CN112565997A - Adaptive noise reduction method and device for hearing aid, hearing aid and storage medium - Google Patents
Adaptive noise reduction method and device for hearing aid, hearing aid and storage medium Download PDFInfo
- Publication number
- CN112565997A CN112565997A CN202011415526.8A CN202011415526A CN112565997A CN 112565997 A CN112565997 A CN 112565997A CN 202011415526 A CN202011415526 A CN 202011415526A CN 112565997 A CN112565997 A CN 112565997A
- Authority
- CN
- China
- Prior art keywords
- noise reduction
- negative feedback
- hearing aid
- reduction algorithm
- algorithm model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04R—LOUDSPEAKERS, MICROPHONES, GRAMOPHONE PICK-UPS OR LIKE ACOUSTIC ELECTROMECHANICAL TRANSDUCERS; DEAF-AID SETS; PUBLIC ADDRESS SYSTEMS
- H04R25/00—Deaf-aid sets, i.e. electro-acoustic or electro-mechanical hearing aids; Electric tinnitus maskers providing an auditory perception
- H04R25/45—Prevention of acoustic reaction, i.e. acoustic oscillatory feedback
- H04R25/453—Prevention of acoustic reaction, i.e. acoustic oscillatory feedback electronically
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04R—LOUDSPEAKERS, MICROPHONES, GRAMOPHONE PICK-UPS OR LIKE ACOUSTIC ELECTROMECHANICAL TRANSDUCERS; DEAF-AID SETS; PUBLIC ADDRESS SYSTEMS
- H04R2225/00—Details of deaf aids covered by H04R25/00, not provided for in any of its subgroups
- H04R2225/49—Reducing the effects of electromagnetic noise on the functioning of hearing aids, by, e.g. shielding, signal processing adaptation, selective (de)activation of electronic parts in hearing aid
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04R—LOUDSPEAKERS, MICROPHONES, GRAMOPHONE PICK-UPS OR LIKE ACOUSTIC ELECTROMECHANICAL TRANSDUCERS; DEAF-AID SETS; PUBLIC ADDRESS SYSTEMS
- H04R2460/00—Details of hearing devices, i.e. of ear- or headphones covered by H04R1/10 or H04R5/033 but not provided for in any of their subgroups, or of hearing aids covered by H04R25/00 but not provided for in any of its subgroups
- H04R2460/01—Hearing devices using active noise cancellation
Landscapes
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Neurosurgery (AREA)
- Otolaryngology (AREA)
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Acoustics & Sound (AREA)
- Signal Processing (AREA)
- Circuit For Audible Band Transducer (AREA)
Abstract
The invention relates to the technical field of hearing aids, and discloses a self-adaptive noise reduction method, a self-adaptive noise reduction device, computer equipment and a computer storage medium of a hearing aid, wherein the method comprises the following steps: inputting a noisy digital signal into a noise reduction algorithm model in the current acoustic environment to obtain a model output result, wherein the noisy digital signal is obtained by encoding an acquired noisy sound signal; based on a preset negative feedback neural network, carrying out noise reduction effect evaluation on the model output result and generating negative feedback; and scheduling a target noise reduction algorithm model from preset noise reduction algorithm models according to the negative feedback to perform noise reduction processing on the noisy data signal to obtain a pure sound signal. The invention avoids the problem that the existing hearing aid cannot obtain good noise reduction effect in a changing noise environment but can additionally increase the power consumption of the hearing aid, improves the noise reduction effect of the hearing aid on noise and enables the hearing aid to obtain longer standby time.
Description
Technical Field
The present invention relates to the field of hearing aid technologies, and in particular, to a method and an apparatus for adaptive noise reduction for a hearing aid, and a computer storage medium.
Background
The hearing aid is used as an auxiliary device for helping people with hearing impairment to normally sense external sound information to carry out normal life, and has the application requirement of ultra-long standby, namely, in a life scene in which the hearing aid is actually applied, the hearing aid is required to be worn by a user for a long time to continuously consume energy to work, and the hearing aid does not need to work only when the user sleeps.
In addition, in order to improve sound clarity and ensure that a user can perceive high-quality effective sound information from the outside, the conventional hearing aids generally integrate a special processing technology to perform noise reduction processing on the external sound, for example, a general noise reduction algorithm is designed on the basis of assuming "noise stationary", however, in an actual acoustic environment where the hearing aid is applied, the "noise stationary" belongs to a very small probability event (a sound field where the hearing aid is located generally continuously changes), so that not only a good noise reduction effect cannot be obtained, but also extra energy consumption is generated when the general noise reduction algorithm is used for calculating and evaluating the sound field environment of the hearing aid, and the overall standby time of the hearing aid is greatly shortened.
In summary, the conventional hearing aid cannot obtain a good noise reduction effect in a changing noise environment where the hearing aid is located based on a general noise reduction algorithm, and additionally increases the power consumption of the hearing aid to shorten the overall standby time of the hearing aid. Therefore, there is a need for a self-adaptive noise reduction method for different noise environments to improve the performance of a hearing aid, so that the hearing aid can achieve a good noise reduction effect in different noise environments.
Disclosure of Invention
The invention mainly aims to provide a self-adaptive noise reduction method, a self-adaptive noise reduction system, self-adaptive noise reduction equipment and a self-adaptive noise reduction storage medium for a hearing aid, and aims to solve the technical problems that the existing hearing aid cannot obtain a good noise reduction effect in a changed noise environment where the hearing aid is located based on a general noise reduction algorithm, and the power consumption of the hearing aid is additionally increased to shorten the whole standby time of the hearing aid.
In order to achieve the above object, an embodiment of the present invention provides an adaptive noise reduction method for a hearing aid, where the adaptive noise reduction method for a hearing aid includes:
inputting a noisy digital signal into a noise reduction algorithm model in the current acoustic environment to obtain a model output result, wherein the noisy digital signal is obtained by encoding an acquired noisy sound signal;
based on a preset negative feedback neural network, carrying out noise reduction effect evaluation on the model output result and generating negative feedback;
and scheduling a target noise reduction algorithm model from preset noise reduction algorithm models according to the negative feedback to perform noise reduction processing on the noisy data signal to obtain a pure sound signal.
Preferably, the step of scheduling a target noise reduction algorithm model from preset noise reduction algorithm models according to the negative feedback to perform noise reduction processing on the noisy data signal to obtain a clean sound signal includes:
determining a noise reduction algorithm model to be selected from each preset noise reduction algorithm model according to the negative feedback, wherein the number of the noise reduction algorithm models to be selected is greater than or equal to one;
if the number of the to-be-selected noise reduction algorithm models is equal to one, taking the to-be-selected noise reduction algorithm models as target noise reduction algorithm models;
if the number of the to-be-selected noise reduction algorithm models is larger than one, determining the target noise reduction algorithm model according to the respective calculation power data of each to-be-selected noise reduction algorithm model;
and calling the target noise reduction algorithm model to perform noise reduction processing on the noisy data signal to obtain a pure sound signal.
Preferably, the step of determining the target noise reduction algorithm model according to the respective computation power data of each of the to-be-selected noise reduction algorithm models includes:
traversing respective calculation force data of each to-be-selected noise reduction algorithm model;
and detecting the minimum target calculation force data in the calculation force data, and determining the noise reduction algorithm model to be selected corresponding to the target calculation force data as the target noise reduction algorithm model.
Preferably, the step of scheduling a target noise reduction algorithm model from preset noise reduction algorithm models according to the negative feedback to perform noise reduction processing on the noisy data signal to obtain a clean sound signal further includes:
detecting a target noise reduction algorithm model corresponding to the negative feedback in each noise reduction algorithm model according to a preset power saving strategy;
and calling the target noise reduction algorithm model to perform noise reduction processing on the noisy data signal to obtain a pure sound signal.
