Noise reduction cochlear implant and method of controlling the same

Abstract

The application discloses a noise reduction cochlear implant and a control method thereof. The cochlear implant comprises a processing module and an implant. The processing module comprises: a microphone array; a noise reduction unit configured to perform voice activity detection based on a sound signal collected by the microphone array; when it is determined that the sound signal is a voice activity signal, determine a noise type and a noise intensity based on a noise signal collected by an auxiliary microphone and a preset noise feature library; determine a noise reduction parameter based on the noise type, the noise intensity and a preset noise reduction level table; output a target voice signal based on the noise reduction parameter and a first voice signal collected by a main microphone; an adjustment unit configured to adjust a sampling rate of the microphone array based on the noise intensity and the intelligibility of the target voice signal; and the implant is configured to stimulate auditory nerves based on the target voice signal. The application has the effect of improving the noise reduction effect of the cochlear implant in a complex environment.

Description

Technical Field

[0001] This application relates to the field of cochlear implant technology, and in particular to a noise-reducing cochlear implant and its control method. Background Technology

[0002] A cochlear implant is an electronic medical implant specifically designed to compensate for sensorineural hearing loss. Its basic working principle involves an externally worn processing module that collects sound signals from the external environment in real time, analyzes, processes, and outputs electrical stimulation, bypassing the damaged inner ear structure and directly acting on the auditory nerve to replace and reconstruct auditory function in the brain's auditory center. Noise reduction processing of the sound signal, as a core step in signal processing, determines the purity and fidelity of the subsequent electrical stimulation signal, directly affecting the patient's auditory experience in daily communication. In current technologies, cochlear implant microphone arrays often use a parallel layout, with signal acquisition points on the same horizontal axis. This makes precise sound source localization difficult, hindering the differentiation between target speech and environmental noise, and limiting noise reduction effectiveness. Furthermore, current noise reduction algorithms are prone to causing speech waveform distortion during processing and lack the ability to differentiate between different types of noise. Therefore, the issue of signal noise reduction in cochlear implants deserves attention. Summary of the Invention

[0003] In view of this, embodiments of this application provide a noise-reducing cochlear implant and its control method, aiming to improve the noise reduction effect of the cochlear implant in complex environments and enhance the user's auditory experience.

[0004] In a first aspect, a noise-reducing cochlear implant is provided, comprising a processing module and an implant. The processing module includes: a microphone array, including a main microphone and an auxiliary microphone, configured to acquire sound signals; a noise reduction unit, connected to the microphone array, configured to perform speech activity detection based on the sound signals acquired by the microphone array; when the sound signal is determined to be a speech activity signal, determining the noise type and noise intensity based on the noise signal acquired by the auxiliary microphone and a preset noise feature library; determining noise reduction parameters based on the noise type, noise intensity, and a preset noise reduction level table; and outputting a target speech signal based on the noise reduction parameters and a first speech signal acquired by the main microphone; an adjustment unit, connected to the microphone array and the noise reduction unit, configured to adjust the sampling rate of the microphone array based on the noise intensity and the clarity of the target speech signal; and the implant is configured to stimulate the auditory nerve based on the target speech signal.

[0005] The above-described noise-reducing cochlear implant significantly improves noise processing in complex environments. A microphone array including main and auxiliary microphones enables accurate spatial sound source localization and targeted sound pickup. Combined with speech activity detection, noise classification and comparison, and parametric noise reduction processing performed by the noise reduction unit, it achieves differentiated suppression of various environmental noises, making the extracted target speech signal closer to natural speech and avoiding speech distortion during the noise reduction process. The introduction of an adjustment unit dynamically adjusts the sampling rate based on noise intensity and speech intelligibility, allowing the device to autonomously sense environmental changes and dynamically allocate computing power. This ensures effective noise reduction while reducing power consumption, effectively improving the cochlear implant's environmental adaptability and battery life.

[0006] Optionally, the main microphone is positioned closer to the target sound source, while the auxiliary microphone is positioned further away from the target sound source.

[0007] Optionally, when the noise reduction unit performs voice activity detection based on the sound signal collected by the microphone array, it is configured to: acquire the short-time energy of the main microphone and the short-time energy of the auxiliary microphone; calculate the voice activity value based on the short-time energy of the main microphone, the short-time energy of the auxiliary microphone, and the instantaneous signal-to-noise ratio; and determine whether the sound signal is a voice activity signal based on the voice activity value and a preset voice threshold.

[0008] Optionally, when the noise reduction unit determines the noise type and noise intensity based on the noise signal collected by the auxiliary microphone and a preset noise feature library, it is configured to: extract the feature value of the noise signal, calculate the noise similarity coefficient based on the feature value and the noise feature library, and determine the noise type; and obtain the noise sound pressure level of the noise signal to determine the noise intensity.

[0009] Optionally, when the noise reduction unit determines the noise reduction parameters based on the noise type, noise intensity, and a preset noise reduction level table, it is configured to: determine the noise reduction level based on the noise sound pressure level and the noise reduction level table; and calculate the noise reduction parameters based on the noise reduction level, the noise similarity coefficient, and the noise intensity.

