Apparatus for robust enhancement of noise-reduction effect for earphone calls and method therefor

Abstract

Disclosed in the present disclosure are an apparatus for robust enhancement of noise-reduction effect for earphone calls, and a method therefor. The apparatus is configured to comprise an earphone body, a first microphone provided on the earphone body, and a control circuit provided in the earphone body, and is configured to comprise a detachable extension rod member, a second microphone being provided at the end of the rod member. The control circuit is configured to use two operating states: a first operating state is the state in which the extension rod member is not connected, and single-microphone speech noise reduction is carried out by means of the first microphone; and a second operating state is the state in which the extension rod member is connected, and double-microphone speech noise reduction is carried out by means of the first microphone and the second microphone. The apparatus for robust enhancement of noise-reduction effect for earphone calls and the method therefor in the present disclosure use the detachable earphone rod member arranged on the earphone to facilitate switching of use modes of the earphone in different background noise environments, thus balancing the wearing comfort and noise-reduction adaptability of earphones.

Description

A device and method for robustly enhancing the noise reduction effect of earphone calls

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This disclosure claims priority to Chinese patent application number 202410135206.9 filed with the China Patent Office on January 31, 2024, entitled “A device and method for robustly enhancing the noise reduction effect of earphone calls,” the entire contents of which are incorporated by reference into this disclosure. Technical Field

[0003] The present disclosure relates to an earphone device, and more particularly to an earphone device that utilizes robust characteristics to enhance the noise reduction effect of earphone calls and an implementation method thereof. Background Art

[0004] In a typical call scenario with existing headsets, the user is in a seated environment, and the surrounding background noise and the voice to be recognized are of the same audio type—that is, they are all human voices. This makes it very difficult to distinguish the target voice from interfering voices and reduce the noise. However, under different levels of ambient noise, the noise reduction algorithms of existing headsets are the same, resulting in insufficient voice noise reduction. The output voice is easily affected by the voices of other seated users, and the headsets can easily cause fatigue.

[0005] Currently, there are two types of devices for reducing noise during headphone calls: one is that the microphone is located in the headphone body, that is, a headphone without an extension stem; the other is a headphone with an extension stem, which places the microphone closer to the human mouth.

[0006] Because headphones with extension rods can place the microphone closer to the human mouth, they can better pick up the wearer's voice, thereby achieving better call noise reduction effects. However, although headphones with extension rods can pick up higher-quality wearer voice signals, they will still pick up interference noise from other operators in the background, especially in complex acoustic environments, which often seriously affects call quality. The conventional approach to solving this problem in existing technologies is to add an auxiliary microphone, which is located away from the target person's mouth. For example, some common call headsets use dual-microphone call noise reduction with extension rods. The audio picked up by the auxiliary microphone is used as background noise to eliminate the background noise in the audio picked up by the main microphone. The wearer's audio can be effectively identified because its pattern can be determined by the time delay between the auxiliary microphone and the main microphone.

[0007] The above-mentioned dual-microphone device with an extension rod can distinguish whether the sound comes from the wearer or the interferer by the difference in amplitude between the wearer's and the interferer's sounds reaching the two microphones. As shown in Figure 1, the distance between the first microphone 101 on the wearer's earphone and the wearer's mouth is much greater than the distance between the second microphone 102 on the extension rod and the wearer's mouth 103. Therefore, the amplitude of the voice received by the first microphone is much greater than that of the first microphone. The distance between the interferer's mouth 104 and the first microphone 101 and the second microphone 102 is close, and the difference in their audio amplitudes is not large. Therefore, the difference between the two can be identified, and then through the signal processing algorithm, the high-quality wearer's voice can be separated.

[0008] However, since the microphone of headphones without extension stems is located farther away from the wearer's mouth than that of headphones with extension stems, the quality of the wearer's voice signal obtained will be poor, resulting in the inability of headphones without extension stems to process high-quality voice data through algorithms in a noisy environment.

[0009] For headphones with an extension stem, higher quality sound from the wearer can usually be obtained, and the algorithm can then process high-quality sound in a noisy environment; but the extension stem also brings defects: on the one hand, the extension stem will reduce the wearer's comfort; on the other hand, the extension stem also has a great negative impact on the aesthetics of the headphones and reduces the convenience of the wearer's operation.

[0010] Therefore, the existing technology still needs to be improved and developed.