Preferably, the output result of the model is the voice data after the noise reduction processing is carried out by the noise reduction algorithm model, the negative feedback neural network is obtained in advance through deep learning training of the nonlinear relation between the voice with noise and the artificial MOS scoring of the noise reduction effect,
the step of evaluating the noise reduction effect and generating negative feedback for the model output result based on a preset negative feedback neural network comprises:
preprocessing the voice data and inputting the voice data into a negative feedback neural network so that the negative feedback neural network can output the noise reduction algorithm model to score the MOS with the noise reduction effect of noise reduction of the noisy data signal;
detecting the difference value obtained by subtracting the preset standard MOS score from the MOS score;
and if the difference value is negative, scoring according to the preset standard MOS to generate negative feedback.
Preferably, after the step of detecting the difference value obtained by subtracting the preset standard MOS score from the MOS score, the method further includes:
and if the difference is positive, adding an execution delay time for performing noise reduction effect evaluation and generating negative feedback based on a preset negative feedback neural network.
Preferably, the execution of the lag time includes: a first hysteresis time and a second hysteresis time, the first hysteresis time being less than the second hysteresis time,
the step of adding the execution lag time for noise reduction effect evaluation and negative feedback generation based on a preset negative feedback neural network comprises:
detecting whether execution delay exists when the current execution carries out noise reduction effect evaluation based on the negative feedback neural network and generates negative feedback;
if not, the first lag time is increased for noise reduction effect evaluation and negative feedback generation based on the negative feedback neural network;
and if so, increasing the second lag time for noise reduction effect evaluation and generation of negative feedback based on the negative feedback neural network.
In addition, to achieve the above object, the present invention provides an adaptive noise reduction device for a hearing aid, including:
the acquisition module is used for inputting a noisy digital signal into a noise reduction algorithm model in the current acoustic environment to obtain a model output result, wherein the noisy digital signal is obtained by encoding an acquired noisy sound signal;
the effect evaluation module is used for carrying out noise reduction effect evaluation on the model output result based on a preset negative feedback neural network and generating negative feedback;
and the noise reduction module is used for scheduling a target noise reduction algorithm model from preset noise reduction algorithm models according to the negative feedback to perform noise reduction processing on the noisy data signal so as to obtain a pure sound signal.
Further, to achieve the above object, the present invention also provides a hearing aid comprising: a memory, a processor, a communication bus and an adaptive noise reduction program for a hearing aid stored on said memory,
the communication bus is used for realizing communication connection between the processor and the memory;
the processor is configured to execute an adaptive noise reduction procedure of the hearing aid to perform the steps of:
inputting a noisy digital signal into a noise reduction algorithm model in the current acoustic environment to obtain a model output result, wherein the noisy digital signal is obtained by encoding an acquired noisy sound signal;
based on a preset negative feedback neural network, carrying out noise reduction effect evaluation on the model output result and generating negative feedback;
and scheduling a target noise reduction algorithm model from preset noise reduction algorithm models according to the negative feedback to perform noise reduction processing on the noisy data signal to obtain a pure sound signal.
Further, to achieve the above object, the present invention also provides a storage medium storing one or more programs executable by one or more processors for:
inputting a noisy digital signal into a noise reduction algorithm model in the current acoustic environment to obtain a model output result, wherein the noisy digital signal is obtained by encoding an acquired noisy sound signal;
based on a preset negative feedback neural network, carrying out noise reduction effect evaluation on the model output result and generating negative feedback;
and scheduling a target noise reduction algorithm model from preset noise reduction algorithm models according to the negative feedback to perform noise reduction processing on the noisy data signal to obtain a pure sound signal.
According to the self-adaptive noise reduction method and device for the hearing aid, the hearing aid and the readable storage medium, a noise-carrying digital signal is input to a noise reduction algorithm model under the current acoustic environment to obtain a model output result, wherein the noise-carrying digital signal is obtained by coding an acquired sound signal with noise; based on a preset negative feedback neural network, carrying out noise reduction effect evaluation on the model output result and generating negative feedback; and scheduling a target noise reduction algorithm model from preset noise reduction algorithm models according to the negative feedback to perform noise reduction processing on the noisy data signal to obtain a pure sound signal.
The invention realizes that the noise reduction effect of noise reduction of the noisy sound collected by the hearing aid based on the noise reduction algorithm starts, the noise reduction effect is evaluated by utilizing the negative feedback neural network to generate the negative feedback to schedule the noise reduction algorithm which is optimal to adapt to the real-time sound field environment to reduce the noisy sound, the problem that the existing hearing aid cannot obtain good noise reduction effect in the changing noise environment but can additionally increase the power consumption of the hearing aid is avoided, the noise reduction effect of the hearing aid on the noisy sound is improved, and the hearing aid can obtain longer standby time.
Drawings
FIG. 1 is a schematic structural diagram of a hardware operating environment of a device according to a method of an embodiment of the present invention;
fig. 2 is a schematic flow chart of an embodiment of an adaptive noise reduction method for a hearing aid according to the present invention;
fig. 3 is a schematic view of an application scenario related to an embodiment of the adaptive noise reduction method for a hearing aid according to the present invention;
fig. 4 is a schematic diagram of another application scenario according to an embodiment of the adaptive noise reduction method for a hearing aid of the present invention;
fig. 5 is a functional block diagram of an adaptive noise reduction device of a hearing aid according to the present invention.
The implementation, functional features and advantages of the objects of the present invention will be further explained with reference to the accompanying drawings.
Detailed Description
It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The main solution of the embodiment of the invention is as follows: inputting a noisy digital signal into a noise reduction algorithm model in the current acoustic environment to obtain a model output result, wherein the noisy digital signal is obtained by encoding an acquired noisy sound signal; based on a preset negative feedback neural network, carrying out noise reduction effect evaluation on the model output result and generating negative feedback; and scheduling a target noise reduction algorithm model from preset noise reduction algorithm models according to the negative feedback to perform noise reduction processing on the noisy data signal to obtain a pure sound signal.
The hearing aid is used as an auxiliary device for helping people with hearing impairment to normally sense external sound information to carry out normal life, and has the application requirement of ultra-long standby, namely, in a life scene in which the hearing aid is actually applied, the hearing aid is required to be worn by a user for a long time to continuously consume energy to work, and the hearing aid does not need to work only when the user sleeps.
In addition, in order to improve sound clarity and ensure that a user can perceive high-quality effective sound information from the outside, the conventional hearing aid generally integrates a special processing technology to perform noise reduction processing on the external sound, for example, a general noise reduction algorithm is designed on the basis of assuming "noise stationary", however, in an actual acoustic environment where the hearing aid is applied, the "noise stationary" belongs to a very small probability event (a sound field where the hearing aid is located generally continuously changes), so that not only a good noise reduction effect cannot be obtained, but also extra energy consumption is generated when the general noise reduction algorithm is used for calculating and evaluating the sound field environment of the hearing aid, and the overall standby time of the hearing aid is greatly shortened.
In summary, the conventional hearing aid cannot obtain a good noise reduction effect in a changing noise environment where the hearing aid is located based on a general noise reduction algorithm, and additionally increases the power consumption of the hearing aid to shorten the overall standby time of the hearing aid.
According to the solution provided by the invention, the noise reduction effect of reducing noise of noisy sound collected by the hearing aid based on the noise reduction algorithm starts, and the noise reduction effect is evaluated by utilizing a negative feedback neural network to generate negative feedback to schedule the noise reduction algorithm which is optimal to adapt to a real-time sound field environment to reduce noise of noisy sound, so that the problem that the existing hearing aid cannot obtain a good noise reduction effect in a changed noise environment but can additionally increase the power consumption of the hearing aid is solved, the noise reduction effect of the hearing aid on noisy sound is improved, and the hearing aid can obtain longer standby time.
As shown in fig. 1, fig. 1 is a schematic diagram of an apparatus configuration of a hardware operating environment of a hearing aid according to an embodiment of the present invention.
