Audio processing method, closed loop pressure regulation system, and computer device

Abstract

The application provides an audio processing method, a closed-loop pressure regulation system and a computer device, and relates to the technical field of intelligent wearing. The method is applied to a wearable device, the wearable device is in communication connection with a storage device, and the wearable device is provided with a sensing module. The method comprises the following steps: in response to receiving a wearing instruction, collecting physiological data of a user through the sensing module; based on the physiological data, analyzing the stress state of the user to obtain a stress state index; based on the stress state index, determining a target audio parameter corresponding to the stress state index; generating a target audio based on the target audio parameter, and playing the target audio. The application enables users in different stress levels to receive audio content corresponding to their current stress level, improves the pertinence of audio pressure regulation for users, and more accurately plays target audio corresponding to different stress state indexes to more accurately regulate the stress of users.

Description

Technical Field

[0001] This application relates to the field of smart wearable technology, and more specifically, to an audio processing method, a closed-loop pressure regulation system, and a computer device. Background Technology

[0002] With the fast pace of life and work, stress monitoring and regulation are gradually becoming an application area of ​​wearable device health functions. Wearable devices are characterized by portability, close fit, and continuous wearability. They can collect physiological data such as heart rate and heart rate variability during daily use and provide users with relaxing, sleep-aiding, or stress-relieving auditory content in conjunction with audio playback functions.

[0003] In related technologies, wearable devices typically play fixed audio clips from a preset audio library or audio from a preset playlist to the user. Since a user's stress levels may vary at different times while using the wearable device, and the content of the fixed audio clips or preset playlists is relatively fixed, there may be a mismatch between the played audio and the user's actual stress level, resulting in insufficient targeting of the stress regulation process. Summary of the Invention

[0004] The purpose of this application is to provide an audio processing method and a closed-loop pressure regulation system, thereby solving to some extent the problem in the related art where the audio being played does not match the user's current pressure, affecting the targeting of the pressure regulation process. This improves the targeting of audio for user pressure regulation, and more accurately plays the target audio corresponding to different pressure state indicators, so as to more accurately regulate the user's pressure.

[0005] Other features and advantages of this application will become apparent from the following detailed description, or may be learned in part from practice of this application.

[0006] According to a first aspect of the embodiments of this application, an audio processing method is provided, applied to a wearable device, wherein the wearable device is communicatively connected to a storage device, and the wearable device is provided with a sensing module; the method includes: In response to a received wearing command, the sensor module collects the user's physiological data, wherein the wearing command is triggered by the user wearing the wearable device; the physiological data includes at least one of the user's heart rate data, respiratory rate data, and sleep state parameters; Based on physiological data, the user's stress state is analyzed to obtain stress state indicators; Based on the stress state index, target audio parameters corresponding to the stress state index are determined. The stress state index is used to characterize the mapping relationship between the user's stress level and the audio parameters. The audio parameters include at least one of rhythm, mode, harmonic complexity, timbre and soundscape element combination. Generate target audio based on target audio parameters, and then play the target audio.

[0007] This application utilizes a wearable device to receive a wearing instruction, collect the user's physiological data, analyze the user's stress state, and obtain a stress state index. Based on the stress state index, target audio parameters corresponding to the stress state index can be determined, so that the target audio parameters corresponding to users under different stress state indices are different. That is, users under different stress state indices will receive audio content that differs in at least one dimension, such as rhythm, mode, harmonic complexity, timbre, and combination of soundscape elements. Then, target audio is generated and played based on the target audio parameters, so that users under different stress levels can receive audio content corresponding to their current stress level, thereby improving the targeting of audio for user stress regulation and more accurately playing the target audio corresponding to different stress state indices to more accurately regulate user stress.

[0008] In some example embodiments of this application, based on the foregoing scheme, generating target audio based on target audio parameters includes: inputting the target audio parameters into an audio generation model, so that the audio generation model synthesizes an audio segment in real time according to the target audio parameters and outputs the audio segment, and using the audio segment as the target audio.

[0009] If the music used to regulate stress is taken from a pre-recorded or stored fixed music library, then the combination of attributes such as rhythm, mode, and harmony of each track in the library is limited and discrete. When the combination of music attributes required by the user's current stress level falls within the range not covered by the music library, the user can only choose the existing track with the closest attributes to replace it, resulting in a discrepancy between the actual music played and the required combination of attributes. In addition, the number of tracks in the music library is limited, and repetition is likely to occur with long-term use.

[0010] In this embodiment, by inputting the target audio parameters into the audio generation model, the audio generation model synthesizes audio segments in real time based on the target audio parameters and uses them as the target audio. This allows the target audio parameters corresponding to any stress state index to be instantly generated into corresponding audio, no longer limited by the limited parameter combinations of the preset audio library. For example, when the combination of rhythm, mode, and harmonic complexity corresponding to a certain stress state index does not exist in the existing tracks, the audio generation model can still instantly synthesize audio that matches the combination. This ensures that the played audio accurately corresponds to the target audio parameters in terms of rhythm, mode, harmonic complexity, timbre, and soundscape element combinations, further improving the audio's relevance to the user's stress state.

[0011] In some example embodiments of this application, based on the foregoing scheme, generating target audio based on target audio parameters includes: obtaining the initial audio currently being played by the wearable device, and the current audio parameters of the currently playing initial audio; determining parameters to be adjusted based on the target audio parameters and the current audio parameters, wherein the parameters to be adjusted include at least one of rhythm, chords, and timbre; and adjusting the initial audio based on the parameters to be adjusted to obtain the target audio.

[0012] If a new piece of music needs to be synthesized from the entire piece when the music being played needs to be changed, on the one hand, the computational load of synthesizing a complete piece of music is large, and on the other hand, there may be obvious jumps in rhythm, chords or timbre between the newly synthesized music and the music that the user is listening to, resulting in an abrupt change in auditory perception, which will disrupt the continuity of the pressure adjustment process.

[0013] In this embodiment, by acquiring the currently playing initial audio and its current audio parameters, the target audio parameters are compared with the current audio parameters to determine the parameters that need adjustment where there are differences between the two. Only the corresponding parameters of the initial audio are adjusted instead of the entire segment is reconstructed. On the one hand, this reduces the amount of computation in audio generation, and on the other hand, it allows the audio to smoothly transition to the target state based on the original initial audio, avoiding auditory breaks caused by starting a new audio. This ensures that the audio remains continuous during the stress regulation process and does not interrupt the user's relaxation due to audio switching, thereby facilitating the continuous and stable regulation of the user's stress.

[0014] In some example embodiments of this application, based on the aforementioned scheme, a pressure state index corresponds to a pressure level, the pressure level is set with a corresponding pressure threshold, and the pressure intensity corresponding to the pressure level decreases sequentially. The number of audio parameters corresponding to each pressure level is positively correlated with the pressure intensity. Generating target audio based on target audio parameters includes: determining the target pressure level corresponding to the pressure state index; and controlling the wearable device to play the target audio corresponding to the target pressure level.

[0015] If the same number of musical elements are used to organize the music for users with different levels of tension, then for users who are more tense and need stronger guidance, there are not enough musical dimensions to adjust, and the intervention is too weak; while for users who are more relaxed, too many musical elements will make the music too complicated and create an extra burden.

[0016] In this embodiment, by categorizing stress status indicators into stress levels that decrease sequentially in intensity, and making the number of audio parameters corresponding to each stress level positively correlated with the stress level, the target stress level corresponding to the stress status indicator is first determined. Then, the wearable device is controlled to play the target audio corresponding to the target stress level. This allows higher stress levels to utilize more dimensions of audio parameters to organize the target audio, while lower stress levels utilize fewer dimensions. For example, the highest stress level utilizes multiple dimensions such as rhythm, harmony, and soundscape elements, while lower stress levels utilize only a few dimensions. This allows the richness of the audio to be matched to the user's stress level, providing sufficient guidance at high stress levels and keeping the audio concise at low stress levels, further improving the audio's targeted adjustment for users with different stress levels.

[0017] In some example embodiments of this application, based on the aforementioned scheme, the audio parameters include at least one of rhythm, number of harmonic levels, harmonic complexity, and number of soundscape element combinations. The stress level corresponding to the stress state index is positively correlated with the stress level, the stress level is positively correlated with the number of harmonic levels, the stress level is positively correlated with the harmonic complexity, and the stress level is positively correlated with the number of soundscape element combinations.

[0018] When organizing music according to the level of tension, if there is a lack of consistent correspondence between the various musical elements and the user's level of tension, such as using a slower tempo and simpler harmony when the level of tension is high, the overall tone of the music will be difficult to form a stable and predictable correspondence with the user's current level of tension. This will also make it difficult for the music to first match the user's current state and then guide them to gradually calm down.

[0019] In this embodiment, by making the rhythm, number of harmonic levels, harmonic complexity, and number of soundscape element combinations positively correlated with the stress level, the higher the stress level, the faster the audio rhythm, the higher the harmonic levels and complexity, and the more soundscape element combinations, resulting in a richer and fuller overall audio. This initially aligns with the user's current high level of physiological arousal, making it easier for the user to accept. As the stress level decreases, the above dimensions decrease accordingly, and the audio becomes more soothing and concise. This allows the audio to change in the same direction as the stress level in multiple dimensions, forming an auditory guidance that is commensurate with the user's stress level and gradually converges, further improving the audio's targeted approach to user stress regulation.

[0020] In some example embodiments of this application, based on the foregoing scheme, the method further includes: during the playback of target audio, collecting current physiological data of the current cycle window according to a preset sliding time window, and determining the current pressure parameter of the current cycle window based on the current physiological data of the current cycle window; obtaining the historical pressure parameter of the previous cycle window, and determining the user's current physiological change trend based on the current pressure parameter of the current cycle window and the historical pressure parameter of the previous cycle window, wherein the current physiological change trend includes one of a positive improvement trend, a no-change trend, and a reverse deterioration trend; adjusting the target audio parameter according to the physiological change trend to obtain the intervention audio parameter of the next cycle window; adjusting the currently played target audio according to the intervention audio parameter, and playing the adjusted target audio.