[0011] Application Contents

[0012] The purpose of the present disclosure is to provide a device and method for robustly enhancing the noise reduction effect of earphone calls, and to provide an earphone noise reduction solution that is convenient for different noise environments in view of the poor voice noise reduction and wearing comfort of the existing earphone technology.

[0013] The technical solutions disclosed in this disclosure are as follows:

[0014] A device for robustly enhancing the noise reduction effect of headset calls, which is configured to include an headset body, a first microphone arranged on the headset body, and a control circuit arranged in the headset body; wherein the device is configured to include a detachable extension rod, with a second microphone arranged at the end of the rod; and the control circuit is configured to adopt two working states: the first working state is a state in which the extension rod is not connected, and single-microphone voice noise reduction is performed through the first microphone; the second working state is a state in which the extension rod is connected, and dual-microphone voice noise reduction is performed through the first microphone and the second microphone.

[0015] The device for robustly enhancing the noise reduction effect of earphone calls is characterized in that the rod is connected to the earphone body by adsorption, a matching joint is correspondingly provided on the end of the rod, and several contact points are provided on the connection end surface.

[0016] A method for implementing the device for robustly enhancing the noise reduction effect of earphone calls includes the following steps: setting a judgment logic for whether the rod is connected in the control circuit, and adopting a noise reduction processing algorithm corresponding to different working states.

[0017] The device implementation method for robustly enhancing the noise reduction effect of earphone calls, wherein, in the first working state, the control circuit adopts a single-microphone noise reduction algorithm to achieve audio noise reduction.

[0018] The device implementation method for robustly enhancing the noise reduction effect of earphone calls, wherein the single-microphone noise reduction algorithm adopts spectral subtraction and includes the following steps:

[0019] S1, perform short-time Fourier transform on the microphone signal to obtain the frequency domain speech spectrum spec;

[0020] S2, taking the amplitude spectrum abs from the frequency domain speech spectrum spec to obtain the original amplitude spectrum mag;

[0021] S3. Get the predicted noise amplitude spectrum est_noise_mag;

[0022] S4, subtract the noise amplitude spectrum from the original amplitude spectrum mag to obtain the noise-reduced amplitude spectrum est_mag;

[0023] S5. Calculate the phase angle of the frequency domain speech spectrum spec to obtain the original phase ori_angle;

[0024] S6, the noise reduction amplitude spectrum est_mag and the original phase ori_angle obtained in the above steps are synthesized and denoised

[0025] Phonetic spectrum;

[0026] S7. Perform a short-time inverse Fourier transform on the noise-reduced speech spectrum obtained above to obtain a noise-reduced speech output.

[0027] The device implementation method for robustly enhancing the noise reduction effect of earphone calls, wherein the step S3 also includes: tracking the minimum value of the original amplitude spectrum mag frequency-by-frequency point, treating the minimum value as the ambient noise floor, and thereby obtaining the predicted noise amplitude spectrum est_noise_mag.

[0028] The method for implementing the device for robustly enhancing the call noise reduction effect of headphones, wherein the step S4 further includes: constraining the amplitude spectrum est_mag after noise reduction, and setting the frequency points where est_mag < 0 to 0.

[0029] The method for implementing the device for robustly enhancing the call noise reduction effect of headphones, wherein in the second working state, the control circuit uses a dual-microphone noise reduction algorithm to achieve audio noise reduction.

[0030] The method for implementing the device for robustly enhancing the call noise reduction effect of headphones, wherein the dual-microphone noise reduction algorithm uses a power spectrum difference algorithm and includes the following steps:

[0031] T1. Perform short-time Fourier transform on the audio data obtained by the first microphone and the second microphone respectively to obtain the first frequency-domain speech spectrum spec1 and the second frequency-domain speech spectrum spec2;

[0032] T2. Perform power spectrum abs processing on the first frequency-domain speech spectrum spec1 and the second frequency-domain speech spectrum spec2 respectively to obtain the first power spectrum pow1 and the second power spectrum pow2;

[0033] T3. Obtain the ratio ratio by the ratio formula ratio = pow2 / pow1, and normalize the ratio ratio to a 0 / 1 mask mask according to the threshold threshold;

[0034] T4. Multiply the second frequency-domain speech spectrum spec2 obtained in the step T1 and the mask mask obtained in the previous step to obtain the noise-reduced speech spectrum;

[0035] T5. Perform short-time inverse Fourier transform on the noise-reduced speech spectrum obtained in the previous step to obtain the noise-reduced speech and output it.

