Active noise control system and method for a smartphone

Abstract

The application discloses an active noise reduction control system and method of a smart phone, comprising an environmental sound acquisition module, a motion state sensing module, a deep learning processing module, a noise reduction signal generation module, a noise reduction execution module, a user preference setting module, a scene recognition module and a power management module; the environmental sound acquisition module collects signals in all directions through a multi-microphone spatial array, and wide frequency response and high sampling rate ensure data quality; the motion state sensing module monitors phone motion parameters in real time to provide a basis for noise reduction strategy adjustment; the deep learning processing module adopts a hybrid model architecture combined with GPU acceleration to accurately analyze noise characteristics; the noise reduction signal generation module can generate adaptive noise reduction signals according to analysis results; and the noise reduction execution module adopts active feedback control and multi-mode sound generation strategies; the application significantly improves the accuracy, adaptability and intelligent level of noise reduction, and greatly improves the auditory experience of users when using the phone in different scenarios.

Description

Technical Field

[0001] This invention relates to the field of smartphone noise reduction, and more particularly to an active noise reduction control system and method for smartphones. Background Technology

[0002] In today's smartphone usage scenarios, users' demand for a quiet and comfortable audio environment is growing. Whether making calls on a noisy street or enjoying music on the subway, environmental noise significantly impacts the user experience. As people's requirements for the audio quality of mobile devices continue to increase, active noise cancellation technology has become key to improving the user experience of smartphones. In various complex environments, such as shopping malls and airports, environmental noise is diverse and dynamically changing. Traditional audio processing methods are insufficient to meet users' expectations for noise reduction effects, making the development of active noise cancellation control systems and methods for smartphones necessary.

[0003] Existing active noise cancellation technologies have several drawbacks in practical applications. Regarding environmental sound acquisition, many systems use simple microphone configurations, failing to capture sound signals from all angles and in a wide range. Furthermore, their narrow frequency response range and low sampling rate result in poor data quality, making it difficult to accurately reflect the true state of complex environmental noise and complicating subsequent noise reduction processing. In terms of noise reduction strategy adjustment, there is a lack of effective monitoring and utilization of the phone's motion status, preventing real-time optimization of the noise reduction strategy based on the phone's dynamic changes. This significantly reduces the noise reduction effect during user movement.

[0004] In the noise reduction signal processing stage, traditional methods often employ simple models and algorithms, making it difficult to accurately analyze noise characteristics and resulting in poor adaptability of the generated noise-reduced signals. For example, a uniform approach to processing broadband signals cannot finely process the characteristics of noise at different frequencies, leading to insufficient noise reduction accuracy. Furthermore, the lack of effective feedback control mechanisms and flexible sound generation strategies during noise reduction execution prevents dynamic adjustments based on environmental changes and user needs, resulting in limited noise reduction effectiveness and a poor user experience. In addition, existing technologies often neglect personalized user needs and fail to fully consider differences in various scenarios, exhibiting shortcomings in computational efficiency and power management, thus failing to achieve efficient and sustained noise reduction functionality. Summary of the Invention

[0005] In order to overcome the shortcomings and deficiencies of the existing technology, the present invention provides an active noise reduction control system and method for smartphones.

[0006] On one hand, this application provides an active noise cancellation control system for a smartphone, the system comprising:

[0007] Ambient sound acquisition module: Multiple high-sensitivity microphones are set at different positions on the mobile phone to form a spatial array, which includes at least three omnidirectional microphones and two directional microphones to collect ambient sound signals from all directions and multiple angles. The microphones have wide frequency response characteristics, with a frequency response range of 20Hz-20kHz and a sampling rate of not less than 48kHz.

[0008] Motion state sensing module: Integrates accelerometer and gyroscope to monitor the phone's acceleration and angular velocity motion parameters in real time. The accelerometer has a measurement range of ±16g and a resolution of 0.001g; the gyroscope has a measurement range of ±2000° / s and a resolution of 0.0625° / s. This module connects to the phone's main processor via I2C or SPI communication interface and transmits the collected motion data at a frequency of no less than 100Hz.

[0009] The deep learning processing module employs a hybrid model architecture combining convolutional neural networks (CNNs) and recurrent neural networks (RNNs). The CNN portion includes convolutional and pooling layers to extract spatial features from environmental sound signals. The convolutional kernels vary in size, such as 3x3 and 5x5, and there are at least three convolutional layers, each followed by a ReLU activation function. The pooling layers use max pooling with a 2x2 kernel. The RNN portion utilizes long short-term memory (LSTM) units with 256 memory units to process the time-series features of the sound signals and capture the dynamic changes in noise. This module leverages the phone's GPU for accelerated computation, performing real-time processing and analysis of the acquired sound and motion data without affecting the phone's normal performance.

[0010] The noise reduction signal generation module generates a noise reduction signal with the same amplitude but opposite phase to the ambient noise, based on the output of the deep learning processing module. An adaptive filter adjusts the noise reduction signal, employing either a least mean square algorithm or a recursive least squares algorithm. The filter coefficients are updated in real-time according to changes in ambient noise to ensure the accuracy and stability of the noise reduction effect. Simultaneously, in conjunction with the phone's audio power amplifier, the noise reduction signal is amplified to a suitable power level to drive the speaker output.

[0011] Noise cancellation module: Composed of multiple miniature speakers, distributed near the ear or in the sound-emitting area of ​​the phone. The speakers feature a flat design to fit within the limited space inside the phone. Their frequency response range matches the microphone of the ambient sound acquisition module, from 20Hz to 20kHz, and they possess good linearity, accurately reproducing the noise-canceling signal and canceling out ambient noise. Furthermore, they can automatically adjust the speaker's sound mode and volume according to different usage scenarios, such as calls, music playback, and video playback.

[0012] Furthermore, the environmental sound acquisition module also includes:

[0013] A barometric pressure sensor is used to detect changes in ambient air pressure, thereby helping to determine the type of environmental noise. For example, the air pressure changes during aircraft flight differ from those on the ground, allowing for more accurate noise identification. This barometric pressure sensor has a measurement range of 300 hPa to 1100 hPa, a resolution of 0.1 hPa, and a measurement accuracy of ±0.3 hPa.

[0014] The temperature sensor monitors the ambient temperature, as temperature affects the propagation characteristics of sound, which in turn affects noise characteristics. The temperature sensor has a measurement range of -40℃ to 125℃ and an accuracy of ±0.5℃. It connects to the main processor via an I2C interface and transmits the collected air pressure and temperature data along with the sound signal to the deep learning processing module for comprehensive analysis.

[0015] Furthermore, the deep learning processing module also includes:

[0016] The online update mechanism utilizes incremental learning algorithms. When the phone collects new environmental sound and motion data, it doesn't need to retrain the entire model; instead, it incrementally updates the existing model parameters. For example, by employing a Bayesian online learning algorithm, the posterior distribution of the model parameters is continuously adjusted based on new data, enabling the model to adapt to environmental changes in real time and improving noise reduction performance.

[0017] Model fusion techniques combine multiple deep learning models with different structures or trained on different datasets. For example, a weighted average fusion method is used, which assigns different weights to each model based on its denoising performance in different scenarios, and combines the outputs of multiple models to improve the accuracy and reliability of denoising decisions.