[0021] If a piece of music is played in a fixed manner after selection without tracking changes in the user's physical state, the music cannot reflect the real-time evolution of the user's tension during listening. Maintaining the original intensity when the music has calmed the user down, or failing to intervene in time when the user's tension rises instead of falling, will cause the adjustment to deviate from the user's real-time needs. Stress regulation will lack feedback and correction on the adjustment effect.

[0022] In this embodiment, physiological data is periodically collected during playback by sliding time windows, and the current pressure parameter of the current period window is determined. This parameter is then compared with the historical pressure parameter of the previous period window to determine the user's physiological change trend. Based on this trend, the target audio parameter is adjusted to obtain the intervention audio parameter for the next period window, and the target audio being played is adjusted accordingly. This ensures that the audio adjustment is based on the actual changes reflected by the physiological data of the user in the two consecutive period windows. This creates a continuously operating closed loop between "collecting physiological data - assessing the change trend - adjusting the audio - collecting physiological data again". The audio is continuously corrected according to the real-time changes in the user's pressure state, achieving closed-loop dynamic regulation of user pressure and further improving the targeting of audio for user pressure regulation.

[0023] In some example embodiments of this application, based on the aforementioned scheme, both the current stress parameter and the historical stress parameter include at least a heart rate variability parameter and a mean heart rate; determining the user's current physiological change trend based on the current stress parameter of the current cycle window and the historical stress parameter of the previous cycle window includes: comparing the heart rate variability parameter of the current cycle window with the heart rate variability parameter of the previous cycle window to determine the amount of change in the heart rate variability parameter between the current cycle window and the previous cycle window; comparing the mean heart rate of the current cycle window with the mean heart rate of the previous cycle window to determine the amount of change in the mean heart rate of the current cycle window; if the heart rate variability... The change in the parameters indicates that the heart rate variability parameter in the current period window is greater than that in the previous period window, and the change in the mean heart rate indicates that the mean heart rate in the current period window is less than that in the previous period window. In this case, the user's current physiological trend is determined to be a positive improvement trend. If neither the change in the heart rate variability parameter nor the change in the mean heart rate exceeds a preset threshold, the user's current physiological trend is determined to be a no-change trend. If the change in the heart rate variability parameter indicates that the heart rate variability parameter in the current period window is less than that in the previous period window, and the change in the mean heart rate indicates that the mean heart rate in the current period window is greater than that in the previous period window, the user's current physiological trend is determined to be a negative deterioration trend.

[0024] If a user's tension is judged to be improving or worsening based solely on a single physiological quantity (such as the speed of heart rate), it is easy to misinterpret such fluctuations as a relief or worsening of tension, since heart rate can fluctuate due to factors unrelated to stress, such as changes in body position and brief activities, and thus cause the music to be adjusted in the wrong direction.

[0025] In this embodiment, the trend is determined by simultaneously using two indicators: heart rate variability and mean heart rate. The two indicators change in opposite directions with pressure—heart rate variability increases and mean heart rate decreases when pressure eases, and heart rate variability decreases and mean heart rate increases when pressure intensifies. A positive improvement is defined only when heart rate variability increases and mean heart rate decreases in the current cycle window; a negative deterioration is defined when heart rate variability decreases and mean heart rate increases in the current cycle window; and no change is defined when neither of the changes exceeds a preset amplitude threshold. This two-indicator approach mutually corroborates each other, eliminating misjudgments caused by short-term fluctuations in a single indicator. This improves the accuracy and anti-interference capability of determining physiological trend changes, providing a reliable basis for correct audio adjustment and thus enhancing the accuracy of the aforementioned closed-loop dynamic adjustment.

[0026] In some example embodiments of this application, based on the aforementioned scheme, the target audio parameters are adjusted according to the physiological change trend to obtain the intervention audio parameters for the next period window. This includes: if the physiological change trend is a positive improvement trend, the target audio parameters are first adjusted to obtain the intervention audio parameters for the next period window. The first adjustment includes lowering the rhythm, reducing the harmonic complexity, and lowering the volume of environmental sounds in the soundscape elements. If the physiological change trend is no change trend, the target audio parameters are second adjusted to obtain the intervention audio parameters for the next period window. The second adjustment includes adjusting the rhythm within a preset range, switching the current instrument timbre or mode, and increasing the harmonic complexity. If the physiological change trend is a negative deterioration trend, the target audio parameters are third adjusted to obtain the intervention audio parameters for the next period window. The third adjustment includes raising the rhythm, raising the harmonic level, switching the current mode to a major mode, and changing the proportion of natural environmental sounds in the soundscape element combination.

[0027] If a uniform approach to adjusting music is used for three different trends in a user's state of tension—that it tends to improve, remain unchanged, or intensify—it will be impossible to adapt to or reverse the user's current state of tension separately: applying strong intervention when tension has eased may be excessive; lack of new stimulation when the state remains unchanged may easily lead to stagnation; and insufficient intervention when tension intensifies will be difficult to reverse.

[0028] In this embodiment, the first, second, and third adjustments are performed for positive improvement trends, no change trends, and reverse deterioration trends, respectively. When there is positive improvement, the rhythm is lowered, the harmonic complexity is reduced, and the volume of ambient sounds is reduced to conform to the relaxation trend and consolidate the existing effect with more soothing audio. When there is no change, the rhythm is adjusted within a preset range of change, the timbre or mode is switched, and the harmonic complexity is increased to break the stagnation with appropriate fresh stimulation. When there is reverse deterioration, the rhythm is raised, the harmonic level is increased, the mode is switched to a bright major key, and the proportion of natural ambient sounds is changed to reverse the deterioration trend with stronger intervention. This makes the direction of audio adjustment match the user's current physiological change trend, thereby improving the effectiveness of the above-mentioned closed-loop dynamic regulation and more accurately regulating the user's stress.

[0029] In some example embodiments of this application, based on the foregoing scheme, the method further includes: if the physiological change trend is a reverse deterioration trend, shortening the duration of the sliding time window; monitoring the user's physiological change trend under the shortened sliding time window; when it is detected that the user's current physiological change trend changes from a reverse deterioration trend to a positive improvement trend or no change trend, restoring the duration of the sliding time window to the duration before shortening, and switching back to the adjustment operation corresponding to the current physiological change trend.

[0030] If the duration of the monitoring time window remains fixed throughout the process, a longer window will result in a lower frequency of status assessment and music adjustment, a slower response, and a greater chance of missing the intervention opportunity when the user's tension increases and timely correction is urgently needed. On the other hand, if a shorter window is always used, the frequency of body data collection and processing will increase when the user's state is stable, resulting in unnecessary resource consumption.

[0031] In this embodiment, by shortening the sliding time window when a reversal worsening trend is detected, trend assessment and audio adjustment are performed at a higher frequency during the deterioration phase, thereby intervening in the deterioration trend more promptly. After the trend changes from reversal to improvement or no change, the sliding time window is restored to its previous length, and the adjustment operation corresponding to the current trend is switched back, allowing the system to return to a lower monitoring frequency when the user's condition is stable. On the one hand, this allows for higher frequency audio correction during the stress deterioration phase, making the aforementioned closed-loop dynamic adjustment more timely in response to the deterioration trend, which is beneficial for more accurate stress regulation. On the other hand, it avoids the resource overhead of high-frequency monitoring throughout the entire process when the user's condition is stable, balancing the timeliness of stress regulation with operational economy.

[0032] On the one hand, there are individual differences in the degree of heart rate variability and the level of heart rate at rest among different users. If a uniform fixed value is used as the standard for judging whether a user has entered a relaxed state, it is difficult to adapt to individual differences. On the other hand, if music is played at the original volume after the user's tension has been fully relieved, it will not only waste the device's power and computing resources, but the continuous sound output itself may also disturb the user who has become calm.

[0033] In some example embodiments of this application, based on the aforementioned scheme, before collecting physiological data of the current cycle window according to a preset sliding time window during the playback of target audio, the method further includes: when the wearable device is first started, collecting the user's physiological data in a resting state, and calculating the user's physiological baseline value based on the physiological data in a resting state, the physiological baseline value including a heart rate variability baseline value and a heart rate baseline value; determining an optimized target value for the heart rate variability parameter based on the heart rate variability baseline value, and determining an optimized target value for the heart rate based on the heart rate baseline value, the optimized target value for the heart rate variability parameter including the user's heart rate variability parameter being higher than a preset first multiple of the heart rate variability baseline value, and the optimized target value for the heart rate including the user's heart rate being lower than a preset second multiple of the heart rate baseline value, the first multiple being greater than the second multiple; when, without receiving a stop command from the user, it is detected that the user's heart rate variability parameter has reached the optimized target value for the heart rate variability parameter, and the user's heart rate has reached the optimized target value for the heart rate, stopping the playback of audio or reducing the volume of the audio.

[0034] In this embodiment, physiological data at rest is collected when the wearable device is first activated, and the user's individual baseline heart rate variability (HRV) and heart rate are calculated. Optimized target values ​​for HRV and heart rate are then determined by setting a first multiple above the HRV baseline and a second multiple below it, respectively. This allows the standard for judging relaxation level to be established on an individual basis, based on each user's own resting level. Furthermore, if no stop command is received from the user, audio playback stops or the volume is reduced once both HRV and heart rate reach their respective optimized target values, automatically ending the audio playback once the user's stress has been sufficiently relieved. On one hand, by setting optimized target values ​​based on the user's own resting baseline, the stress adjustment target is made individualized, more accurately reflecting the relaxation state that each user can actually achieve. On the other hand, the automatic termination of audio playback after reaching the optimized target value avoids unnecessary resource consumption and continuous disturbance to users who have already calmed down. In addition, setting the first multiple to be greater than the second multiple is in line with the physiological law that heart rate variability has a greater relative range of change than heart rate, making the two optimization target values ​​closer to the relaxation state that users can actually achieve.

[0035] If the pressure regulation process can only be initiated by operating on the wearable device itself, the user's operation is not convenient, given the small size of the wearable device and the limited area available for operation; at the same time, the storage device, which serves as the carrier for the wearable device, is also unable to participate in the regulation process.

