A denoising processing method for self-adaptive adjustment of hearing aid electric signal

Abstract

The present application relates to the technical fields of signal processing, and specifically to a denoising processing method for self-adaptive adjustment of hearing aid electric signals, which comprises: evaluating the speech analysis feasibility of the input electric signals of the hearing aid; if the effective speech analysis probability is lower than the preset probability threshold, generating a denoising output signal with speech credibility rating in response to the local frequency band attention area. The technical problems of fixed technical index configuration of multi-microphone array and spectrum analysis module, insufficient capture ability of weak speech signals, and inability to adapt to dynamic adjustment of complex acoustic environment are solved, multi-dimensional signal characteristics are collected to construct a scientific speech detectability benchmark, the speech analysis feasibility is accurately evaluated, the effective speech misfiltering rate is reduced, the sampling rate and frequency band range of the signal processing unit are adjusted, the precise registration of spectrum plane data and discrete frequency point data is synchronously completed, and the technical effects of real-time adaptation to different acoustic environments, noise suppression, and effective speech signal preservation are achieved.

Description

Technical Field

[0001] This invention relates to the field of signal processing technology, specifically to a noise reduction method for adaptive adjustment of electrical signals in hearing aids. Background Technology

[0002] As a core device for people with hearing impairments to improve their auditory experience and integrate into daily communication, hearing aids directly determine the user's auditory comfort and speech recognition efficiency through their electrical signal processing accuracy and adaptive adjustment capabilities. With the diversification of life scenarios, the acoustic environment in which users are located is becoming increasingly complex. Indoor and outdoor noise, multi-source interference and other factors can easily cause the speech signals received by hearing aids to be contaminated, seriously affecting the speech resolution effect.

[0003] In hearing aids with adaptive electrical signal adjustment and noise reduction, the use of single-source audio or noise signals limits the reliability of feasibility assessment for audio parsing in complex environments. This can easily lead to issues such as false filtering of effective audio or residual noise. Furthermore, the technical specifications of multi-microphone arrays and spectrum analysis modules lack specificity, making it impossible to dynamically adjust the sampling rate and frequency band based on the local frequency band audio distribution. This results in insufficient ability to capture weak audio signals, making it difficult to meet the auditory needs in complex scenarios.

[0004] In summary, existing technologies suffer from fixed technical specifications for multi-microphone arrays and spectrum analysis modules, insufficient ability to capture weak speech signals, and a tendency to misfilter effective speech and leave noise residue. They also fail to adapt to dynamic adjustments in complex acoustic environments. Summary of the Invention

[0005] This application provides a noise reduction processing method for adaptive adjustment of electrical signals in hearing aids, aiming to solve the technical problems in the prior art where the technical specifications of multi-microphone arrays and spectrum analysis modules are fixed, the ability to capture weak speech signals is insufficient, which easily leads to the false filtering of effective speech and noise residue, and the inability to adapt to dynamic adjustment in complex acoustic environments.

[0006] In view of the above problems, the technical solution to achieve the present application is as follows:

[0007] This application provides a noise reduction processing method for adaptive adjustment of hearing aid electrical signals. The method includes: acquiring a hearing aid input electrical signal, including speech signal energy, speech fundamental frequency characteristics, and background noise intensity, in the acoustic environment of the target user, and configuring a speech detectability criterion; connecting to a signal processing unit to evaluate the speech parsing feasibility of the hearing aid input electrical signal using the speech detectability criterion, wherein the signal processing unit includes a multi-microphone transceiver array and a real-time spectrum analysis processing module; when the effective speech parsing probability of the hearing aid input electrical signal is lower than a preset probability threshold, determining a local frequency band interest region under the speech signal energy distribution marker in the hearing aid input electrical signal; responding to the local frequency band interest region, adjusting the sampling rate and frequency band analysis range of the signal processing unit, registering and correcting the spectral surface data under the speech detectability criterion with the discrete frequency point data of the signal processing unit, and generating a noise-reduced output signal with a speech credibility rating.

[0008] In one possible implementation, an acoustic environment monitoring unit is used to obtain noise interference distribution information of the acoustic environment in which the target user is located; the noise interference distribution information is uploaded to an edge computing node, and the time-frequency domain filtering processing of the hearing aid input electrical signal is performed using the noise interference distribution information.

[0009] In a possible implementation, based on the multi-microphone transceiver array in the signal processing unit, microphone subarrays are deployed at equal intervals along the incident direction of the sound source. The first type of technical indicators corresponding to the microphone subarrays include directional gain and spatial resolution. Based on the real-time spectrum analysis processing module in the signal processing unit, fixed-point analysis channels are set in key speech segments. The second type of technical indicators corresponding to the fixed-point analysis channels include frequency band analysis bandwidth and spectrum sampling accuracy. The first type of technical indicators and the second type of technical indicators are jointly configured according to the speech signal energy and speech fundamental frequency characteristics in the hearing aid input electrical signal.

[0010] In a possible implementation, the speech enhancement trajectory is time-frequency aligned with the frequency sampling sequence of the signal processing unit, the audio processing protocol of the signal processing unit is adjusted synchronously, and the inter-frame processing delay is determined; the time-frequency synchronization accuracy is obtained through the inter-frame processing delay; based on the speech signal energy and background noise intensity in the hearing aid input electrical signal, the signal interpolation mechanism of the signal processing unit is configured with the time-frequency synchronization accuracy.

[0011] In one possible implementation, a multimodal sensor network is deployed in the acoustic environment where the target user is located. The multimodal sensor network is used to collect the hearing aid input electrical signal and auxiliary perception information. Based on the hearing aid input electrical signal and auxiliary perception information, a speech activity distribution map is generated. The speech activity distribution map and the spatial-semantic calibration information of the multimodal sensor network are used to perform time-frequency alignment between the speech enhancement trajectory and the frequency sampling sequence of the signal processing unit.

[0012] In possible implementations, the auxiliary perception information includes at least one of user motion state, spatial location, and scene semantic labels, and the auxiliary perception information is used to assist in identifying the acoustic environment.