[0010] Optionally, the microphone array includes at least two main microphones. When the noise reduction unit outputs the target speech signal based on the noise reduction parameters and the first speech signal acquired by the main microphones, it is configured to: acquire the time difference between the acquisition of the first speech signal by the main microphones; calculate the azimuth angle of the target speech in the first speech signal based on the sound speed parameter, the spacing parameter of the main microphones, and the time difference; perform beamforming processing on the first speech signal based on the azimuth angle to generate a second speech signal; and perform noise reduction on the noise frequency band in the second speech signal based on the noise reduction parameters to output the target speech signal.

[0011] Optionally, when adjusting the sampling rate of the microphone array based on noise intensity and the clarity of the target speech signal, the adjustment unit is configured to: acquire a first variance of the target speech signal and a second variance of the noise signal; calculate the clarity of the target speech signal based on the first and second variances; calculate a target sampling rate based on the clarity, noise sound pressure level, a preset minimum sampling rate, and a preset maximum sampling rate; and control the microphone array to switch to the target sampling rate within a preset switching time.

[0012] Optionally, the adjustment unit is further configured to: increase the proportion of computing power allocated to processing speech signals in the noise reduction unit when the noise intensity is greater than or equal to a first intensity threshold; and control the auxiliary microphone to enter a sleep state and reduce the proportion of computing power allocated to processing speech signals in the noise reduction unit when the noise intensity is less than a second intensity threshold; wherein the first intensity threshold is greater than the second intensity threshold.

[0013] Secondly, a control method for a noise-reducing cochlear implant is provided, comprising: acquiring sound signals through a microphone array, wherein the microphone array includes a main microphone and an auxiliary microphone; performing speech activity detection based on the sound signals acquired by the microphone array; when the sound signal is determined to be a speech activity signal, determining the noise type and noise intensity based on the noise signal acquired by the auxiliary microphone and a preset noise feature library; determining noise reduction parameters based on the noise type, noise intensity, and a preset noise reduction level table; outputting a target speech signal based on the noise reduction parameters and a first speech signal acquired by the main microphone; adjusting the sampling rate of the microphone array based on the noise intensity and the clarity of the target speech signal; and stimulating the auditory nerve based on the target speech signal.

[0014] Optionally, based on noise reduction parameters and a first speech signal acquired by the main microphone, a target speech signal is output, including: acquiring the time difference between the acquisition of the first speech signal by the main microphone; calculating the azimuth angle of the target speech segment in the first speech signal based on the sound speed parameter, the spacing parameter of the main microphone, and the time difference; performing beamforming processing on the first speech signal based on the azimuth angle to generate a second speech signal; and performing noise reduction on the noise frequency band in the second speech signal based on the noise reduction parameters to output the target speech signal. Attached Figure Description

[0015] The accompanying drawings used in the description of the embodiments of this disclosure are briefly introduced below: Figure 1 The present application provides a schematic diagram of the structure of a noise-reducing cochlear implant according to some embodiments; Figure 2 This invention provides a schematic diagram of the structure of another noise-reducing cochlear implant according to some embodiments of the present application; Figure 3 The diagram shows a flowchart of a noise-reducing cochlear implant control method provided in some embodiments of this application. Detailed Implementation

[0016] To more clearly illustrate the technical solutions in the embodiments of this disclosure, examples of implementation methods of this disclosure will be described below with reference to the accompanying drawings. The accompanying drawings described below are merely some embodiments of this disclosure. For those skilled in the art, other drawings and other implementation methods can be obtained based on these drawings without creative effort. Adjustments and improvements made without departing from the concept of this disclosure are all within the protection scope of this disclosure.

[0017] To keep the drawings simple, each figure only schematically shows the parts relevant to the embodiment, and they do not represent the actual structure of the product. In addition, for the sake of clarity and ease of understanding, some figures only schematically show parts of components with the same structure or function, and there may actually be more or fewer components with the same structure or function.

[0018] In this disclosure, unless otherwise expressly specified and limited, ordinal numbers, such as “first”, “second”, etc., are used only to distinguish and describe related objects, and should not be construed as indicating or implying the relative importance or order between related objects; furthermore, they do not represent the quantity of related objects. “Multiple” includes two or more, and other quantifiers are similar. “ / ” is used to describe the relationship between related objects, indicating an “or” relationship between them. “And / or” is used to describe the relationship between related objects, including any combination relationship between them, such as “a and / or b” including: “a alone”, “b alone”, or “a and b”. “One or more” or “at least one” of multiple objects refers to any object or any combination of multiple objects, such as “one or more of a1, a2, a3” or “at least one of a1, a2, a3” including: “a1 alone”, “a2 alone”, “a3 alone”, “a1 and a2”, “a1 and a3”, “a2 and a3”, or “a1, a2 and a3”.