[0036] In some example embodiments of this application, based on the foregoing scheme, the method further includes: receiving an activation command from the storage device, wherein the activation command is obtained by the storage device in response to a preset touch operation performed by the user on the touch area; and in response to the activation command, performing the step of collecting the user's physiological data through the sensing module in response to the received wearing command.

[0037] In this embodiment, by setting a touch area on the storage device, the storage device generates an activation command in response to the user's preset touch operation on the touch area. The wearable device then responds to the activation command by performing the step of collecting the user's physiological data. On the one hand, this allows the user to conveniently start the pressure regulation process using the more easily operable touch area on the storage device; on the other hand, it allows the storage device, which originally only served a storage function, to participate in the initiation of the regulation process, forming a closed loop of pressure regulation in collaboration with the wearable device.

[0038] If a user's stress level is judged based on a single and crude indicator such as average heart rate, the physiological information reflected by such indicators is limited and it is difficult to accurately depict the stress level corresponding to autonomic nervous activity, resulting in insufficient accuracy.

[0039] In some example embodiments of this application, based on the aforementioned scheme, the physiological data further includes a heartbeat signal, which includes at least one of an electrocardiogram signal and a photoplethysmography (PPG) signal; based on the physiological data, the user's stress state is analyzed to obtain a stress state index, including: preprocessing the heartbeat signal to obtain time-domain and frequency-domain sub-parameters of the heart rate variability parameter; inputting the time-domain and frequency-domain sub-parameters into a pre-trained stress analysis model to output the user's stress state and obtain a stress state index, wherein the stress analysis model is a classification model trained based on a labeled physiological dataset.

[0040] In this embodiment, by preprocessing cardiac signals such as electrocardiogram (ECG) or photoplethysmography (PPG) signals, time-domain and frequency-domain sub-parameters of heart rate variability (HRV) parameters are extracted. These two types of sub-parameters are then input into a stress analysis classification model trained on a labeled physiological dataset to output stress status. This allows the determination of stress status to utilize the multidimensional information of HRV reflecting autonomic nervous activity in both the time and frequency domains. By using a classification model trained on samples for comprehensive discrimination, the accuracy of stress status indicators in characterizing the user's actual stress level is improved. This provides a more accurate basis for subsequently determining the target audio parameters that match the stress level, thereby improving the targeting of audio for user stress regulation.

[0041] According to a second aspect of the embodiments of this application, a closed-loop pressure regulation system is provided, including a storage device and a wearable device, wherein the storage device is used to house the wearable device; the wearable device is configured to perform an audio processing method as described in any of the first aspects.

[0042] Because the storage device in this system houses the wearable device, the two are physically integrated, facilitating pairing and storage. The collection of physiological data, analysis of stress levels, and generation and playback of target audio are all handled by the wearable device, which fits snugly against the user's body. This can be combined with touch activation and other interactive features from the storage device. The storage device and wearable device work together via communication, forming a closed loop from physiological data collection and stress level analysis to target audio generation and playback and physiological feedback adjustment. On one hand, the wearable device can generate and dynamically adjust target audio based on the user's real-time stress level, ensuring that users at different stress levels receive audio content corresponding to their current stress level, thus improving the audio's targeted approach to stress regulation. On the other hand, the storage device's integration of the wearable device with the storage device facilitates pairing and storage and allows it to participate in the initiation of the adjustment process, ensuring the stable operation of the aforementioned closed-loop stress regulation.

[0043] According to a third aspect of the embodiments of this application, a computer device is provided, including: a processor; and a memory storing computer-readable instructions, which, when executed by the processor, implement an audio processing method as described in any of the first aspects.

[0044] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0045] 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.

[0046] Figure 1 The illustration shows a schematic flowchart of an audio processing method provided in an exemplary embodiment of this application.

[0047] Figure 2 The illustration shows a flowchart of an audio processing method provided in another exemplary embodiment of this application.

[0048] Figure 3 The schematic diagram illustrates the structure of a closed-loop pressure regulation system provided in an exemplary embodiment of this application.

[0049] In the accompanying drawings, the same or corresponding reference numerals indicate the same or corresponding parts. Detailed Implementation

[0050] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided to make this application more comprehensive and complete, and to fully convey the concept of the exemplary embodiments to those skilled in the art.

[0051] Furthermore, the described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. Numerous specific details are provided in the following description to give a thorough understanding of embodiments of this application. However, those skilled in the art will recognize that the technical solutions of this application can be practiced without one or more of the specific details, or other methods, components, apparatuses, steps, etc., can be employed. In other instances, well-known methods, apparatuses, implementations, or operations are not shown or described in detail to avoid obscuring various aspects of this application.

[0052] Furthermore, the accompanying drawings are for illustrative purposes only and are not necessarily drawn to scale. The block diagrams shown in the drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0053] With the fast pace of life and work, monitoring and regulating users' stress levels using wearable devices is gradually becoming an application area for wearable device health functions. These wearable devices are typically worn close to the user's body, such as headphones worn in the ear, and can collect physiological data such as heart rate and heart rate variability during daily use, combined with audio playback functions to provide users with relaxing, sleep-aiding, or stress-relieving auditory content. To facilitate charging and storage, wearable devices are usually equipped with storage devices, such as charging cases.

[0054] In related technologies, wearable devices typically play fixed audio clips from a preset audio library or audio from a preset playlist to the user. Since a user's stress levels may vary at different times while using the wearable device, and the content of the fixed audio clips or preset playlists is relatively fixed, there may be a mismatch between the played audio and the user's actual stress level, resulting in insufficient targeting of the stress regulation process.

[0055] Therefore, in some embodiments of this application, an audio processing method, a closed-loop pressure regulation system, and a computer device are provided. The overall architecture of the closed-loop pressure regulation system upon which the audio processing method relies will be described first, followed by a description of the steps of the audio processing method in conjunction with the accompanying drawings.

[0056] See Figure 2 , Figure 2 The illustration schematically depicts a closed-loop pressure regulation process provided in an exemplary embodiment of this application. This process can be applied to a closed-loop pressure regulation system including a storage device and a wearable device, wherein the wearable device can be an in-ear headphone, and the storage device can be a charging case for housing the wearable device. The process may include steps S201 to S206.

[0057] In step S201, the user performs a preset touch operation through the touch area of ​​the storage device, triggering the pressure adjustment process. The preset touch operation can be continuous tapping of the touch area, long pressing of the touch area, or other touch operations that can be recognized by the storage device. After recognizing the preset touch operation, the storage device can generate a trigger command.

[0058] In step S202, the wearable device responds to a trigger command and activates the PPG sensor used for collecting physiological signals. Specifically, the storage device can send the trigger command to the wearable device via a Bluetooth Low Energy communication link; after receiving the trigger command, the wearable device activates the PPG sensor in the sensing module, so that the PPG sensor enters the physiological signal collection state.

[0059] In step S203, the user's heartbeat signal is acquired using a PPG sensor, and HRV data is obtained based on the heartbeat signal. Specifically, the PPG sensor can acquire the user's photoplethysmography (PPG) signal, and the processing unit of the wearable device can preprocess the PPG signal and obtain heart rate variability data based on the preprocessed heartbeat signal.

[0060] In step S204, the user's current stress state is analyzed based on HRV data to obtain a stress state index. Specifically, the wearable device can determine the user's current stress level, stress score, or relaxation score based on the time-domain and / or frequency-domain parameters in the HRV data, and use these as a stress state index.

[0061] In step S205, the corresponding AI audio is generated or adjusted based on the stress state index. Specifically, the wearable device can determine the corresponding target audio parameters based on the stress state index and generate AI audio based on the target audio parameters; if audio has already been played, the wearable device can also adjust the currently playing AI audio based on the new stress state index.

[0062] In step S206, the generated or adjusted AI audio is played through the wearable device to provide audio intervention to the user. Specifically, the wearable device can play the generated or adjusted AI audio through its audio playback module to apply auditory intervention to the user.

[0063] During audio intervention, the wearable device can continuously perform cyclical feedback. That is, after playing the AI ​​audio, it returns to step S203 to continue collecting the user's heart rate signal through the PPG sensor, and obtains new HRV data based on the newly collected heart rate signal. Then, it re-analyzes the stress state based on the new HRV data, generates or adjusts the AI ​​audio, and continues playing. In this way, a closed-loop regulation process of "collecting HRV data - analyzing stress state - generating or adjusting AI audio - playing audio intervention - collecting HRV data again" can be formed, enabling the AI ​​audio to dynamically adjust with changes in the user's physiological state.

[0064] In some embodiments, the touch area may be formed by a touch sensor disposed on the surface of the storage device. The touch sensor is used to detect a triple tap, double tap, or long press operation performed by the user on the touch area, and sends the detection result to the interaction control module, so that the interaction control module generates an activation command based on the detection result.

[0065] The wearable device and the storage device are connected via a communication link. In some embodiments, this communication link is a Bluetooth Low Energy (BLE) link; in other embodiments, it can also be Bluetooth Classic, Near Field Communication (NFC), or other wireless communication links, and this application is not limited to these. Physiological data collected by the wearable device and activation commands generated by the storage device can be transmitted between the two devices via this communication link. It is understood that the physiological data collection, stress state analysis, target audio generation and playback by the wearable device, and touch activation by the storage device can all be completed independently between the wearable device and the storage device, without the need for mandatory binding to an application on an external terminal such as a mobile phone, thereby lowering the barrier to entry and enhancing scenario adaptability.

[0066] Wearable devices refer to devices that can be worn on a user's body to collect the user's physiological data and play audio. In some embodiments, a wearable device is an in-ear headphone worn on the user's ear; in other embodiments, a wearable device may also be a headset, a neckband device, or other wearable devices capable of collecting physiological data and playing audio, and is not limited thereto. Storage devices refer to devices used to house wearable devices and capable of charging or storing them.