[0013] In a possible implementation, when the acoustic interference fluctuation rate under the background noise intensity exceeds the preset fluctuation range, and the proportion of signal interpolation operation under the signal interpolation mechanism to the total number of frames is greater than the proportion threshold, a multi-channel cross-validation mechanism is triggered; the multi-channel cross-validation mechanism is used to determine that the speech has a confidence interval, and the confidence interval is used to update the speech credibility rating.

[0014] In a possible implementation, the spatial coordinates of the sound source are located in the multimodal sensing network; at the spatial coordinates of the sound source, the signal processing unit and the acoustic environment monitoring unit work together; the dynamic features of speech are collected through the spatial coordinates of the sound source, the impact of the acoustic interference fluctuation rate under the background noise intensity on speech parsing is analyzed, and the preset fluctuation range is configured.

[0015] In a possible implementation, when the acoustic interference fluctuation rate under the background noise intensity exceeds a preset fluctuation range, a noise avoidance control command is generated. The noise avoidance control command is used to trigger the reconfiguration of the spatial filtering parameters of the hearing aid. The spatial filtering parameters are determined based on the spatial coordinates of the sound source and the spatial azimuth angle of the strong interference sound source.

[0016] In a possible implementation, an edge computing node is invoked based on the spatial coordinates of the sound source, and the edge computing node is communicatively connected to the cloud audio processing center; the edge computing node receives the spatial coordinates of the sound source and the spatial azimuth angle of the strong interference sound source, performs sound field propagation simulation and prediction to determine the energy diffusion path; based on the energy diffusion path, the coverage range of the null region of beamforming is dynamically expanded, and gain suppression is applied to the frequency bands within the null region.

[0017] In summary, one or more technical solutions provided in this application achieve the following technical effects: collecting multi-dimensional signal features to construct a scientific speech detectability benchmark, accurately assessing the feasibility of speech parsing, reducing the effective speech false filtering rate, adjusting the sampling rate and frequency band range of the signal processing unit, simultaneously completing the accurate registration of spectral surface data and discrete frequency point data, adapting to different acoustic environments in real time, and preserving the effective speech signal while suppressing noise. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0019] Figure 1 This application provides a flowchart illustrating a noise reduction processing method for adaptive adjustment of electrical signals in hearing aids. Detailed Implementation

[0020] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. It should be understood that this application is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application. It should also be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, not all of them.

[0021] The embodiments are described in detail below with reference to the accompanying drawings, such as... Figure 1 As shown, this application provides a noise reduction processing method for adaptive adjustment of hearing aid electrical signals, wherein the method includes:

[0022] S1: In the acoustic environment where the target user is located, collect the hearing aid input electrical signal, including speech signal energy, speech fundamental frequency characteristics, and background noise intensity, and configure the speech detectability judgment criteria.

[0023] Specifically, speech signal energy refers to the amount of energy a speech signal possesses per unit time, usually measured by the square of the signal amplitude, reflecting the strength of the speech signal; speech fundamental frequency characteristics refer to the characteristics of the lowest frequency component in the speech signal. The fundamental frequency is the determining factor of speech pitch and plays an important role in distinguishing different people's voices and the intonation of speech; background noise intensity refers to the intensity of other sound signals besides the speech signal in the acoustic environment in which the target user is located. These noise signals may interfere with the clarity of the speech signal; the speech detectability criterion is a standard set based on the multi-dimensional signal characteristics such as the collected speech signal energy, speech fundamental frequency characteristics, and background noise intensity, used to evaluate whether the speech signal can be effectively parsed.

[0024] Execution steps: In the acoustic environment of the target user, the hearing aid's input electrical signals are collected using sensors and other devices, including speech signal energy, speech fundamental frequency characteristics, and background noise intensity. Collecting these multi-dimensional signal features is to construct a comprehensive benchmark for speech detectability assessment. This benchmark can more accurately evaluate the resolvability of speech signals in the current acoustic environment. For example, in a noisy restaurant environment, the background noise intensity may be high, while the speech signal energy may be relatively weak. In this case, the speech signal can be better identified through the speech fundamental frequency characteristics, thereby improving the accuracy of speech detectability assessment.

[0025] Furthermore, it also includes: using an acoustic environment monitoring unit to obtain noise interference distribution information of the acoustic environment in which the target user is located; uploading the noise interference distribution information to an edge computing node; and using the noise interference distribution information to perform time-frequency domain filtering processing on the hearing aid input electrical signal.

[0026] Specifically, an acoustic environment monitoring unit refers to a device used to monitor and analyze the acoustic environment of a target user, capable of detecting and recording the noise level and distribution in the environment; noise interference distribution information refers to the distribution of noise at different frequencies and spatial locations obtained by the acoustic environment monitoring unit, including information such as noise intensity, frequency range, and direction; edge computing nodes refer to computing nodes that perform data processing and analysis at the network edge, capable of quickly processing local data and reducing data transmission latency; time-frequency domain filtering processing refers to filtering the input electrical signal of the hearing aid in both time and frequency dimensions to remove noise interference and retain effective speech signals.

[0027] Execution steps: Obtain noise interference distribution information of the target user's acoustic environment through the acoustic environment monitoring unit. For example, in an environment with traffic noise and background music, the acoustic environment monitoring unit can detect that traffic noise is mainly concentrated in the low frequency band, while background music covers a wider frequency range. Upload the obtained noise interference distribution information to the edge computing node. The edge computing node can quickly process this data, avoiding the transmission of large amounts of data to the cloud or central server, thereby reducing processing time and network latency.

[0028] By performing time-frequency domain filtering on the hearing aid input signal through edge computing nodes, noise can be precisely removed based on noise interference distribution information. Specifically, if the noise is mainly concentrated in the low-frequency band, a low-pass filter can be set to remove this low-frequency noise; if the noise is intermittent, thresholding can be performed in the time domain to retain only the active periods of the speech signal. This time-frequency domain filtering can significantly improve the quality of the speech signal and reduce noise interference. Preferably, by monitoring and analyzing noise interference in the acoustic environment, precise noise filtering can be provided for the hearing aid input signal, thereby improving the clarity and intelligibility of the speech signal.