[0019] A cochlear implant is an electronic medical implant specifically designed to compensate for sensorineural hearing loss. Its basic principle is to pick up external sound signals, analyze and process them, and then output electrical stimulation directly to the auditory nerve, bypassing the damaged inner ear structures to achieve auditory replacement. Noise reduction processing of the sound signal, as a core step in signal processing, determines the purity and fidelity of the subsequent electrical stimulation signal, directly affecting the patient's auditory experience in daily communication. Current cochlear implants mostly use parallel microphone arrays based on single or dual microphones to extract sound signals, primarily relying on one-dimensional spatial orientation to suppress background noise. While this method can provide basic speech perception in quiet environments, it cannot achieve precise spatial sound source localization, especially in noisy environments where it struggles to effectively distinguish target speech from environmental noise, resulting in generally low target speech recognition rates. Furthermore, existing noise reduction methods generally neglect the coordination between microphone spatial layout and noise reduction algorithms, lacking differentiated processing for different noise types and intensities, severely impacting the patient's speech comprehension ability and wearing experience in real-world auditory scenarios. For example, current cochlear implants primarily achieve noise reduction by filtering fixed frequency bands, lacking protection for core audio frequencies. This results in speech distortion and reduced clarity while filtering out noise. Furthermore, current cochlear implants often maintain a fixed sampling rate, constantly operating at high power consumption, significantly shortening the device's battery life.

[0020] Therefore, this application provides a noise-reducing cochlear implant and its control method. Accurate spatial sound source localization is achieved through a microphone array including main and auxiliary microphones. This is combined with speech activity detection, noise classification and comparison, and parameterized noise reduction processing. Furthermore, the sampling rate is dynamically adjusted based on noise intensity and speech intelligibility to enhance the cochlear implant's speech discrimination capability. While ensuring noise reduction effectiveness, this method also reduces device power consumption, effectively improving the cochlear implant's environmental adaptability and battery life.

[0021] The following description is in conjunction with the accompanying drawings: Figure 1A schematic diagram of a noise-reducing cochlear implant provided in some embodiments of this application is shown. The noise-reducing cochlear implant includes a processing module 100 and an implant 200. The processing module 100 includes: a microphone array 110, including a main microphone and an auxiliary microphone, configured to collect sound signals; a noise reduction unit 120, connected to the microphone array 110, configured to perform speech activity detection based on the sound signals collected by the microphone array 110; when the sound signal is determined to be a speech activity signal, determining the noise type and noise intensity based on the noise signal collected by the auxiliary microphone and a preset noise feature library; determining noise reduction parameters based on the noise type, noise intensity, and a preset noise reduction level table; outputting a target speech signal based on the noise reduction parameters and a first speech signal collected by the main microphone; an adjustment unit 130, connected to the microphone array 110 and the noise reduction unit 120, configured to adjust the sampling rate of the microphone array 110 based on the noise intensity and the clarity of the target speech signal; and the implant 200 configured to stimulate the auditory nerve based on the target speech signal.

[0022] The noise-reducing cochlear implant described above includes a processing module 100 and an implant 200. The implant 200 is surgically implanted under the patient's skin, consisting of a stimulation receiver and an electrode array extending into the scala tympani of the inner ear. It receives audio and control signals from the external processing module 100 and converts them into electrical pulse sequences that stimulate the patient's auditory nerve fibers. The processing module 100, typically worn behind the patient's ear, integrates a microphone array 110 for acquiring sound signals, a noise-reducing unit 120 for processing sound signals, and an adjustment unit 130 for adjusting the sampling frequency of the microphone array 110. It also transmits the target speech signal to the implant 200.

[0023] First, upon activation of the cochlear implant, the device performs a self-test, checking aspects such as the operational status of the microphone array 110, the noise reduction unit 120, and the response sensitivity of the adjustment unit 130. After passing the self-test, it enters a standby state, awaiting sound signal input. When the sound signal intensity acquired by the main microphone reaches a preset recognition threshold, the auxiliary microphone is activated and begins acquiring ambient noise signals. The sound signal may include ambient white noise, human voices, or other sounds. The core difference between the auxiliary and main microphones lies in their physical functions and pickup focus. The main microphone can be positioned directly in front of the patient to target and pick up speech signals containing target semantic information, such as the speaker's actual voice. The auxiliary microphone, on the other hand, can be deployed in areas more easily enveloped by the surrounding ambient sound field, focusing on picking up omnidirectional background noise from around the patient. The initial sound signal acquired by the microphone array 110 is a continuous analog signal with complex components, including the natural speech of the target speaker, as well as ambient white noise, traffic noise, wind noise, or other background noise. To facilitate computation by the processing module 100, the analog sound signal can be converted into a digital audio signal in real time using the built-in analog-to-digital converter of the processing module 100. In subsequent processing, the core frequency bands of the target language are primarily preserved, while non-target noise that interferes with semantic understanding is separated out.

[0024] The noise reduction unit 120 can perform speech activity detection based on the sound signals collected by the microphone array 110, determining whether the current sound signal belongs to a speech activity signal. This allows for the early screening of environmental noise that does not contain human speech, preventing noise from entering subsequent processing and thus avoiding wasted computing power. For example, it can identify whether the current audio stream contains human language activity features by analyzing the energy fluctuations and signal-to-noise ratio of the signal in real time. For example, it can use a traditional threshold discrimination method based on short-time energy and zero-crossing rate or a deep learning detection model based on lightweight neural networks. If the current sound signal is determined to contain only noise such as wind or traffic noise, it can be directly intercepted. Only when the current sound signal is clearly identified as a speech activity signal can it proceed to the subsequent noise reduction processing.