[0067] In some embodiments, in-ear headphones are used simultaneously as a physiological signal acquisition carrier and an audio intervention carrier. That is, the in-ear headphones acquire the user's heartbeat signals, heart rate data, or heart rate variability-related data through a sensing module, and simultaneously play target audio generated or adjusted based on stress state indicators through an audio playback module. Thus, physiological data acquisition and auditory intervention can be completed on the same wearable device, eliminating the need to separate the physiological monitoring device and the audio playback device.

[0068] In some embodiments, the storage device is a charging case that comes with the earphones; in other embodiments, the storage device may also be a storage box, a base, or other device capable of accommodating the wearable device, and is not limited thereto. The storage device has an internal cavity in which the wearable device can be placed for storage or charging.

[0069] The sensing module refers to a module located in the wearable device for collecting user physiological data. In some embodiments, the sensing module includes a photoplethysmography (PPG) sensor, which can be attached to the user's ear canal or auricle to collect the user's heartbeat signal; in other embodiments, the sensing module may also include an electrocardiogram (ECG) sensor or other sensors capable of collecting heartbeat signals, and is not limited thereto. The physiological data collected by the sensing module is transmitted to the processing unit of the wearable device for subsequent processing. The processing unit refers to a unit located in the wearable device for performing processes such as stress state analysis and audio generation. In some embodiments, the processing unit is a microcontroller unit or an embedded chip; in other embodiments, the processing unit may also be a digital signal processor or other processing device with computing power, and is not limited thereto. The aforementioned virtual modules, such as the sensing module, processing unit, audio playback module, and interactive control module, can all be implemented by corresponding to specific circuits, chips, or processing units in the wearable device or storage device. This application does not limit the specific physical implementation of each module.

[0070] See Figure 1 This is a flowchart illustrating an exemplary embodiment of an audio processing method provided in this application. The audio processing method is applied to a wearable device, which is communicatively connected to a storage device. The wearable device is equipped with a sensing module. The method may include the following steps S101 to S104: Step S101: In response to the received wearing instruction, the user's physiological data is collected through the sensing module. The wearing instruction is triggered by the user wearing the wearable device. The physiological data includes at least one of the user's heart rate data, respiratory rate data, and sleep state parameters. Step S102: Based on physiological data, analyze the user's stress state to obtain stress state indicators; Step S103: Based on the stress state index, determine the target audio parameters corresponding to the stress state index. The stress state index is used to characterize the mapping relationship between the user's stress level and the audio parameters. The audio parameters include at least one of rhythm, mode, harmonic complexity, timbre and soundscape element combination. Step S104: Generate target audio based on target audio parameters and play the target audio.

[0071] The wearing instruction refers to an instruction used to indicate that the user has worn the wearable device and intends to begin pressure adjustment. In some embodiments, after the user wears the wearable device, the wearing instruction is triggered by double-clicking the wearable device, long-pressing a touch area on the wearable device, or by the wearable device automatically detecting the wearing status when the wearable device is removed from the storage cavity of the storage device and worn. In other embodiments, the wearing instruction can also be triggered by the user performing a preset operation on the storage device. This application does not limit the specific triggering method of the wearing instruction.

[0072] Physiological data refers to data that characterizes the user's current physical state, and in this application, it includes at least one of the user's heart rate data, respiratory rate data, and sleep state parameters. Specifically, heart rate data refers to data characterizing the number of heartbeats per unit time; respiratory rate data refers to data characterizing the number of breaths per unit time; and sleep state parameters refer to parameters characterizing the user's current sleep depth or sleep stage. In some embodiments, physiological data further includes the user's heartbeat signal, which includes at least one of electrocardiogram (ECG) signals and photoplethysmography (PPG) signals.

[0073] The stress state index refers to an indicator obtained by quantifying the user's current stress state, used to characterize the mapping relationship between the user's stress level and audio parameters. In some embodiments, the stress state index is reflected as a graded stress level, such as dividing stress into several levels from high to low; in other embodiments, the stress state index can also be reflected as a continuously measured quantitative score, such as a relaxation score between 0 and 1, where a lower score indicates a higher level of user stress, and a higher score indicates a more relaxed user. This application does not limit the specific quantification form of the stress state index.

[0074] Audio parameters refer to parameters used to characterize the auditory properties of the generated audio, including at least one of rhythm, mode, harmonic complexity, timbre, and soundscape element combination. Rhythm can be characterized by beats per minute (BPM); mode refers to the tonal system used in the audio, such as major or minor; harmonic complexity refers to the richness of harmony in the audio, such as the number of harmonic levels or the complexity of chord arrangements; timbre refers to the type of instrument or sound used in the audio; and soundscape element combination refers to the combination of ambient or background sounds superimposed on the audio, such as a single instrument or a combination of an instrument and natural ambient sounds.

[0075] In step S101, after receiving the wearing instruction, the wearable device collects the user's physiological data through the sensing module. In some embodiments, after the PPG sensor in the sensing module is activated, it collects the user's heartbeat signal in real time and continuously. The heartbeat signal includes at least the electrocardiogram signal or photoplethysmography (PPG) signal required to calculate heart rate variability (HRV). The collected signal is then transmitted to the processing unit for further processing.

[0076] In step S102, the wearable device analyzes the user's stress state based on the collected physiological data to obtain a stress state index. In step S103, the wearable device determines the corresponding target audio parameters based on the stress state index. In step S104, the wearable device generates target audio based on the target audio parameters and plays the target audio through the audio playback module.

[0077] In this embodiment, the wearable device, in response to a received wearing command, collects the user's physiological data and analyzes the user's stress state to obtain a stress state index. Based on this index, it determines the target audio parameters corresponding to the stress state index. Since the stress state index characterizes the mapping relationship between the user's stress level and the audio parameters, users under different stress state indices will receive different target audio parameters. Target audio is then generated and played based on these parameters, allowing users at different stress levels to receive audio content corresponding to their current stress level, thereby improving the targeted nature of the audio for stress regulation. Furthermore, the audio parameters are not pre-set but are determined and generated based on the user's real-time physiological data. Compared to playing generic template audio, this approach better matches the user's individual needs, overcoming the problems of generalized intervention content and lack of dynamic adjustment based on the user's current stress state in related technologies, thus improving the personalization of stress regulation. Moreover, since the target audio is generated in real-time according to the user's current physiological state and continuously adjusted based on changes in the user's physiological state during the closed-loop feedback process, the audio received by the user in different regulation cycles is not a repetitive playback of a fixed template but rather dynamic audio content related to the user's own state. This can reduce auditory fatigue caused by repeated playback of fixed audio and increase users' willingness to continue using the pressure adjustment function.

[0078] Meanwhile, since each round of audio adjustment is based on the physiological change trend after the previous round of audio playback, the target audio parameters can be continuously corrected along the direction corresponding to the user's current actual feedback, thereby reducing the time of continuous playback of ineffective audio intervention and improving the intervention efficiency in the stress regulation process.

[0079] According to some exemplary embodiments of this application, based on the foregoing scheme, step S104, which generates target audio based on target audio parameters, may include: inputting the target audio parameters into an audio generation model so that the audio generation model synthesizes an audio segment in real time according to the target audio parameters and outputs the audio segment, and using the audio segment as the target audio.

[0080] The audio generation model refers to a model used to generate audio based on target audio parameters. In some embodiments, the audio generation model employs a lightweight embedded Transformer generation model, which includes an input layer, an encoding layer, and an audio generation layer. The input layer encodes the acoustic parameters corresponding to the target audio parameters into feature vectors, for example, 32-dimensional feature vectors. The encoding layer encodes parameter-related features from the feature vectors. In some embodiments, the encoding layer includes two sets of cascaded Transformer encoding units, each set of Transformer encoding units having four attention heads in its multi-head attention mechanism, and the hidden layer dimension of the feedforward network can be 64. The audio generation layer generates audio based on the encoded features. In some embodiments, the audio generation layer combines a lightweight WaveNet vocoder to output an audio segment, such as an audio segment with a sampling rate of 44.1 kHz and a bit depth of 16 bits. It is understood that the number of encoding units, the dimension of the feature vector, the number of attention heads, the dimension of the feedforward network, the sampling rate, and the bit depth are all exemplary values, and those skilled in the art can select other values ​​according to actual needs, and are not limited thereto.

[0081] In some embodiments, the audio segment output by the audio generation layer can be a lossless audio segment. The audio generation module can also output the target audio as a PCM audio stream and send the PCM audio stream to the local player of the wearable device, whereby the local player plays the PCM audio stream in real time, thereby applying auditory intervention to the user through the headphone unit.

[0082] In some embodiments, the audio generation model generates a native audio segment of a preset duration in one go based on the target audio parameters, such as generating an audio segment of about 30 seconds in one go, with a synthesis delay of no more than about 1 second. In other embodiments, the audio generation model may also be other pre-trained generative neural network models, or a parametric synthesizer may be used to directly synthesize audio based on the target audio parameters. This application does not limit the specific implementation of the audio generation model.

[0083] In this embodiment, by inputting the target audio parameters into the audio generation model, the model synthesizes audio segments in real time based on the target audio parameters and uses them as the target audio. This allows any target audio parameter corresponding to a stress state index to be instantly generated into the corresponding audio, no longer limited by the finite and discrete parameter combinations of a pre-set audio library. This ensures that the played audio precisely corresponds to the target audio parameters. Furthermore, the audio generation model employs a lightweight embedded design, enabling audio generation to be completed on the embedded chip with limited computing power in wearable devices, while meeting the latency requirements of real-time synthesis. Therefore, personalized audio corresponding to the user's current stress state can be generated locally on the device in real time without relying on an external terminal.

[0084] According to some exemplary embodiments of this application, based on the foregoing scheme, step S104, generating target audio based on target audio parameters, may further include: obtaining the initial audio currently being played by the wearable device, and the current audio parameters of the currently playing initial audio; determining parameters to be adjusted based on the target audio parameters and the current audio parameters, wherein the parameters to be adjusted include at least one of rhythm, chord, and timbre; and performing audio adjustment on the initial audio based on the parameters to be adjusted to obtain the target audio.