[0036] The method for implementing the device for robustly enhancing the call noise reduction effect of headphones, wherein the normalization process in the step T3 further includes: normalizing to 1 when the ratio ratio >= threshold, and normalizing to 0 when the ratio ratio < threshold, where the threshold threshold is obtained by collecting the noise ratio ratio_n of the interfering person's speech obtained through the previous steps and the target ratio ratio_s of the target person's speech obtained through the previous steps, and satisfies the threshold threshold <= ratio_s and threshod >= ratio_n.

[0037] The present disclosure provides a device and method for robustly enhancing the noise reduction effect of earphone calls. Due to the use of a detachable earphone rod provided on the earphone, it is possible to switch between different working states of single microphone and dual microphones, thereby facilitating the switching of the use mode of the call earphone in different background noise environments, and taking into account the wearing comfort and noise reduction adaptability of the earphone. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] FIG1 is a schematic diagram of a usage environment of a telephone headset in the prior art.

[0039] FIG2 is a schematic diagram showing the working state selection of a preferred embodiment of the apparatus and method for robustly enhancing the noise reduction effect of earphone calls disclosed in the present invention.

[0040] FIG3 is a schematic diagram of the appearance of a preferred embodiment of the device for robustly enhancing the noise reduction effect of earphone calls as disclosed in the present invention.

[0041] FIG4 is a schematic diagram of a single-microphone noise reduction processing flow in a first working state of a preferred embodiment of the apparatus and method for robustly enhancing the noise reduction effect of earphone calls disclosed in the present invention.

[0042] FIG5 is a schematic diagram of a dual-microphone noise reduction processing flow in the second working state of a preferred embodiment of the apparatus and method for robustly enhancing the noise reduction effect of earphone calls disclosed in the present invention. DETAILED DESCRIPTION

[0043] The preferred embodiments of the present disclosure are described in detail below.

[0044] The present disclosure provides a preferred embodiment of a device and method for robustly enhancing the noise reduction effect of earphone calls, which is mainly provided on a telephone headset. A telephone headset is generally worn on the user's head, and the two ears use earplugs or earmuffs to achieve monitoring and playback functions, and have at least one microphone configured to pick up sound. Operator seats are generally multiple seats set up in a space, and each operator accepts consultations and answers external calls separately. In the improved technical embodiment scheme of the present disclosure, as shown in Figure 3, a wearable module body 200 can be provided, which is hung on one or both ears in an ear-hanging manner and has an extension rod 300.

[0045] A first microphone 201 is provided on the headset body 200. The headset body can have various shapes, but generally has a side facing outward in the area between the ear and the mouth, and a microphone opening is provided on the side, corresponding to which is provided a receiving slot for the first microphone 201, and corresponding connecting wires, etc., to connect to the control circuit inside the body.

[0046] A second microphone 202 is disposed at the end of the rod 300. The second microphone 202 is detachably mounted on the earphone body 200, for example, by a magnetic connection structure, but this is not limited to the above structure and may also be connected in other ways, such as screwing or snap-on connections. There are many different connection structures. As shown in FIG3 , a power contact 203 configured to electrically connect to the rod 300 is disposed on the lower side of the earphone body 200. This contact may be disposed in a recess and corresponding to the rod power contact 301 at the end of the rod 300 opposite the second microphone. An in-ear earplug protrusion 204 for insertion into the ear canal is disposed on the other side of the earphone body 200. Of course, in other embodiments, a circumaural structure may also be employed.

[0047] In low-noise environments, such as during off-peak hours and when fewer calls are received, the disclosed apparatus and method embodiments can utilize only the first microphone 201 on the headset body to achieve sound pickup, i.e., the first operating state. In high-noise environments, such as during peak call times, the rod-shaped second microphone 202 can be directly attached to the headset body and connected via pre-installed connectors, thereby achieving a dual-microphone operating mode. Based on the volume picked up by the second microphone (sound amplitude comparison), the operator's voice spectrum is subjected to noise reduction. The spectrum and amplitude of the second microphone are compared with those of the first microphone to determine whether the difference in the operator's voice amplitude meets the requirements. If so, the voice is amplified; otherwise, it is amplified, thereby achieving dual-microphone noise reduction, i.e., the second operating state.