[0025] Next, the noise signal collected by the auxiliary microphone is extracted and compared with the system's preset noise feature library to determine the type of noise in the current environment, such as wind or horn noise, and the noise intensity is calculated simultaneously. After obtaining the environmental indicators, the noise reduction unit 120 can match noise reduction parameters to the current scene based on an internally preset noise reduction level table. The preset noise reduction level table can be obtained through acoustic experiments and clinical hearing tests. Furthermore, it can also be personalized by considering the individual patient's hearing range and comfort level, thereby providing appropriate noise reduction intensities for different types of noise.

[0026] Subsequently, the noise reduction unit 120 applies the selected noise reduction parameters to the first speech signal acquired by the main microphone. The first speech signal is the original sound signal acquired by the main microphone, which contains valid speech information from the target direction and environmental background noise. The noise reduction unit 120 can first utilize the signal differences between the main microphones to perform spatial directional processing such as beamforming to initially reduce interference from other directions. Then, using the selected noise reduction parameters as weights, it combines an adaptive filtering algorithm to suppress residual noise frequency bands. This process removes noise while preserving core speech features to the greatest extent possible, ultimately outputting the target speech signal.

[0027] While the noise reduction unit 120 performs noise reduction processing, unlike existing cochlear implants which use a fixed sampling rate, the adjustment unit 130 in this application can monitor the noise intensity and the clarity of the target speech signal in real time, and evaluate and adjust the sampling rate of the microphone array 110 based on these two dimensions. For example, when the patient is in a noisy environment where the speech is difficult to recognize, the sampling rate can be increased to ensure sufficient sound source data for the noise reduction algorithm. Conversely, when the patient is in a relatively quiet environment where the target speech is relatively clear, the sampling rate can be appropriately reduced to extend the cochlear implant's battery life. Finally, the implant 200 converts the target speech signal sent by the processing module 100 into an electrical pulse sequence, precisely stimulating the patient's auditory nerve to achieve auditory reconstruction.

[0028] The above-described noise-reducing cochlear implant significantly improves noise processing in complex environments. A microphone array including main and auxiliary microphones enables accurate spatial sound source localization and targeted sound pickup. Combined with speech activity detection, noise classification and comparison, and parametric noise reduction processing performed by the noise reduction unit, it achieves differentiated suppression of various environmental noises, making the extracted target speech signal closer to natural speech and avoiding speech distortion during the noise reduction process. The introduction of an adjustment unit 130 dynamically adjusts the sampling rate based on noise intensity and speech intelligibility, enabling the device to autonomously sense environmental changes and dynamically allocate computing power. This reduces power consumption while maintaining noise reduction effectiveness, effectively improving the cochlear implant's environmental adaptability and battery life.

[0029] Figure 2 A schematic diagram of another noise-reducing cochlear implant provided in some embodiments of this application is shown. (Reference) Figure 2The processing module 100 may include a main body 140 mounted behind the auricle and a wireless control unit 160 attached to the outer side of the scalp. The main body 140 is connected to the wireless control unit 160 via a cable 150. The main body 140 may include a battery and a microphone array 110 for picking up ambient sound. The wireless control unit 160 may be attached to the outer side of the patient's scalp and contains a radio frequency transmitting antenna for receiving digital audio transmitted from the main body 140, converting it into high-frequency electromagnetic waves, and transmitting it to a subcutaneous implant. The noise reduction unit 120 and the adjustment unit 130 may be highly integrated digital signal processors, microcontrollers, or application-specific integrated circuits, primarily packaged and integrated on a main control circuit board inside the main body 140, and powered by a nearby battery. The microphone array 110 is arranged on the outer shell of the main body 140 of the processing module 100, for example... Figure 1 The diagram shows the layout including main microphone 110A, main microphone 110B, and auxiliary microphone 110C. It should be noted that... Figure 1 The structural form, component division, and distribution of the main and auxiliary microphones shown are only one optional implementation of the processing module 100 of this application. Those skilled in the art can make adjustments according to actual product requirements. For example, in an integrated or off-ear cochlear implant implementation, the processing module 100 can eliminate the main body 140 and integrate the battery, microphone array 110, noise reduction unit 120, and adjustment unit 130 inside the housing of the wireless control unit 160 to form an integrated structure.

[0030] like Figure 2 As shown, in some embodiments of this application, the main microphone is positioned close to the target sound source, while the auxiliary microphone is positioned far away from the target sound source.

[0031] Traditional cochlear implant microphone arrays 110 are typically arranged in a horizontal straight line. However, this collinear layout can easily lead to spatial localization confusion when resolving the sound source location, either front-back or up-down. To avoid this problem, a main microphone can be positioned close to the target sound source to perform directional tracking of the target speech directly in front, while the auxiliary microphone can be positioned further away from the target sound source to focus on capturing ambient background noise from the surroundings, providing a data basis for subsequent noise classification. Furthermore, the microphone array 110 can include two main microphones and one auxiliary microphone, employing a triangular asymmetrical layout. For example, see reference... Figure 2In the structure shown, main microphones 110A and 110B are symmetrically arranged on the upper sides of the processing module 100 housing, with a horizontal distance between them that can be configured to be 15mm to 20mm. The auxiliary microphone 110C is separately located at the lower center of the processing module 100, with a vertical distance of 10mm to 12mm between it and the midpoint of the line connecting the two main microphones above. Because the auxiliary microphone 110C is positioned relatively close to the bottom of the ear, it attenuates sound from directly in front to a certain extent, which helps to pick up background noise from the surrounding environment.