[0018] Furthermore, the noise reduction signal generation module further includes:

[0019] The sub-band decomposition unit divides the acquired ambient sound signal and the generated noise-reduced signal into multiple sub-bands according to frequency. For example, a uniform filter bank can be used to divide the frequency range of 0Hz-20kHz into 16 sub-bands. Noise reduction processing is performed separately in each sub-band. The adaptive filter coefficients are adjusted independently according to the noise characteristics of different sub-bands, thereby eliminating noise in different frequency bands more precisely and improving the overall noise reduction effect.

[0020] The phase compensation circuit, due to the phase delay during sound propagation, is used to accurately compensate for the phase of the noise-reducing signal, ensuring precise time-domain alignment between the noise-reducing signal and ambient noise for optimal noise cancellation. The phase compensation range is ±180°, with a resolution of 0.1°.

[0021] Furthermore, the noise reduction execution module also includes:

[0022] An active feedback control mechanism is employed, with an additional microphone placed near the speaker to collect the mixed signal of the noise-reduced signal output by the speaker and the residual ambient noise in real time. This mixed signal is then fed back to the noise-reduced signal generation module for comparison and analysis with the original ambient sound signal. The amplitude and phase of the noise-reduced signal are further adjusted to form a closed-loop control, continuously optimizing the noise reduction effect.

[0023] The multi-mode sound strategy allows for switching between different sound modes based on the phone's usage scenario and user needs. For example, in a call scenario, a mono sound mode is used to focus on canceling out noise from the direction of the call; in a music playback scenario, it switches to a stereo sound mode to maintain the stereo effect of the music while reducing noise; and in noisy outdoor environments, it automatically increases the speaker volume and enhances the low-frequency noise reduction effect.

[0024] Furthermore, the system also includes:

[0025] The user preference settings module allows users to personalize the active noise cancellation system via a mobile application. For example, users can select different noise reduction intensity levels, such as light, medium, and heavy noise reduction; they can also customize the noise reduction priority for certain specific frequency noises. The system will adjust the parameters of the deep learning processing module and the noise reduction signal generation module according to the user's settings to meet their personalized needs.

[0026] The scene recognition module uses machine learning algorithms to identify the phone's environment, such as airport, subway, office, or indoors, based on data collected by the ambient sound acquisition module, motion perception module, and other sensors (e.g., barometric pressure and temperature sensors). Different noise reduction parameter configuration files are pre-stored for different scenarios, and the system automatically calls the appropriate configuration file to optimize the noise reduction effect.

[0027] Furthermore, the system adopts a distributed computing architecture:

[0028] Some computational tasks are handled by the phone's local processor, such as the initial data processing of the ambient sound acquisition module and the data acquisition and transmission of the motion state perception module. The local processor uses a high-performance mobile CPU, combined with GPU acceleration, to ensure that tasks with high real-time requirements can be executed quickly.

[0029] Complex deep learning model computation tasks, such as the operations of convolutional neural networks and recurrent neural networks, can be transmitted to cloud servers for processing via wireless networks (such as Wi-Fi or 5G). Cloud servers possess powerful computing resources, enabling them to quickly complete deep learning model computations and return the results to the mobile phone for generating noise-reduced signals. This distributed computing architecture fully utilizes the local resources of the mobile phone while leveraging the powerful computing capabilities of the cloud to improve overall system performance.

[0030] Furthermore, the system also includes:

[0031] The power management module employs dynamic power management strategies tailored to the power consumption characteristics of each module in the active noise cancellation system. For example, when ambient noise is low or the phone is stationary, the operating frequency of the deep learning processing module is reduced to decrease GPU power consumption; when the amount of data collected by the microphone and sensors is small, their sampling frequency is reduced to save power. Simultaneously, the power output of execution modules such as the speaker is optimized, dynamically adjusting the output power according to noise cancellation requirements. This ensures effective noise cancellation while minimizing overall system power consumption and extending battery life.

[0032] On the other hand, this application also provides an active noise cancellation control method for smartphones, the method comprising the following steps:

[0033] Data acquisition steps: The ambient sound acquisition module, motion sensing module, and other auxiliary sensors (such as air pressure and temperature sensors) synchronously acquire ambient sound signals, mobile phone motion data, and environmental parameter data at set frequencies. The ambient sound acquisition module acquires sound signals at a sampling rate of no less than 48kHz, the motion sensing module acquires motion data at a frequency of no less than 100Hz, and other sensors acquire data at their respective optimal response frequencies. This data is then transmitted to the mobile phone's main processor via the corresponding communication interfaces.

[0034] Data preprocessing steps: The collected raw data is preprocessed, including noise removal, data calibration, and normalization. For ambient sound signals, digital filtering techniques are used to remove high-frequency or low-frequency noise; motion data is calibrated to eliminate sensor drift errors; all data is normalized to ensure they are within the same numerical range, facilitating subsequent processing by deep learning models.

[0035] Deep learning processing steps: The preprocessed data is input into the deep learning processing module. A hybrid model combining convolutional neural networks and recurrent neural networks is used to extract spatial and temporal features from the data, analyzing the type, intensity, and changing trends of environmental noise. The model continuously learns from training data, optimizes its parameters, and improves its ability to identify and predict noise.

[0036] The noise reduction signal generation steps are as follows: Based on the output of the deep learning processing module, a noise reduction signal with the same amplitude but opposite phase to the ambient noise is generated. Using an adaptive filter, the filter coefficients are updated according to the real-time changes in ambient noise using either the least mean square algorithm or the recursive least squares algorithm to adjust the noise reduction signal and make it more accurately match the ambient noise.

[0037] Noise Reduction Execution Steps: The generated noise-reduced signal is amplified to a suitable power level by an audio power amplifier, driving the miniature speaker output of the noise reduction execution module to cancel out ambient noise. Simultaneously, according to an active feedback control mechanism, the mixed signal output from the speaker is acquired in real time and fed back to the noise-reduced signal generation step for further optimization and adjustment.

[0038] Furthermore, the method also includes the following steps:

[0039] User preference application steps: The system reads the user's personalized settings in the user preference settings module, such as noise reduction intensity level and noise reduction priority for specific frequencies. Based on the user settings, it adjusts the relevant parameters in the deep learning processing steps and the noise reduction signal generation steps to ensure the active noise reduction system meets the user's personalized needs.

[0040] Scene recognition application steps: Use the scene recognition module to identify the environment scene in which the mobile phone is located. Based on the recognition results, call the corresponding configuration from the pre-stored noise reduction parameter configuration files of different scenes, and optimize and adjust the parameters in the entire active noise reduction control process to adapt to the noise reduction needs of different environmental scenes.