As shown in fig. 1, the hearing aid may comprise: a processor 1001, such as a CPU, a communication bus 1002, a user interface 1003, and a memory 1004. Wherein a communication bus 1002 is used to enable connective communication between these components. The user interface 1003 may include an input unit such as a Keyboard (Keyboard), and the optional user interface 1003 may further include a wired charging interface, a wireless charging interface. The memory 1004 may be a high-speed RAM memory or a non-volatile memory (non-volatile memory), such as Flash memory. The memory 1004 may alternatively be a storage device separate from the processor 1001.
Optionally, the hearing aid may also comprise sensors such as light sensors and other sensors. In particular, the light sensor comprises a proximity sensor that can be switched from other modes, such as sleep/standby/power saving, to an active mode when the device is moved to the ear.
It will be appreciated by those skilled in the art that the device configuration shown in fig. 1 does not constitute a limitation of the hearing aid, and in other embodiments the hearing aid may comprise more or less components than those shown, or some components may be combined, or a different arrangement of components.
As shown in fig. 1, the memory 1004, which is a kind of computer storage medium, may include therein an operating system, a user interface module, and an adaptive noise reduction program of the hearing aid.
In the computer device shown in fig. 1, the user interface 1003 is mainly used for user interaction, including turning on and off, adjusting the set volume, and adjusting the set parameters; and the processor 1001 may be adapted to call up the adaptive noise reduction program of the hearing aid stored in the memory 1004 and perform the following steps:
inputting a noisy digital signal into a noise reduction algorithm model in the current acoustic environment to obtain a model output result, wherein the noisy digital signal is obtained by encoding an acquired noisy sound signal;
based on a preset negative feedback neural network, carrying out noise reduction effect evaluation on the model output result and generating negative feedback;
and scheduling a target noise reduction algorithm model from preset noise reduction algorithm models according to the negative feedback to perform noise reduction processing on the noisy data signal to obtain a pure sound signal.
Further, the processor 1001 may be configured to invoke an adaptive noise reduction program of the hearing aid stored in the memory 1004, and further perform the following steps:
determining a noise reduction algorithm model to be selected from each preset noise reduction algorithm model according to the negative feedback, wherein the number of the noise reduction algorithm models to be selected is greater than or equal to one;
if the number of the to-be-selected noise reduction algorithm models is equal to one, taking the to-be-selected noise reduction algorithm models as target noise reduction algorithm models;
if the number of the to-be-selected noise reduction algorithm models is larger than one, determining the target noise reduction algorithm model according to the respective calculation power data of each to-be-selected noise reduction algorithm model;
and calling the target noise reduction algorithm model to perform noise reduction processing on the noisy data signal to obtain a pure sound signal.
Further, the processor 1001 may be configured to invoke an adaptive noise reduction program of the hearing aid stored in the memory 1004, and further perform the following steps:
traversing respective calculation force data of each to-be-selected noise reduction algorithm model;
and detecting the minimum target calculation force data in the calculation force data, and determining the noise reduction algorithm model to be selected corresponding to the target calculation force data as the target noise reduction algorithm model.
Further, the processor 1001 may be configured to invoke an adaptive noise reduction program of the hearing aid stored in the memory 1004, and further perform the following steps:
detecting a target noise reduction algorithm model corresponding to the negative feedback in each noise reduction algorithm model according to a preset power saving strategy;
and calling the target noise reduction algorithm model to perform noise reduction processing on the noisy data signal to obtain a pure sound signal.
Further, the model output result is speech data after noise reduction processing is performed on the noise reduction algorithm model, the negative feedback neural network is obtained in advance through deep learning training of a nonlinear relationship between noisy speech and noise reduction effect artificial MOS scoring, and the processor 1001 may be configured to invoke an adaptive noise reduction program of the hearing aid stored in the memory 1005, and further perform the following steps:
preprocessing the voice data and inputting the voice data into a negative feedback neural network so that the negative feedback neural network can output the noise reduction algorithm model to score the MOS with the noise reduction effect of noise reduction of the noisy data signal;
detecting the difference value obtained by subtracting the preset standard MOS score from the MOS score;
and if the difference value is negative, scoring according to the preset standard MOS to generate negative feedback.
Further, the processor 1001 may be configured to call an adaptive noise reduction program of the hearing aid stored in the memory 1004, and after performing the step of detecting the difference between the MOS score and the preset standard MOS score, further perform the following steps:
and if the difference is positive, adding an execution delay time for performing noise reduction effect evaluation and generating negative feedback based on a preset negative feedback neural network.
Further, the performing a lag time comprises: a first lag time and a second lag time, said first lag time being smaller than said second lag time, the processor 1001 may be configured to invoke an adaptive noise reduction routine of the hearing aid stored in the memory 1004, further performing the following steps:
detecting whether execution delay exists when the current execution carries out noise reduction effect evaluation based on the negative feedback neural network and generates negative feedback;
if not, the first lag time is increased for noise reduction effect evaluation and negative feedback generation based on the negative feedback neural network;
and if so, increasing the second lag time for noise reduction effect evaluation and generation of negative feedback based on the negative feedback neural network.
The specific embodiment of the computer device related to the adaptive noise reduction method of the hearing aid of the present invention is substantially the same as each specific embodiment of the adaptive noise reduction method of the hearing aid described below, and therefore, no further description is given herein, and for convenience of description, a terminal device is used instead of the computer device for explanation.
The invention provides an adaptive noise reduction method for a hearing aid.
Referring to fig. 2, fig. 2 is a schematic flowchart illustrating a first embodiment of an adaptive noise reduction method for a hearing aid according to the present invention, in this embodiment, the adaptive noise reduction method for a hearing aid includes:
step S100, inputting a noisy digital signal into a noise reduction algorithm model in the current acoustic environment to obtain a model output result, wherein the noisy digital signal is obtained by encoding an acquired noisy sound signal;
in the present embodiment, the sound signal with noise may be specifically a sound signal obtained by mixing the valid sound and the invalid sound-noise together. After the hearing aid device has simultaneously detected the desired sound emitted by the signal source and the undesired sound-noise generated by the noise source from the acoustic environment in which it is currently located, the hearing aid device first converts the noisy sound signal (mixed desired sound and noise) together into a noisy digital signal with noise.
Under the current acoustic environment, the hearing aid equipment inputs the noise-carrying digital signal of the acquired noise-carrying sound signal into a noise reduction algorithm model directly corresponding to the hearing aid equipment at the current moment, so that a model output result of the noise reduction algorithm model after noise reduction processing is carried out on the noise-carrying digital signal is obtained.
It should be noted that, in this embodiment, the current acoustic environment may specifically be any scene where a user wearing a hearing aid device may be in real life, such as an indoor scene, a road, a station, a dish market, a street, and the like, and it should be understood that, depending on different needs of practical applications, in other embodiments, the current acoustic environment may also be other scenes not listed above, and the adaptive noise reduction method of the hearing aid of the present invention is not limited to the kind of the current acoustic environment.
A plurality of noise reduction algorithm models are simultaneously preset in the hearing aid device, and can be flexibly called based on the specific needs of the hearing aid device. The noise reduction algorithm model directly corresponding to the hearing aid equipment at the current moment specifically comprises the following steps: the method comprises the following steps that a noise reduction algorithm model called by the hearing aid device when noise reduction processing is carried out on a noisy digital signal at the previous time or in the previous period; the noise reduction algorithm model may be any mature noise reduction algorithm at any time, such as wiener filtering noise reduction algorithm, forward neural network noise reduction algorithm, and the like, and it should be understood that, based on different design requirements of practical applications, in other embodiments, the noise reduction algorithm model may be different from each noise reduction algorithm listed in this embodiment.