[0085] Here, the initial audio refers to the audio currently being played by the wearable device, and the current audio parameters refer to the audio parameters corresponding to the initial audio. In some embodiments, the wearable device compares the target audio parameters with the current audio parameters of the initial audio item by item, and determines the audio parameters that differ between the two as parameters to be adjusted. For example, when there is a difference in rhythm between the target audio parameters and the current audio parameters, the rhythm is determined as a parameter to be adjusted. In some embodiments, when the wearable device adjusts the initial audio, it adopts a frame-level dynamic correction method, for example, using an adjustment duration of approximately 0.5 seconds per frame, and only gradually fine-tuning the rhythm, chords, timbre, and other adjustment parameters of the initial audio without interrupting the current audio playback stream. In some embodiments, the audio adjustment process uses an interpolation algorithm to achieve a transition between adjacent frames, so that the parameters to be adjusted change continuously between adjacent frames, thereby achieving a seamless transition. It is understood that the above-mentioned single-frame adjustment duration and specific transition algorithms are exemplary and not limited thereto.

[0086] In this embodiment, by adjusting only the parameters that differ from the target audio parameters in the initial audio, rather than regenerating the entire audio segment, the computational load of audio generation is reduced, and the audio smoothly transitions to the target state based on the original initial audio. Furthermore, by using frame-level dynamic correction combined with interpolation algorithms to achieve seamless transitions between adjacent frames, the audio playback stream is not interrupted during adjustment, and popping sounds and rhythmic breaks are avoided. This prevents abrupt audio switching for the user and maintains audio continuity during pressure adjustment.

[0087] According to some exemplary embodiments of this application, based on the aforementioned scheme, a pressure state index corresponds to a pressure level, the pressure level is set with a corresponding pressure threshold, and the pressure degree corresponding to the pressure level decreases sequentially. The number of audio parameters corresponding to each pressure level is positively correlated with the pressure degree. Step S104, which generates target audio based on target audio parameters, may include: determining the target pressure level corresponding to the pressure state index; and controlling the wearable device to play the target audio corresponding to the target pressure level.

[0088] In this context, pressure level refers to the classification of pressure states based on their intensity. In some embodiments, pressure levels are divided into multiple levels, such as five levels. Each pressure level has a corresponding pressure threshold. The wearable device compares the pressure state index with the pressure thresholds of each pressure level to determine the target pressure level to which the pressure state index falls. The intensity of pressure varies across different pressure levels. The number of audio parameters corresponding to each pressure level is positively correlated with the pressure intensity. For example, higher pressure levels correspond to a larger number of audio parameters, simultaneously utilizing multiple dimensions such as rhythm, harmony, and soundscape elements to organize the audio; lower pressure levels correspond to a smaller number of audio parameters, using only some of these dimensions. In some embodiments, the wearable device pre-configures the correspondence between each pressure level and the target audio parameters, for example, storing it in the form of a graded mapping table. The wearable device uses the pressure level as a criterion, retrieves this graded mapping table to assign values ​​to the target audio parameters, and then controls the playback of the target audio corresponding to the target pressure level.

[0089] In this embodiment, by categorizing stress status indicators into different stress levels and making the number of audio parameters corresponding to each stress level positively correlated with the stress level, the richness of the audio is matched to the user's stress level. This provides sufficient guidance when the stress level is high and maintains audio simplicity when the stress level is low. Furthermore, fewer audio parameter dimensions are used when the stress level is low, correspondingly reducing the number of parameters involved in audio generation and adjustment, which helps to reduce the computational burden on the wearable device.

[0090] According to some exemplary embodiments of this application, based on the aforementioned scheme, the audio parameters include at least one of rhythm, number of harmonic levels, harmonic complexity, and number of soundscape element combinations. The stress level corresponding to the stress state index is positively correlated with the stress level, the stress level is positively correlated with the number of harmonic levels, the stress level is positively correlated with the harmonic complexity, and the stress level is positively correlated with the number of soundscape element combinations.

[0091] In some embodiments, the wearable device is pre-configured with a graded mapping table between pressure levels and various audio parameters. For example, the pressure levels are divided into 1 to 5, with the pressure intensity represented by levels 1 to 5 increasing sequentially; the BPM corresponding to the rhythm increases with the pressure intensity, with the BPM corresponding to each pressure level falling sequentially in the range of approximately 60 to 110, meaning the higher the pressure intensity, the higher the BPM and the faster the rhythm; each pressure level uses a major key harmonic system; the higher the pressure intensity, the more harmonic levels and the more complex the chord arrangement; the soundscape element combination gradually superimposes natural ambient sounds from a single instrument, meaning the higher the pressure intensity, the more soundscape element combinations. It is understood that the number of pressure levels, BPM ranges, keys, number of harmonic levels, and specific values ​​and combinations of soundscape elements described above are all exemplary, and those skilled in the art can select other values ​​according to actual needs, and are not limited thereto.

[0092] In this embodiment, by making the rhythm, number of harmonic levels, harmonic complexity, and number of soundscape element combinations positively correlated with the stress level, the higher the stress level, the faster the audio rhythm, the higher the harmonic levels and complexity, and the more soundscape element combinations, resulting in a richer and fuller overall audio. This initially aligns with the user's current high level of physiological arousal, making it easier for the user to accept. As the stress level decreases, the above dimensions decrease accordingly, and the audio becomes more soothing and concise, thus forming an auditory guidance that is commensurate with the user's stress level and gradually converges.

[0093] According to some exemplary embodiments of this application, based on the foregoing scheme, the method may further include: during the playback of target audio, collecting current physiological data of the current cycle window according to a preset sliding time window, and determining the current pressure parameter of the current cycle window based on the current physiological data of the current cycle window; obtaining the historical pressure parameter of the previous cycle window, and determining the user's current physiological change trend based on the current pressure parameter of the current cycle window and the historical pressure parameter of the previous cycle window, wherein the current physiological change trend includes one of a positive improvement trend, a no-change trend, and a reverse deterioration trend; adjusting the target audio parameter according to the physiological change trend to obtain the intervention audio parameter of the next cycle window; adjusting the currently played target audio according to the intervention audio parameter, and playing the adjusted target audio.

[0094] The sliding time window refers to a time window used to periodically collect and statistically analyze physiological data; one sliding time window corresponds to one cycle window. In some embodiments, the duration of the sliding time window is approximately 5 seconds. While playing the target audio, the wearable device continuously collects physiological data within the current cycle window and determines the current stress parameter for that window. The current stress parameter is a parameter derived from the physiological data within the current cycle window, characterizing the user's current stress level. The historical stress parameter is the corresponding parameter derived from the physiological data within the previous cycle window. The physiological trend refers to the direction of change in the user's stress level over time. In some embodiments, the wearable device uses a comparative difference method to compare the current stress parameter of the current cycle window with the historical stress parameter of the previous cycle window, determining the user's current physiological trend based on the difference. The intervention audio parameter refers to the audio parameter used to generate or adjust the audio played in the next cycle window after adjusting the target audio parameter according to the physiological trend.

[0095] In some embodiments, after playing the target audio, the wearable device returns to the step of collecting physiological data, continuously monitors the changes in the user's physiological data under auditory intervention, and continuously adjusts the audio played subsequently, thereby forming a continuously operating closed loop between physiological data collection, stress state analysis, target audio generation and playback, and audio adjustment based on physiological feedback.

[0096] In this embodiment, physiological data is periodically collected and current stress parameters are determined by sliding time windows during the playback of target audio. These parameters are then compared with historical stress parameters from the previous period to determine the trend of physiological changes. Based on this comparison, the target audio parameters are adjusted to obtain intervention audio parameters, and the played audio is adjusted accordingly. This ensures that audio adjustments are based on the actual changes reflected in the user's physiological data from the two consecutive period windows, thus forming a closed loop between data collection, analysis, playback generation, and feedback adjustment. This achieves closed-loop dynamic regulation of the user's stress. Furthermore, the generation and playback of audio, along with subsequent physiological monitoring and audio adjustment, are continuously and seamlessly integrated locally on the wearable device. This eliminates the separation between monitoring and intervention, enabling real-time linkage and overcoming the problems of separation between monitoring and intervention and insufficient real-time intervention in related technologies, thereby improving the immediacy of stress regulation.

[0097] According to some exemplary embodiments of this application, based on the aforementioned scheme, both the current stress parameters and historical stress parameters include at least heart rate variability parameters and mean heart rate; determining the user's current physiological change trend based on the current stress parameters of the current cycle window and the historical stress parameters of the previous cycle window may include: comparing the heart rate variability parameters of the current cycle window with those of the previous cycle window to determine the amount of change in the heart rate variability parameters between the two; comparing the mean heart rate of the current cycle window with that of the previous cycle window to determine the amount of change in the mean heart rate between the two; if the current cycle window... If the heart rate variability parameter of the current period is greater than that of the previous period window, and the mean heart rate of the current period window is less than that of the previous period window, then the user's current physiological trend is determined to be a positive improvement trend. If neither the change in the heart rate variability parameter nor the change in the mean heart rate exceeds a preset amplitude threshold, then the user's current physiological trend is determined to be a no-change trend. If the heart rate variability parameter of the current period window is less than that of the previous period window, and the mean heart rate of the current period window is greater than that of the previous period window, then the user's current physiological trend is determined to be a negative deterioration trend.

[0098] Here, heart rate variability (HRV) refers to a parameter characterizing the variability of a user's heart rate, and mean heart rate refers to the average heart rate within a given period. In some embodiments, a positive improvement trend corresponds to a state where user stress is easing, in which case the user's HRV increases, mean heart rate decreases, and the values ​​approach the optimization target value; a negative deterioration trend corresponds to a state where user stress is increasing, in which case the user's HRV decreases, mean heart rate increases; and no change trend corresponds to a state where user stress is relatively stable, in which case the fluctuation range of HRV and mean heart rate is small. In some embodiments, a preset amplitude threshold can be set to approximately 5%, that is, when the change range of both HRV and mean heart rate does not exceed approximately 5%, it is determined to be a state of no change trend. It should be understood that the above preset amplitude threshold is an exemplary value and is not limited thereto.