[0041] Beneficial effects:

[0042] This invention proposes an active noise cancellation control system and method for smartphones. The system comprises multiple modules working collaboratively. An ambient sound acquisition module uses a multi-microphone spatial array to collect signals from all directions, ensuring data quality through wide bandwidth response and high sampling rate. A motion state perception module monitors the phone's motion parameters in real time, providing a basis for adjusting the noise cancellation strategy. A deep learning processing module employs a hybrid model architecture combined with GPU acceleration to accurately analyze noise characteristics. A noise cancellation signal generation module generates an adapted noise cancellation signal based on the analysis results, with sub-band decomposition and phase compensation improving noise cancellation accuracy. The noise cancellation execution module's active feedback control and multi-mode sound generation strategy effectively enhance noise cancellation performance and user experience. Simultaneously, the system also includes modules for user preference settings, scene recognition, distributed computing, and power management to meet personalized needs and improve computational efficiency and battery life. At the methodological level, a complete and efficient process is formed, from data acquisition and preprocessing to analyzing, generating, and adjusting the noise cancellation signal using a deep learning model, then optimizing based on feedback mechanisms and adapting parameters by combining user preferences and scene recognition. Overall, this application significantly improves the accuracy, adaptability, and intelligence of noise reduction, greatly enhancing the user's auditory experience when using a mobile phone in different scenarios. Attached Figure Description

[0043] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 This is a diagram showing the modular composition of the system of the present invention;

[0045] Figure 2 This is a flowchart illustrating the operation of the method of the present invention. Detailed Implementation

[0046] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0047] like Figure 1 As shown, this embodiment provides an active noise cancellation control system for a smartphone, characterized in that the system includes:

[0048] Ambient sound acquisition module: Multiple high-sensitivity microphones are set at different positions on the mobile phone to form a spatial array, including three omnidirectional microphones and two directional microphones, to collect ambient sound signals from all directions and multiple angles. The microphones have wide frequency response characteristics, with a frequency response range of 20Hz-20kHz and a sampling rate of no less than 48kHz.

[0049] Specifically, the environmental sound acquisition module is the data source for the active noise cancellation control system. Its ability to acquire environmental sound signals from all directions and multiple angles lays the foundation for accurate analysis and noise reduction. By placing multiple microphones at different locations on the phone to form a spatial array, it can capture sounds from all directions, ensuring comprehensive perception of complex environmental noise. This module uses a spatial array consisting of three omnidirectional microphones and two directional microphones. The omnidirectional microphones can receive sounds from all directions, while the directional microphones can selectively acquire sounds from specific directions; the combination of the two enables more detailed acquisition of environmental sounds. The microphones have wide frequency response characteristics, with a frequency response range of 20Hz-20kHz, covering the main frequency range of human hearing, enabling complete acquisition of various sounds in the environment. The sampling rate is no less than 48kHz; a high sampling rate ensures more accurate and detailed sound signals and reduces information loss. This module plays a crucial role in various complex environments. For example, on the street, it can collect the sounds of cars driving and crowds; in shopping malls, it can capture promotional announcements from merchants and people talking; in the subway, it can also collect the noise of trains running and the footsteps of passengers. These rich sound signals will be transmitted to subsequent modules for processing.

[0050] Motion state sensing module: Integrates accelerometer and gyroscope to monitor the phone's acceleration and angular velocity motion parameters in real time. The accelerometer has a measurement range of ±16g and a resolution of 0.001g; the gyroscope has a measurement range of ±2000° / s and a resolution of 0.0625° / s. This module connects to the phone's main processor via I2C or SPI communication interface and transmits the collected motion data at a frequency of no less than 100Hz.

[0051] Specifically, mobile phones are often in different states of motion during use, and these changes in motion can affect ambient sound and noise reduction performance. The motion state sensing module monitors the phone's acceleration and angular velocity in real time, allowing the system to dynamically adjust its noise reduction strategy based on the phone's movement, thus improving the accuracy and adaptability of noise reduction.

[0052] Model: This module integrates an accelerometer and a gyroscope. The accelerometer measures the phone's acceleration, while the gyroscope measures its angular velocity. Through the collaborative work of these two sensors, comprehensive and accurate motion status information of the phone can be obtained. The accelerometer has a measurement range of ±16g and a resolution of 0.001g, enabling precise measurement of the phone's motion under different accelerations. The gyroscope has a measurement range of ±2000° / s and a resolution of 0.0625° / s, accurately sensing the phone's rotation speed. This module connects to the phone's main processor via an I2C or SPI communication interface and transmits the collected motion data at a frequency of no less than 100Hz to ensure real-time data transmission. When the user walks while holding the phone, the accelerometer and gyroscope detect the phone's shaking and movement. The system can adjust the noise reduction strategy based on this motion data to avoid additional noise from motion affecting the noise reduction effect. In scenarios such as riding in a car or running, the noise reduction parameters can also be dynamically optimized based on the phone's motion status, providing a more stable noise reduction experience.

[0053] The deep learning processing module employs a hybrid model architecture combining convolutional neural networks (CNNs) and recurrent neural networks (RNNs). The CNN portion includes convolutional and pooling layers to extract spatial features from environmental sound signals, with each convolutional layer followed by a ReLU activation function. The pooling layers use max pooling with a 2x2 kernel size. The RNN portion utilizes long short-term memory (LSTM) units with 256 memory units to process the time-series features of the sound signals and capture the dynamic changes in noise. This module leverages the phone's GPU for accelerated computation, enabling real-time processing and analysis of the collected sound and motion data without affecting the phone's normal performance.

[0054] Specifically, the deep learning processing module is the core of the entire active noise reduction control system. It is responsible for the deep analysis and processing of the acquired sound and motion data. By extracting spatial and temporal features from the environmental sound signal, it can accurately capture the dynamic changes in noise, providing a basis for the subsequent generation of effective noise reduction signals.

[0055] Model: This module employs a hybrid model architecture combining Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs). The CNN portion includes convolutional layers and pooling layers. Convolutional layers extract spatial features from environmental sound signals using convolutional kernels, and each convolutional layer is followed by a ReLU activation function to introduce non-linearity and enhance the model's expressive power. The pooling layers use max pooling with a 2x2 kernel size to reduce the dimensionality of the feature maps and decrease computational cost. The RNN portion uses Long Short-Term Memory (LSTM) units with 256 memory units, capable of processing the time-series features of sound signals and capturing the dynamic changes in noise. Utilizing the mobile phone's GPU for accelerated computation, real-time processing and analysis of the collected data are achieved without affecting the phone's normal performance. The parallel computing capabilities of the GPU significantly improve the model's computation speed, ensuring the system can respond promptly to changes in environmental noise. In scenarios with complex and variable environmental noise, such as construction sites and airport waiting halls, the deep learning processing module can quickly and accurately analyze the type, intensity, and trend of noise. For example, at a construction site, there may be multiple noises such as the roar of machines and the shouts of workers. This module can identify the characteristics of different types of noise by extracting and analyzing the features of the sound signals, providing accurate information for subsequent noise reduction processing.

[0056] Noise Reduction Signal Generation Module: Based on the output of the deep learning processing module, a noise reduction signal with the same amplitude but opposite phase to the ambient noise is generated. The noise reduction signal is adjusted by an adaptive filter. The filter adopts the least mean square algorithm or the recursive least square algorithm and updates the filter coefficients in real time according to the changes in ambient noise. At the same time, combined with the mobile phone's audio power amplifier, the noise reduction signal is amplified to a suitable power level to drive the speaker output.

[0057] Specifically, the main function of the noise reduction signal generation module is to generate a noise reduction signal with the same amplitude but opposite phase to the ambient noise, based on the output of the deep learning processing module. In this way, when the noise reduction signal and ambient noise are superimposed, they cancel each other out, thus reducing ambient noise. This module uses an adaptive filter to adjust the noise reduction signal based on the output of the deep learning processing module. The adaptive filter employs either the Least Mean Square (LMS) algorithm or the Recursive Least Squares (RLS) algorithm, which updates the filter coefficients in real time according to changes in ambient noise, allowing the noise reduction signal to better adapt to different noise environments. Combined with the phone's audio power amplifier, the noise reduction signal is amplified to a suitable power level to drive the speaker output. The power amplifier enhances the strength of the noise reduction signal, ensuring that the speaker can output a sufficiently strong noise reduction signal to cancel out ambient noise. This module plays a crucial role in various scenarios requiring noise reduction. For example, during calls, it can generate corresponding noise reduction signals based on the surrounding ambient noise, reducing external noise interference and improving call quality; when listening to music, it effectively reduces ambient noise, allowing users to focus more on enjoying the music.