[0029] S2: Connect to the signal processing unit to evaluate the feasibility of speech analysis of the hearing aid input electrical signal based on the speech detectability determination benchmark. The signal processing unit includes a multi-microphone transceiver array and a real-time spectrum analysis and processing module.

[0030] Specifically, the signal processing unit refers to the device that transmits the acquired hearing aid input electrical signal to the signal processing unit. The signal processing unit includes a multi-microphone transceiver array and a real-time spectrum analysis and processing module. The multi-microphone transceiver array is an array of multiple microphones used to receive sound signals from different directions, enhancing the reception of speech signals. The real-time spectrum analysis and processing module performs real-time spectrum analysis on the signal, decomposing the signal into different frequency components to better understand the frequency characteristics of the signal.

[0031] Execution steps: The acquired hearing aid input electrical signal is transmitted to the signal processing unit, which includes a multi-microphone transceiver array and a real-time spectrum analysis and processing module. The multi-microphone transceiver array can use multiple microphones to receive sound signals from different angles, and enhance the directionality of the speech signal through beamforming and other technologies to reduce background noise interference. The real-time spectrum analysis and processing module performs real-time spectrum analysis on the signal, decomposing the signal into different frequency components to better understand the frequency characteristics of the signal, thereby providing more detailed information for subsequent feasibility assessment of speech parsing.

[0032] By working together with a multi-microphone transceiver array and a real-time spectrum analysis and processing module, the feasibility of speech signal parsing can be evaluated more accurately, providing a basis for noise reduction processing. Furthermore, in environments with multiple sound sources, the multi-microphone transceiver array can locate the direction of the speech signal, and the real-time spectrum analysis and processing module can analyze the frequency components of the speech signal, thereby more accurately identifying the speech signal, ensuring that noise reduction processing can effectively suppress noise while retaining the effective speech signal, and improving the feasibility of speech parsing.

[0033] Furthermore, the connected signal processing unit also includes: deploying microphone subarrays at equal intervals along the incident direction of the sound source based on the multi-microphone transceiver array in the signal processing unit, wherein the first type of technical indicators corresponding to the microphone subarrays includes directional gain and spatial resolution; setting fixed-point analysis channels in key speech segments based on the real-time spectrum analysis processing module in the signal processing unit, wherein the second type of technical indicators corresponding to the fixed-point analysis channels includes frequency band analysis bandwidth and spectrum sampling accuracy; and jointly configuring the first type of technical indicators and the second type of technical indicators according to the speech signal energy and speech fundamental frequency characteristics in the hearing aid input electrical signal.

[0034] Specifically, a multi-microphone transceiver array (MTA) refers to an array composed of multiple microphones used to receive sound signals from different directions, enhancing the reception of speech signals. A microphone subarray refers to a group of microphones deployed at equal intervals along a specific direction in a MTA, used to improve signal reception capability in that specific direction. Directional gain refers to the signal gain capability of a microphone subarray in a specific direction, used to enhance speech signals in that specific direction. Spatial resolution refers to the ability of a microphone subarray to distinguish signals from different spatial locations, used to improve the accuracy of sound source localization.

[0035] A real-time spectrum analysis processing module refers to a module that performs real-time spectrum analysis on a signal, decomposing the signal into different frequency components. A fixed-point analysis channel refers to a dedicated analysis channel set up in key speech segments, used for more detailed analysis of these frequency bands. The bandwidth of the frequency band analysis is the frequency range that the fixed-point analysis channel can analyze. The spectrum sampling accuracy refers to the precision of sampling frequency components in spectrum analysis. Joint configuration refers to configuring directional gain, spatial resolution, bandwidth of the frequency band analysis, and spectrum sampling accuracy simultaneously based on the speech signal energy and fundamental frequency characteristics to optimize the signal processing effect.

[0036] Execution steps: In the signal processing unit, the multi-microphone transceiver array is a key component for enhancing the voice signal. Furthermore, by deploying microphone subarrays at equal intervals along the direction of sound source incidence, the signal reception capability in a specific direction can be improved. For example, if the sound source is located in front, the reception effect of the voice signal in front can be enhanced by deploying microphone subarrays at equal intervals. Directional gain and spatial resolution are important technical indicators of microphone subarrays. Directional gain can enhance the voice signal in a specific direction and reduce noise interference in other directions. Spatial resolution can improve the accuracy of sound source localization and help users perceive the sound source more accurately.

[0037] Furthermore, in environments with multiple sound sources, directional gain can separate the speech signal from background noise, and spatial resolution helps users distinguish the location of different sound sources. The real-time spectrum analysis processing module is responsible for performing real-time spectrum analysis on the signal; fixed-point analysis channels are set up in key audio segments to perform more detailed analysis on these frequency bands. For example, key audio segments contain most of the speech information, the frequency band analysis bandwidth determines the frequency range of the analysis, and the spectrum sampling accuracy determines the accuracy of the analysis.

[0038] By setting a wider frequency band analysis bandwidth and higher spectrum sampling accuracy, the frequency characteristics of speech signals can be captured more accurately. Based on the speech signal energy and fundamental frequency characteristics in the hearing aid's input electrical signal, the directional gain, spatial resolution, frequency band analysis bandwidth, and spectrum sampling accuracy are jointly configured to optimize signal processing. Furthermore, if the speech signal energy is low, the directional gain and spectrum sampling accuracy can be increased to enhance the reception of the speech signal. If the fundamental frequency characteristics indicate that the speech signal is mainly concentrated in a specific frequency band, the frequency band analysis bandwidth can be adjusted to focus on the analysis of that frequency band. This joint configuration can significantly improve the clarity and intelligibility of the speech signal and reduce noise interference. Through this joint configuration, the technical specifications of the microphone subarray and spectrum analysis processing module are optimized, improving the signal-to-noise ratio of the speech signal.