[0048] In the headset device of the preferred embodiment of the device and method for robustly enhancing the noise reduction effect of headset calls disclosed in the present invention, an extension rod can be connected when making calls in a noisy environment to form a dual-microphone structure and utilize a dual-microphone noise reduction algorithm to improve call quality, which is better than the call quality of headsets without an extension rod. When making calls in a non-noisy environment, the extension rod can be disconnected, and the first microphone on the headset body can be used to perform single-microphone call noise reduction processing, taking into account both comfort and aesthetics. In non-call scenarios, such as when listening to music, the extension rod can be removed, taking into account both comfort and aesthetics. When using the extension rod, the first microphone on the headset body is reused to form a dual-microphone structure, which can apply more effective dual-microphone call noise reduction, reduce structural complexity, and reduce hardware costs.

[0049] The implementation methods of the detachable extension rod described in the present disclosure include but are not limited to the following: the detachable extension rod can be detached by means of snaps, magnets, etc., and power contacts and signal contact points are set at the contact position between the extension rod and the earphone body, so that the circuit can determine whether the extension rod is currently connected, and then switch to different noise reduction algorithms in the control circuit.

[0050] As shown in Figure 2, in a preferred embodiment of the device and method for robustly enhancing the noise reduction effect of earphone calls described in the present disclosure, a call noise reduction algorithm is selected according to the connection status of the extension rod during a call: if the extension rod is not connected, only the first microphone data is obtained, and then the single-microphone noise reduction algorithm is called to output the noise-reduced voice; if the extension rod is connected, the audio data picked up by the first microphone and the second microphone are obtained at the same time, and then the dual-microphone noise reduction algorithm is called to output the noise-reduced voice.

[0051] During a call, in a preferred embodiment of the device and method for robustly enhancing the noise reduction effect of earphone calls disclosed in the present invention, the dual-microphone noise reduction algorithm and the single-microphone noise reduction algorithm are respectively called for processing when the extension rod is connected and when it is not connected. According to the second microphone placed on the extension rod close to the wearer's mouth and the first microphone set on the earphone body away from the wearer's mouth, the wearer's voice is identified and judged by the audio recognition amplitude and phase difference between the two microphones, thereby facilitating the user's wearing comfort and noise reduction accuracy.

[0052] In the apparatus and method for robustly enhancing headset call noise reduction in a preferred embodiment of the present disclosure, when the rod and second microphone are removed, only the first microphone on the headset body is used for noise reduction processing. This employs a single-microphone noise reduction algorithm, which can be implemented using algorithms such as spectral subtraction, Wiener filtering, and deep learning. Spectral subtraction offers the advantage of simple and fast processing, and the corresponding implementation method is described below using spectral subtraction as an example.

[0053] The processing steps of spectral subtraction are shown in Figure 4, which mainly include the following steps:

[0054] S1, perform short-time Fourier transform on the microphone signal to obtain the frequency domain speech spectrum spec;

[0055] S2. Take the amplitude spectrum abs (take the absolute value) of the frequency domain speech spectrum spec to obtain the original amplitude spectrum mag;

[0056] S3. Track the minimum value of the original amplitude spectrum mag frequency-by-frequency. Since the minimum value is tracked, it can be regarded as the ambient noise floor (because the energy increases when there is speech), thereby obtaining the predicted noise amplitude spectrum est_noise_mag;

[0057] S4. Subtract the noise amplitude spectrum (mag est_noise_mag) from the original amplitude spectrum (mag) to obtain the noise-reduced amplitude spectrum (est_mag). Typically, constraints are applied to the noise-reduced amplitude spectrum (est_mag). For example, frequencies where est_mag < 0 are set to 0, using the formula est_mag = max(est_mag, 0). This is because, according to physical meaning, amplitude cannot be negative.

[0058] S5. In addition, the phase angle is processed on the frequency domain speech spectrum spec to obtain the original phase ori_angle;

[0059] S6. Synthesize the denoised speech spectrum using the denoised amplitude spectrum est_mag and the original phase ori_angle obtained in the above steps;

[0060] Optionally, this merging process means that the audio signal acquired by the microphone contains two parameters, amplitude and phase. In the processing process of the preferred embodiment of the present disclosure, only the amplitude is processed, while its original phase is retained; in the noise reduction and restoration process, the phase corresponding to the noise-reduced amplitude signal is merged and restored to obtain the noise-reduced speech spectrum.