[0058] Noise reduction module: Composed of multiple miniature speakers, distributed near the ear or sound-emitting area of ​​the phone. The speakers adopt a flat design to adapt to the limited space inside the phone. Its frequency response range is matched with the microphone of the ambient sound acquisition module, which is 20Hz-20kHz. It automatically adjusts the speaker's sound mode and volume according to different usage scenarios of the phone.

[0059] Specifically, the noise reduction execution module is the key link in converting the noise reduction signal into actual noise reduction effect. It outputs the noise reduction signal through multiple miniature speakers, which cancels out ambient noise, thereby reducing environmental noise. This module consists of multiple miniature speakers distributed near the ear or the sound-emitting area of ​​the phone. The speakers adopt a flat design, which can adapt to the limited space inside the phone while ensuring good sound output. The frequency response range of the speakers is matched with the microphone of the ambient sound acquisition module, 20Hz-20kHz, ensuring accurate output of noise reduction signals corresponding to ambient noise. Depending on the different usage scenarios of the phone, this module can automatically adjust the speaker's sound mode and volume to achieve the best noise reduction effect. The noise reduction execution module automatically adjusts its working mode in different usage scenarios. For example, in a call scenario, it uses a mono sound mode to focus on canceling noise in the direction of the call; in a music playback scenario, it switches to stereo sound mode to maintain the stereo effect of the music while reducing noise; in noisy outdoor environments, it automatically increases the speaker volume and enhances the low-frequency noise reduction effect to effectively reduce interference from external noise.

[0060] Preferably, the environmental sound acquisition module further includes:

[0061] The barometric pressure sensor is used to detect changes in ambient air pressure and help determine the type of ambient noise. The barometric pressure sensor has a measurement range of 300hPa-1100hPa, a resolution of 0.1hPa, and a measurement accuracy of ±0.3hPa.

[0062] Specifically, barometric pressure sensors play a crucial role in the active noise cancellation control system of smartphones. Changes in ambient air pressure affect sound propagation; different air pressure environments result in varying sound propagation characteristics and attenuation levels, leading to different features of environmental noise. By detecting changes in ambient air pressure, the system can help determine the type of environmental noise. For example, in high-altitude areas with lower air pressure, the characteristics of wind noise differ from those in low-altitude areas; air pressure is relatively stable in enclosed spaces, but changes significantly in scenarios involving rapid ascent and descent, such as elevators, resulting in different noise levels. Therefore, barometric pressure sensors provide additional environmental information to the system, helping to more accurately analyze and process environmental noise. A barometric pressure sensor is a device that converts the physical quantity of ambient air pressure into a measurable electrical signal. It senses changes in air pressure through internal pressure-sensitive elements, such as piezoresistive or capacitive sensors, and converts this into an electrical signal output. The measurement range is 300 hPa to 1100 hPa, covering common air pressure values ​​in most parts of the world, effectively measuring everything from low air pressure in high-altitude areas to high air pressure in low-altitude areas. A resolution of 0.1 hPa means it can accurately detect minute changes in air pressure, providing the system with more detailed environmental information. A measurement accuracy of ±0.3 hPa ensures the reliability of the measurement results, enabling the system to determine the type of noise based on accurate air pressure data. The pressure-sensitive element inside the air pressure sensor senses changes in ambient air pressure and converts them into an electrical signal. This signal is amplified and filtered by internal signal conditioning circuitry, and then transmitted to the phone's main processor via a specific interface (such as I2C or SPI). The main processor then transmits this air pressure data, along with other data such as sound signals, to the deep learning processing module for comprehensive analysis. The air pressure sensor plays a vital role when the user is in different geographical environments or scenarios. For example, during mountain climbing, as altitude increases, air pressure gradually decreases. After detecting the air pressure change, the system can combine it with sound signals to determine the noise characteristics such as wind noise and adjust noise reduction strategies accordingly. When traveling by plane, air pressure changes drastically during takeoff and landing; the sensor can capture these changes in time, helping the system adapt to environmental noise at different stages. During the elevator's ascent and descent, rapid changes in air pressure can also be detected by sensors, enabling the system to process the noise generated by the elevator's operation more accurately.

[0063] The temperature sensor monitors the ambient temperature. The temperature sensor has a measurement range of -40℃ to 125℃ and an accuracy of ±0.5℃. It connects to the main processor via an I2C interface and transmits the collected air pressure and temperature data along with the sound signal to the deep learning processing module for comprehensive analysis.

[0064] Specifically, temperature also affects sound propagation and noise characteristics. Under different temperature conditions, air density, sound speed, and other factors change, thus altering the propagation and attenuation patterns of sound and affecting the characteristics of environmental noise. By monitoring ambient temperature, the system can comprehensively consider the impact of temperature factors on noise, further improving its ability to analyze and process environmental noise. For example, the same environmental noise may exhibit different characteristics in hot summers and cold winters due to temperature differences. A temperature sensor is a device that senses ambient temperature and converts it into an electrical signal. Common temperature sensors include thermocouples and thermistors, which utilize the temperature characteristics of materials to measure temperature. The measurement range is -40℃ to 125℃, covering the temperature range of most natural environments and everyday usage scenarios. The accuracy is ±0.5℃, ensuring the accuracy of the measurement results, enabling the system to perform noise analysis and processing based on relatively accurate temperature data. The temperature sensor senses changes in ambient temperature through its internal sensitive element, converting the temperature signal into an electrical signal. This electrical signal is processed by a signal conditioning circuit and then transmitted to the phone's main processor via an I2C interface. The main processor transmits the collected temperature data, along with sound signals and air pressure data, to the deep learning processing module. This module performs comprehensive analysis to account for the impact of temperature on environmental noise. Temperature sensors play a crucial role in different seasons and environments. In hot summer conditions, sound propagation speed may increase, altering noise propagation characteristics. Upon detecting high temperatures, the system can adjust its noise reduction strategy based on other data to adapt to the noise characteristics of the high-temperature environment. In cold winter conditions, lower temperatures and higher air density can affect sound propagation. Sensors allow the system to promptly detect temperature changes and optimize noise reduction. Furthermore, in specific scenarios, such as near heat-generating equipment (e.g., computers, engines), temperatures rise. Temperature sensors can detect these localized temperature changes, helping the system to process surrounding noise more accurately.

[0065] Preferably, the deep learning processing module further comprises:

[0066] The online update mechanism utilizes an incremental learning algorithm. When the mobile phone collects new environmental sound and motion data, it does not need to retrain the entire model, but instead performs incremental updates based on the existing model parameters.