[0099] In this embodiment, the physiological trend is determined by simultaneously using two indicators: heart rate variability and mean heart rate. The two indicators change in opposite directions with stress—heart rate variability increases and mean heart rate decreases when stress is relieved, and heart rate variability decreases and mean heart rate increases when stress is aggravated—so that the two indicators can be used to corroborate each other. This eliminates misjudgments caused by short-term fluctuations of a single indicator due to factors such as changes in body position or brief activities, improves the accuracy of determining the physiological trend, and provides a reliable basis for the correct adjustment of audio.

[0100] According to some exemplary embodiments of this application, based on the aforementioned scheme, adjusting the target audio parameters according to the physiological change trend to obtain the intervention audio parameters for the next cycle window may include: if the physiological change trend is a positive improvement trend, then a first adjustment is made to the target audio parameters to obtain the intervention audio parameters for the next cycle window, the first adjustment including lowering the rhythm, reducing the harmonic complexity, and lowering the volume of environmental sounds in the soundscape elements; if the physiological change trend is no change trend, then a second adjustment is made to the target audio parameters to obtain the intervention audio parameters for the next cycle window, the second adjustment including adjusting the rhythm within a preset range, switching the current instrument timbre or mode, and increasing the harmonic complexity; if the physiological change trend is a negative deterioration trend, then a third adjustment is made to the target audio parameters to obtain the intervention audio parameters for the next cycle window, the third adjustment including raising the rhythm, raising the harmonic level, switching the current mode to a major mode, and changing the proportion of natural environmental sounds in the soundscape element combination.

[0101] In some embodiments, the first adjustment is an adjustment that consolidates existing effects with more soothing audio to align with the relaxation trend. For example, the first adjustment may be: maintaining the current key, decreasing the BPM by approximately 2 to 3 units per cycle, reducing the first-order harmonic complexity, and decreasing the volume of ambient sounds by approximately 5%, so that the audio smoothly approaches the parameters of relaxing audio. In some embodiments, the second adjustment is an adjustment that breaks the stagnation with moderate fresh stimulation. For example, the second adjustment may be: ensuring that the BPM fluctuation does not exceed approximately ±3 units, switching the current instrument timbre or key, increasing the first-order harmonic complexity, and fine-tuning the proportion of soundscape elements to strengthen the intervention stimulation. In some embodiments, the third adjustment is an adjustment that reverses the deteriorating trend with stronger intervention. For example, the third adjustment may be: increasing the BPM by approximately 5 to 8 units, increasing the harmonic level, switching the current key to a more soothing and bright major key, and replacing it with a low-stimulation natural soundscape. It is understood that the adjustment range of BPM, the number of adjustment levels of harmonic complexity or harmonic hierarchy, and the adjustment ratio of volume in the above adjustments are all exemplary values. Those skilled in the art can select other values ​​according to actual needs, and are not limited to these.

[0102] In this embodiment, by performing the first adjustment, the second adjustment, and the third adjustment respectively for positive improvement trends, no change trends, and reverse deterioration trends, the direction of audio adjustment is adapted to the user's current physiological change trend: when there is positive improvement, more soothing audio is used to consolidate the existing effect; when there is no change, moderate new stimulation is used to break the stagnation; and when there is reverse deterioration, stronger intervention is used to reverse the deterioration trend, thereby improving the effectiveness of closed-loop dynamic regulation and more accurately regulating the user's stress.

[0103] According to some exemplary embodiments of this application, based on the foregoing scheme, the method may further include: if the physiological change trend is a reverse deterioration trend, shortening the duration of the sliding time window; monitoring the user's physiological change trend within the shortened sliding time window; when the user's current physiological change trend is detected to change from a reverse deterioration trend to a positive improvement trend or no change trend, restoring the duration of the sliding time window to the duration before shortening, and switching back to the adjustment operation corresponding to the current physiological change trend.

[0104] In some embodiments, the sliding time window is approximately 5 seconds during normal adjustment. When a reversal trend is detected, the sliding time window is compressed to approximately 3 seconds to increase the frequency of sampling and trend determination. After the user's physiological trend changes from reversal to improvement or no change, the sliding time window is restored from approximately 3 seconds to approximately 5 seconds, and the adjustment operation corresponding to the current trend is switched back. It should be understood that the window durations before and after the shortening are exemplary values ​​and are not limited to these.

[0105] In this embodiment, by shortening the sliding time window when a reversal deterioration trend is detected, trend assessment and audio adjustment are performed at a higher frequency during the deterioration phase, thereby intervening in the deterioration trend more promptly. The sliding time window duration is restored after the trend turns positive or remains unchanged, allowing the system to return to a lower monitoring frequency when the user's condition stabilizes. On the one hand, this allows for higher frequency audio correction during the stress deterioration phase, making the closed-loop dynamic adjustment respond more promptly to the deterioration trend. On the other hand, it avoids the resource overhead of high-frequency monitoring throughout the entire process when the user's condition stabilizes, balancing the timeliness of stress adjustment with operational economy.

[0106] According to some exemplary embodiments of this application, based on the aforementioned scheme, before collecting physiological data of the current cycle window according to a preset sliding time window during the playback of target audio, the method may further include: when the wearable device is first started, collecting the user's physiological data in a resting state, and calculating the user's physiological baseline value based on the physiological data in the resting state, the physiological baseline value including a heart rate variability baseline value and a heart rate baseline value; determining an optimized target value for the heart rate variability parameter based on the heart rate variability baseline value, and determining an optimized target value for the heart rate based on the heart rate baseline value, the optimized target value for the heart rate variability parameter including the user's heart rate variability parameter being higher than a preset first multiple of the heart rate variability baseline value, and the optimized target value for the heart rate including the user's heart rate being lower than a preset second multiple of the heart rate baseline value, the first multiple being greater than the second multiple; when, without receiving a stop command from the user, it is detected that the user's heart rate variability parameter has reached the optimized target value for the heart rate variability parameter, and the user's heart rate has reached the optimized target value for the heart rate, stopping the playback of audio or reducing the volume of the audio.

[0107] The physiological baseline value refers to a reference value reflecting the user's resting level. In some embodiments, the wearable device collects physiological data from the user for approximately 3 minutes at rest upon initial startup, and calculates the user's individual heart rate variability baseline value and heart rate baseline value as the user's personal baseline. The optimization target value refers to the target value used to determine whether the user has reached a relaxed state. In some embodiments, the optimization target value for the heart rate variability parameter is that the user's heart rate variability parameter is not lower than approximately 1.1 times the heart rate variability baseline value, and the optimization target value for the heart rate is that the user's heart rate is not higher than approximately 0.9 times the heart rate baseline value, i.e., the first multiple is approximately 1.1 and the second multiple is approximately 0.9, with the first multiple being greater than the second multiple. In some embodiments, the user stop command can be a command triggered after the user performs a preset touch operation on the touch area of ​​the storage device again; in other embodiments, stopping audio playback or reducing audio volume can also be triggered after detecting that the heart rate variability parameter and heart rate simultaneously reach their respective optimization target values ​​and remain stable for a period of time. It is understood that the above-mentioned resting collection duration, the first multiple, and the second multiple are all exemplary values ​​and are not limited thereto.

[0108] In this embodiment, resting physiological data is collected upon initial device startup of the wearable device, and the user's individual baseline heart rate variability (HRV) and heart rate are calculated. Optimization target values ​​are then determined by setting a first multiple above the HRV baseline and a second multiple below the HRV baseline, respectively. This ensures that the standard for judging relaxation level is tailored to each user's individual resting state. Furthermore, if no stop command is received from the user, audio playback stops or the volume is reduced once both HRV and heart rate reach their respective optimization target values. This allows the audio to automatically terminate after the user's stress has been sufficiently relieved, thus achieving personalized stress management while avoiding unnecessary resource consumption and continuous disturbance to a user who has already calmed down. Additionally, setting the first multiple to be greater than the second multiple aligns with the physiological principle that HRV has a larger relative variation than heart rate, making the two optimization target values ​​closer to the user's actual achievable relaxation state.

[0109] According to some exemplary embodiments of this application, based on the foregoing scheme, the method may further include: receiving an activation command from a storage device, wherein the activation command is obtained by the storage device in response to a preset touch operation performed by a user on a touch area; and in response to the activation command, performing the step of collecting the user's physiological data through a sensing module in response to a received wearing command.

[0110] The touch area refers to the area located on the storage device for receiving user touch operations, such as a touch sensing area on the surface of the storage device. The preset touch operation refers to a pre-set touch operation used to trigger an activation command. In some embodiments, the preset touch operation is a three-hit operation on the touch area. After recognizing the three-hit operation, the interaction control module of the storage device generates an activation command and sends the activation command to the wearable device via a communication link. In other embodiments, the preset touch operation may also be a double-click, long press, or other pre-set touch operation; this application does not limit the specific form of the preset touch operation. After receiving the activation command, the activation sensing module of the wearable device begins to collect the user's physiological data.

[0111] In this embodiment, by setting a touch area on the storage device, the storage device generates an activation command in response to a preset touch operation performed by the user on the touch area. The wearable device then begins to collect the user's physiological data in response to the activation command. On the one hand, this allows the user to conveniently initiate the pressure regulation process using the more easily operable touch area on the storage device, and allows the storage device, which originally only served a storage function, to participate in the initiation of the regulation process, forming a closed loop of pressure regulation in collaboration with the wearable device. On the other hand, the user only needs to perform a touch operation on the storage device to initiate the entire regulation process, without the need for applications on external terminals such as mobile phones, making the interaction process simpler and more private.

[0112] According to some exemplary embodiments of this application, based on the aforementioned scheme, the physiological data also includes a heartbeat signal, which includes at least one of an electrocardiogram signal and a photoplethysmography (PPG) signal; step S102, which analyzes the user's stress state based on the physiological data to obtain a stress state index, may include: preprocessing the heartbeat signal to obtain time-domain and frequency-domain sub-parameters of the heart rate variability parameter; inputting the time-domain and frequency-domain sub-parameters into a pre-trained stress analysis model to output the user's stress state and obtain a stress state index, wherein the stress analysis model is a classification model trained based on a labeled physiological dataset.

