-high frequency band (2kHz to 8kHz) of the dynamic equalizer by 6dB, making the beat guidance sound more prominent; at the same time, the wind noise attenuation filter is activated to achieve wind noise suppression of no less than 32dB; in terms of spatial sound image reconstruction, the safety sound channel in the 180° rear direction is retained without attenuation to ensure that the sound of vehicles approaching from behind can be clearly perceived.

[0046] In road cycling mode, the scenario-based sound field reconstruction module 103 prioritizes the safety sound control mechanism to the highest level. This mechanism continuously monitors the detection results of transient safety noise. Once the noise classification unit 1021 reports "transient safety noise" and the sound pressure level exceeds 70dB, the module immediately reduces the volume of the motion audio by 30%, while simultaneously increasing the gain in the direction of the safety sound source by 12dB and focusing the spatial acoustic phase in that direction, enabling the wearer to accurately perceive the source of the threat. The gain of the navigation voice channel is simultaneously increased by 10dB.

[0047] In gym training mode, the scenario-based sound field reconstruction module 103 increases the gain of the low-frequency band (60Hz to 500Hz) of the dynamic equalizer by 8dB, so that the bass rhythm resonates with the exertion and recovery cycle of strength training; at the same time, the closed sound field reconstruction algorithm is activated to reduce the ambient noise by no less than 28dB through active noise reduction, creating an immersive training auditory environment.

[0048] In mountain hiking mode, the scene-based sound field reconstruction module 103 bypasses the noise reduction path of the natural sound audio signal collected by the environmental noise microphone, deliberately preserving the complete spectrum of natural environmental sounds such as birdsong, streams, and wind; at the same time, it enhances the ability to focus on the shouts of teammates, so that the wearer can enjoy the natural sounds without losing voice communication with teammates.

[0049] The scenario-based sound field reconstruction module 103 of this invention adopts a fusion scheme of dynamic equalization, spatial acoustic reconstruction, and safety sound priority control, realizing adaptive switching of exclusive sound field modes under different sports scenarios. In outdoor running mode, it focuses on mid-to-high frequency gain while retaining the rear safety sound channel; in road cycling mode, it activates omnidirectional safety sound focusing and sets transient safety noise as the highest priority; in gym training mode, it enhances low-frequency gain to match the rhythm of strength training; and in mountain hiking mode, it uses wideband mild noise reduction to retain natural ambient sound. This solves the structural contradiction between noise reduction performance and safety assurance in existing technologies.

[0050] In this embodiment of the invention, the motion safety warning module 104 is connected to the scene-based sound field reconstruction module 103, and is used to send sound field adjustment commands to the scene-based sound field reconstruction module 103 to drive the scene-based sound field reconstruction module 103 to perform corresponding sound field parameter adjustments. (See also...) Figure 4 As shown, the motion safety warning module 104 includes an intention-based dynamic audio control unit 1041, a fatigue and safety-based dual warning unit 1042, and a personalized sound field iteration unit 1043.

[0051] The intention-based dynamic audio control unit 1041 is used to generate dynamic sound field control parameters according to the real-time motion intention, so as to adjust the audio rhythm, frequency band gain or spatial sound field distribution in the sound field reconstruction information; the fatigue and safety-based dual warning unit 1042 is used to perform graded sound field warnings according to the sound information of the surrounding environment and the muscle fatigue state; the sound field control command includes the dynamic sound field control parameters and the graded sound field warning command.

[0052] The personalized sound field iteration unit 1043 is used to establish a sound field parameter model based on the wearer's initial test data, and to optimize the sound field parameter model by combining the wearer's feedback data on sound field reconstruction information based on different scenarios and the data detected by the smart sports headphone system during the wearer's exercise at a preset period (e.g., update period ≤ 7 days).