[0067] Specifically, in the active noise cancellation control system of smartphones, the environment is complex and ever-changing, with new noise scenarios constantly emerging. Traditional model training methods often require retraining the entire model when encountering new data, which is time-consuming, consumes significant computing resources, and may even affect the normal use of the phone. The online update mechanism, however, utilizes an incremental learning algorithm to incrementally update the existing model parameters as the phone collects new environmental sound and motion data, without retraining the entire model. This allows the system to quickly adapt to new noise environments, adjust noise cancellation strategies promptly, improve noise cancellation performance, and save computing resources and time, ensuring the normal performance of the phone. This mechanism is based on an incremental learning algorithm. Incremental learning is a machine learning method that allows the model to continuously learn as it receives new data, rather than learning all data at once. In this scenario, when new environmental sound and motion data arrive, the model fine-tunes its existing parameters based on this new data, rather than completely discarding the old model and retraining. While specific module parameters are not explicitly mentioned, the performance of the incremental learning algorithm may be affected by parameters such as the learning rate. The learning rate controls the magnitude of parameter adjustments made by the model with each update. A suitable learning rate allows the model to quickly adapt to new data while avoiding over-adjustment that could lead to model instability. When the phone collects new environmental sound and motion data, this data is transmitted to the deep learning processing module. The module first preprocesses the new data to ensure compatibility with the existing model's data format. Then, the incremental learning algorithm calculates the impact of the new data on the model parameters and updates them based on certain rules (such as gradient descent). The updated model parameters are then used for subsequent noise reduction processing. For example, when a user travels from one city to another, the environmental noise levels may differ, such as different traffic noise patterns and the characteristics of crowd noise. An online update mechanism allows the phone system to quickly update the noise reduction model based on newly collected environmental data upon arriving in a new city, adapting to the new noise environment. Similarly, environmental noise varies with the seasons; for example, there may be more cicada chirping in summer and more noticeable wind in winter. This mechanism allows the system to adjust in a timely manner to ensure good noise reduction performance.

[0068] Model fusion technology combines multiple deep learning models with different structures or trained on different datasets. Based on the denoising performance of each model in different scenarios, different weights are assigned to each model, and the outputs of multiple models are combined. Deep learning models with different structures or trained on different datasets may have different advantages in different scenarios. A single model often struggles to achieve optimal denoising performance in all scenarios. Model fusion technology combines multiple such models, assigning different weights to each model based on its denoising performance in different scenarios, and combining the outputs of multiple models. This fully leverages the strengths of each model, improving the overall denoising performance of the system in various complex scenarios and enhancing the system's robustness and adaptability. This technology involves multiple deep learning models with different structures or trained on different datasets, such as convolutional neural networks (CNNs) or recurrent neural networks (RNNs) with different numbers of layers and neurons. These models can be trained on different environmental noise datasets; for example, some models are trained on urban traffic noise datasets, while others are trained on shopping mall noise datasets. The main module parameter is the weight assigned to each model. The weight allocation needs to be determined based on the denoising performance of each model in different scenarios. For example, in a specific scenario, model A performs better in noise reduction, so its weight in that scenario will be relatively higher; while model B performs worse, so its weight will be lower. First, multiple different deep learning models are integrated into a deep learning processing module. When processing environmental sound and motion data, each model processes the data and outputs its own noise reduction result. Then, the system evaluates the performance of each model in the current scenario according to preset evaluation metrics (such as a noise reduction effect evaluation function). Based on the evaluation results, appropriate weights are assigned to each model. Finally, the outputs of each model are weighted and summed according to their assigned weights to obtain the final noise reduction result. In complex and changing environments, model fusion technology plays a crucial role. For example, in a city square scenario with both traffic noise and crowd noise, different models may have different processing capabilities for traffic noise and crowd noise. Through model fusion technology, higher weights can be assigned to models more effective at handling traffic noise, and appropriate weights can be assigned to models more effective at handling crowd noise, thereby combining the advantages of each model to achieve better noise reduction results. For example, on an airplane, there is both low-frequency noise from the engine and high-frequency noise such as conversations in the cabin. Model fusion technology can help the system better cope with this complex noise environment.

[0069] Preferably, the noise reduction signal generation module further includes:

[0070] The sub-band decomposition unit divides the collected ambient sound signal and the generated noise-reduced signal into multiple sub-bands according to frequency. The frequency range of 0Hz-20kHz is divided into 16 sub-bands, and noise reduction processing is performed in each sub-band. The adaptive filter coefficients are independently adjusted according to the noise characteristics of different sub-bands.

[0071] Specifically, environmental noise typically contains rich frequency components, with different frequencies exhibiting different characteristics and origins. The subband decomposition unit divides the acquired environmental sound signal and the generated noise-reduced signal into multiple subbands according to frequency, enabling refined processing of the noise characteristics of each subband. This approach avoids the limitations of uniformly processing the entire broadband signal, significantly improving the targeting and effectiveness of noise reduction. By independently adjusting the adaptive filter coefficients within each subband, the system can more accurately match the variations in noise at different frequencies, thereby achieving superior noise reduction performance. This unit employs a frequency domain decomposition model, using mathematical methods such as Fourier transform to divide the 0Hz-20kHz audio signal spectrum into multiple subbands. In this design, 16 subbands are chosen, each covering a specific frequency range. This division method achieves a good balance between computational complexity and noise reduction performance, ensuring refined processing of noise at different frequencies without excessive computation affecting the system's real-time performance. The main parameters are the number of subbands (16 subbands) and the frequency range corresponding to each subband. The frequency range of 0Hz-20kHz is uniformly divided into 16 sub-bands, each with a frequency span of approximately 1.25kHz (20kHz ÷ 16). In practical applications, this uniform division method can adapt well to most common noise spectrum distributions. However, for some noise scenarios with specific frequency characteristics, non-uniform division can be performed as needed to further optimize noise reduction performance. First, the raw sound signal acquired by the environmental sound acquisition module enters the sub-band decomposition unit. Here, the signal is converted from the time domain to the frequency domain through Fourier transform, and then divided into 16 sub-bands according to the predetermined frequency range. For each sub-band, its noise characteristics, such as noise intensity and frequency distribution, are calculated. Based on these characteristics, the coefficients of the adaptive filter are independently adjusted so that the filter can generate a noise-reduced signal with the same amplitude but opposite phase as the noise in that sub-band. Finally, the noise-reduced signals of each sub-band are converted back to the time domain through inverse Fourier transform and synthesized to obtain the complete noise-reduced signal output. The sub-band decomposition unit demonstrates strong advantages in various complex noise environments. For example, in a factory workshop, environmental noise includes low-frequency rumbling from machine operation, high-frequency sharp noise from tool collisions, and mid-frequency noise caused by equipment vibration. Through subband decomposition, the system can process noise in different frequency bands separately, effectively reducing interference from various types of noise. Inside a subway car, there is both low-frequency noise from train operation and high-frequency sounds such as announcements and passenger conversations. The subband decomposition unit can precisely process noise in different frequency ranges, improving passenger auditory comfort.

[0072] The phase compensation circuit is used to accurately compensate the phase of the noise-reduced signal, so that the noise-reduced signal and the ambient noise are accurately aligned in the time domain. The phase compensation range is ±180° and the resolution is 0.1°.