[0032] In some embodiments of this application, when the noise reduction unit 120 performs voice activity detection based on the sound signal collected by the microphone array 110, it is configured to: acquire the short-time energy of the main microphone and the short-time energy of the auxiliary microphone; calculate the voice activity value based on the short-time energy of the main microphone, the short-time energy of the auxiliary microphone, and the instantaneous signal-to-noise ratio; and determine whether the sound signal is a voice activity signal based on the voice activity value and a preset voice threshold.

[0033] In microphone array 110, the physical position of the main microphone facilitates the pickup of sound directly in front, while the auxiliary microphone focuses on picking up ambient noise. Therefore, when a patient is conversing with others, the target speech picked up by the main microphone exhibits a distinctly fluctuating envelope phenomenon, with its energy amplitude significantly higher than that of the auxiliary microphone. Conversely, if the environment only contains wind noise or traffic noise, the short-time energy distributions of the main and auxiliary microphones are quite similar. Based on these characteristics, the short-time energy distributions of the main and auxiliary microphones can be compared to preliminarily determine whether the sound signal contains the target speech signal. The instantaneous signal-to-noise ratio (SNR) refers to the ratio between the energy of the target speech signal and the noise energy within a very short time window, reflecting the dynamic changes in the sound signal. Therefore, the instantaneous SNR of the sound signal can be combined with the short-time energy of the microphones as a basis for judging the target speech signal. Furthermore, the probability of the target speech signal appearing in the sound signal can be represented by the speech activity value. For example, the speech activity value can be calculated using Formula 1:

[0034] in, For voice activity value; For example, as the main source of short-term energy in wheat. , These are the short-time energy values ​​of the main microphone 110A and the main microphone 110B, respectively. To provide short-term energy for the auxiliary microphone 110C; This represents the instantaneous signal-to-noise ratio. A small constant can be chosen to avoid the denominator being zero.

[0035] The noise reduction unit 120 compares the calculated speech activity value with a preset speech threshold. If the speech activity value is greater than or equal to the preset speech threshold, the current sound signal is determined to be a speech activity signal and the subsequent noise reduction process begins; otherwise, it enters a low-power mode. The preset speech threshold can be a fixed value pre-set based on acoustic experiments and environmental tests, or it can be a dynamically adjustable value. For example, the noise reduction unit 120 can statistically analyze the average energy of background noise over a period of time and adjust the preset speech threshold upwards or downwards based on the average energy level to adapt to different noise intensities.

[0036] In some embodiments of this application, when the noise reduction unit 120 determines the noise type and noise intensity based on the noise signal collected by the auxiliary microphone and a preset noise feature library, it is configured to: extract the feature value of the noise signal, calculate the noise similarity coefficient based on the feature value and the noise feature library, and determine the noise type; obtain the noise sound pressure level of the noise signal, and determine the noise intensity.

[0037] Compared to the main microphone, the auxiliary microphone can receive background noise from the surrounding environment more comprehensively, and the noise signal it collects can reflect the acoustic properties of the current background noise. The noise reduction unit 120 can analyze the noise signal and extract multi-dimensional feature values ​​that characterize the noise, such as short-time zero-crossing rate, spectral centroid, and frequency domain energy distribution. The system's internal preset noise feature library contains pre-loaded standard acoustic templates for various typical noises, such as white noise, traffic noise, wind noise, or mechanical knocking sounds. The noise reduction unit 120 compares the real-time extracted feature values ​​with each template in the preset noise feature library to obtain a noise similarity coefficient. For example, the noise similarity coefficient can be calculated using Formula 2:

[0038] in, It is the noise similarity coefficient, ranging from 0 to 1; It is the k-th feature value in the collected noise signal; It is the k-th feature value of a certain type of noise in the noise feature library; This refers to the feature dimension. The feature dimension is the number of noise features extracted when analyzing a noise signal. For example, when extracting two noise features—the spectral centroid and the frequency domain energy—the feature dimension is set to 2. The noise reduction unit 120 can calculate the similarity coefficient between the noise signal and all noise types, and select the template category with the highest similarity coefficient to determine the noise type of the current environment.

[0039] Noise sound pressure level is a quantitative indicator of sound energy. While determining the noise type, the noise reduction unit 120 can extract the noise sound pressure level of the noise signal and determine it as noise intensity. For example, noise intensity can be calculated using Formula 3:

[0040] in, Noise sound pressure level, measured in dB; The root mean square sound pressure level of the noise signal can represent the original physical sound wave energy collected by the auxiliary microphone, and its unit is Pa. This is the reference sound pressure level. The reference sound pressure level is the smallest sound level that the human ear can hear, which can be taken as 2 × 10⁻⁶ in this case. -5 Pa is used as a standard reference value. After determining the noise intensity and noise type, the noise reduction unit 120 can calculate and allocate the most suitable noise reduction parameters based on these indicators.

[0041] In some embodiments of this application, when the noise reduction unit 120 determines the noise reduction parameters based on the noise type, noise intensity, and a preset noise reduction level table, it is configured to: determine the noise reduction level based on the noise sound pressure level and the noise reduction level table; and calculate the noise reduction parameters based on the noise reduction level, the noise similarity coefficient, and the noise intensity.