Specifically, for example, referring to the application scenario shown in fig. 3, in the current market acoustic environment, the hearing aid device collects the valid sound-human output speech from the signal source based on the preset microphone, and synchronously collects some noisy noise from the noise source, then the hearing aid device encodes the speech-with-noise signal mixed with noisy noise by presetting a mature sound signal transform processing technique, so as to transform the noisy signal into a noisy digital signal, and finally, the hearing aid device directly inputs the noisy digital signal transformed at the current moment into a noise reduction model by calling a noise reduction algorithm model for noise reduction processing on the noisy digital signal at the previous time, i.e. a wiener filter noise reduction algorithm with a parameter model, so as to reduce the noisy digital signal by using the wiener filter noise reduction algorithm, and further obtaining a noise-reduced model output result output by the model.
S200, evaluating the noise reduction effect aiming at the model output result based on a preset negative feedback neural network and generating negative feedback;
in this embodiment, the model outputs the voice data after the noise reduction processing is performed on the voice data, which is the noisy digital signal obtained by converting the noisy sound signal, by the noise reduction algorithm model.
After the hearing aid equipment obtains a model output result obtained by carrying out noise reduction processing on a noise-carrying digital signal through a noise reduction algorithm model in the current acoustic environment, the hearing aid equipment evaluates the noise reduction effect of the model output result obtained by carrying out noise reduction processing on the noise-carrying digital signal through a preset negative feedback neural network, and generates negative feedback facing the noise reduction algorithm model based on the evaluation of the noise reduction effect.
It should be noted that, in this embodiment, the preset negative feedback neural network may be obtained by deep learning training of a nonlinear relationship between a noisy speech and a noise reduction effect artificial MOS (Mean Opinion Score) Score in advance. Specifically, for example, the hearing aid device combines a clean speech library and a collected classification noise library as training data in advance, and then, based on deep learning training, the negative feedback neural network sufficiently learns the nonlinear relationship between the noisy speech and the artificial MOS score, so that the negative feedback neural network can simulate the artificial MOS score based on the noise reduction effect of the noise reduction algorithm model on the noisy digital signal.
Further, in a possible embodiment, the step S200 may include:
step S201, preprocessing the voice data and inputting the preprocessed voice data into a negative feedback neural network, so that the negative feedback neural network outputs the noise reduction algorithm model to score the MOS with the noise reduction effect of noise reduction of the noisy data signal;
in this embodiment, the preprocessing may be specifically discrete fourier transform for voice data.
The hearing aid device performs discrete Fourier transform on voice data subjected to noise reduction processing on noise-carrying digital signals through a noise reduction algorithm model in the current acoustic environment to enable the voice data to become a frequency spectrum, then the hearing aid extracts amplitude values from the frequency spectrum and inputs the amplitude values into a negative feedback neural network, and the negative feedback neural network performs deep learning training on the basis of the amplitude values to simulate manual MOS scoring on noise reduction effects of noise reduction of the noise-carrying digital signals through the noise reduction algorithm model and outputs the MOS scoring.
Step S202, detecting the difference value of the MOS scoring minus the preset standard MOS scoring;
and step S203, if the difference value is negative, generating negative feedback according to the preset standard MOS score.
It should be noted that, in this embodiment, the preset standard MOS score may be specifically configured in advance for a development designer of the hearing aid device to identify voice data obtained after the noise reduction module performs noise reduction on the noisy digital signal, and after the voice data is output to a user of the hearing aid device, the user can clearly obtain effective sound, so as to ensure a MOS score of good use experience of the user, specifically, for example, if the design developer of the hearing aid device configures the preset standard MOS score to 4.0 scores based on a long development experience, the identification is: if the MOS scoring performed on the noise reduction effect of the noise reduction algorithm model for the noise reduction of the noisy digital signal reaches 4.0 points or is more than 4.0 points, the sound data of the noise-reduced digital signal currently passing through the noise reduction algorithm model can enable a hearing aid user to clearly know effective sound, and if the MOS scoring does not reach 4.0 points, the sound data of the noise-reduced digital signal currently passing through the noise reduction algorithm model cannot enable the hearing aid user to clearly know the effective sound.
After the hearing aid device inputs the voice data subjected to noise reduction by the noise reduction algorithm model into the negative feedback neural network so as to obtain the MOS score output by the negative feedback neural network, the hearing aid device further subtracts the preset standard MOS score from the MOS score so as to obtain a difference value, and then when the difference value is negative, namely the MOS score is not reached to the preset standard MOS score, the hearing aid device generates negative feedback facing a plurality of noise reduction algorithm models according to the preset standard MOS score.
Specifically, for example, referring to the application scenario shown in fig. 4, the hearing aid device calls a noise reduction algorithm model of the current acoustic environment to perform noise reduction on a noisy digital signal to obtain voice data, performs discrete fourier transform to convert the voice data into a frequency spectrum, extracts the amplitude of the frequency spectrum, inputs the frequency spectrum into a pre-trained negative feedback neural network, so as to obtain that the MOS score performed by the negative feedback neural network on the noise reduction effect of the noise reduction algorithm model on the noisy digital signal is 2.5 minutes, the hearing aid device subtracts a preset standard MOS score from-2.5 minutes to-4.0 minutes to obtain a negative difference value of 1.5, so that the hearing aid device starts to perform negative feedback on a plurality of noise reduction algorithm models pre-configured for the hearing aid device according to the preset standard MOS score of-4.0, namely, a target noise reduction algorithm model which can be artificially graded by a negative feedback neural network to reach the preset standard MOS grade of-4.0 and consumes the least electric energy can be selected from the plurality of noise reduction algorithm models based on the negative feedback.
In this embodiment, each of the noise reduction algorithm models configured in advance in the hearing aid device indicates power calculation data corresponding to the power consumption of the hearing aid device in a positive correlation, that is, if the power calculation data of one noise reduction algorithm model is larger, the more power the model consumes for the hearing aid device.
And step S300, scheduling a target noise reduction algorithm model from preset noise reduction algorithm models according to the negative feedback, and carrying out noise reduction processing on the noisy data signal to obtain a pure sound signal.
After the hearing aid equipment carries out noise reduction effect evaluation based on a negative feedback neural network to generate negative feedback facing a noise reduction algorithm model, the hearing aid equipment determines a target noise reduction algorithm model which can be manually simulated by the negative feedback neural network to carry out MOS scoring to reach the preset standard MOS scoring and consumes the minimum electric energy from a plurality of noise reduction algorithm models which are configured in advance of the hearing aid equipment based on the negative feedback, and finally calls the target noise reduction algorithm model to carry out noise reduction processing on noisy digital signals again to obtain pure sound signals to output.
It should be noted that, in this embodiment, the clean sound signal may be a sound signal that can be directly output to the user based on a speaker of the hearing aid device, so that the user can clearly obtain effective sound information.
Further, in a possible embodiment, the step S300 may include:
step S301, determining noise reduction algorithm models to be selected from preset noise reduction algorithm models according to the negative feedback, wherein the number of the noise reduction algorithm models to be selected is greater than or equal to one.
The hearing aid device determines one or more noise reduction algorithm models to be selected, which can be evaluated by a negative feedback neural network to reach the preset standard MOS scoring, in the noise reduction algorithm models with noise reduction effect of noise-carrying digital signals based on the respective calculation data of the noise reduction algorithm models, among a plurality of noise reduction algorithm models which are configured in advance.
Specifically, for example, the hearing aid device may previously determine, based on the calculated force data owned by the noise reduction algorithm model itself, the noise reduction effect of the noise reduction algorithm model for reducing noise of the noisy digital signal, which can be evaluated by the negative feedback neural network so as to simulate the value of manually-performed MOS scoring, and integrate and form a corresponding relationship between the calculated force data of the noise reduction algorithm model and the MOS scoring, and store the corresponding relationship, so that when the hearing aid device determines the noise reduction algorithm model to be selected from the plurality of noise reduction algorithm models, it may determine one or more noise reduction effects for reducing noise of the noisy digital signal, which can be manually-performed MOS scoring by the negative feedback network to reach the noise reduction algorithm model to be selected of-4.0 minutes of the preset standard MOS scoring based on the corresponding relationship between the calculated force data of the plurality of noise reduction algorithm models and the MOS scoring.