[0061] S7. Perform a short-time inverse Fourier transform on the noise-reduced speech spectrum obtained above to obtain a noise-reduced speech output, which is the speech audio obtained by the first microphone.

[0062] In a preferred embodiment of the device and method for robustly enhancing the noise reduction effect of headphone calls disclosed herein, when the ambient noise is relatively noisy, for example, when there are many agents sitting nearby conducting voice consultations and exchanges, the single-microphone algorithm has no way to distinguish other background noises that are also human voices. At this time, the rod can be adsorbed and connected to the system, and the first microphone and the second microphone work at the same time, and the dual-microphone noise reduction algorithm is automatically started in the headphone control circuit.

[0063] Optionally, the implementation of the dual-microphone noise reduction algorithm includes the following main steps.

[0064] Specifically, dual-microphone noise reduction can be implemented using algorithms such as power level difference (PLD) and deep learning. As shown in FIG5 , the implementation steps of a preferred embodiment of the present disclosure are described using PLD as an example:

[0065] T1, performing short-time Fourier transform on the audio data obtained by the first microphone and the second microphone, respectively, to obtain a first frequency domain speech spectrum spec1 and a second frequency domain speech spectrum spec2;

[0066] T2, respectively calculating power spectrum processing for the first frequency domain speech spectrum spec1 and the second frequency domain speech spectrum spec2, and obtaining a first power spectrum pow1 and a second power spectrum pow2;

[0067] T3, the ratio is obtained by the ratio formula ratio = pow2 / pow1, and the ratio is normalized to 0 / 1 mask according to the threshold threshold, that is, the ratio ratio> = threshold > 1, the ratio ratio<threshold > 0, wherein the threshold threshold is obtained by collecting the noise ratio ratio_n obtained by the aforementioned steps of the interfering person's voice and the target ratio ratio_s obtained by collecting the target person's voice, and the threshold threshold <= ratio_s and threshold >= ratio_n must be satisfied; the noise ratio ratio_n and the target ratio ratio_s can be obtained by pre-processing and setting during the product design process.

[0068] Optionally, the noise ratio and target ratio can be initialized after the headset is powered on. This means two pre-processing steps can be set, initiated by the user. For example, during one period, such as 1-15 minutes, the target person is prompted to perform sufficient voice reading recording and processing, and the target ratio is calculated. During another period, such as another 1-15 minutes, the target person is prompted to remain silent while normal ambient noise is recorded and processed to obtain the noise ratio value. These two ratio values ​​are recorded and locked in the system's configuration parameters.

[0069] Optionally, it can also be set that when the environment changes or the background noise changes and the target person changes, the above noise ratio and target ratio can be restarted by the user to perform initialization update so as to obtain noise reduction processing configuration parameters that match the current target person and background noise.

[0070] T4, multiplying the second frequency domain speech spectrum spec2 obtained in step T1 by the mask mask obtained in the above step to obtain a noise-reduced speech spectrum;

[0071] T5. Perform a short-time inverse Fourier transform on the noise-reduced speech spectrum obtained in the above steps to obtain the noise-reduced speech and output it.

[0072] In the preferred embodiment of the device and method for robustly enhancing the noise reduction effect of earphone calls disclosed above, it can be selected whether to add a second microphone to perform more accurate noise reduction processing according to different background noise conditions. The preferred embodiment of the present disclosure achieves a balanced use of earphone comfort and noise reduction processing by switching between two different working states.

[0073] It should be understood that those skilled in the art can make improvements or changes based on the above description, and all such improvements and changes should fall within the scope of protection of the claims appended to this disclosure. Industrial Applicability:

[0074] The embodiments of the present disclosure provide a device and method for robustly enhancing the noise reduction effect of earphone calls. A detachable earphone rod is provided on the earphone to switch between different working states of single microphone and dual microphones, which facilitates the switching of the use mode of the call earphone in different background noise environments and takes into account the wearing comfort and noise reduction adaptability of the earphone.

Claims

1. A device for robustly enhancing the noise reduction effect of earphone calls, comprising an earphone body, a first microphone disposed on the earphone body, and a control circuit disposed within the earphone body; characterized in that: The arrangement includes a detachable extension rod, with a second microphone provided at the end of the rod; and a control circuit, which is arranged to adopt two working states: the first working state is a state in which the extension rod is not connected, and single-microphone voice noise reduction is performed through the first microphone; the second working state is a state in which the extension rod is connected, and dual-microphone voice noise reduction is performed through the first microphone and the second microphone.