[0073] Specifically, precise time-domain alignment between the noise-reduced signal and ambient noise is crucial for effective noise reduction. Even if the amplitude of the noise-reduced signal is equal to that of the ambient noise, a phase deviation will prevent complete cancellation during superposition, thus affecting the noise reduction effect. Phase compensation circuits precisely compensate for the phase of the noise-reduced signal, ensuring accurate time-domain alignment and maximizing mutual cancellation, thereby significantly improving the performance of the noise reduction system. The phase compensation circuit employs a phase adjustment model, adjusting the phase of the noise-reduced signal through specific circuit structures or digital signal processing algorithms. Its core principle is to calculate the phase value to be compensated based on the phase difference between the ambient noise and the noise-reduced signal, and then adjust the phase of the noise-reduced signal through corresponding circuit components or algorithm operations. Key parameters include the phase compensation range (±180°) and resolution (0.1°). The ±180° phase compensation range ensures that the circuit can handle various possible phase deviations, effectively compensating for both leading and lagging phase differences. A resolution of 0.1° ensures the accuracy of phase compensation, enabling fine-tuning of the phase to meet the stringent requirements for phase alignment under varying noise environments. During the noise reduction signal generation process, the phase compensation circuit monitors the phase difference between the ambient noise signal and the generated noise-reduced signal in real time. A specialized phase detection algorithm calculates the current phase deviation. Based on this deviation, the circuit uses internal phase adjustment elements (such as variable delay lines and digital phase shifters) to adjust the phase of the noise-reduced signal. The adjusted noise-reduced signal is aligned with the ambient noise as precisely as possible in the time domain before being output to cancel out the ambient noise. Throughout this process, phase detection and compensation are dynamic and real-time, adapting to the constantly changing characteristics of ambient noise. In practical applications, the phase compensation circuit plays a crucial role in various scenarios. For example, in indoor environments, due to factors such as sound reflection, there may be complex phase relationships between ambient noise and the sound signal captured by the mobile phone. The phase compensation circuit can precisely adjust the phase of the noise-reduced signal so that it can effectively cancel out the reflected ambient noise, improving the indoor noise reduction effect. In noisy outdoor environments, such as on the street, noise from different directions may undergo different phase changes during propagation. The phase compensation circuit can dynamically adjust the phase of the noise reduction signal based on the phase difference detected in real time, ensuring effective noise reduction under various complex conditions.

[0074] Preferably, the noise reduction execution module further comprises:

[0075] An active feedback control mechanism is used to place an additional microphone near the speaker to collect the mixed signal of the noise reduction signal output by the speaker and the residual ambient noise in real time. This mixed signal is fed back to the noise reduction signal generation module, compared and analyzed with the original ambient sound signal, and the amplitude and phase of the noise reduction signal are further adjusted to form a closed-loop control.

[0076] The multi-mode sound strategy switches between different sound modes according to the phone's usage scenarios and user needs. In call scenarios, it adopts a mono sound mode to focus on canceling noise in the direction of the call; in music playback scenarios, it switches to stereo sound mode to ensure the stereo effect of the music while reducing noise; in noisy outdoor environments, it automatically increases the speaker volume and enhances the low-frequency noise reduction effect.

[0077] Specifically, in the active noise cancellation control system of smartphones, these two characteristics of the noise cancellation execution module greatly improve the noise cancellation effect and user experience.

[0078] The active feedback control mechanism is of great significance. It adds an extra microphone near the speaker to collect the mixed signal of the noise-reduced signal output by the speaker and the residual ambient noise in real time. This mixed signal is fed back to the noise-reducing signal generation module for comparison and analysis with the original ambient sound signal. The process is as follows: the extra microphone continuously collects and transmits the mixed signal; the noise-reducing signal generation module compares it with the original ambient sound signal, calculates the difference between the two, and further adjusts the amplitude and phase of the noise-reducing signal accordingly, forming a closed-loop control. In practical applications, such as on a subway, where ambient noise is constantly changing, the active feedback control mechanism can continuously optimize the noise-reducing signal, continuously reduce noise interference, and ensure consistently good noise reduction performance.

[0079] The multi-mode sound strategy is equally practical, flexibly switching sound modes based on the phone's usage scenario and user needs. In implementation, the system analyzes data from the phone's sensors and user actions to determine the surrounding environment. During calls, the system automatically switches to mono mode to focus on canceling out directional noise. For example, in noisy street calls, it effectively reduces ambient noise interference, improving call clarity. In music playback, it switches to stereo mode, reducing environmental noise while maintaining a stereo effect, allowing users to immerse themselves in the music. In noisy outdoor environments, the system automatically increases speaker volume and enhances low-frequency noise reduction; for instance, near construction sites, it effectively blocks low-frequency machine noise, ensuring a superior listening experience.

[0080] Preferably, the system also includes:

[0081] The user preference settings module allows users to personalize the active noise cancellation system through a mobile application. Users can select different noise reduction intensity levels, such as light, medium, and heavy noise reduction, and customize the noise reduction priority for certain specific frequency noises. The system will adjust the parameters of the deep learning processing module and the noise reduction signal generation module according to the user's settings to meet the user's personalized needs.

[0082] The scene recognition module uses machine learning algorithms to identify the environment in which the phone is located based on data collected by the ambient sound acquisition module, motion state perception module and other sensors. Different noise reduction parameter configuration files are pre-stored for different scenes, and the system automatically calls the corresponding configuration file.

[0083] Specifically, in this smartphone active noise cancellation control system, the user preference setting module and scene recognition module greatly enhance the system's personalization and intelligence.

[0084] The user preference settings module is of great significance, as it empowers users to customize their active noise cancellation experience. Users can easily select different noise cancellation levels—light, moderate, and heavy—via a mobile application, and also customize the noise cancellation priority for specific frequencies. The process involves the user completing these settings in the application, which are then transmitted to the system. Based on the user's settings, the system precisely adjusts the parameters of the deep learning processing module and the noise cancellation signal generation module to match the user's individual needs. For example, some users are extremely sensitive to high-frequency sharp noise; they can increase the noise cancellation priority for specific high-frequency noise in the settings, and the system will then focus on processing this type of noise, creating a more comfortable listening environment for the user.

[0085] The scene recognition module is equally crucial. Leveraging machine learning algorithms, it integrates data from the environmental sound acquisition module, motion perception module, and other sensors to accurately identify the phone's surrounding environment. The process involves the system continuously collecting data from various sensors, which the machine learning algorithm analyzes, compares, and performs pattern recognition to determine the current environment. For different scenarios, the system pre-stores corresponding noise reduction parameter configuration files. Once a scene is identified, the appropriate configuration file is automatically invoked. In real-world usage, when a user moves from a quiet library to a noisy street, the scene recognition module quickly identifies the change and automatically calls the noise reduction parameter configuration file suitable for noisy street environments. The system then adjusts its noise reduction strategy accordingly, increasing noise reduction intensity and optimizing the noise reduction frequency range to ensure optimal noise reduction performance in various scenarios and improve the user experience.

[0086] Preferably, the system adopts a distributed computing architecture:

[0087] Some computing tasks are handled by the phone's local processor, which uses a high-performance mobile CPU combined with GPU acceleration to quickly execute tasks with high real-time requirements.

[0088] Complex deep learning model computation tasks, such as the operations of convolutional neural networks and recurrent neural networks, are transmitted wirelessly to a cloud server for processing, and the processing results are returned to the mobile phone to generate noise-reduced signals.

[0089] Specifically, the distributed computing architecture adopted by the active noise cancellation control system of this smartphone greatly optimizes the system performance, and has outstanding performance in terms of significance, implementation process and application scenarios.