[0042] The preset noise reduction level table can be set based on acoustic experiments and clinical hearing test data, which specifies the noise reduction intensity benchmarks corresponding to different sound pressure level ranges, such as dividing them into mild, moderate, and severe intervention levels. The noise reduction unit 120 can compare and match the currently acquired noise sound pressure level with the noise reduction level table to establish a macroscopic noise reduction level. For example, the noise reduction level table can refer to Formula 4:

[0043] in, The noise reduction level ranges from 0 to 5. This represents the noise sound pressure level. Through the aforementioned graded processing, noise can be quickly suppressed, avoiding distortion of the target speech caused by excessive noise reduction. Furthermore, the noise reduction unit 120 calculates the noise reduction parameters by combining the noise similarity coefficient and noise intensity of the current noise signal. For example, refer to Formula 5:

[0044] in, This is a noise reduction parameter, representing the amount of noise gain attenuation; a negative value indicates that the noise is suppressed or attenuated. This is a weighting constant; its value range can be found by referring to [reference needed]. ; The correction coefficient for the noise type can be fine-tuned according to different noise types, and the value range can be from 0.05 to 0.15. Based on the noise reduction parameters, the noise reduction unit 120 can selectively suppress the noise frequency band in the first speech signal collected by the main microphone, thereby achieving differentiated noise reduction effects without damaging the target speech.

[0045] In some embodiments of this application, the microphone array 110 includes at least two main microphones. When the noise reduction unit 120 outputs a target speech signal based on the noise reduction parameters and the first speech signal acquired by the main microphones, it is configured to: acquire the time difference between the acquisition of the first speech signal by the main microphones; calculate the azimuth angle of the target speech in the first speech signal based on the sound speed parameters, the spacing parameters of the main microphones, and the time difference; perform beamforming processing on the first speech signal based on the azimuth angle to generate a second speech signal; and perform noise reduction on the noise frequency band in the second speech signal based on the noise reduction parameters to output the target speech signal.

[0046] In an array structure containing multiple main microphones, due to the physical spacing between the microphones, sound waves emitted from the same sound source will have a time difference when they reach different microphones. The noise reduction unit 120 can extract this time difference value and, combined with the sound speed parameters of the current environment and the physical spacing parameters between the microphones, use a geometric calculation model to convert this time difference into a three-dimensional spatial azimuth angle of the target speech source relative to the cochlear implant device. For example, see Formula 6:

[0047] in, It is the azimuth angle of the target speech, in degrees, ranging from -90° to 90°; The speed of sound can be taken as 340 m / s at room temperature; It is the time difference between the target speech captured by the two main microphones, in seconds; It is the distance between the two main microphones.

[0048] After determining the azimuth angle, the noise reduction unit 120 can perform beamforming processing on the first speech signal containing the target speech and background noise based on that azimuth angle. By adjusting the delay time and amplitude weighting coefficients of different microphone signal paths, a high-gain receiving beam is constructed in the direction of the target azimuth angle, while spatial attenuation is applied to interfering sound sources deviating from that azimuth angle, generating a second speech signal. For example, refer to Formula 7:

[0049] in, For beamforming gain; The weighting coefficient for the i-th microphone ranges from 0 to 1; For the frequency of the voice signal; The time delay of the i-th microphone relative to the reference microphone is expressed in seconds. The second speech signal generated after spatial directional filtering has significantly reduced noise from non-target sound source directions. Since noise interference still exists in the target direction, the noise reduction unit 120 can suppress the residual noise frequency bands in the second speech signal based on the noise reduction parameters, and finally output a high signal-to-noise ratio, high-fidelity target speech signal, which can then be converted into electrical impulses by the implant 200 to stimulate the auditory nerve.

[0050] In some embodiments of this application, when adjusting the sampling rate of the microphone array 110 based on noise intensity and the clarity of the target speech signal, the adjustment unit 130 is configured to: acquire a first variance of the target speech signal and a second variance of the noise signal; calculate the clarity of the target speech signal based on the first variance and the second variance; calculate a target sampling rate based on the clarity, noise sound pressure level, a preset minimum sampling rate, and a preset maximum sampling rate; and control the microphone array 110 to switch to the target sampling rate within a preset switching time.

[0051] Due to the envelope characteristics of natural human speech, the target speech signal has more energy than noise, resulting in a more pronounced first variance compared to the second variance. For example, the intelligibility of the target speech signal can be calculated using Equation 8:

[0052] in, The intelligibility of the target speech, ranging from 0 to 1; This is the variance of the noise signal, i.e., the second variance; Let V be the variance of the target speech signal, i.e., the first variance. By calculating the ratio of the two, we can objectively determine whether the current target speech signal is masked by environmental noise.