2. The device for robustly enhancing the noise reduction effect of earphone calls according to claim 1, characterized in that: The rod is connected to the earphone body by an adsorption method, and a matching joint is correspondingly provided on the end of the rod, and a plurality of contact points are provided on the connection end surface.

3. A method for implementing the device for robustly enhancing the noise reduction effect of earphone calls as claimed in claim 1 or 2, characterized in that: The following steps are involved: The control circuit is provided with a judgment logic for whether the rods are connected, and a noise reduction processing algorithm corresponding to different working states is adopted.

4. The method for implementing the device for robustly enhancing the noise reduction effect of earphone calls according to claim 3, characterized in that: In the first working state, the control circuit adopts a single-microphone noise reduction algorithm to achieve voice noise reduction.

5. The method for implementing the device for robustly enhancing the noise reduction effect of earphone calls according to claim 4, characterized in that: The single-microphone noise reduction algorithm uses spectral subtraction and includes the following steps: S1, perform short-time Fourier transform on the microphone signal to obtain the frequency domain speech spectrum spec; S2, taking the amplitude spectrum abs from the frequency domain speech spectrum spec to obtain the original amplitude spectrum mag; S3. Get the predicted noise amplitude spectrum est_noise_mag; S4, subtract the noise amplitude spectrum from the original amplitude spectrum mag to obtain the noise-reduced amplitude spectrum est_mag; S5. Calculate the phase angle of the frequency domain speech spectrum spec to obtain the original phase ori_angle; S6. Synthesize the denoised speech spectrum using the denoised amplitude spectrum est_mag and the original phase ori_angle obtained in the above steps; S7. Perform a short-time inverse Fourier transform on the noise-reduced speech spectrum obtained above to obtain a noise-reduced speech output.

6. The method for implementing the device for robustly enhancing the noise reduction effect of earphone calls according to claim 5, characterized in that: Step S3 further includes: tracking the minimum value of the original amplitude spectrum mag frequency-by-frequency point, and regarding the minimum value as the ambient noise floor, thereby obtaining a predicted noise amplitude spectrum est_noise_mag.

7. The method for implementing the device for robustly enhancing the noise reduction effect of earphone calls according to claim 6, characterized in that: The step S4 further includes: constraining the amplitude spectrum est_mag after noise reduction, and setting the frequency points where est_mag<0 to 0.

8. The method for implementing the device for robustly enhancing the noise reduction effect of earphone calls according to any one of claims 3 to 7, characterized in that: In the second working state, the control circuit adopts a dual-microphone noise reduction algorithm to achieve audio noise reduction.

9. The method for implementing the device for robustly enhancing the noise reduction effect of earphone calls according to claim 8, characterized in that: The dual-microphone noise reduction algorithm adopts a power spectrum difference algorithm and includes the following steps: T1, performing short-time Fourier transform on the audio data obtained by the first microphone and the second microphone, respectively, to obtain a first frequency domain speech spectrum spec1 and a second frequency domain speech spectrum spec2; T2, respectively calculating the power spectrum pow of the first frequency domain speech spectrum spec1 and the second frequency domain speech spectrum spec2, and obtaining the first power spectrum pow1 and the second power spectrum pow2; T3. The ratio is obtained by the ratio formula ratio = pow2 / pow1, and the ratio is normalized to a 0 / 1 mask according to the threshold threshold; T4, multiplying the second frequency domain speech spectrum spec2 obtained in step T1 by the mask mask obtained in the above step to obtain a noise-reduced speech spectrum; T5. Perform a short-time inverse Fourier transform on the noise-reduced speech spectrum obtained in the above steps to obtain the noise-reduced speech and output it.

10. The method for implementing the device for robustly enhancing the noise reduction effect of earphone calls according to claim 9, characterized in that: The normalization process in the step T3 further includes: when the ratio ratio >= threshold, it is normalized to 1, and when the ratio ratio < threshold, it is normalized to 0, where the threshold threshold is obtained by collecting the noise ratio ratio_n of the interfering person's voice through the foregoing steps and the target ratio ratio_s of the target person's voice through the foregoing steps, and satisfies the threshold threshold <= ratio_s and threshod > ratio_n.

Images