[0053] For example, the personalized sound field iteration unit 1043 performs long-term adaptive optimization of the system. Upon initial system use, the user can be guided through a human-computer interface to complete a motion type preference test and an audio preference test. Simultaneously, electromyography (EMG) signals are collected under resting conditions and standard exercise loads for individual baseline calibration, establishing an initial sound field parameter model. In subsequent use, the personalized sound field iteration unit 1043 periodically collects the user's subjective feedback scores on the sound field reconstruction information, as well as objective motion data detected by the system, including the number of fatigue occurrences, the number of safety warning triggers, and the confidence distribution of motion intention recognition. It then uses a gradient descent algorithm to periodically optimize the sound field parameter model, ensuring that the sound field reconstruction strategy continuously approaches the user's individual optimal preferences.

[0054] In this embodiment of the invention, when the intention-based dynamic audio control unit 1041 executes the generation of dynamic sound field control parameters according to the real-time motion intention, it includes: When an acceleration intention is detected, the dynamic sound field modulation parameters are used to enhance the rhythm parameters and high-frequency gain of the motion audio. When a deceleration intention is detected, the dynamic sound field modulation parameters are used to gradually reduce the rhythm parameters and low-frequency gain of the motion audio. When a turning intention is detected, the dynamic sound field control parameters are used to spatially focus the safety sound channel in the turning direction and reduce the audio volume in the opposite direction. When an emergency stop intention is detected, the dynamic sound field control parameters are used to pause the motion audio output within a preset response time and amplify the omnidirectional environmental safety sound signal. When the intention to exert force is detected, the dynamic sound field control parameters are used to synchronize the accent nodes of the motion audio with the timing of the force exertion and to enhance the low-frequency sound field feedback.

[0055] In a practical application scenario of this invention, the intention-based dynamic audio control unit 1041 polls the output of the motion intention perception unit 1012 at millisecond intervals. Once a new motion intention is detected, corresponding dynamic control parameters are immediately generated and pushed to the scene-based sound field reconstruction module 103 for execution. Taking a turning intention as an example: when a left turning intention is detected, the intention-based dynamic audio control unit 1041 generates a dynamic sound field control parameter combination of "6dB gain increase in the left safety sound channel, 3dB attenuation in the right audio volume, and left front spatial acoustic focus," enabling the wearer to clearly perceive the environmental acoustic information of the turning direction during the turning process.

[0056] In this embodiment of the invention, when the fatigue- and safety-based dual early warning unit 1042 performs the graded sound field early warning based on the sound information of the surrounding environment and the muscle fatigue state, it includes: When the muscle fatigue state is mild, a mild warning is triggered, and a prompt sound is superimposed on the sound field reconstruction signal; When the muscle fatigue state is severe fatigue, a severe warning is triggered, the volume of the exercise audio is automatically reduced, and a fatigue reminder voice is played. When the intensity of the detected transient safety noise exceeds the preset threshold, a safety warning is triggered. The motion audio output is paused within the preset response time, and the location information of the ambient sound source where the wearer is located is focused and a location prompt sound is output.

[0057] In a practical application scenario of this invention, the dual early warning unit 1042 based on fatigue and safety performs hierarchical decision-making. The first layer of the decision logic is safety priority judgment: if the noise classification unit 1021 reports the detection of transient safety noise and the sound pressure level is not lower than 70dB, then regardless of the fatigue state, the system immediately triggers a safety warning and generates a sequence of instructions to "pause motion audio output, amplify the ambient sound in that direction, and output a direction prompt sound", with a preset response time of less than 15 milliseconds.

[0058] The second layer of the decision-making logic is fatigue grading: if there is no safety threat, the fatigue level is graded based on the muscle fatigue index output by the fatigue recognition unit 1013. When the muscle fatigue index is between 0.6 and 0.8, a mild warning is triggered, and a low-volume, wide-bandwidth, gentle prompt tone is superimposed on the sound field reconstruction signal to remind the wearer to pay attention to the adjustment of the exercise rhythm without interrupting the continuity of the audio. When the muscle fatigue index is below 0.5, a severe warning is triggered, generating a sequence of instructions to "automatically reduce the volume of the exercise audio, pause music playback, play voice reminders, and push stretching suggestions to the APP".