[0090] In terms of significance, the distributed computing architecture fully leverages the advantages of the phone's local processor and cloud servers. The phone's local processor uses a high-performance mobile CPU combined with GPU acceleration, enabling it to quickly execute tasks with high real-time requirements. This ensures the system's timely processing of data such as environmental sound acquisition and motion perception, guaranteeing the real-time performance of noise reduction. Meanwhile, transmitting complex deep learning model computation tasks, such as the operations of convolutional neural networks and recurrent neural networks, to the cloud server for processing via wireless network reduces the computational burden on the phone's local storage, utilizes the powerful computing resources of the cloud to improve computational efficiency, and ultimately generates high-quality noise-reduced signals.

[0091] The implementation process is as follows: When the system starts working, after the environmental sound acquisition module and motion state perception module collect data, basic data processing tasks with high real-time requirements, such as preliminary screening and format conversion of the collected data, are processed quickly by the phone's high-performance mobile CPU combined with GPU acceleration. For complex computational tasks requiring deep analysis, such as convolutional neural network and recurrent neural network operations, these tasks are transmitted to a cloud server via a wireless network. The cloud server, with its powerful computing capabilities, processes these tasks and returns the results to the phone via the network. The phone then uses these results to generate a noise-reduced signal.

[0092] In specific use cases, when a user is on the subway, the surrounding environmental noise is complex and changes in real time. In this situation, the phone's local processor quickly processes data from the environmental sound acquisition module and motion perception module, promptly capturing instantaneous changes in noise and the phone's movement. Complex deep learning model computation tasks, such as analyzing the characteristics of noise at different frequencies and predicting noise trends, are handled by the cloud server. The cloud server quickly completes the calculations and returns the results, which the phone uses to generate accurate noise reduction signals, effectively reducing subway noise and providing a quiet auditory environment for the user. In even more complex noise environments, such as outdoor construction sites, this distributed computing architecture can also operate efficiently, ensuring the system performs excellently in noise reduction tasks in various complex environments.

[0093] Preferably, the system also includes:

[0094] The power management module employs a dynamic power management strategy to assess the power consumption characteristics of each module in the active noise cancellation system. When the ambient noise is low or the phone is stationary, it reduces the computing frequency of the deep learning processing module to decrease GPU power consumption. When the amount of data collected by the microphone and sensors is small, it reduces their sampling frequency to save power. At the same time, it optimizes the power output of the speaker execution module and dynamically adjusts the output power according to the noise cancellation requirements.

[0095] Specifically, the power management module plays an indispensable role in this smartphone active noise cancellation control system. Its significance lies in the fact that while the various modules of the active noise cancellation system consume power during operation, the power management module employs a dynamic power management strategy to effectively improve the phone's battery life and prevent excessive power consumption due to the active noise cancellation function.

[0096] In terms of implementation, the system continuously monitors ambient noise and the phone's status. When low ambient noise is detected, indicating a reduced need for real-time noise reduction and processing intensity, or when the phone is stationary and the motion sensing module's data changes little, the power management module reduces the computational frequency of the deep learning processing module, decreasing the GPU's resource consumption for complex calculations and thus reducing power consumption. When the amount of data collected by the microphone and sensors is small, their sampling frequency is reduced, satisfying basic data acquisition needs while minimizing power consumption. For the speaker module, the power management module dynamically adjusts its power output based on current noise reduction requirements; if the noise reduction requirement is low, the output power is reduced.

[0097] In specific use cases, such as in a quiet library with extremely low ambient noise, the power management module activates a dynamic strategy to reduce the computational frequency of the deep learning processing module, correspondingly reducing the sampling frequency of the microphone and sensors, and decreasing the speaker output power. This significantly saves power while maintaining basic noise cancellation functionality. The same principle applies when the user places the phone aside and it remains stationary, avoiding unnecessary power consumption, extending battery life, and ensuring the phone still has sufficient power when the user needs to use the active noise cancellation function.

[0098] On the other hand, this embodiment also provides an active noise cancellation control method for smartphones, the method comprising:

[0099] Step S1: Through the ambient sound acquisition module, motion state perception module and other auxiliary sensors, the ambient sound signal, mobile phone motion data and environmental parameter data are synchronously acquired at the set frequency. The ambient sound acquisition module acquires the sound signal at a sampling rate of not less than 48kHz, the motion state perception module acquires the motion data at a frequency of not less than 100Hz, and other sensors acquire data at their respective optimal response frequencies. The data are then transmitted to the mobile phone main processor through the corresponding communication interface.

[0100] Step S2: Preprocess the collected raw data, including noise removal, data calibration, and normalization. For ambient sound signals, digital filtering technology is used to remove high-frequency or low-frequency noise. The motion data is calibrated to eliminate sensor drift error. All data are normalized to ensure that the data are within the same numerical range.

[0101] Step S3: Input the preprocessed data into the deep learning processing module. Use a hybrid model combining convolutional neural networks and recurrent neural networks to extract spatial and temporal features from the data, analyze the type, intensity and trend of environmental noise, and optimize model parameters by continuously learning from training data.

[0102] Step S4: Based on the output of the deep learning processing module, generate a noise reduction signal with the same amplitude and opposite phase as the environmental noise. Using an adaptive filter, update the filter coefficients using the least mean square algorithm or recursive least squares algorithm according to the real-time changes in the environmental noise, and adjust the noise reduction signal.

[0103] Step S5: The generated noise reduction signal is amplified to a suitable power level by an audio power amplifier to drive the output of the miniature speaker of the noise reduction execution module to cancel out the ambient noise. At the same time, according to the active feedback control mechanism, the mixed signal after the speaker output is collected in real time and fed back to the noise reduction signal generation step to further optimize and adjust the noise reduction signal.

[0104] Step S6: Read the user's personalized settings in the user preference settings module, including noise reduction intensity level and noise reduction priority for specific frequencies. Based on the user settings, adjust the relevant parameters in the deep learning processing steps and the noise reduction signal generation steps so that the active noise reduction system meets the user's personalized needs.

[0105] Step S7: Use the scene recognition module to identify the environment scene where the mobile phone is located. Based on the recognition result, call the corresponding configuration from the pre-stored noise reduction parameter configuration files of different scenes to optimize and adjust the parameters in the entire active noise reduction control process to adapt to the noise reduction needs of different environmental scenes.

[0106] Obviously, those skilled in the art can make various modifications and variations to the embodiments of this application without departing from the spirit and scope of this application. Therefore, if these modifications and variations to the embodiments of this application fall within the scope of the claims of this application and their equivalents, this application also intends to include these modifications and variations.