[0053] The sampling rate is the frequency at which a microphone captures sound. A higher sampling rate captures more comprehensive sound information, but also increases the device's power consumption. When a patient is in a noisy environment such as a subway station or market, the noise sound pressure level is high and speech intelligibility is low. To ensure the normal operation of the noise reduction algorithm, the sampling rate can be appropriately increased. Conversely, when a patient is in a quiet environment, the noise sound pressure level is low and the target speech intelligibility is high. In this case, the microphone's operating frequency can be actively reduced, thereby saving power to the cochlear implant while ensuring normal communication. For example, the target sampling rate can be calculated based on the noise sound pressure level, a preset minimum sampling rate, and a preset maximum sampling rate, as shown in Formula 9:

[0054] in, The target sampling rate; The lowest sampling rate; This is the highest sampling rate; This is an adjustment coefficient used to control the rate of change of the sampling rate; The threshold sound pressure level can be taken here as the boundary between moderate noise and quiet environments, i.e., 40dB. After calculating the target sampling rate, it can be adjusted in conjunction with the intelligibility. For example, when the target speech intelligibility is <60%, the target sampling rate can be appropriately increased, with a reference range of 32 to 48kHz; when the target speech intelligibility is 60% to 80%, the target sampling rate can be controlled at around 24kHz; when the target speech intelligibility is ≥80%, the target sampling rate can be appropriately decreased, and only the main microphone can be used to collect the sound signal. In addition, information such as noise suppression ratio and signal-to-noise ratio can also be used as reference factors for adjusting the sampling rate. After determining the target sampling rate, the adjustment unit 130 can control the microphone array 110 to smoothly transition and switch to the calculated target sampling rate within a preset switching time, avoiding popping or sound stuttering, and ensuring the patient's auditory experience. For example, the smooth switching of the sampling rate can refer to Formula 10:

[0055] in, The sampling rate at time t during the switching process; The sampling rate before the switch; The target sampling rate after the switch; Preset switching time; To switch time, and 0≤t≤T .

[0056] In some embodiments of this application, the adjustment unit 130 is further configured to: increase the computing power allocation ratio for processing speech signals in the noise reduction unit 120 when the noise intensity is greater than or equal to a first intensity threshold; and control the auxiliary microphone to enter a sleep state and reduce the computing power allocation ratio for processing speech signals in the noise reduction unit 120 when the noise intensity is less than a second intensity threshold; wherein the first intensity threshold is greater than the second intensity threshold.

[0057] Unlike traditional cochlear implants that maintain a fixed power consumption and computing power allocation throughout their entire operating cycle, the adjustment unit 130 in this application can adjust the operating status of the cochlear implant device according to changes in the external environment. When the adjustment unit 130 detects that the current noise intensity is greater than or equal to a first intensity threshold, it indicates that the patient is in a noisy environment. It can proactively increase the proportion of computing power allocated to speech signal processing in the noise reduction unit 120, concentrating computing resources on the noise reduction task to ensure that clear and intelligible target speech is still output to the patient even in harsh environments. Conversely, when the noise intensity is less than a second intensity threshold, it indicates that the patient has entered a relatively quiet environment. The adjustment unit 130 can, on the one hand, control the auxiliary microphone responsible for picking up background noise to enter a sleep state, and on the other hand, simultaneously reduce the proportion of computing power allocated to speech signal processing in the noise reduction unit 120, maintaining only basic functions and reducing power consumption and resource waste. Furthermore, the noise reduction level can be adjusted based on the noise intensity, thereby specifically improving the noise reduction effect.

[0058] Based on the same technical concept, this application also proposes a control method for noise-reducing cochlear implants. Figure 3 A flowchart illustrating a noise-reducing cochlear implant control method provided in some embodiments of this application is shown. The control method includes: S310 acquires sound signals through a microphone array, which includes a main microphone and an auxiliary microphone; S320 performs voice activity detection based on sound signals acquired by a microphone array; S330, when the sound signal is determined to be a speech activity signal, the noise type and noise intensity are determined based on the noise signal collected by the auxiliary microphone and the preset noise feature library; S340 determines noise reduction parameters based on noise type, noise intensity, and a preset noise reduction level table; S350 outputs the target speech signal based on the noise reduction parameters and the first speech signal collected by the main microphone. S360 adjusts the sampling rate of the microphone array based on noise intensity and the clarity of the target speech signal; S370 stimulates the auditory nerve based on the target speech signal.

[0059] The noise-reducing cochlear implant control method provided in this application, through an asymmetrical layout of main and auxiliary microphones and combined with environmental noise indicators, calculates and matches differentiated noise reduction parameters in real time, suppressing noise interference while restoring the natural listening experience of the target speech. Furthermore, this application dynamically adjusts the sampling rate and computing power allocation based on environmental noise intensity and speech intelligibility, reducing power consumption while ensuring performance, thus balancing noise reduction effect, battery life, and a good listening experience.

[0060] In some embodiments of this application, outputting a target speech signal based on noise reduction parameters and a first speech signal acquired by the main microphone includes: acquiring the time difference between the acquisition of the first speech signal by the main microphone; calculating the azimuth angle of the target speech segment in the first speech signal based on the sound speed parameter, the spacing parameter of the main microphone, and the time difference; performing beamforming processing on the first speech signal based on the azimuth angle to generate a second speech signal; and performing noise reduction on the noise frequency band in the second speech signal based on the noise reduction parameters to output the target speech signal.

[0061] In the above embodiments, the descriptions of each embodiment have their own emphasis. Parts not described in detail or in a particular embodiment can be referred to in the relevant descriptions of other embodiments. Furthermore, the above embodiments can be freely combined as needed.