[0039] Furthermore, it also includes: time-frequency alignment of the speech enhancement trajectory with the frequency sampling sequence of the signal processing unit, synchronous adjustment of the audio processing protocol of the signal processing unit, and determination of the inter-frame processing delay; obtaining the time-frequency synchronization accuracy through the inter-frame processing delay; and configuring the signal interpolation mechanism of the signal processing unit with the time-frequency synchronization accuracy based on the speech signal energy and background noise intensity in the hearing aid input electrical signal.

[0040] Specifically, the speech enhancement trajectory refers to the trajectory of the signal after speech enhancement processing in terms of time and frequency, reflecting the enhancement effect of the speech signal at different time and frequency points; the signal processing unit frequency sampling sequence refers to the sequence of signal samples collected by the signal processing unit at different frequency points; time-frequency alignment refers to aligning the speech enhancement trajectory with the signal processing unit frequency sampling sequence in terms of time and frequency to ensure that the two are synchronized in time and matched in frequency.

[0041] Audio processing protocols refer to the rules and standards followed by signal processing units when processing audio signals, including sampling rate and encoding format; inter-frame processing delay refers to the time delay between adjacent frames during signal processing; time-frequency synchronization accuracy refers to the precision of alignment in time and frequency, reflecting the accuracy and reliability of signal processing; signal frame interpolation mechanism refers to the mechanism for supplementing and repairing lost or damaged frames during signal processing to ensure the integrity and continuity of the signal.

[0042] Execution steps: Time-frequency alignment is required between the speech enhancement trajectory and the frequency sampling sequence of the signal processing unit. Furthermore, the speech enhancement trajectory may show a significant change in the frequency components of the speech signal at a certain point in time, while the frequency sampling sequence of the signal processing unit records the actual signal samples collected at that point. Time-frequency alignment ensures consistency between the two in time and frequency, thereby improving the accuracy of signal processing. For example, the speech enhancement trajectory may show a significant speech signal enhancement at a certain point, while the frequency sampling sequence may not have a corresponding signal at that frequency. Time-frequency alignment can adjust the frequency sampling sequence to match the speech enhancement trajectory.

[0043] The audio processing protocol of the signal processing unit is adjusted synchronously to adapt to the signal characteristics after time-frequency alignment. Furthermore, if the speech enhancement trajectory shows that the frequency range of the speech signal is wide, the sampling rate in the audio processing protocol can be adjusted to ensure that these high-frequency signals can be accurately captured. The effect of signal processing can be further improved by adjusting the audio processing protocol. At the same time as adjusting the audio processing protocol, the inter-frame processing delay needs to be determined. By determining the inter-frame processing delay, the rhythm and synchronization of signal processing can be better controlled.

[0044] The time-frequency synchronization accuracy is obtained by processing the inter-frame delay. Furthermore, a smaller inter-frame processing delay indicates better signal processing synchronization and higher time-frequency synchronization accuracy. The level of time-frequency synchronization accuracy directly affects the quality of signal processing and the user's auditory experience. Based on the speech signal energy and background noise intensity in the hearing aid's input electrical signal, the signal interpolation mechanism of the signal processing unit is configured with time-frequency synchronization accuracy. Specifically, if the speech signal energy is low and the background noise intensity is high, signal interpolation needs to be performed more frequently to ensure the integrity and intelligibility of the speech signal. Furthermore, the signal interpolation mechanism improves the processing effect of the speech signal, reduces signal loss or damage, ensures the integrity and continuity of the speech signal, and provides users with a clearer and more reliable speech signal.

[0045] Furthermore, the method includes: deploying a multimodal sensor network in the acoustic environment where the target user is located, wherein the multimodal sensor network is used to collect the hearing aid input electrical signal and auxiliary perception information; generating a speech activity distribution map based on the hearing aid input electrical signal and auxiliary perception information; and using the speech activity distribution map and the spatial-semantic calibration information of the multimodal sensor network to perform time-frequency alignment of the speech enhancement trajectory and the frequency sampling sequence of the signal processing unit.

[0046] Specifically, a multimodal sensor network refers to a network composed of various types of sensors that can collect multiple types of information, such as sound, images, and motion. Auxiliary perception information refers to other perception information besides the hearing aid input electrical signal, such as the user's motion state, spatial location, and scene semantic labels. A speech activity distribution map is a visual representation generated based on the hearing aid input electrical signal and auxiliary perception information, showing the activity level of the speech signal at different times and spatial locations. Spatial-semantic calibration information refers to the annotation information in the multimodal sensor network regarding sensor location and the semantic content of perception information, used to determine the sensor's spatial location and the specific meaning of the perception information. Time-frequency alignment refers to aligning the speech enhancement trajectory with the frequency sampling sequence of the signal processing unit in time and frequency to ensure time synchronization and frequency matching.

[0047] Execution steps: Deploy a multimodal sensor network in the acoustic environment of the target user to comprehensively collect the hearing aid input electrical signals and auxiliary perception information. Furthermore, the multimodal sensor network may include a microphone for collecting sound signals, an accelerometer for sensing the user's motion state, and a camera for acquiring scene images. Through these sensors, rich environmental information is obtained, providing more comprehensive data support for speech processing. Based on the hearing aid input electrical signals and auxiliary perception information, a speech activity distribution map is generated. Furthermore, by analyzing the sound signals collected by the microphone and the scene images acquired by the camera, the activity level of the speech signal in different spatial locations can be determined. For example, in a conference room environment, the speech activity distribution map can display the active area of ​​the speech signal around the conference table.

[0048] By combining speech activity distribution maps and spatial-semantic calibration information from multimodal sensor networks, time-frequency alignment is performed between speech enhancement trajectories and the frequency sampling sequences of the signal processing unit. For example, based on the speech activity distribution map, enhancement trajectories of the speech signal at specific time and spatial locations are determined. Using spatial-semantic calibration information, these trajectories are precisely aligned with the frequency sampling sequences of the signal processing unit. This alignment ensures that the signal processing unit processes the speech signal at the correct time and frequency points, improving the processing accuracy and reliability. For instance, in environments with multiple sound sources, time-frequency alignment ensures that the signal processing unit only enhances the target speech signal while ignoring other interfering signals. Preferably, comprehensive perceptual information is collected to generate a speech activity distribution map, and this information is used for precise time-frequency alignment, thereby improving the speech signal processing effect.

[0049] Furthermore, the auxiliary perception information includes at least one of user motion state, spatial location, and scene semantic tags, and the auxiliary perception information is used to assist in identifying the acoustic environment.

[0050] Specifically, user motion state refers to the user's activity in the acoustic environment, such as whether the user is moving, the speed and direction of movement; spatial location refers to the user's specific location in the acoustic environment, which can be determined by positioning technologies such as GPS or indoor positioning systems; scene semantic label refers to the semantic description of the user's environment, such as whether the user is indoors or outdoors, in a quiet library or a noisy street; assisted recognition refers to using this auxiliary perception information to enhance the understanding and judgment of the acoustic environment, helping to more accurately identify and adapt to environmental changes.

[0051] Execution steps: In the process of acoustic environment recognition and processing of the hearing aid, the auxiliary perception information further includes at least one of user motion state, spatial location, and scene semantic label. For example, the user's motion state can be obtained through sensors such as accelerometers and gyroscopes. If the user is in motion, it means that the user is moving from one environment to another, such as moving from a quiet indoor area to a noisy street. This motion state information can help the hearing aid adjust its processing strategy in advance to adapt to the upcoming change in acoustic environment.

[0052] Spatial location information can be obtained through positioning technology. For example, in an indoor environment, Wi-Fi positioning or Bluetooth beacons can determine the user's specific location in the room. If the user is near a window, they will be affected by outdoor noise. The hearing aid can adjust its noise suppression strategy based on this location information. Scene semantic tags provide a higher level of environmental description. For example, if image recognition technology identifies that the user is in a restaurant environment, the hearing aid can automatically switch to a noise reduction mode suitable for the restaurant environment to enhance the clarity of the speech signal. The combined use of these auxiliary perception information significantly improves the hearing aid's accuracy in recognizing the acoustic environment.

[0053] Preferably, in complex environments with multiple sound sources, by combining the user's motion state and spatial location information, the hearing aid can more accurately locate the source of the speech signal, improve the reception quality of the speech signal, and at the same time, scene semantic tags help the hearing aid select the preset parameters most suitable for the current environment; through multi-dimensional auxiliary perception information, the hearing aid's adaptability and processing effect to the acoustic environment are enhanced, the clarity and intelligibility of the speech signal are improved, and a more natural and comfortable auditory experience is provided to the user.

[0054] S3: When the effective speech resolution probability of the hearing aid input electrical signal is lower than a preset probability threshold, a local frequency band interest area under the speech signal energy distribution mark in the hearing aid input electrical signal is determined; S4: In response to the local frequency band interest area, the sampling rate and frequency band analysis range of the signal processing unit are adjusted, and the spectral surface data under the speech detectability determination benchmark and the discrete frequency point data of the signal processing unit are registered and corrected to generate a denoised output signal with speech credibility rating.

[0055] Specifically, the effective speech resolution probability refers to the probability that a speech signal in the input electrical signal of a hearing aid can be correctly resolved. It is the result of evaluating the input signal using a speech detectability criterion. The preset probability threshold is a pre-set probability value used to determine whether the speech signal is within an acceptable resolution range. Speech signal energy distribution marking refers to marking the distribution of speech signal energy in different frequency bands to identify areas where speech signal energy is concentrated. The local frequency band interest area refers to the frequency band area with relatively high speech signal energy that needs to be focused on under the speech signal energy distribution marking.

[0056] Sampling rate refers to the frequency at which the signal processing unit collects signal samples per unit time. A higher sampling rate can provide more accurate signal information. Frequency band analysis range refers to the range of frequency components of the signal that the signal processing unit analyzes. Registration correction refers to aligning and correcting the spectral surface data under the speech detectability judgment benchmark with the discrete frequency point data of the signal processing unit to ensure that the two are synchronized in frequency and time. Speech credibility rating refers to rating the credibility of the speech signal in the denoised output signal, reflecting the reliability of the speech signal after denoising.

[0057] Execution steps: When the effective speech resolution probability of the hearing aid input electrical signal is lower than the preset probability threshold, it indicates that the current speech signal resolution effect is poor, and there is a situation where the speech signal is submerged by noise or misjudged. At this time, by identifying the local frequency band interest area through speech signal energy distribution marking, we can focus on the frequency band area with higher speech signal energy, avoid unnecessary processing of the entire frequency band, thereby improving processing efficiency and reducing speech distortion. For example, in an environment with background music and multiple people talking, the speech signal may be concentrated in certain specific frequency bands. By identifying the local frequency band interest area, these frequency bands can be processed in a targeted manner to improve the accuracy of speech resolution.

[0058] By adjusting the sampling rate and frequency band analysis range of the signal processing unit in response to the local frequency band area of ​​interest, speech signals can be captured and analyzed more accurately. For example, increasing the sampling rate can better process high-frequency speech signals; focusing the frequency band analysis range on key speech segments can improve the resolution accuracy of speech signals; and registering and correcting the spectral surface data under the speech detectability determination benchmark with the discrete frequency point data of the signal processing unit can ensure the synchronization of the two in frequency and time, further improving the resolution accuracy of speech signals.

[0059] Through registration correction, the speech signal in the spectral surface data is precisely aligned with the speech signal in the discrete frequency point data, reducing errors. A denoised output signal with speech reliability rating is generated, providing users with a more reliable and clearer speech signal and a reference index for speech reliability rating, thereby judging the reliability of the speech signal and enabling a deeper understanding of the speech content. Preferably, by focusing on the local frequency band area of ​​interest and adjusting signal processing parameters, the resolution accuracy and reliability of the speech signal are improved, while noise interference is reduced, providing users with a better listening experience.

[0060] Furthermore, the process of registering and correcting the spectral data under the speech detectability determination benchmark with the discrete frequency data of the signal processing unit to generate a denoised output signal with a speech credibility rating also includes: triggering a multi-channel cross-validation mechanism when the acoustic interference fluctuation rate under the background noise intensity exceeds a preset fluctuation range, and the proportion of signal interpolation operation under the signal interpolation mechanism to the total number of frames is greater than a proportion threshold; using the multi-channel cross-validation mechanism to determine that the speech has a confidence interval, and using the confidence interval to update the speech credibility rating.

[0061] Specifically, acoustic interference fluctuation rate refers to the variation range of background noise intensity over a certain period of time, reflecting the stability and dynamic characteristics of the noise; preset fluctuation range is a pre-set range of noise intensity fluctuation used to determine whether the noise is within an acceptable fluctuation range; signal frame interpolation mechanism refers to the mechanism for supplementing and repairing lost or damaged frames during signal processing; proportion threshold is a pre-set ratio value used to determine whether the frequency of signal frame interpolation is too high; multi-channel cross-validation mechanism refers to verifying the speech signal through multiple signal processing channels to improve the accuracy and reliability of speech signal detection; speech presence confidence interval refers to the probability interval of speech signal presence determined by the multi-channel cross-validation mechanism, reflecting the credibility of the speech signal's existence; speech credibility rating refers to rating the credibility of the speech signal, reflecting the reliability of the speech signal after denoising processing.

[0062] Execution steps: When the acoustic interference fluctuation rate under the background noise intensity exceeds the preset fluctuation range, it indicates that the dynamic changes of the noise are large, which may affect the stability and intelligibility of the speech signal. This means that the noise environment is very unstable, such as at a busy intersection or in a place with sudden noise. At the same time, if the proportion of signal frame interpolation operation under the signal interpolation mechanism to the total number of frames is greater than the proportion threshold, for example, the proportion threshold is set to 10%, but the actual proportion of frame interpolation operation reaches 15%, it indicates that the signal loss or damage is relatively serious, and it may be necessary to further verify the reliability of the speech signal.

[0063] To further explain, in this scenario, a multi-channel cross-validation mechanism is triggered. This mechanism validates the speech signal through multiple signal processing channels. For example, it can utilize different microphone channels in a multi-microphone transceiver array or combine different frequency band analysis channels in the signal processing unit to validate the speech signal from multiple angles. In this way, the presence or absence of a speech signal can be determined more accurately, improving the accuracy and reliability of speech signal detection. For instance, in an environment with multiple sound sources, the multi-channel cross-validation mechanism can distinguish which signals are genuine speech signals and which are noise interference. After determining the existence of a confidence interval for the speech signal through the multi-channel cross-validation mechanism, this confidence interval is used to update the speech credibility rating.

[0064] For example, a multi-channel cross-validation mechanism determines the presence of speech with a confidence interval of 90% to 95%, meaning that the speech signal is highly reliable and the speech credibility rating can be improved accordingly. This update mechanism can provide users with more accurate speech signal quality information, helping them to better understand and trust the output signal of the hearing aid. By monitoring noise fluctuations and signal interpolation, potential speech signal quality problems can be detected in a timely manner. The accuracy of speech signal detection is improved through the multi-channel cross-validation mechanism, and the speech credibility rating is updated to provide users with more reliable and clearer speech signals, thereby improving the user's auditory experience.

[0065] Furthermore, the method includes: locating the spatial coordinates of the sound source in the multimodal sensing network; coordinating the signal processing unit and the acoustic environment monitoring unit at the spatial coordinates of the sound source; acquiring dynamic speech features through the spatial coordinates of the sound source; analyzing the impact of acoustic interference fluctuation rate under the background noise intensity on speech parsing; and configuring the preset fluctuation range.

[0066] Specifically, the spatial coordinates of the sound source refer to the specific location of the sound source in three-dimensional space, usually represented by a coordinate system; the collaborative work between the signal processing unit and the acoustic environment monitoring unit refers to the data sharing and functional cooperation between these two units to achieve more efficient signal processing and environmental monitoring; the dynamic characteristics of speech refer to the dynamic changes of the speech signal in time and frequency, including the intensity, frequency variation, and duration of the speech signal; the acoustic interference fluctuation rate refers to the amplitude of the change in background noise intensity over a certain period of time, reflecting the stability and dynamic characteristics of the noise; the preset fluctuation range is a pre-set range of noise intensity fluctuation, used to determine whether the noise is within an acceptable fluctuation range.

[0067] Execution steps: The spatial coordinates of the sound source are located using sensors. Specifically, beamforming technology is used to determine the direction and distance of the sound source based on an array of multiple microphones, thereby obtaining the spatial coordinates of the sound source. After obtaining the spatial coordinates of the sound source, the signal processing unit and the acoustic environment monitoring unit work together. The signal processing unit can adjust the directivity of the microphones according to the spatial coordinates of the sound source to enhance the reception of the target sound source. The acoustic environment monitoring unit can monitor the noise level and distribution at the coordinates, providing noise information to the signal processing unit.

[0068] By acquiring dynamic features of speech through the spatial coordinates of the sound source, the characteristics of the speech signal can be analyzed more accurately. Furthermore, dynamic features such as the intensity and frequency changes of the speech signal are acquired at the spatial coordinates of the sound source. Simultaneously, the impact of acoustic interference fluctuation rate under background noise intensity on speech resolution can be analyzed, allowing for a better understanding of the degree of noise interference on the speech signal. For example, a large noise fluctuation rate may cause certain parts of the speech signal to be masked or distorted. Based on these analysis results, preset fluctuation ranges are configured. Preferably, through precise sound source localization and dynamic feature acquisition, combined with noise analysis, the target speech signal can be separated more effectively, optimizing the parameter configuration of the signal processing unit and improving the resolution accuracy and reliability of the speech signal.

[0069] Furthermore, it also includes: when the acoustic interference fluctuation rate under the background noise intensity exceeds the preset fluctuation range, generating a noise avoidance control command, the noise avoidance control command being used to trigger the reconfiguration of the spatial filtering parameters of the hearing aid; the spatial filtering parameters are determined based on the spatial coordinates of the sound source and the spatial azimuth angle of the strong interference sound source.

[0070] Specifically, noise avoidance control commands are instructions generated by the system when the acoustic interference fluctuation rate exceeds the preset fluctuation range, used to instruct the hearing aid to take measures to avoid noise interference; spatial filter parameter reconfiguration refers to readjusting the spatial filter parameters of the hearing aid according to the current acoustic environment and sound source location to optimize the reception of speech signals; spatial azimuth refers to the position angle of the strong interference sound source relative to the target sound source or user, usually used to describe the direction of the sound source in space.

[0071] Execution steps: When the acoustic interference fluctuation rate under background noise intensity exceeds the preset fluctuation range, it indicates that the noise interference in the current environment has reached a level that may affect the clarity of the speech signal. At this time, a noise avoidance control command is generated. For example, if the preset fluctuation range is ±10dB, but the actual fluctuation rate reaches ±15dB, it indicates that the noise environment is very unstable and measures need to be taken to reduce noise interference. Using the noise avoidance control command, the spatial filtering parameters of the hearing aid are reconfigured. The adjustment of the spatial filtering parameters is determined based on the spatial coordinates of the sound source and the spatial azimuth angle of the strong interference sound source. For example, if the target sound source is located in front of the user and the strong interference sound source is located to the left of the user, by determining these spatial azimuth angles, the hearing aid can adjust its spatial filtering parameters to enhance the signal reception of the sound source in front while suppressing the signal of the interference sound source on the left.

[0072] Furthermore, spatial filtering parameters can be adjusted using various techniques, such as beamforming, which adjusts the microphone sensitivity according to the direction of the sound source to form a beam pointing towards the target sound source while reducing the reception of interfering sound sources. Preferably, by dynamically adjusting the spatial filtering parameters of the hearing aid, the target speech signal can be separated more effectively, the influence of strong interfering sound sources can be effectively avoided, the performance of the hearing aid in complex acoustic environments can be significantly improved, and the clarity and intelligibility of the speech signal can be enhanced.

[0073] Furthermore, it also includes: calling an edge computing node based on the spatial coordinates of the sound source, the edge computing node being communicatively connected to the cloud audio processing center; the edge computing node receiving the spatial coordinates of the sound source and the spatial azimuth angle of the strong interference sound source, performing sound field propagation simulation prediction to determine the energy diffusion path; and dynamically expanding the coverage range of the null region of beamforming based on the energy diffusion path, and applying gain suppression to the frequency bands within the null region.

[0074] Specifically, edge computing nodes refer to computing nodes that perform data processing and analysis at the network edge, capable of rapidly processing local data and reducing data transmission latency; cloud audio processing centers refer to servers or data centers that perform audio processing in the cloud, possessing powerful computing and storage capabilities; sound field propagation simulation prediction refers to predicting the propagation path and energy distribution of sound in space by simulating the propagation characteristics of a sound field; energy diffusion path refers to how energy diffuses in space as sound propagates from the sound source to the receiving point; beamforming null regions refer to the regions formed in beamforming technology that suppress signals in specific directions by adjusting the direction and shape of the beam; gain suppression refers to attenuating the amplitude of a signal to reduce its intensity.

[0075] Execution steps: Based on the spatial coordinates of the sound source, an edge computing node is invoked. The edge computing node communicates with the cloud audio processing center. For example, once the spatial coordinates of the sound source are determined, the edge computing node can receive this coordinate information and communicate with the cloud audio processing center to obtain more powerful computing support. The edge computing node receives the spatial coordinates of the sound source and the spatial azimuth angle of the strong interference sound source, performs sound field propagation simulation prediction, and determines the energy diffusion path. Furthermore, by simulating sound field propagation, the propagation path of the sound from the sound source to the hearing aid and how the energy diffuses in space are predicted. If the strong interference sound source is located to the left of the user, the energy diffusion path of the interference sound source can be determined through sound field propagation simulation prediction.

[0076] Based on the energy diffusion path, the coverage area of ​​the beamforming null region is dynamically expanded, and gain suppression is applied to the frequency bands within the null region. For example, if it is predicted that the energy of the interfering sound source mainly diffuses along a specific path, the beamforming null region can be dynamically adjusted to cover this path, and gain suppression can be applied to the frequency bands within the null region, thereby reducing the impact of the interfering sound source. Preferably, through the collaborative work of edge computing nodes and cloud audio processing centers, sound field propagation simulation prediction and beamforming technology are used to dynamically adjust the signal processing strategy of the hearing aid to effectively suppress the influence of strong interfering sound sources and improve the clarity and intelligibility of the speech signal.

[0077] In summary, the beneficial effects of the embodiments of this application are:

[0078] This approach involves collecting the hearing aid input electrical signal, including speech signal energy, speech fundamental frequency characteristics, and background noise intensity, within the target user's acoustic environment and configuring a speech detectability criterion. A signal processing unit, comprising a multi-microphone transceiver array and a real-time spectrum analysis module, is then connected to evaluate the feasibility of speech parsing of the hearing aid input electrical signal based on this criterion. When the effective speech parsing probability of the hearing aid input electrical signal is lower than a preset probability threshold, a local frequency band region of interest is determined under the speech signal energy distribution marker in the hearing aid input electrical signal. In response to the local frequency band region of interest, the sampling rate and frequency band analysis range of the signal processing unit are adjusted, and the spectral surface data under the speech detectability criterion are registered and corrected with the discrete frequency point data of the signal processing unit to generate a denoised output signal with a speech credibility rating. This application provides a noise reduction processing method for adaptive adjustment of hearing aid electrical signals. It achieves the technical effect of collecting multi-dimensional signal features to construct a scientific speech detectability benchmark, accurately assessing the feasibility of speech parsing, reducing the effective speech false filtering rate, adjusting the sampling rate and frequency band range of the signal processing unit, simultaneously completing the accurate registration of spectral surface data and discrete frequency point data, adapting to different acoustic environments in real time, and preserving the effective speech signal while suppressing noise.

[0079] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0080] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application also intends to include such modifications and variations.

Claims

1. A noise reduction processing method for adaptive adjustment of electrical signals in hearing aids, characterized in that, The method includes: In the acoustic environment of the target user, the hearing aid input electrical signal, including speech signal energy, speech fundamental frequency characteristics, and background noise intensity, is collected, and a speech detectability judgment benchmark is configured. The signal processing unit is connected to evaluate the feasibility of speech analysis of the input electrical signal of the hearing aid based on the speech detectability determination criterion. The signal processing unit includes a multi-microphone transceiver array and a real-time spectrum analysis and processing module. When the effective speech resolution probability of the hearing aid input electrical signal is lower than a preset probability threshold, a local frequency band interest area under the speech signal energy distribution mark in the hearing aid input electrical signal is determined. In response to the local frequency band region of interest, the sampling rate and frequency band analysis range of the signal processing unit are adjusted, and the spectral surface data under the speech detectability determination benchmark and the discrete frequency point data of the signal processing unit are registered and corrected to generate a denoised output signal with speech credibility rating.

2. The noise reduction processing method for adaptive adjustment of electrical signals in hearing aids as described in claim 1, characterized in that, Using an acoustic environment monitoring unit, noise interference distribution information of the acoustic environment in which the target user is located is obtained; The noise interference distribution information is uploaded to the edge computing node, and the time-frequency domain filtering processing of the hearing aid input electrical signal is performed using the noise interference distribution information.

3. The noise reduction processing method for adaptive adjustment of electrical signals in hearing aids as described in claim 2, characterized in that, The method further includes connecting the signal processing unit: Based on the multi-microphone transceiver array in the signal processing unit, microphone sub-arrays are deployed at equal intervals along the incident direction of the sound source. The technical indicators corresponding to the microphone sub-arrays include directional gain and spatial resolution. Based on the real-time spectrum analysis and processing module in the signal processing unit, a fixed-point analysis channel is set in the key audio segments. The two types of technical indicators corresponding to the fixed-point analysis channel include frequency band analysis bandwidth and spectrum sampling accuracy. Based on the speech signal energy and speech fundamental frequency characteristics in the input electrical signal of the hearing aid, the first type of technical indicators and the second type of technical indicators are jointly configured.

4. The noise reduction processing method for adaptive adjustment of electrical signals in hearing aids as described in claim 3, characterized in that, The method further includes: The speech enhancement trajectory is aligned with the frequency sampling sequence of the signal processing unit in time and frequency, the audio processing protocol of the signal processing unit is adjusted synchronously, and the inter-frame processing delay is determined. The time-frequency synchronization accuracy is obtained by using the inter-frame processing delay. Based on the speech signal energy and background noise intensity in the input electrical signal of the hearing aid, the signal interpolation mechanism of the signal processing unit is configured with the time-frequency synchronization accuracy.

5. The noise reduction processing method for adaptive adjustment of electrical signals in hearing aids as described in claim 4, characterized in that, The method includes: In the acoustic environment where the target user is located, a multimodal sensor network is deployed, which is used to collect the hearing aid input electrical signals and auxiliary sensing information; Based on the input electrical signal of the hearing aid and the auxiliary sensing information, a speech activity distribution map is generated. Then, the speech enhancement trajectory and the frequency sampling sequence of the signal processing unit are time-frequency aligned using the speech activity distribution map and the spatial-semantic labeling information of the multimodal sensor network.

6. The noise reduction processing method for adaptive adjustment of electrical signals in hearing aids as described in claim 5, characterized in that, The method includes: The auxiliary perception information includes at least one of user motion state, spatial location, and scene semantic tags, and is used to assist in identifying the acoustic environment.

7. The noise reduction processing method for adaptive adjustment of electrical signals in hearing aids as described in claim 5, characterized in that, The method further includes registering and correcting the spectral surface data under the speech detectability determination benchmark with the discrete frequency point data of the signal processing unit to generate a denoised output signal with speech credibility rating. When the acoustic interference fluctuation rate under the background noise intensity exceeds the preset fluctuation range, and the proportion of signal interpolation operation under the signal interpolation mechanism to the total number of frames is greater than the proportion threshold, the multi-channel cross-validation mechanism is triggered. Using the multi-channel cross-validation mechanism, a confidence interval is determined for the speech, and the speech credibility rating is updated using the confidence interval.

8. The noise reduction processing method for adaptive adjustment of electrical signals in hearing aids as described in claim 7, characterized in that, The method includes: In the multimodal sensing network, the spatial coordinates of the sound source are located; At the spatial coordinates of the sound source, the signal processing unit and the acoustic environment monitoring unit work together. By acquiring dynamic speech features through the spatial coordinates of the sound source, analyzing the impact of acoustic interference fluctuation rate under the background noise intensity on speech parsing, and configuring the preset fluctuation range.

9. A noise reduction processing method for adaptive adjustment of electrical signals in hearing aids as described in claim 8, characterized in that, When the acoustic interference fluctuation rate under the background noise intensity exceeds the preset fluctuation range, a noise avoidance control command is generated. The noise avoidance control command is used to trigger the reconfiguration of the spatial filtering parameters of the hearing aid. The spatial filtering parameters are determined based on the spatial coordinates of the sound source and the spatial azimuth of the strong interference sound source.

10. A noise reduction processing method for adaptive adjustment of electrical signals in hearing aids as described in claim 9, characterized in that, Based on the spatial coordinates of the sound source, an edge computing node is invoked, and the edge computing node is communicatively connected to the cloud audio processing center. The edge computing node receives the spatial coordinates of the sound source and the spatial azimuth angle of the strong interference sound source, and performs sound field propagation simulation and prediction to determine the energy diffusion path. Based on the energy diffusion path, the coverage of the null region of beamforming is dynamically expanded, and gain suppression is applied to the frequency bands within the null region.

Images