Step S302, if the number of the noise reduction algorithm models to be selected is equal to one, taking the noise reduction algorithm models to be selected as target noise reduction algorithm models;
after the hearing aid device determines one or more noise reduction algorithm models to be selected from the plurality of noise reduction algorithm models based on the respective computational power data of the plurality of noise reduction algorithm models, if the hearing aid device detects that there is only one noise reduction algorithm model to be selected, the hearing aid device directly determines the one noise reduction algorithm model to be selected as a target noise reduction algorithm model for carrying out noise reduction processing on noisy digital signals again.
Step S303, if the number of the to-be-selected noise reduction algorithm models is greater than one, determining the target noise reduction algorithm model according to respective calculation data of each to-be-selected noise reduction algorithm model;
after the hearing aid device determines one or more noise reduction algorithm models to be selected from the noise reduction algorithm models based on the respective computing power data of the noise reduction algorithm models, if the hearing aid device detects that a plurality of noise reduction algorithm models to be selected exist, the hearing aid device determines the noise reduction algorithm model to be a target noise reduction algorithm model for carrying out noise reduction processing on the noise digital signal again according to the noise reduction algorithm model to be selected with the lowest power consumption in the noise reduction algorithm models.
Further, in a possible embodiment, the step S303 may include:
step S3031, traversing respective computational power data of each to-be-selected noise reduction algorithm model;
step S3032, detecting the minimum target calculation force data in each calculation force data, and determining the noise reduction algorithm model to be selected corresponding to the target calculation force data as the target noise reduction algorithm model.
After the hearing aid device determines a plurality of noise reduction algorithm models to be selected from the plurality of noise reduction algorithm models based on respective computing power data of the plurality of noise reduction algorithm models, the hearing aid device traverses the computing power data of each noise reduction algorithm model to be selected from the plurality of noise reduction algorithm models, so that a positive correlation is formed between the computing power data of the noise reduction algorithm models and the power consumption of the hearing aid device, a minimum target computing power data is determined from the plurality of computing power data, and the noise reduction algorithm model to be selected corresponding to the target computing power data is determined as a target noise reduction algorithm model for noise reduction processing of the noise-added digital signal again.
Step S304, calling the target noise reduction algorithm model to perform noise reduction processing on the noisy data signal to obtain a pure sound signal.
The hearing aid device determines a selected noise reduction algorithm model determined from the plurality of noise reduction algorithm models as a target noise reduction algorithm model, or determines a plurality of selected noise reduction algorithm models from the plurality of noise reduction algorithm models, and selects the selected noise reduction algorithm model with the minimum computational power data as the target noise reduction algorithm model, and then the hearing aid device starts to recall the target noise reduction algorithm model, and re-inputs the noisy digital signal into the target noise reduction algorithm model, so that the target noise reduction algorithm model performs noise reduction processing on the noisy digital signal again and outputs a pure sound signal directly output to a user.
In this embodiment, the hearing aid device inputs the noise-carrying digital signal of the noise-carrying sound signal acquired by conversion to the noise reduction algorithm model directly corresponding to the hearing aid device at the current moment in the current acoustic environment, so as to obtain a model output result of the noise reduction algorithm model after performing noise reduction processing on the noise-carrying digital signal; then, the hearing aid device evaluates the noise reduction algorithm model to perform noise reduction processing on the noisy digital signal by using a preset negative feedback neural network to obtain the noise reduction effect of the output result of the model, and generates negative feedback facing the noise reduction algorithm model based on the evaluation of the noise reduction effect; and finally, the hearing aid device determines a target noise reduction algorithm model which can be artificially simulated by a negative feedback neural network to perform MOS scoring to reach the preset standard MOS scoring and has the minimum consumed electric energy based on the negative feedback from a plurality of noise reduction algorithm models which are configured in advance by the hearing aid device, and finally calls the target noise reduction algorithm model to perform noise reduction processing on the noisy digital signal again to obtain a pure sound signal to output.
The invention realizes that the noise reduction effect of noise reduction of the noisy sound collected by the hearing aid based on the noise reduction algorithm starts, the noise reduction effect is evaluated by utilizing the negative feedback neural network to generate the negative feedback to schedule the noise reduction algorithm which is optimal to adapt to the real-time sound field environment to reduce the noisy sound, the problem that the existing hearing aid cannot obtain good noise reduction effect in the changing noise environment but can additionally increase the power consumption of the hearing aid is avoided, the noise reduction effect of the hearing aid on the noisy sound is improved, and the hearing aid can obtain longer standby time.
Further, on the basis of the first embodiment of the adaptive noise reduction method for a hearing aid of the present invention, a second embodiment of the adaptive noise reduction method for a hearing aid is provided, in this embodiment, the step S300 of scheduling a target noise reduction algorithm model from preset noise reduction algorithm models according to the negative feedback to perform noise reduction processing on the noisy data signal to obtain a clean sound signal may further include:
step S305, detecting a target noise reduction algorithm model corresponding to the negative feedback in each noise reduction algorithm model according to a preset power saving strategy;
step S306, calling the target noise reduction algorithm model to perform noise reduction processing on the noisy data signal to obtain a pure sound signal.
It should be noted that, in this embodiment, the preset power saving strategy is generated by a design developer of the hearing aid device in advance based on respective calculation data and noise reduction effect data configuration of each noise reduction algorithm model, and the preset power saving strategy can achieve a good balance between the power consumption of the noise reduction algorithm model to the hearing aid and the noise reduction effect of the noise reduction algorithm model to the noisy digital signal. In addition, the noise reduction effect data may specifically be an MOS value that is obtained by the hearing aid device performing noise reduction on the noisy digital signal according to the noise reduction algorithm model and performing MOS scoring by a negative feedback neural network simulation, and correspondingly generate data that is in a positive correlation with the MOS value, for example, if the MOS score of a certain noise reduction algorithm model performing MOS scoring on the noisy digital signal is 4.0 minutes, the hearing aid device may generate the noise reduction effect data of the noise reduction algorithm model to be 80%.
The method comprises the steps that a hearing aid device selects a noise reduction effect aiming at noise reduction of a noise-carrying digital signal from a plurality of noise reduction algorithm models which are configured in advance according to a preset power saving strategy as a target noise reduction algorithm model corresponding to negative feedback, the noise reduction effect can be evaluated by a negative feedback neural network to reach a preset standard MOS score, and the noise reduction algorithm model aiming at the minimum power consumption of the hearing aid device is used as the target noise reduction algorithm model, then the hearing aid device immediately calls the target noise reduction algorithm model, re-inputs the noise-carrying digital signal into the target noise reduction algorithm model, and the target noise reduction algorithm model carries out noise reduction processing on the noise-carrying digital signal again and then outputs a pure sound signal directly output to a user.
In the embodiment, a power saving strategy is configured and generated in advance based on the long experience of the design developer of the hearing aid device between the noise reduction effect achieved by the noise reduction algorithm model and the power consumption generated by the hearing aid device, therefore, after the noise reduction effect of the noise reduction algorithm model of the current acoustic environment aiming at noise reduction of the noisy digital signal is evaluated by the negative feedback neural network and generates negative feedback, the hearing aid equipment can also automatically select one noise reduction algorithm model from a plurality of preset noise reduction algorithm models according to the power saving strategy, wherein the noise reduction algorithm model can not only meet the noise reduction effect on the noisy digital signals and reach the preset standard, and the power consumption for the hearing aid device is also minimized to ensure that the hearing aid device can obtain a target noise reduction algorithm model for longer standby time, and calling the target algorithm model to perform noise reduction processing on the noisy digital signal again, and outputting a pure sound signal which can be directly output to a user by the hearing aid equipment.