Claims

1. A noise-reducing cochlear implant, characterized in that, It includes a processing module and an implant, the processing module including: A microphone array, including a main microphone and an auxiliary microphone, is configured to capture sound signals; A noise reduction unit, connected to the microphone array, is configured to perform speech activity detection based on the sound signal collected by the microphone array; when the sound signal is determined to be a speech activity signal, it determines the noise type and noise intensity based on the noise signal collected by the auxiliary microphone and a preset noise feature library; it determines noise reduction parameters based on the noise type, the noise intensity, and a preset noise reduction level table; and it outputs a target speech signal based on the noise reduction parameters and a first speech signal collected by the main microphone. An adjustment unit, connected to the microphone array and the noise reduction unit, is configured to adjust the sampling rate of the microphone array based on the noise intensity and the clarity of the target speech signal; The implant is configured to stimulate the auditory nerve based on the target speech signal.

2. The noise-reducing cochlear implant according to claim 1, characterized in that, The main microphone is positioned close to the target sound source, while the auxiliary microphone is positioned away from the target sound source.

3. The noise-reducing cochlear implant according to claim 1, characterized in that, The noise reduction unit is configured to perform voice activity detection based on the sound signals collected by the microphone array as follows: Acquire the short-time energy of the main microphone and the short-time energy of the auxiliary microphone; Based on the short-term energy of the main microphone, the short-term energy of the auxiliary microphone, and the instantaneous signal-to-noise ratio, the speech activity value is calculated; Based on the voice activity value and the preset voice threshold, it is determined whether the sound signal is a voice activity signal.

4. The noise-reducing cochlear implant according to claim 1, characterized in that, When determining the noise type and noise intensity based on the noise signal collected by the auxiliary microphone and a preset noise feature library, the noise reduction unit is configured as follows: Extract the feature values ​​of the noise signal, calculate the noise similarity coefficient based on the feature values ​​and the noise feature library, and determine the noise type; Obtain the noise sound pressure level of the noise signal and determine the noise intensity.

5. The noise-reducing cochlear implant according to claim 4, characterized in that, When determining the noise reduction parameters based on the noise type, noise intensity, and a preset noise reduction level table, the noise reduction unit is configured as follows: The noise reduction level is determined based on the noise sound pressure level and the noise reduction level table; The noise reduction parameters are calculated based on the noise reduction level, the noise similarity coefficient, and the noise intensity.

6. The noise-reducing cochlear implant according to claim 5, characterized in that, The microphone array includes at least two main microphones, and the noise reduction unit is configured to output a target speech signal based on the noise reduction parameters and the first speech signal acquired by the main microphones: Obtain the time difference between the acquisition of the first voice signal by the main microphone; Based on the sound speed parameter, the spacing parameter of the main microphone, and the time difference, the azimuth angle of the target speech in the first speech signal is calculated; Beamforming is performed on the first speech signal based on the azimuth angle to generate a second speech signal. The noise frequency band in the second speech signal is denoised based on the denoising parameters, and the target speech signal is output.

7. The noise-reducing cochlear implant according to claim 4, characterized in that, The adjustment unit is configured to adjust the sampling rate of the microphone array based on the noise intensity and the clarity of the target speech signal as follows: Obtain the first variance of the target speech signal and the second variance of the noise signal, and calculate the clarity of the target speech signal based on the first variance and the second variance; The target sampling rate is calculated based on the clarity, the noise sound pressure level, the preset minimum sampling rate, and the preset maximum sampling rate, and the microphone array is controlled to switch to the target sampling rate within a preset switching time.

8. The noise-reducing cochlear implant according to claim 7, characterized in that, The adjustment unit is further configured to: When the noise intensity is greater than or equal to the first intensity threshold, the proportion of computing power allocated to processing the speech signal in the noise reduction unit is increased; When the noise intensity is less than the second intensity threshold, the auxiliary microphone is controlled to enter a sleep state and the computing power allocation ratio for processing the voice signal in the noise reduction unit is reduced; Wherein, the first intensity threshold is greater than the second intensity threshold.

9. A control method for a noise-reducing cochlear implant, characterized in that, include: Sound signals are acquired through a microphone array, wherein the microphone array includes a main microphone and an auxiliary microphone; Voice activity detection is performed based on the sound signals collected by the microphone array; When the sound signal is determined to be a speech activity signal, the noise type and noise intensity are determined based on the noise signal collected by the auxiliary microphone and a preset noise feature library. Based on the noise type, the noise intensity, and the preset noise reduction level table, the noise reduction parameters are determined; Based on the noise reduction parameters and the first speech signal collected by the main microphone, the target speech signal is output. The sampling rate of the microphone array is adjusted based on the noise intensity and the clarity of the target speech signal. The auditory nerve is stimulated based on the target speech signal.

10. The control method according to claim 9, characterized in that, The step of outputting a target speech signal based on the noise reduction parameters and the first speech signal acquired by the main microphone includes: Obtain the time difference between the acquisition of the first voice signal by the main microphone; Based on the sound speed parameter, the spacing parameter of the main microphone, and the time difference, the azimuth angle of the target audio segment in the first speech signal is calculated; Beamforming is performed on the first speech signal based on the azimuth angle to generate a second speech signal. The noise frequency band in the second speech signal is denoised based on the denoising parameters, and the target speech signal is output.

Images