Claims

1. An active noise control system for a smartphone, the system comprising: Comprise: The system comprises: Ambient sound acquisition module: set multiple high-sensitivity microphones in different positions of the mobile phone to form a spatial array, including three omnidirectional microphones and two directional microphones, collect ambient sound signals in all directions and multiple angles, the microphone has wide frequency response characteristics, the frequency response range is 20Hz-20kHz, and the sampling rate is not less than 48kHz; Motion state sensing module: integrate accelerometer and gyroscope to monitor the acceleration and angular velocity motion parameters of the mobile phone in real time, the accelerometer measurement range is ±16g, and the resolution is 0.001g; the gyroscope measurement range is ±2000° / s, and the resolution is 0.0625° / s, the module is connected with the mobile phone main processor through I2C or SPI communication interface, and the collected motion data is transmitted at a frequency of not less than 100Hz; Deep learning processing module: a hybrid model architecture combining convolutional neural network and recurrent neural network is adopted, the convolutional neural network part includes convolutional layer and pooling layer, which is used to extract spatial features in ambient sound signals, and each convolutional layer is connected with ReLU activation function; the pooling layer adopts maximum pooling method, and the pooling kernel size is 2x2; the recurrent neural network part uses long short-term memory network unit, which has 256 memory units, and is used to process time sequence features of sound signals and capture dynamic change rule of noise; the module uses mobile phone GPU to accelerate calculation, and analyzes the collected sound and motion data in real time without affecting the normal use performance of the mobile phone; The system also comprises a noise reduction signal generation module: based on the output results of the deep learning processing module, a noise reduction signal with equal amplitude and opposite phase to the ambient noise is generated, the noise reduction signal is adjusted through an adaptive filter, the filter uses least mean square algorithm or recursive least square algorithm, and the filter coefficients are updated in real time according to the change of ambient noise; at the same time, combined with the audio power amplifier of the mobile phone, the noise reduction signal is amplified to an appropriate power level to drive the loudspeaker to output; The noise reduction execution module is composed of multiple micro loudspeakers, which are distributed near the ear or sound area of the mobile phone, the loudspeaker adopts flat design to adapt to the limited space inside the mobile phone, the frequency response range matches that of the microphone of the ambient sound acquisition module, which is 20Hz-20kHz, and the loudspeaker automatically adjusts the sound mode and volume according to different use scenarios of the mobile phone; The ambient sound acquisition module further comprises: Atmospheric pressure sensor for detecting environmental pressure changes to assist in judging the type of environmental noise, the atmospheric pressure sensor measures in the range of 300hPa-1100hPa, with a resolution of 0.1hPa and a measurement accuracy of ±0.3hPa; Temperature sensor for monitoring ambient temperature, the temperature sensor measures in the range of-40°C-125°C with an accuracy of ±0.5°C, and is connected with the main processor through I2C interface to transmit the collected pressure and temperature data to the deep learning processing module for comprehensive analysis together with the sound signal; The deep learning processing module also has: The online updating mechanism uses an incremental learning algorithm, and when new environmental sound and motion data are collected by the mobile phone, the entire model does not need to be retrained, but is incrementally updated based on the existing model parameters; The model fusion technology fuses multiple deep learning models of different structures or trained on different data sets, assigns different weights to each model according to the noise reduction performance of each model in different scenes, and integrates the output results of multiple models; The noise reduction signal generation module further comprises: A sub-band decomposition unit divides the collected environmental sound signal and the generated noise reduction signal into multiple sub-bands according to frequency, divides the frequency range of 0Hz-20kHz into 16 sub-bands, and performs noise reduction processing in each sub-band. For different noise characteristics of sub-bands, the adaptive filter coefficients are independently adjusted; A phase compensation circuit is used to accurately compensate the phase of the noise reduction signal, and to accurately align the noise reduction signal and the environmental noise in the time domain. The phase compensation range is ±180°, and the resolution is 0.1°. The noise reduction execution module further comprises: An active feedback control mechanism is provided near the loudspeaker, which collects the mixed signal of the noise reduction signal output by the loudspeaker and the residual environmental noise in real time, feeds the mixed signal back to the noise reduction signal generation module, compares and analyzes the original environmental sound signal, further adjusts the amplitude and phase of the noise reduction signal, and forms a closed loop control; A multi-mode sound emission strategy is used to switch different sound emission modes according to the use scene and user demand of the mobile phone. In the call scene, a single-channel sound emission mode is used to concentrate on canceling the noise in the call direction. In the music playing scene, the stereo sound emission mode is switched to, and the stereo effect of the music is ensured while the noise is reduced. In a noisy outdoor environment, the loudspeaker volume is automatically increased, and the low-frequency noise reduction effect is enhanced. The system further comprises: A user preference setting module, in which the user sets the active noise reduction system through the mobile phone application program. The user selects different noise reduction intensity levels, light, medium and heavy noise reduction, and customizes the noise reduction priority of certain specific frequency noise. The system adjusts the parameters of the deep learning processing module and the noise reduction signal generation module according to the user settings to meet the user's individual needs. A scene recognition module uses machine learning algorithms to identify the environment scene of the mobile phone according to the data collected by the environmental sound collection module, the motion state perception module and other sensors. Different noise reduction parameter configuration files are pre-stored for different scenes, and the system automatically calls the corresponding configuration file. The system uses a distributed computing architecture: Some computing tasks are handled by the local processor of the mobile phone. The local processor uses a high-performance mobile CPU combined with GPU acceleration to quickly execute tasks with high real-time requirements. Complex deep learning model calculation tasks, convolutional neural network and recurrent neural network operations, are transmitted to the cloud server for processing through a wireless network, and the processing results are returned to the mobile phone for generating noise reduction signals. The system further comprises: The power management module adopts a dynamic power management strategy for the power consumption characteristics of each module of the active noise reduction system. When the ambient noise is low or the mobile phone is in a stationary state, the operation frequency of the deep learning processing module is reduced to reduce the power consumption of the GPU. When the amount of data collected by the microphone and sensor is small, the sampling frequency is reduced to save power. At the same time, the power output of the speaker execution module is optimized, and the output power is dynamically adjusted according to the noise reduction demand.

2. An active noise control method of a smartphone, characterized by, The method comprises: Step S1, synchronously collecting environmental sound signals, mobile phone motion data and environmental parameter data at a set frequency through an environmental sound collection module, a motion state perception module and other auxiliary sensors. The environmental sound collection module collects sound signals at a sampling rate of not less than 48 kHz, the motion state perception module collects motion data at a frequency of not less than 100 Hz, and other sensors collect data at their respective optimal response frequencies, and these data are transmitted to the mobile phone main processor through corresponding communication interfaces; Step S2, preprocessing the collected original data, including removing noise interference, data calibration, normalization operation, for the environmental sound signals, using digital filtering technology to remove high or low frequency noise; calibrating the motion data to eliminate sensor drift error; normalizing all data to make the data in the same numerical range; Step S3, inputting the preprocessed data into the deep learning processing module, using a hybrid model combining convolutional neural network and recurrent neural network to extract spatial and temporal features in the data, analyzing the type, intensity and trend of environmental noise, and continuously learning and training the data to optimize the model parameters; Step S4, generating a noise reduction signal with the same amplitude and opposite phase as the environmental noise according to the output result of the deep learning processing module, using an adaptive filter to update the filter coefficients according to the real-time changes of the environmental noise using the least mean square algorithm or the recursive least square algorithm to adjust the noise reduction signal; Step S5, amplifying the generated noise reduction signal to an appropriate power level through an audio power amplifier to drive the micro speaker of the noise reduction execution module to output and cancel the environmental noise, and at the same time, according to the active feedback control mechanism, real-time collection of the mixed signal after the speaker output is fed back to the noise reduction signal generation step for further optimization and adjustment of the noise reduction signal; Step S6, reading the individualized settings of the user in the user preference setting module, including the noise reduction intensity level and the priority of specific frequency noise reduction, adjusting the related parameters in the deep learning processing step and the noise reduction signal generation step according to the user settings, so that the active noise reduction system meets the individualized needs of the user; Step S7, identifying the environment scene where the mobile phone is located using the scene recognition module, and according to the recognition result, calling the corresponding configuration from the pre-stored different scene noise reduction parameter configuration file to optimize and adjust the parameters in the whole active noise reduction control process, which is used to adapt to the noise reduction demand in different environment scenes.

Images