[0113] Preprocessing the heartbeat signal refers to filtering and denoising the acquired electrocardiogram (ECG) or photoplethysmography (PPG) signal, and then calculating the time-domain and frequency-domain sub-parameters of the heart rate variability parameters. In some embodiments, the time-domain sub-parameters include at least one of the standard deviation of adjacent normal heartbeat intervals (SDNN) and the root mean square of successive differences (RMSSD); the frequency-domain sub-parameters include the ratio of low-frequency power to high-frequency power (LF / HF), etc. It is understood that the above-mentioned time-domain and frequency-domain sub-parameters are exemplary and not limited thereto.

[0114] The stress analysis model is a classification model trained on a labeled physiological dataset. In some embodiments, the stress analysis model uses a lightweight one-dimensional convolutional neural network (1D-CNN) combined with a fully connected neural network hierarchical structure. The model includes a feature extraction layer, a fully connected mapping layer, and a classification output layer, and adopts an embedded lightweight design to adapt to the embedded chips with limited computing power in wearable devices. The feature extraction layer consists of three cascaded convolutional blocks. Each convolutional block includes a one-dimensional convolutional layer, a batch normalization layer, a Rectified Linear Unit (ReLU) activation layer, and a max pooling layer. The convolutional kernel size can be 3×1, the convolutional stride can be 1, and the pooling kernel size can be 2×1. This is used for noise reduction and deep feature extraction of multidimensional features in the time and frequency domains of heart rate variability parameters. The fully connected mapping layer consists of two hidden fully connected layers. The first layer can have 64 neurons, and the second layer can have 32 neurons. Both layers use the ReLU activation function to perform nonlinear dimensionality reduction mapping of high-dimensional features. The classification output layer uses the Softmax normalized exponential function for activation and outputs a multidimensional feature vector, such as a 5-dimensional feature vector, corresponding to stress levels 1 to 5, and outputting a quantitative indicator to represent the user's current stress level. In some embodiments, the stress analysis model is trained using a cross-entropy loss function, and convergence optimization is achieved based on a labeled physiological dataset. Its inference latency can be no higher than approximately 50ms, and the accuracy of embedded operation can be no lower than approximately 92%. In other embodiments, the stress analysis model may also employ other types of classification neural networks or machine learning models. This application does not limit the specific structure of the stress analysis model. It is understood that the above-mentioned convolutional block levels, convolutional kernel and pooling kernel sizes, number of neurons in the fully connected layer, output dimension, inference latency, and accuracy are all exemplary values. Those skilled in the art can select other values ​​according to actual needs, and are not limited thereto.

[0115] In this embodiment, by preprocessing the electrocardiogram (ECG) signal or photoplethysmography (PPG) pulse wave signal, time-domain and frequency-domain sub-parameters of heart rate variability (HRV) parameters are extracted. These two types of sub-parameters are then input into a stress analysis classification model trained on a labeled physiological dataset to output the stress state. This allows the determination of the stress state to utilize the multidimensional information of HRV reflecting autonomic nervous activity in both the time and frequency domains, and to comprehensively discriminate using a sample-trained classification model. This improves the accuracy of the stress state index in characterizing the user's actual stress level, providing a more accurate basis for subsequently determining the matching target audio parameters. Furthermore, the stress analysis model employs a lightweight embedded design, enabling stress state analysis to be completed with low inference latency on embedded chips with limited computing power in wearable devices, meeting the real-time requirements of closed-loop regulation.

[0116] See Figure 3 , Figure 3 The diagram schematically illustrates the structure of a closed-loop pressure regulation system provided in an exemplary embodiment of this application. The closed-loop pressure regulation system may include a storage device and a wearable device, which are communicatively connected via a communication link. The storage device can be used to house the wearable device, which can be used to collect the user's physiological data, analyze the user's stress state, generate or adjust AI audio, and play AI audio for audio intervention on the user.

[0117] In some embodiments, the storage device may include a touch area, a touch sensor, an interaction control module, and a first communication module. The touch area may be disposed on the outer surface of the storage device and is used to receive preset touch operations performed by the user. The touch sensor is disposed corresponding to the touch area and is used to detect the preset touch operation performed by the user on the touch area. The preset touch operation may be a triple tap, a double tap, a long press, or other operations recognizable by the touch sensor. The interaction control module is connected to the touch sensor and is used to generate a trigger command based on the preset touch operation detected by the touch sensor. The first communication module is connected to the interaction control module and is used to send the trigger command to the wearable device via a communication link.

[0118] In some embodiments, the communication link between the first communication module and the second communication module can be a Bluetooth Low Energy communication link. The first communication module can send a trigger command to the second communication module through the Bluetooth Low Energy communication link, so that the wearable device can start the subsequent physiological signal acquisition and audio intervention process according to the trigger command received by the second communication module.

[0119] In some embodiments, the wearable device may include a second communication module, a sensing module, a processing unit, and an audio playback module. The second communication module establishes a communication connection with a first communication module of the storage device and receives trigger commands from the storage device. The sensing module may include a PPG sensor, which is activated after the wearable device responds to the trigger command and collects the user's photoplethysmography (PPG) signal. The sensing module can output the PPG signal collected by the PPG sensor to the HRV data calculation module in the processing unit, so that the HRV data calculation module can obtain HRV data based on the PPG signal. The processing unit is connected to the second communication module, the sensing module, and the audio playback module, respectively, and processes the physiological signals collected by the sensing module and controls the audio playback module to play generated or adjusted AI audio.

[0120] In some embodiments, the processing unit may include an HRV data calculation module, a stress state analysis module, and an audio generation and adjustment module. The HRV data calculation module calculates HRV data based on the photoplethysmography (PPG) signal acquired by the PPG sensor. The stress state analysis module, connected to the HRV data calculation module, analyzes the user's current stress state based on the HRV data to obtain a stress state index. The audio generation and adjustment module, connected to the stress state analysis module, determines target audio parameters based on the stress state index and generates AI audio based on the target audio parameters; if the wearable device is already playing AI audio, the audio generation and adjustment module can also adjust the currently playing AI audio based on the new stress state index.

[0121] In some embodiments, the audio playback module is connected to the processing unit and is used to play AI audio generated or adjusted by the audio generation and adjustment module for audio intervention on the user. While the audio playback module plays the AI ​​audio, the sensing module can continue to collect the user's physiological signals and input the collected physiological signals into the processing unit. The processing unit can recalculate HRV data, analyze stress state, and adjust the AI ​​audio based on the newly collected physiological signals, thus forming a loop feedback path from the audio playback module to the sensing module and the processing unit. In this loop feedback path, the processing unit can evaluate the current audio intervention effect based on changes in the user's HRV and heart rate data after playing the AI ​​audio, and use this effect as the basis for generating or adjusting the target audio parameters for the next round. Therefore, the closed-loop stress regulation system can, after being triggered by the storage device, complete physiological signal collection, HRV data calculation, stress state analysis, AI audio generation or adjustment, and audio playback intervention through the wearable device, and continue to perform the next round of regulation based on physiological changes after playback.

[0122] In this embodiment, the storage device is used to house the wearable device, making the storage device and the wearable device physically integrated, facilitating pairing and storage. The wearable device handles the collection of physiological data, analysis of stress states, and generation and playback of target audio, and can interact with the storage device via touch activation. The storage device and the wearable device cooperate through a communication link, forming a closed loop between them, from physiological data collection and stress state analysis to target audio generation and playback and physiological feedback adjustment. On one hand, the wearable device can generate and dynamically adjust target audio based on the user's real-time stress state, improving the audio's relevance to stress regulation. On the other hand, the aforementioned closed loop can be completed independently between the storage device and the wearable device, without the need for mandatory binding to an external terminal application, ensuring stable operation of stress regulation and enhancing scenario adaptability.

[0123] Based on the above audio processing method, this application also provides a computer device, including a processor and a memory. The memory stores computer-readable instructions, which, when executed by the processor, implement the audio processing method as described in any of the above embodiments. In some embodiments, the processor may be a microcontroller unit, an embedded chip, or a digital signal processor located in a wearable device, and the memory may be a read-only memory, random access memory, or flash memory located in the wearable device. In some embodiments, this application also provides a computer-readable storage medium storing a computer program, which, when executed by the processor, implements the audio processing method as described in any of the above embodiments. The computer-readable storage medium may be a read-only memory, random access memory, flash memory, optical disk, or other medium capable of storing a computer program; this application does not limit this.

[0124] For ease of understanding, the following example uses in-ear headphones as the wearable device and a charging case as the storage device, illustrating the complete process of the aforementioned audio processing method when several weighted features are enabled, within a specific application scenario. In this example, the stress state index is represented by a relaxation score ranging from 0 to 1, with lower scores indicating higher user stress levels. When a user feels stressed, they put the in-ear headphones in their ears and tap the touch area on the charging case three times consecutively. The charging case's interaction control module recognizes this triple-tap operation and generates an activation command, which is sent to the in-ear headphones via Bluetooth Low Energy. The in-ear headphones, responding to the activation command, activate their internal PPG sensor, which begins real-time acquisition of the user's photoplethysmography (PPG) signal and transmits the raw signal or the resulting heart rate interval sequence to the processing unit. The heart rate interval sequence obtained from the PPG signal may include an RR interval sequence, and the sensing module can transmit the acquired raw signal or RR interval sequence to the processing unit of the in-ear headphones in real time.

[0125] The processing unit preprocesses the signal and calculates the time-domain and frequency-domain sub-parameters of the heart rate variability parameter. This data is then input into the stress analysis model, which outputs the user's current stress state and converts it into a relaxation score. For example, a relaxation score of 0.2 indicates a high level of stress. The audio generation module receives this relaxation score, determines the corresponding target audio parameters based on the mapping relationship, and uses the audio generation model to synthesize a low-pitched, smooth audio clip simulating a deep-sea environment in real time. This clip is then played locally through the headphone's audio playback module. During this process, the audio generated by the audio generation model can be a PCM audio stream, which can be played in real time by the local player on the in-ear headphones.