The invention realizes that the noise reduction effect is evaluated by using a negative feedback neural network to generate negative feedback to schedule a noise reduction algorithm which is suitable for the real-time sound field environment to be optimal to reduce noise of noise, thereby avoiding the problem that the existing hearing aid cannot obtain good noise reduction effect in the changing noise environment but can additionally increase the power consumption of the hearing aid, improving the noise reduction effect of the hearing aid on the noise and enabling the hearing aid to obtain longer standby time.
Further, on the basis of the first embodiment of the adaptive noise reduction method for a hearing aid of the present invention, a third embodiment of the adaptive noise reduction method for a hearing aid is provided, in this embodiment, after the step S202 detects a difference value obtained by subtracting a preset standard MOS score from the MOS score, the adaptive noise reduction method for a hearing aid of the present invention may further include:
step S204, if the difference value is positive, adding a preset negative feedback neural network to evaluate the noise reduction effect and generate execution delay time of negative feedback;
the hearing aid equipment inputs voice data subjected to noise reduction through the noise reduction algorithm model into a negative feedback neural network so as to obtain MOS scoring output by the negative feedback neural network, then the hearing aid equipment further subtracts preset standard MOS scoring from the MOS scoring so as to obtain a difference value, and then, when the difference value is positive, namely the MOS scoring is proved to reach the preset standard MOS scoring, the hearing aid equipment adds a delay for calling the negative feedback neural network for noise reduction effect evaluation next time on the basis of executing noise reduction effect evaluation on the noise reduction algorithm model of the current acoustic environment by calling the negative feedback neural network currently so as to generate execution lag time of the negative feedback.
It should be noted that, in the present embodiment, the execution hysteresis time includes, but is not limited to, a first hysteresis time and a second hysteresis time, wherein the first hysteresis time is smaller than the second hysteresis time. Specifically, for example, the hearing aid device may be configured to generate the first lag time in a size of 15ms (milliseconds) and the second lag time in a size of 25ms in advance based on the design developer configuration.
Further, in a possible embodiment, the step S204 may include:
step S2041, detecting whether execution delay exists when noise reduction effect evaluation is performed based on the negative feedback neural network and negative feedback is generated in the current execution;
the hearing aid device calls the negative feedback neural network to evaluate the noise reduction effect of noise reduction on the noise-carrying digital signal aiming at the noise reduction algorithm model of the current acoustic environment, and the obtained MOS score of the negative feedback neural network imitating manual work is larger than the preset standard MOS score, so that the difference value obtained by subtracting the preset standard MOS score from the MOS score is used as the right time for the hearing aid device, and the hearing aid device immediately detects whether the calling execution delay exists when the noise reduction effect evaluation is carried out on the noise reduction algorithm model by calling the negative feedback neural network at the current time to judge and generate the negative feedback.
Step S2042, if not, performing noise reduction effect evaluation based on the negative feedback neural network and generating negative feedback to increase the first lag time;
specifically, for example, if the hearing aid device detects that the negative feedback neural network is called at the present time to evaluate the noise reduction effect of the noise reduction algorithm model to determine to generate the negative feedback, and there is no execution delay of the call, that is, if the hearing aid device performs noise reduction processing on the noisy digital signal and outputs a model output result (i.e., the sound data after noise reduction) in the present acoustic environment, then the hearing aid device performs discrete fourier transform on the model output result to form a frequency spectrum, and then inputs the amplitude of the frequency spectrum into the negative feedback neural network to evaluate the noise reduction effect, then the hearing aid device determines to generate the negative feedback by calling the negative feedback neural network to evaluate the noise reduction effect of the noise reduction algorithm model for the next time, and increases the first delay time to 15ms to make the negative feedback neural network called for the next time to evaluate the noise reduction effect of the noise reduction algorithm model to determine to generate the negative feedback, execution will be invoked after a delay of 15 ms.
Step S2043, if yes, performing noise reduction effect evaluation based on the negative feedback neural network and generating negative feedback to increase the second lag time.
Specifically, for example, if the hearing aid device detects that the negative feedback neural network is called at the present time to evaluate the noise reduction effect of the noise reduction algorithm model to determine to generate the negative feedback, the called execution delay already exists, that is, if the hearing aid device performs noise reduction processing on the noisy digital signal and outputs the model output result (i.e., the sound data after noise reduction) in the present acoustic environment, the discrete fourier transform is performed on the model output result to form a frequency spectrum after 50ms delay, and then the amplitude of the frequency spectrum is input to the negative feedback neural network to evaluate the noise reduction effect, then the hearing aid device calls the negative feedback neural network for the next time to evaluate the noise reduction effect of the noise reduction algorithm model to determine to generate the negative feedback, and increases the second delay time by 25ms to make the negative feedback neural network called for the next time to evaluate the noise reduction effect of the noise reduction algorithm model to determine to generate the negative feedback, execution will be invoked after a delay of 15ms +25 ms.
In this embodiment, after the hearing aid device inputs the voice data after the noise reduction of the noise reduction algorithm model to the negative feedback neural network to obtain the MOS score output by the negative feedback neural network, the hearing aid device further subtracts the preset standard MOS score from the MOS score to obtain a difference, and then, when the difference is positive, that is, when the MOS score reaches the preset standard MOS score, the hearing aid device adds a delay to next call the negative feedback neural network for noise reduction evaluation to generate execution delay time of the negative feedback on the basis of the current call of the negative feedback neural network for noise reduction evaluation of the noise reduction algorithm model of the current acoustic environment.
The noise reduction algorithm model under the current acoustic environment can reach the standard condition aiming at the noise reduction effect of the noise-carrying digital signal, namely, the hearing aid equipment can output a sound signal which can be clearly obtained to a user, and then the hearing aid equipment delays to further utilize the negative feedback neural network to evaluate the noise reduction effect when calling the noise reduction algorithm model for noise reduction next time, so that the power consumption of the hearing aid equipment for evaluating the noise reduction effect by utilizing the negative feedback neural network is reduced, and the hearing aid can be further ensured to obtain longer standby time.
In addition, the present invention further provides an adaptive noise reduction device for a hearing aid, referring to fig. 5, fig. 5 is a schematic diagram of functional modules of the adaptive noise reduction device for a hearing aid according to the present invention, the adaptive noise reduction device for a hearing aid includes:
the reading module 101 is configured to obtain a module, and is configured to input a noisy digital signal to a noise reduction algorithm model in a current acoustic environment to obtain a model output result, where the noisy digital signal is obtained by encoding an acquired noisy sound signal;
the effect evaluation module 102 is configured to perform noise reduction effect evaluation on the model output result based on a preset negative feedback neural network and generate negative feedback;
and the noise reduction module 103 is configured to schedule a target noise reduction algorithm model from preset noise reduction algorithm models according to the negative feedback, and perform noise reduction processing on the noisy data signal to obtain a pure sound signal.
Preferably, the noise reduction module 103 comprises:
the determining unit is used for determining a noise reduction algorithm model to be selected from all the preset noise reduction algorithm models according to the negative feedback, wherein the number of the noise reduction algorithm models to be selected is greater than or equal to one;
the first selection unit is used for taking the noise reduction algorithm model to be selected as a target noise reduction algorithm model if the number of the noise reduction algorithm model to be selected is equal to one;
the second selection unit is used for determining the target noise reduction algorithm model according to the respective calculation force data of each to-be-selected noise reduction algorithm model if the number of the to-be-selected noise reduction algorithm models is larger than one;
and the first calling unit is used for calling the target noise reduction algorithm model to perform noise reduction processing on the noisy data signal so as to obtain a pure sound signal.
Preferably, the second selection unit includes:
the traversal subunit is used for traversing the respective computational power data of each to-be-selected noise reduction algorithm model;
and the detection subunit is used for detecting the minimum target calculation force data in the calculation force data and determining the noise reduction algorithm model to be selected corresponding to the target calculation force data as the target noise reduction algorithm model.