[0059] This invention overcomes the limitations of existing technologies that rely on a single heart rate threshold to trigger warnings, and establishes a multi-dimensional risk assessment system based on muscle fatigue index and environmental sound source characteristics. The fatigue recognition unit 1013 calculates the muscle fatigue index by monitoring the rate of decrease in the median frequency of electromyographic signals over time, achieving dynamic quantitative monitoring of muscle fatigue status. The dual warning unit 1042, based on fatigue and safety, performs graded warnings according to muscle fatigue status and transient safety noise intensity. For mild fatigue, a warning sound is superimposed on the sound field reconstruction signal; for severe fatigue, the volume of the motion audio is automatically reduced and a fatigue reminder voice is played; when high-intensity transient safety noise is detected, the motion audio output is paused within a preset response time, and a directional warning sound is output focusing on the danger direction.

[0060] In a practical application scenario of this invention, a user is wearing the smart sports headphones of this invention while cycling on a road. The system has determined the current exercise scenario to be "road cycling" through the scene recognition unit 1022, and the music is playing normally. At this time: The electromyography (EMG) sensing module 101 continuously collects the user's surface EMG signals. The motion intention sensing unit 1012 analyzes the EMG characteristics through a long short-term memory network and determines that the current motion intention is "straight-line at a constant speed". The fatigue recognition unit 1013 calculates the muscle fatigue index as 0.85 based on the rate of decrease of the median frequency of EMG, indicating that the user is in a fatigue-free state.

[0061] The ambient noise microphone of the multi-source sensing fusion module 102 detects a sudden pulse sound coming from behind. The noise classification unit 1021 classifies it as "transient safety noise". The orientation positioning unit 1023 combines the binaural time difference and GPS heading angle data to calculate that the absolute azimuth angle of the sound source is 180° (i.e. directly behind the wearer) and the sound pressure level is 85dB.

[0062] The dual warning unit of the motion safety warning module 104 intervenes with safety priority logic to trigger a safety warning. The warning command is sent to the scene-based sound field reconstruction module 103, which performs the following sound field parameter adjustments within 12 milliseconds: reduces the motion audio volume by 35%, increases the sound gain in the 180° direction by 12dB, and performs spatial sound phase focusing.

[0063] Users can clearly perceive the specific location of the horn sound from behind through the reconstructed sound field, and take timely action to move to the side to avoid it, thus avoiding safety risks.

[0064] In another practical application scenario of the present invention, a user wears the smart sports headphones of the present invention to perform dumbbell squat training in a gym.

[0065] The electromyography sensing unit 1011 of the electromyography sensing module 101 continuously collects temporalis muscle linkage signals related to quadriceps movement. The movement intention sensing unit 1012 identifies the alternating occurrence of multiple "force exertion intentions". The fatigue recognition unit 1013 monitors that the median frequency of the electromyography signal shows a continuous accelerating downward trend. The calculated muscle fatigue index drops to 0.45, entering the severe fatigue range.

[0066] Based on the dual early warning unit 1042 for fatigue and safety, in the absence of any safety threat input, a severe warning is triggered according to fatigue grading logic. The warning command is sent to the contextualized sound field reconstruction module 103, which automatically reduces the volume of the exercise audio and pauses music playback, playing a voice reminder through the headphones: "Muscle fatigue, please stop training and stretch." At the same time, the accompanying application pushes a targeted stretching tutorial for the quadriceps.

[0067] Users who accept voice advice to stop training and stretch can effectively avoid the risk of sports injuries caused by excessive muscle exertion.

[0068] In summary, at the level of motion intention perception, this invention integrates an eight-channel flexible dry electrode electromyography array at the ear hook and headband, combined with the temporal modeling capability of long short-term memory networks, to achieve accurate recognition of various motion intentions, enabling audio control to synchronize in real time with changes in limb movements.

[0069] At the sound field reconstruction level, this invention dynamically adjusts the audio rhythm, frequency band gain and spatial sound field distribution based on real-time motion intention, motion scene type, environmental noise category and external sound source location information, realizing a differentiated noise reduction strategy. It effectively suppresses steady-state interference such as wind noise and equipment noise, while selectively amplifying and spatially focusing transient safety sounds such as vehicle horns and teammates' shouts.

[0070] At the level of safety early warning and closed-loop intervention, this invention establishes a graded early warning mechanism that simultaneously monitors internal fatigue and external threats from two dimensions. The internal dimension calculates the muscle fatigue index using the rate of decline of the median frequency of electromyographic signals; the external dimension uses the multi-source perception fusion module 102 to locate external sound sources. Upon triggering an early warning, the motion safety early warning module 104 directly drives the scenario-based sound field reconstruction module 103 to enforce intervention on audio parameters. Simultaneously, the personalized sound field iteration unit 1043 automatically optimizes the sound field parameter model by combining the wearer's subjective evaluation with objective motion data, achieving the system's adaptive capability to continuously evolve with the user's exercise habits.

[0071] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0072] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0073] Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be embraced within the invention. No appended diagram markings in the claims should be construed as limiting the scope of the claims.

[0074] Furthermore, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices recited in a system claim may also be implemented by a single unit or device through software or hardware. The terms "first," "second," etc., are used to indicate names and do not indicate any specific order.

[0075] Finally, it should be noted that the above 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 preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A smart sports headphone system based on electromyography intention recognition and contextualized sound field reconstruction, characterized in that, include: The electromyography (EMG) sensing module includes an EMG sensing unit and a motor intention sensing unit. The electromyography (EMG) sensing unit is located on both sides of the ear hooks and headband of the sports headphones and is used to collect the wearer's surface EMG signals. The motion intention sensing unit is used to identify the wearer's real-time motion intention based on the surface electromyography signal; A multi-source sensing fusion module is installed in the earcups of the sports headphones to acquire sound information of the wearer's environment, sports scene information, and location information; The scene-based sound field reconstruction module is used to adjust the sound field parameters according to the real-time motion intention, the sound information of the surrounding environment, the motion scene information and the orientation information, so as to generate sound field reconstruction information corresponding to the scene.

2. The intelligent sports headphone system based on electromyography intention recognition and scene-based sound field reconstruction according to claim 1, characterized in that, The motion intention sensing unit is used to extract integral electromyography (EMG) values, root mean square (RMS) and zero-crossing number features based on the surface EMG signals, and to identify the wearer's motion intention based on the integral EMG values, RMS and zero-crossing number features using a long short-term memory network algorithm.

3. The intelligent sports headphone system based on electromyography intention recognition and scene-based sound field reconstruction according to claim 1, characterized in that, The electromyography sensing unit integrates a flexible electromyography sensor and uses a multi-channel flexible dry electrode electromyography array to collect the wearer's surface electromyography signals during exercise. The electromyography sensing module also includes a signal preprocessing unit, which performs adaptive notch filtering on the surface electromyography signal to eliminate motion artifacts and power frequency interference, and uses wavelet packet decomposition to extract electromyography signal features to obtain the preprocessed surface electromyography signal.

4. The intelligent sports headphone system based on electromyography intention recognition and scene-based sound field reconstruction according to claim 1, characterized in that, The electromyography sensing module also includes a fatigue recognition unit, which is used to assess the wearer's muscle fatigue state in real time based on the surface electromyography signal. The fatigue recognition unit is used to calculate the muscle fatigue index based on the rate of decrease of the median frequency of the surface electromyography signal over time, so as to characterize the wearer's muscle fatigue state.

5. The intelligent sports headphone system based on electromyography intention recognition and scene-based sound field reconstruction according to claim 4, characterized in that, It also includes a motion safety warning module, which is connected to the scene-based sound field reconstruction module and is used to send sound field control commands to the scene-based sound field reconstruction module to drive the scene-based sound field reconstruction module to perform corresponding sound field parameter adjustments.

6. The intelligent sports headphone system based on electromyography intention recognition and scene-based sound field reconstruction according to claim 5, characterized in that, The sports safety early warning module includes an intention-based dynamic audio control unit, a fatigue and safety-based dual early warning unit, and a personalized sound field iteration unit. The intention-based dynamic audio control unit is used to generate dynamic sound field control parameters according to the real-time motion intention, so as to adjust the audio rhythm, frequency band gain or spatial sound field distribution in the sound field reconstruction information. The dual early warning unit based on fatigue and safety is used to provide graded sound field early warning based on the sound information of the surrounding environment and the muscle fatigue state. The sound field control command includes the dynamic sound field control parameters and the graded sound field warning command. The personalized sound field iteration unit is used to establish a sound field parameter model based on the wearer's initial test data, and to optimize the sound field parameter model at a preset period by combining the wearer's feedback data on sound field reconstruction information based on different scenarios and the data detected by the smart sports headphone system during the wearer's exercise.

7. The intelligent sports headphone system based on electromyography intention recognition and scene-based sound field reconstruction according to claim 6, characterized in that, When the intent-based dynamic audio control unit executes the generation of dynamic sound field control parameters based on the real-time motion intent, it includes: When an acceleration intention is detected, the dynamic sound field modulation parameters are used to enhance the rhythm parameters and high-frequency gain of the motion audio. When a deceleration intention is detected, the dynamic sound field modulation parameters are used to gradually reduce the rhythm parameters and low-frequency gain of the motion audio. When a turning intention is detected, the dynamic sound field control parameters are used to spatially focus the safety sound channel in the turning direction and reduce the audio volume in the opposite direction. When an emergency stop intention is detected, the dynamic sound field control parameters are used to pause the motion audio output within a preset response time and amplify the omnidirectional environmental safety sound signal. When the intention to exert force is detected, the dynamic sound field control parameters are used to synchronize the accent nodes of the motion audio with the timing of the force exertion and to enhance the low-frequency sound field feedback.

8. The intelligent sports headphone system based on electromyography intention recognition and scene-based sound field reconstruction according to claim 6, characterized in that, When the fatigue- and safety-based dual early warning unit executes the graded sound field early warning based on the sound information of the surrounding environment and the muscle fatigue state, it includes: When the muscle fatigue state is mild, a mild warning is triggered, and a prompt sound is superimposed on the sound field reconstruction signal; When the muscle fatigue state is severe fatigue, a severe warning is triggered, the volume of the exercise audio is automatically reduced, and a fatigue reminder voice is played. When the intensity of the detected transient safety noise exceeds the preset threshold, a safety warning is triggered. The motion audio output is paused within the preset response time, and the location information of the ambient sound source where the wearer is located is focused and a location prompt sound is output.

9. The intelligent sports headphone system based on electromyography intention recognition and scene-based sound field reconstruction according to claim 1, characterized in that, The multi-source perception fusion module integrates several hardware units, including an environmental noise microphone, a motion scene recognition sensor, and a GPS positioning unit. The ambient noise microphone is used to collect ambient acoustic data, the motion scene recognition sensor is used to collect the wearer's motion posture data, and the GPS positioning unit is used to collect spatial positioning data.

10. The intelligent sports headphone system based on electromyography intention recognition and scene-based sound field reconstruction according to claim 9, characterized in that, The multi-source perception fusion module includes a noise classification unit, a scene recognition unit, and a location positioning unit. The noise classification unit is used to identify the sound information of the wearer's environment based on the data collected by the hardware unit. The sound information of the environment includes steady-state motion noise and transient safety noise. The scene recognition unit is used to identify motion scene information based on the data collected by the hardware unit; The orientation positioning unit is used to determine the wearer's direction of movement and the orientation information of the ambient sound source based on the data collected by the hardware unit.

Images