[0126] In the subsequent process, the in-ear headphones continuously collect the user's physiological data and assess its physiological trends by sliding a time window: if the user's heart rate variability increases, the mean heart rate decreases, and the relaxation score improves from 0.2 to 0.4, it is considered a positive improvement trend. The audio generation module smoothly adds a soft synthesizer padding tone to the current audio soundscape to consolidate the relaxation effect with more soothing audio; if the user's physiological data does not improve, it is considered a no-change trend, and the audio generation module switches strategies, such as slowing down the tempo to guide the user to take deep breaths; if the user's physiological data deteriorates in the opposite direction, the sliding time window is compressed to increase the judgment frequency while enhancing intervention. The above process continues to cycle until the user taps the charging case three times to stop, or the system detects that the user's heart rate variability parameter and heart rate simultaneously reach their respective optimization target values ​​and stabilize for a period of time before automatically ending. It is understood that the specific object, relaxation score value, and audio format described in this paragraph are only typical examples, and the enabled combination of subordinate features is also only a typical combination, and does not constitute a limitation on the scope of protection of this application.

[0127] It should be noted that although several modules or units have been mentioned in the detailed description above, this division is not mandatory. In fact, according to the embodiments of this application, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0128] Furthermore, the above figures are merely illustrative of the processes included in the method according to exemplary embodiments of this application, and are not intended to be limiting. It is readily understood that the processes shown in the above figures do not indicate or limit the temporal order of these processes. Additionally, it is readily understood that these processes may be executed synchronously or asynchronously, for example, in multiple modules.

[0129] Through the description of the above embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, touch terminal, or network device, etc.) to execute the methods according to the embodiments of this application.

[0130] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of this application are indicated by the claims.

[0131] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. An audio processing method, characterized in that, Applied to wearable devices, wherein the wearable device is communicatively connected to a storage device, and the wearable device is equipped with a sensing module, the method includes: In response to a received wearing instruction, the user's physiological data is collected through the sensing module, wherein the wearing instruction is triggered by the user wearing the wearable device; the physiological data includes at least one of the user's heart rate data, respiratory rate data, and sleep state parameters; Based on the physiological data, the user's stress state is analyzed to obtain a stress state index; Based on the stress state index, target audio parameters corresponding to the stress state index are determined, wherein the stress state index is used to characterize the mapping relationship between the user's stress level and the audio parameters, and the audio parameters include at least one of rhythm, mode, harmonic complexity, timbre and soundscape element combination; The target audio is generated based on the target audio parameters, and then the target audio is played.

2. The audio processing method according to claim 1, characterized in that, The process of generating target audio based on the target audio parameters includes: The target audio parameters are input into the audio generation model, so that the audio generation model synthesizes an audio segment in real time based on the target audio parameters, outputs the audio segment, and uses the audio segment as the target audio.

3. The audio processing method according to claim 1, characterized in that, The process of generating target audio based on the target audio parameters includes: Obtain the initial audio currently being played by the wearable device, and the current audio parameters of the currently playing initial audio; Based on the target audio parameters and the current audio parameters, parameters to be adjusted are determined, wherein the parameters to be adjusted include at least one of rhythm, chord, and timbre; Based on the parameters to be adjusted, the initial audio is adjusted to obtain the target audio.

4. The audio processing method according to claim 1, characterized in that, The pressure status index corresponds to a pressure level, the pressure level is set with a corresponding pressure threshold, and the pressure intensity corresponding to the pressure level decreases sequentially. The number of audio parameters corresponding to each pressure level is positively correlated with the pressure intensity. The process of generating target audio based on the target audio parameters includes: Determine the target pressure level corresponding to the pressure status index; Control the wearable device to play target audio corresponding to the target pressure level.

5. The audio processing method according to claim 4, characterized in that, The audio parameters include at least one of rhythm, number of harmonic levels, harmonic complexity, and number of soundscape element combinations. The stress level corresponding to the stress state index is positively correlated with the rhythm, the stress level is positively correlated with the number of harmonic levels, the stress level is positively correlated with the harmonic complexity, and the stress level is positively correlated with the number of soundscape element combinations.

6. The audio processing method according to claim 1, characterized in that, The method further includes: During the playback of the target audio, the current physiological data of the current cycle window is collected according to the preset sliding time window, and the current pressure parameter of the current cycle window is determined based on the current physiological data of the current cycle window. The historical pressure parameters of the previous cycle window are obtained, and the current pressure parameters of the current cycle window and the historical pressure parameters of the previous cycle window are used to determine the current physiological change trend of the user; wherein, the current physiological change trend includes one of the following: a positive improvement trend, a no-change trend, and a reverse deterioration trend. Based on the physiological change trend, the target audio parameters are adjusted to obtain the intervention audio parameters for the next cycle window; The target audio being played is adjusted according to the intervention audio parameters, and the adjusted target audio is then played.

7. The audio processing method according to claim 6, characterized in that, Both the current stress parameters and the historical stress parameters include at least heart rate variability parameters and mean heart rate; determining the user's current physiological trend based on the current stress parameters of the current cycle window and the historical stress parameters of the previous cycle window includes: The heart rate variability parameter of the current cycle window is compared with the heart rate variability parameter of the previous cycle window to determine the amount of change in the heart rate variability parameter between the current cycle window and the previous cycle window. The average heart rate of the current cycle window is compared with the average heart rate of the previous cycle window to determine the change in the average heart rate between the current cycle window and the previous cycle window. If the change in the heart rate variability parameter indicates that the heart rate variability parameter of the current cycle window is greater than that of the previous cycle window, and the change in the mean heart rate indicates that the mean heart rate of the current cycle window is less than that of the previous cycle window, then the user's current physiological change trend is determined to be the positive improvement trend. If the change in the heart rate variability parameter and the change in the mean heart rate do not exceed the preset amplitude threshold, then the user's current physiological trend is determined to be no trend. If the change in the heart rate variability parameter indicates that the heart rate variability parameter of the current cycle window is less than that of the previous cycle window, and the change in the mean heart rate indicates that the mean heart rate of the current cycle window is greater than that of the previous cycle window, then the user's current physiological trend is determined to be a reverse deterioration trend.

8. The audio processing method according to claim 6 or 7, characterized in that, The step of adjusting the target audio parameters based on the physiological change trend to obtain the intervention audio parameters for the next cycle window includes: If the physiological change trend is the positive improvement trend, then the target audio parameters are adjusted first to obtain the intervention audio parameters for the next cycle window; the first adjustment includes lowering the rhythm, reducing the harmonic complexity, and lowering the volume of environmental sounds in the soundscape elements; If the physiological change trend is the no-change trend, then the target audio parameters are adjusted a second time to obtain the intervention audio parameters for the next cycle window; the second adjustment includes adjusting the rhythm, switching the current instrument timbre or mode, and increasing the harmonic complexity within a preset range of change. If the physiological change trend is the reverse deterioration trend, then the target audio parameters are adjusted a third time to obtain the intervention audio parameters for the next cycle window; the third adjustment includes raising the rhythm, raising the harmonic level, switching the current mode to a major mode, and changing the proportion of natural environmental sounds in the combination of soundscape elements.

9. The audio processing method according to claim 6 or 7, characterized in that, The method further includes: If the physiological change trend is the reverse deterioration trend, then the duration of the sliding time window is shortened; Monitor the physiological change trends of the user within the shortened sliding time window; When the user's current physiological change trend is detected to change from the reverse deterioration trend to the positive improvement trend or no change trend, the duration of the sliding time window is restored to the duration before it was shortened, and the adjustment operation corresponding to the current physiological change trend is switched back.

10. The audio processing method according to claim 6, characterized in that, Before collecting physiological data for the current period window according to a preset sliding time window during the playback of the target audio, the method further includes: When the wearable device is first activated, it collects the user's physiological data in a resting state and calculates the user's physiological baseline value based on the physiological data in a resting state. The physiological baseline value includes the heart rate variability baseline value and the heart rate baseline value. The optimization target value of the heart rate variability parameter is determined based on the heart rate variability baseline value, and the optimization target value of the heart rate is determined based on the heart rate baseline value; the optimization target value of the heart rate variability parameter includes the user's heart rate variability parameter being higher than the heart rate variability baseline value by a preset first multiple, and the optimization target value of the heart rate includes the user's heart rate being lower than the heart rate baseline value by a preset second multiple, wherein the first multiple is greater than the second multiple; If, without receiving a stop command from the user, the system detects that the user's heart rate variability parameter has reached the optimized target value of the heart rate variability parameter, and the user's heart rate has reached the optimized target value of the heart rate, then the audio playback is stopped or the audio volume is reduced.

11. The audio processing method according to claim 1, characterized in that, The method further includes: The device receives an activation command from the storage device, the activation command being obtained by the storage device in response to a preset touch operation performed by the user on the touch area; In response to the activation command, the step of collecting the user's physiological data through the sensing module in response to the received wearing command is executed.

12. The audio processing method according to claim 1, characterized in that, The physiological data also includes heart rate signals, which include at least one of electrocardiogram (ECG) signals and photoplethysmography (PPG) pulse wave signals. Based on the physiological data, the analysis of the user's stress state to obtain stress state indicators includes: The heartbeat signal is preprocessed to obtain the time-domain and frequency-domain sub-parameters of the heart rate variability parameters; The time-domain sub-parameters and the frequency-domain sub-parameters are input into a pre-trained stress analysis model, which outputs the user's stress state to obtain a stress state index; wherein, the stress analysis model is a classification model trained based on a labeled physiological dataset.

13. A closed-loop pressure regulation system, characterized in that, The device includes a storage device and a wearable device, the storage device being used to house the wearable device; the wearable device is configured to perform the audio processing method as described in any one of claims 1 to 12.

14. A computer device, characterized in that, include: processor; as well as A memory storing computer-readable instructions that, when executed by the processor, implement the audio processing method as described in any one of claims 1 to 12.

Images