Preferably, the noise reduction module 103 further includes:
a third selecting unit, configured to detect, according to a preset power saving policy, a target noise reduction algorithm model corresponding to the negative feedback in each of the noise reduction algorithm models;
and the second calling unit is used for calling the target noise reduction algorithm model to perform noise reduction processing on the noisy data signal so as to obtain a pure sound signal.
Preferably, the model output result is speech data subjected to noise reduction processing by the noise reduction algorithm model, the negative feedback neural network is obtained in advance through deep learning training of a nonlinear relationship between noisy speech and noise reduction effect artificial MOS scoring, and the effect evaluation module 102 includes:
the input unit is used for inputting the preprocessed voice data into a negative feedback neural network so that the negative feedback neural network can output the MOS scoring of the noise reduction effect of the noise reduction algorithm model for noise reduction of the noisy data signal;
the first detection unit is used for detecting the difference value obtained by subtracting the preset standard MOS mark from the MOS mark;
and the generating unit is used for generating negative feedback according to the preset standard MOS scoring if the difference value is negative.
Preferably, the hysteresis module 104 is configured to add an execution hysteresis time for performing noise reduction effect evaluation based on a preset negative feedback neural network and generating negative feedback if the difference is positive.
Preferably, the execution of the lag time includes: a first hysteresis time and a second hysteresis time, the first hysteresis time being less than the second hysteresis time, a hysteresis module 104 comprising:
the second detection unit is used for detecting whether execution delay exists when the noise reduction effect evaluation is performed on the basis of the negative feedback neural network and negative feedback is generated in the current execution;
a first increasing unit, configured to, if not, increase the first lag time for noise reduction effect evaluation and generation of negative feedback based on the negative feedback neural network;
and if so, increasing the second lag time for noise reduction effect evaluation and negative feedback generation based on the negative feedback neural network.
Furthermore, the present invention also provides a computer storage medium storing one or more programs, the one or more programs further executable by one or more processors for:
inputting a noisy digital signal into a noise reduction algorithm model in the current acoustic environment to obtain a model output result, wherein the noisy digital signal is obtained by encoding an acquired noisy sound signal;
based on a preset negative feedback neural network, carrying out noise reduction effect evaluation on the model output result and generating negative feedback;
and scheduling a target noise reduction algorithm model from preset noise reduction algorithm models according to the negative feedback to perform noise reduction processing on the noisy data signal to obtain a pure sound signal.
The specific implementation of the computer storage medium of the present invention is substantially the same as the embodiments of the adaptive noise reduction method for a hearing aid, and is not described herein again.
It should be noted that the blockchain in the present invention is a novel application mode of computer technologies such as distributed data storage, point-to-point transmission, consensus mechanism, and encryption algorithm. A block chain (Blockchain), which is essentially a decentralized database, is a series of data blocks associated by using a cryptographic method, and each data block contains information of a batch of network transactions, so as to verify the validity (anti-counterfeiting) of the information and generate a next block. The blockchain may include a blockchain underlying platform, a platform product service layer, an application service layer, and the like. Further, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments.
Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner. Based on such understanding, the technical solutions of the present invention may be embodied in the form of a software product, which is stored in a storage medium (such as ROM/RAM, magnetic disk, optical disk) and includes instructions for enabling a terminal device (such as a mobile phone, a computer, a server, an air conditioner, or a network device) to execute the method according to the embodiments of the present invention.
The above description is only a preferred embodiment of the present invention, and not intended to limit the scope of the present invention, and all modifications of equivalent structures and equivalent processes, which are made by using the contents of the present specification and the accompanying drawings, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.
Claims (10)
1. An adaptive noise reduction method for a hearing aid, the adaptive noise reduction method comprising:
inputting a noisy digital signal into a noise reduction algorithm model in the current acoustic environment to obtain a model output result, wherein the noisy digital signal is obtained by encoding an acquired noisy sound signal;
based on a preset negative feedback neural network, carrying out noise reduction effect evaluation on the model output result and generating negative feedback;
and scheduling a target noise reduction algorithm model from preset noise reduction algorithm models according to the negative feedback to perform noise reduction processing on the noisy data signal to obtain a pure sound signal.
2. The adaptive noise reduction method for hearing aids according to claim 1, wherein said step of scheduling a target noise reduction algorithm model from preset noise reduction algorithm models according to said negative feedback to perform noise reduction processing on said noisy data signal to obtain a clean sound signal comprises:
determining a noise reduction algorithm model to be selected from each preset noise reduction algorithm model according to the negative feedback, wherein the number of the noise reduction algorithm models to be selected is greater than or equal to one;
if the number of the to-be-selected noise reduction algorithm models is equal to one, taking the to-be-selected noise reduction algorithm models as target noise reduction algorithm models;
if the number of the to-be-selected noise reduction algorithm models is larger than one, determining the target noise reduction algorithm model according to the respective calculation power data of each to-be-selected noise reduction algorithm model;
and calling the target noise reduction algorithm model to perform noise reduction processing on the noisy data signal to obtain a pure sound signal.
3. The adaptive noise reduction method for a hearing aid according to claim 2, wherein said step of determining said target noise reduction algorithm model based on respective computational power data of each of said candidate noise reduction algorithm models comprises:
traversing respective calculation force data of each to-be-selected noise reduction algorithm model;
and detecting the minimum target calculation force data in the calculation force data, and determining the noise reduction algorithm model to be selected corresponding to the target calculation force data as the target noise reduction algorithm model.
4. The adaptive noise reduction method for a hearing aid according to claim 1, wherein said step of scheduling a target noise reduction algorithm model from preset noise reduction algorithm models according to said negative feedback to perform noise reduction processing on said noisy data signal to obtain a clean sound signal further comprises:
detecting a target noise reduction algorithm model corresponding to the negative feedback in each noise reduction algorithm model according to a preset power saving strategy;
and calling the target noise reduction algorithm model to perform noise reduction processing on the noisy data signal to obtain a pure sound signal.
5. The adaptive noise reduction method for hearing aids according to claim 1, wherein the model output result is speech data after noise reduction processing by the noise reduction algorithm model, the negative feedback neural network is obtained in advance through deep learning training of a nonlinear relationship between noisy speech and noise reduction effect artificial MOS scoring,
the step of evaluating the noise reduction effect and generating negative feedback for the model output result based on a preset negative feedback neural network comprises:
preprocessing the voice data and inputting the voice data into a negative feedback neural network so that the negative feedback neural network can output the noise reduction algorithm model to score the MOS with the noise reduction effect of noise reduction of the noisy data signal;
detecting the difference value obtained by subtracting the preset standard MOS score from the MOS score;
and if the difference value is negative, scoring according to the preset standard MOS to generate negative feedback.
6. The adaptive noise reduction method for a hearing aid according to claim 5, wherein after the step of detecting the difference of the MOS score minus a preset standard MOS score, the method further comprises:
and if the difference is positive, adding an execution delay time for performing noise reduction effect evaluation and generating negative feedback based on a preset negative feedback neural network.
7. The adaptive noise reduction method for a hearing aid according to claim 6, wherein said performing a lag time comprises: a first hysteresis time and a second hysteresis time, the first hysteresis time being less than the second hysteresis time,
the step of adding the execution lag time for noise reduction effect evaluation and negative feedback generation based on a preset negative feedback neural network comprises:
detecting whether execution delay exists when the current execution carries out noise reduction effect evaluation based on the negative feedback neural network and generates negative feedback;
if not, the first lag time is increased for noise reduction effect evaluation and negative feedback generation based on the negative feedback neural network;
and if so, increasing the second lag time for noise reduction effect evaluation and generation of negative feedback based on the negative feedback neural network.
8. An adaptive noise reduction device for a hearing aid, characterized in that the adaptive noise reduction device comprises:
the acquisition module is used for inputting a noisy digital signal into a noise reduction algorithm model in the current acoustic environment to obtain a model output result, wherein the noisy digital signal is obtained by encoding an acquired noisy sound signal;
the effect evaluation module is used for carrying out noise reduction effect evaluation on the model output result based on a preset negative feedback neural network and generating negative feedback;
and the noise reduction module is used for scheduling a target noise reduction algorithm model from preset noise reduction algorithm models according to the negative feedback to perform noise reduction processing on the noisy data signal so as to obtain a pure sound signal.
9. A hearing aid, characterized in that the hearing aid comprises: a memory, a processor, a communication bus and an adaptive noise reduction program for a hearing aid stored on said memory,
the communication bus is used for realizing communication connection between the processor and the memory;
the processor is adapted to execute an adaptive noise reduction program of the internet based hearing aid to implement the steps of the adaptive noise reduction method of the hearing aid according to any of the claims 1 to 7.
10. A computer storage medium, characterized in that the computer storage medium has stored thereon an adaptive noise reduction program for a hearing aid, which when executed by a processor implements the steps of the adaptive noise reduction method for a hearing aid according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011415526.8A CN112565997B (en) | 2020-12-04 | 2020-12-04 | Adaptive noise reduction method and device for hearing aid, hearing aid and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011415526.8A CN112565997B (en) | 2020-12-04 | 2020-12-04 | Adaptive noise reduction method and device for hearing aid, hearing aid and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112565997A true CN112565997A (en) | 2021-03-26 |
| CN112565997B CN112565997B (en) | 2022-03-22 |
Family
ID=75058954
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011415526.8A Active CN112565997B (en) | 2020-12-04 | 2020-12-04 | Adaptive noise reduction method and device for hearing aid, hearing aid and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112565997B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113660594A (en) * | 2021-08-21 | 2021-11-16 | 武汉左点科技有限公司 | Self-adjusting noise reduction method and device for hearing aid system |
| CN114666092A (en) * | 2022-02-16 | 2022-06-24 | 奇安信科技集团股份有限公司 | Real-time behavior safety baseline data noise reduction method and device for safety analysis |
| CN114859419A (en) * | 2022-04-26 | 2022-08-05 | 陕西物科微达光学仪器有限公司 | Doppler high-frequency signal noise reduction method and system |
| CN116132899A (en) * | 2023-04-18 | 2023-05-16 | 杭州汇听科技有限公司 | Remote verification and adjustment system of hearing aid |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1326479A2 (en) * | 1997-04-16 | 2003-07-09 | DSPFactory Ltd. | Method and apparatus for noise reduction, particularly in hearing aids |
| KR101231866B1 (en) * | 2012-09-11 | 2013-02-08 | (주)알고코리아 | Hearing aid for cancelling a feedback noise and controlling method therefor |
| CN107229227A (en) * | 2017-06-30 | 2017-10-03 | 邢优胜 | A kind of active noise controlling method based on fuzzy neural network, system and the tank helmet |
| CN109712628A (en) * | 2019-03-15 | 2019-05-03 | 哈尔滨理工大学 | A kind of voice de-noising method and audio recognition method based on RNN |
| CN109859767A (en) * | 2019-03-06 | 2019-06-07 | 哈尔滨工业大学（深圳） | A kind of environment self-adaption neural network noise-reduction method, system and storage medium for digital deaf-aid |
| CN111192599A (en) * | 2018-11-14 | 2020-05-22 | 中移（杭州）信息技术有限公司 | Noise reduction method and device |
-
2020
- 2020-12-04 CN CN202011415526.8A patent/CN112565997B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1326479A2 (en) * | 1997-04-16 | 2003-07-09 | DSPFactory Ltd. | Method and apparatus for noise reduction, particularly in hearing aids |
| KR101231866B1 (en) * | 2012-09-11 | 2013-02-08 | (주)알고코리아 | Hearing aid for cancelling a feedback noise and controlling method therefor |
| CN107229227A (en) * | 2017-06-30 | 2017-10-03 | 邢优胜 | A kind of active noise controlling method based on fuzzy neural network, system and the tank helmet |
| CN111192599A (en) * | 2018-11-14 | 2020-05-22 | 中移（杭州）信息技术有限公司 | Noise reduction method and device |
| CN109859767A (en) * | 2019-03-06 | 2019-06-07 | 哈尔滨工业大学（深圳） | A kind of environment self-adaption neural network noise-reduction method, system and storage medium for digital deaf-aid |
| CN109712628A (en) * | 2019-03-15 | 2019-05-03 | 哈尔滨理工大学 | A kind of voice de-noising method and audio recognition method based on RNN |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113660594A (en) * | 2021-08-21 | 2021-11-16 | 武汉左点科技有限公司 | Self-adjusting noise reduction method and device for hearing aid system |
| CN113660594B (en) * | 2021-08-21 | 2024-05-17 | 武汉左点科技有限公司 | Self-adjusting noise reduction method and device for hearing aid system |
| CN114666092A (en) * | 2022-02-16 | 2022-06-24 | 奇安信科技集团股份有限公司 | Real-time behavior safety baseline data noise reduction method and device for safety analysis |
| CN114859419A (en) * | 2022-04-26 | 2022-08-05 | 陕西物科微达光学仪器有限公司 | Doppler high-frequency signal noise reduction method and system |
| CN116132899A (en) * | 2023-04-18 | 2023-05-16 | 杭州汇听科技有限公司 | Remote verification and adjustment system of hearing aid |
| CN116132899B (en) * | 2023-04-18 | 2023-06-16 | 杭州汇听科技有限公司 | Remote verification and adjustment system of hearing aid |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112565997B (en) | 2022-03-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112565997B (en) | Adaptive noise reduction method and device for hearing aid, hearing aid and storage medium | |
| CN111179961B (en) | Audio signal processing method and device, electronic equipment and storage medium | |
| CN110428808B (en) | Voice recognition method and device | |
| CN105009204B (en) | Speech recognition power management | |
| CN110310628B (en) | Method, device and equipment for optimizing wake-up model and storage medium | |
| CN107799126A (en) | Sound end detecting method and device based on Supervised machine learning | |
| CN107734126A (en) | voice adjusting method, device, terminal and storage medium | |
| CN107564532A (en) | Awakening method, device, equipment and the computer-readable recording medium of electronic equipment | |
| CN113015073A (en) | Method for adjusting a hearing instrument and related hearing system | |
| CN114338623A (en) | Audio processing method, device, equipment, medium and computer program product | |
| CN115775564B (en) | Audio processing method, device, storage medium and intelligent glasses | |
| CN115810356A (en) | Voice control method, device, storage medium and electronic equipment | |
| CN114333774A (en) | Speech recognition method, speech recognition device, computer equipment and storage medium | |
| CN116156439B (en) | Intelligent wearable electronic intercom interaction system | |
| CN114822573B (en) | Voice enhancement method, device, earphone device and computer readable storage medium | |
| CN116132875A (en) | Multi-mode intelligent control method, system and storage medium for hearing-aid earphone | |
| CN112954570B (en) | Hearing assistance method, device, equipment and medium integrating edge computing and cloud computing | |
| CN113808566B (en) | Vibration noise processing method and device, electronic equipment and storage medium | |
| CN118524337A (en) | Method, device, equipment and storage medium for managing hearing assistance equipment | |
| CN113889109B (en) | Voice wake-up mode adjusting method, storage medium and electronic equipment | |
| CN118200829A (en) | Hearing aid method, device, equipment and storage medium | |
| CN115331672B (en) | Device control method, device, electronic device and storage medium | |
| CN116367063B (en) | Bone conduction hearing aid equipment and system based on embedded | |
| CN118368561B (en) | Bluetooth headset noise reduction processing method, device, equipment and storage medium | |
| CN112771608A (en) | Voice information processing method and device, storage medium and electronic equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |