Wind noise reduction method for over-ear earbuds, electronic device, and readable storage medium

Abstract

The application discloses an anti-wind noise method of an OWS ear clip type earphone, electronic equipment and a readable storage medium, wherein the anti-wind noise method of the OWS ear clip type earphone can make the voice signal and the wind noise signal have obvious distinguishability, reduce the influence of wind noise on the use process of the earphone, facilitate the user to clearly and accurately hear the content of voice communication in a wind noise scene, make the earphone have higher clarity when communicating and listening to music, and make the user experience better.

Description

Technical Field

[0001] This invention relates to the field of headphone technology, and in particular to an OWS clip-on headphone wind noise reduction method, electronic device, and readable storage medium. Background Technology

[0002] OWS (Open Wearable Stereo) headphones offer a different wearing experience compared to traditional in-ear headphones, especially during exercise, allowing users to perceive external sounds and improving safety. However, OWS headphones also face some technical challenges in their design, with wind noise being a particularly prominent issue.

[0003] Wind noise refers to non-speech wind noise signals captured by the headphone microphone in windy environments, especially noticeable during high-speed movement or outdoor activities. Due to their open-back design, OWS headphones have microphones closer to the external environment, making them more susceptible to wind noise interference. Wind noise during use severely impacts call and music clarity, resulting in a poor user experience. Therefore, it is necessary to research a new technical solution to address these issues. Summary of the Invention

[0004] In view of this, the present invention addresses the deficiencies of the prior art, and its main objective is to provide an anti-wind noise method, electronic device, and readable storage medium for OWS clip-on headphones, which can effectively solve the problems of wind noise interference, poor call and music clarity, and poor user experience of existing OWS headphones.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] A method for reducing wind noise in an OWS clip-on earphone, the OWS clip-on earphone including a feedforward microphone, an auxiliary microphone, and a call microphone, wherein the feedforward microphone is used to directly acquire wind noise signals, the auxiliary microphone is used to collect wind noise signals after physical noise reduction, and the call microphone is used to collect voice signals during a call; the method for reducing wind noise includes the following steps:

[0007] Fourier transforms are performed on the first wind noise signal collected by the feedforward microphone and the second wind noise signal collected by the auxiliary microphone to obtain the frequency domain of the first wind noise signal and the frequency domain of the second wind noise signal.

[0008] Based on the frequency domain of the first wind noise signal and the frequency domain of the second wind noise signal, the first power spectral density of the first wind noise signal and the second power spectral density of the second wind noise signal are determined respectively.

[0009] The signal segment in the first wind noise signal with a first power spectral density greater than a preset first threshold is marked as the first designated wind noise signal, and the signal segment in the second wind noise signal with a second power spectral density less than a preset second threshold is marked as the second designated wind noise signal.

[0010] The first specified wind noise signal and the second specified wind noise signal are analyzed respectively. When the peak value of the first specified wind noise signal and the second specified wind noise signal in the frequency domain is greater than the preset frequency domain threshold, the corresponding peak point is taken as the main frequency component, and the frequency of the first specified wind noise signal and the second specified wind noise signal are obtained respectively.

[0011] The phases of the first and second specified wind noise signals are determined based on the autocorrelation function.

[0012] The voice signal acquired by the microphone is analyzed to determine the frequency and phase of the voice signal;

[0013] The speech signal is subjected to Fourier transform, and the calculated speech signal is divided into multiple signal segments. These segments are then compared with the first specified wind noise signal and the second specified wind noise signal. The signal segments with high similarity to the first specified wind noise signal and the second specified wind noise signal are marked as signal segments to be converted, and the remaining signal segments are marked as unconverted signal segments.

[0014] The target wind noise signal segment is obtained by enhancing the signal segment to be converted. The target wind noise signal segment and the unconverted signal segment are then recombined in their original order, and the recombined speech signal is output.

[0015] As a preferred embodiment, determining the first power spectral density of the first wind noise signal and the second power spectral density of the second wind noise signal based on the frequency domains of the first and second wind noise signals respectively includes:

[0016] Based on the frequency domains of the first wind noise signal and the second wind noise signal, the frame rate and frequency point of the first wind noise signal and the second wind noise signal are determined respectively.

[0017] Based on the frame rate and frequency of the first wind noise signal and the second wind noise signal, the first wind noise signal and the second wind noise signal are respectively divided into multiple overlapping signal segments;

[0018] Select a window function, multiply the window function by each signal segment, and then perform windowing processing on each signal segment;

[0019] Fourier transform is performed on the windowed signal segments to obtain the power spectral density corresponding to each signal segment.

[0020] As a preferred approach, the power spectral density X(k) is calculated using the following formula:

[0021]

[0022] Where x[n] is the nth signal segment after windowing; N is the total number of signal segments; k is the frequency point, which usually ranges from 0 to N-1; j is the imaginary unit, which satisfies j2=-1.

[0023] As a preferred approach, the step of enhancing the signal segment to be converted to obtain the target wind noise signal segment includes:

[0024] Calculate the frequency difference and phase difference between the first specified wind noise signal and the second specified wind noise signal to obtain the reference frequency difference and reference phase difference;

[0025] The signal to be converted is enhanced based on the reference frequency difference and reference phase difference to obtain the target wind noise signal segment to be determined;

[0026] The target wind noise signal segment to be determined is compared with the second specified wind noise signal. If the similarity between the two is less than or equal to the preset similarity threshold, the target wind noise signal segment to be determined is taken as the final target wind noise signal segment. If the similarity between the two is greater than the preset similarity threshold, the target wind noise signal segment to be determined is enhanced again until the enhanced target wind noise signal segment to be determined is compared with the second specified wind noise signal. If the similarity between the two is less than or equal to the preset similarity threshold, the enhanced target wind noise signal segment to be determined is taken as the final target wind noise signal segment.

[0027] As a preferred approach, the calculation formula for the target wind noise signal segment to be determined is as follows;

[0028] F(n)=f(n)·e j2πΔf

[0029] F(n)=f(n)·e j2πΔw

[0030] Where F(n) is the target wind noise signal segment to be determined, f(n) is the signal segment to be converted before enhancement, Δf is the reference frequency difference, Δw is the reference phase difference, and j is the imaginary unit, satisfying j 2 =-1.

[0031] As a preferred option, the shell of the OWS clip-on earphone is provided with an air guide groove, and the auxiliary microphone corresponds to the air guide groove.

[0032] As a preferred option, the feedforward microphone is mounted on the side of the housing of the OWS clip-on headphones away from the ear.

[0033] As a preferred embodiment, the call microphone is mounted on the side of the housing of the OWS clip-on headset closer to the mouth.

[0034] A computer-readable storage medium storing a plurality of instructions adapted for loading by a processor and executing the steps of the above-described OWS clip-on headphone wind noise reduction method.

[0035] An electronic device includes a memory, a processor, and program instructions stored in the memory and executable on the processor, wherein the program instructions, when executed by the processor, implement the wind noise reduction method for the OWS clip-on headphones described above.

[0036] Compared with the prior art, the present invention has obvious advantages and beneficial effects. Specifically, as can be seen from the above technical solution:

[0037] The wind noise reduction method for the OWS clip-on headphones of this application first performs a Fourier transform on the acquired first wind noise signal and second wind noise signal to obtain the frequency domain of the first wind noise signal and the frequency domain of the second wind noise signal; based on the frequency domain of the first wind noise signal and the frequency domain of the second wind noise signal, the first power spectral density of the first wind noise signal and the second power spectral density of the second wind noise signal are determined; then, the autocorrelation function of the first specified wind noise signal and the second specified wind noise signal is determined; and the phase characteristics of the first specified wind noise signal and the second specified wind noise signal are inferred by analyzing the shape and peak position of the autocorrelation function. The method involves performing a Fourier transform on the speech signal and enhancing the segment of the speech signal that is highly similar to the wind noise signal to obtain the target wind noise signal segment. Then, the target wind noise signal segment is reassembled in its original order, and the reassembled speech signal is output. Therefore, this application can make the speech signal and the wind noise signal clearly distinguishable, reduce the impact of wind noise on the use of the headphones, and make it easier for users to hear the content of voice calls more clearly and accurately in windy scenarios. This results in higher clarity for the headphones when making calls and listening to music, and a better user experience.

[0038] To more clearly illustrate the structural features and effects of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. Attached Figure Description

[0039] Figure 1 This is a flowchart of an OWS clip-on headphone wind noise reduction method according to an embodiment of the present invention;

[0040] Figure 2 This is a schematic diagram of the structure of the OWS clip-on earphone in one embodiment of the present invention;

[0041] Figure 3 This is a schematic diagram of the structure of a feedforward microphone mounted on an OWS clip-on earphone in one embodiment of the present invention;

[0042] Figure 4This is a schematic diagram of the structure of an auxiliary microphone mounted on an OWS clip-on earphone in one embodiment of the present invention.

[0043] Figure captions: 101, OWS clip-on headphones; 102, feedforward microphone; 103, air duct; 104, auxiliary microphone. Detailed Implementation

[0044] The present invention will now be further described with reference to the accompanying drawings and specific embodiments.

[0045] like Figures 1-4 As shown in the figure, this embodiment discloses an anti-wind noise method for OWS clip-on headphones, which is mainly used to process the wind noise of the voice signal heard by the user when making a voice call using the headphones.

[0046] The OWS ear clip-on headset 101 includes a feedforward microphone 102, an auxiliary microphone 104, and a call microphone. The feedforward microphone 102 is used to directly acquire wind noise signals, the auxiliary microphone 104 is used to collect wind noise signals that have undergone physical noise reduction, and the call microphone is used to collect voice signals during calls.

[0047] The wind noise reduction method includes the following steps:

[0048] Step S100: Perform Fourier transform on the first wind noise signal collected by the feedforward microphone 102 and the second wind noise signal collected by the auxiliary microphone 104 respectively to obtain the frequency domain of the first wind noise signal and the frequency domain of the second wind noise signal.

[0049] Step S200: Determine the first power spectral density of the first wind noise signal and the second power spectral density of the second wind noise signal based on the frequency domain of the first wind noise signal and the frequency domain of the second wind noise signal, respectively.

[0050] Step S300: Mark the signal segment in the first wind noise signal whose first power spectral density is greater than a preset first threshold as the first designated wind noise signal, and mark the signal segment in the second wind noise signal whose second power spectral density is less than a preset second threshold as the second designated wind noise signal; wherein, the first wind noise signal is a directly acquired wind noise signal, therefore, the intensity and frequency of the first designated wind noise signal are both large, which can form a significant wind noise effect; while the second wind noise signal is a wind noise signal collected by the headphones after physical noise reduction, therefore, the second designated wind noise signal is a wind noise signal that is difficult to eliminate by physical noise reduction, and its frequency is low and its stability is good;

[0051] Step S400: Analyze the first specified wind noise signal and the second specified wind noise signal respectively. When the peak value of the first specified wind noise signal and the second specified wind noise signal in the frequency domain is greater than the preset frequency domain threshold, take the corresponding peak point as the main frequency component and obtain the frequency of the first specified wind noise signal and the second specified wind noise signal respectively.

[0052] Step S500: Determine the phase of the first specified wind noise signal and the second specified wind noise signal based on the autocorrelation function;

[0053] Step S600: Analyze the voice signal acquired by the microphone to determine the frequency and phase of the voice signal;

[0054] Step S700: Perform Fourier transform on the speech signal, and divide the calculated speech signal into multiple signal segments. Then compare them with the first specified wind noise signal and the second specified wind noise signal respectively. Mark the signal segments with high similarity to the first specified wind noise signal and the second specified wind noise signal as signal segments to be converted, and mark the remaining signal segments as unconverted signal segments.

[0055] Step S800: Enhance the signal segment to be converted to obtain the target wind noise signal segment, reassemble the target wind noise signal segment and the unconverted signal segment in the original order, and output the reassembled speech signal.

[0056] In this application, through the processing described above, when a user makes a voice call using headphones in a wind-noise environment (e.g., when the wind is strong outdoors, or when the user is simultaneously cycling, running, or engaging in other activities), the system can identify signal segments in the voice signal that have a high similarity to the frequency and phase characteristics of the wind noise signal based on the frequency and phase characteristics of the wind noise signal acquired by the feedforward microphone and the auxiliary microphone. Then, the corresponding signal segments are enhanced to convert the received voice signal into a call voice signal with a low similarity to the spectral characteristics of the wind noise signal for output. This allows for a clear distinction between the voice signal and the wind noise signal, reducing the impact of wind noise on the use of the headphones. This makes it easier for users to hear the content of voice calls clearly and accurately even in wind-noise environments, resulting in higher clarity for calls and music playback and a better user experience.

[0057] In this embodiment, in step S200, determining the first power spectral density of the first wind noise signal and the second power spectral density of the second wind noise signal based on the frequency domain of the first wind noise signal and the frequency domain of the second wind noise signal respectively includes:

[0058] Step 201: Determine the frame rate and frequency point of the first wind noise signal and the second wind noise signal respectively based on the frequency domain of the first wind noise signal and the frequency domain of the second wind noise signal;

[0059] Step 202: Based on the frame rate and frequency of the first wind noise signal and the second wind noise signal, divide the first wind noise signal and the second wind noise signal into multiple overlapping signal segments respectively;

[0060] Step 203: Select a window function and multiply it by each signal segment to perform windowing processing on each signal segment; among which, Hanning window, Hamming window, and Blackman window can be used for windowing processing.

[0061] Step 204: Perform a Fourier transform on the windowed signal segments to obtain the power spectral density corresponding to each signal segment.

[0062] Furthermore, the power spectral density is the power distribution of a signal at different frequencies, which can be obtained through the discrete Fourier transform. The formula for calculating the power spectral density X(k) is as follows:

[0063]

[0064] Where x[n] is the nth signal segment after windowing; N is the total number of signal segments; k is the frequency point, which usually ranges from 0 to N-1; j is the imaginary unit, which satisfies j2=-1.

[0065] In this embodiment, before step S600, it is first determined whether the user is in a call state. This can be determined by combining the application state of electronic devices such as mobile phones (for example, the voice call state of the mobile phone has changed to the input state). Once determined, the call microphone is activated to obtain the user's voice signal.

[0066] In this embodiment, in step S600, the frequency and phase of the received voice signal can be determined in the following way.

[0067] First, regarding frequency determination: Bluetooth headsets receive digital audio signals from mobile devices (such as mobile phones, smartwatches, etc.). The audio source specifies the sampling frequency when encoding the audio data. After receiving these signals, the Bluetooth headset can determine the basic frequency parameters of the audio (Bluetooth headsets themselves have a certain frequency response range specification in their design. For example, the frequency response range of common Bluetooth headsets may be 20Hz-20kHz. In most cases, Bluetooth headsets output sound according to this designed frequency range, and when playing audio sources normally, they will try to accurately reproduce the sound within this frequency range). Simultaneously, the DSP chip inside the Bluetooth headset processes the received digital audio signal. It can adjust and optimize the audio frequency according to preset algorithms and parameters (for example, to enhance the bass effect, the DSP chip may amplify the low-frequency part of the audio signal). Therefore, the digital audio signal is converted into an analog audio signal output by a DAC. The DAC can more accurately reproduce the various frequency components in the digital audio signal, thus directly determining the frequency of the voice signal. Moreover, the output audio frequency is usually consistent with the input digital audio signal; that is, the frequency of the received voice signal is usually consistent with the frequency of the digital audio signal transmitted by the mobile device.

[0068] Secondly, regarding how the phase is determined: During the encoding and transmission of the audio source, the phase information of the received voice signal is included in the digital audio data. Therefore, when Bluetooth headphones receive this data, they also receive the phase information. For example, the Bluetooth audio transmission protocol strives to ensure the phase integrity of the audio signal so that Bluetooth headphones can accurately reproduce the phase of the audio; that is, the phase of the received voice signal is usually consistent with the phase of the digital audio signal transmitted by the mobile device.

[0069] In this embodiment, step S800, which involves enhancing the signal segment to be converted to obtain the target wind noise signal segment, includes:

[0070] Step S801: Calculate the frequency difference and phase difference between the first specified wind noise signal and the second specified wind noise signal to obtain the reference frequency difference and reference phase difference;

[0071] Step S802: Enhance the signal of the segment to be converted based on the reference frequency difference and reference phase difference to obtain the target wind noise signal segment to be determined;

[0072] Step S803: Perform a similarity analysis between the target wind noise signal segment to be determined and the second specified wind noise signal; if the similarity between the two is less than or equal to a preset similarity threshold, then the target wind noise signal segment to be determined is taken as the final target wind noise signal segment; if the similarity between the two is greater than the preset similarity threshold, then the target wind noise signal segment to be determined is enhanced again until the enhanced target wind noise signal segment to be determined is compared with the second specified wind noise signal and the similarity between the two is less than or equal to the preset similarity threshold, then the enhanced target wind noise signal segment to be determined is taken as the final target wind noise signal segment.

[0073] Therefore, after processing steps S801 to S803, the target wind noise signal segment with enhanced signal can be obtained.

[0074] It is worth noting that in step S803, a similarity analysis is performed between the target wind noise signal segment to be determined and the second specified wind noise signal. The main reason for this is that the first specified wind noise signal has a relatively high intensity and frequency, which can create a significant wind noise impact; the second specified wind noise signal is a wind noise signal that is difficult to eliminate through physical noise reduction, and its frequency is lower and its stability is better. In the application scenario of this application, users need to conduct voice calls in windy environments. Although the first specified wind noise signal has a relatively high intensity and frequency, it can be largely filtered out through physical noise reduction by the air duct. Even if it is not completely filtered out, the remaining first specified wind noise signal is unlikely to form a continuous wind noise signal segment, and its impact on the user's ability to hear the content is minimal. However, the second specified wind noise signal has good stability and is difficult to filter out through physical noise reduction. Conventional noise reduction methods use algorithms to filter or suppress it, which can easily lead to the loss of some details in the voice content, affecting the user's ability to obtain complete voice information. Therefore, this application enhances a target wind noise signal that is similar to the second specified wind noise signal, thereby making the target wind noise signal significantly different from the second specified wind noise signal, so that the user can clearly hear the voice information in the target wind noise signal segment while retaining the second specified wind noise signal.

[0075] Furthermore, the calculation formula for the target wind noise signal segment to be determined is as follows;

[0076] F(n)=f(n)·e j2πΔf

[0077] F(n)=f(n)·e j2πΔw

[0078] Where F(n) is the target wind noise signal segment to be determined, f(n) is the signal segment to be converted before enhancement, Δf is the reference frequency difference, Δw is the reference phase difference, and j is the imaginary unit, satisfying j 2 =-1.

[0079] In this application, after the signal enhancement processing in step S800, the target wind noise signal segment is significantly different from the first specified wind noise signal and the second specified wind noise signal. After the target wind noise signal segment is reconstructed into a voice signal, it can be ensured that the answering voice signal output by the Bluetooth headset can be distinguished from the wind noise when the user makes a call, so that the answering voice signal is significantly different from the wind noise signal, making it easier for the user to hear the content of the voice call more accurately in wind noise scenarios.

[0080] Furthermore, this application primarily addresses wind noise reduction for the voice signal heard by the user during a call using the headset. Similarly, due to wind noise, the voice signal received by the microphone when the user speaks is also somewhat blurred. Therefore, by applying the wind noise reduction method of this application to the voice signal, the voice signal sent by the Bluetooth headset (i.e., the OWS clip-on headset of this application) to the mobile device also has reduced wind noise, thereby ensuring that the user on the other end of the call can also hear the content of the voice call more accurately.

[0081] Please see Figures 2-4 The feedforward microphone 102 is installed on the side of the housing of the OWS ear clip-on earphone 101 away from the ear. A portion of the structure of the feedforward microphone 102 is exposed outside the earphone housing, so it can directly acquire external wind noise signals, and the intensity and frequency of the wind noise signals are relatively large.

[0082] Please see Figures 2-4 The OWS clip-on earphone has an air duct on its shell, located on the curved surface of the earphone shell side, and an auxiliary microphone corresponding to the air duct. The auxiliary microphone can be installed inside the air duct or inside the earphone shell and corresponding to the air duct (see [link]). Figure 4 Specifically, the air duct is located on the curved wall of the earphone shell, which allows the air to quickly disperse along the curved wall around the air duct, reducing the energy of the air blowing into the air duct and thus reducing wind noise, thereby achieving the effect of wind noise reduction. At the same time, the design of the air duct can reduce wind noise entering the earphone shell, thereby reducing the wind noise signal acquired by the auxiliary microphone. Therefore, the wind noise signal acquired by the auxiliary microphone is a second wind noise signal after physical noise reduction processing. Thus, the second specified wind noise signal is a wind noise signal that is difficult to eliminate by physical noise reduction, and its frequency is low and its stability is good.

[0083] Furthermore, the call microphone (not shown in the figure) is installed on the side of the housing of the OWS ear clip headset 101 near the mouth, so as to facilitate the pickup of the user's voice signal.

[0084] The key design feature of the wind noise reduction method for OWS clip-on headphones provided by this invention is as follows: First, the acquired first wind noise signal and second wind noise signal are subjected to Fourier transform to obtain the frequency domains of the first and second wind noise signals. Based on the frequency domains of the first and second wind noise signals, the first power spectral density of the first wind noise signal and the second power spectral density of the second wind noise signal are determined. Then, the autocorrelation functions of the first and second specified wind noise signals are determined. By analyzing the shape and peak position of the autocorrelation functions, the autocorrelation functions of the first and second specified wind noise signals are inferred. The second step involves determining the phase characteristics of the wind noise signal; then performing a Fourier transform on the speech signal, and enhancing the segment of the speech signal that is highly similar to the wind noise signal to obtain the target wind noise signal segment. Subsequently, the target wind noise signal segment is reassembled in its original order, and the reassembled speech signal is output. Therefore, in this application, the speech signal and the wind noise signal can be clearly distinguished, reducing the impact of wind noise on the use of the headphones. This allows users to hear the content of voice calls more clearly and accurately even in windy environments, resulting in higher clarity for the headphones during calls and music playback, and a better user experience.

[0085] It is worth noting that both the first and second wind noise signals are continuous noise signals with corresponding autocorrelation functions, and this is existing technology, so it will not be elaborated here.

[0086] This application also provides a computer-readable storage medium storing a plurality of instructions adapted for loading by a processor and executing the steps of the above-described OWS clip-on headphone wind noise reduction method.

[0087] This application also provides an electronic device, including a memory, a processor, and program instructions stored in the memory and executable on the processor, wherein the program instructions, when executed by the processor, implement the wind noise reduction method of the OWS clip-on headphones described above.

[0088] In one embodiment, the processor may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip. The processor is typically used to control the overall operation of an electronic device. In this embodiment, the processor is used to run program code stored in a readable storage medium or to process data.

[0089] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a readable storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0090] The above description is merely a preferred embodiment of the present invention and does not constitute any limitation on the technical scope of the present invention. Therefore, any minor modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention shall still fall within the scope of the technical solution of the present invention.

Claims

1. A method for wind noise reduction in an OWS clip-on earphone, the OWS clip-on earphone comprising a feedforward microphone, an auxiliary microphone, and a call microphone, wherein the feedforward microphone is used to directly acquire wind noise signals, the auxiliary microphone is used to collect wind noise signals after physical noise reduction, and the call microphone is used to collect voice signals during a call; characterized in that: The wind noise reduction method includes the following steps: Fourier transforms are performed on the first wind noise signal collected by the feedforward microphone and the second wind noise signal collected by the auxiliary microphone to obtain the frequency domain of the first wind noise signal and the frequency domain of the second wind noise signal. Based on the frequency domain of the first wind noise signal and the frequency domain of the second wind noise signal, the first power spectral density of the first wind noise signal and the second power spectral density of the second wind noise signal are determined respectively. The signal segment in the first wind noise signal with a first power spectral density greater than a preset first threshold is marked as the first designated wind noise signal, and the signal segment in the second wind noise signal with a second power spectral density less than a preset second threshold is marked as the second designated wind noise signal. The first specified wind noise signal and the second specified wind noise signal are analyzed respectively. When the peak value of the first specified wind noise signal and the second specified wind noise signal in the frequency domain is greater than the preset frequency domain threshold, the corresponding peak point is taken as the main frequency component, and the frequency of the first specified wind noise signal and the second specified wind noise signal are obtained respectively. The phases of the first and second specified wind noise signals are determined based on the autocorrelation function. The voice signal acquired by the microphone is analyzed to determine the frequency and phase of the voice signal; The speech signal is subjected to Fourier transform, and the calculated speech signal is divided into multiple signal segments. These segments are then compared with the first specified wind noise signal and the second specified wind noise signal. The signal segments with high similarity to the first specified wind noise signal and the second specified wind noise signal are marked as signal segments to be converted, and the remaining signal segments are marked as unconverted signal segments. The target wind noise signal segment is obtained by enhancing the signal segment to be converted. The target wind noise signal segment and the unconverted signal segment are then recombined in their original order, and the recombined speech signal is output.

2. The wind noise reduction method for OWS clip-on headphones according to claim 1, characterized in that: The step of determining the first power spectral density of the first wind noise signal and the second power spectral density of the second wind noise signal based on the frequency domains of the first and second wind noise signals includes: Based on the frequency domains of the first wind noise signal and the second wind noise signal, the frame rate and frequency point of the first wind noise signal and the second wind noise signal are determined respectively. Based on the frame rate and frequency of the first wind noise signal and the second wind noise signal, the first wind noise signal and the second wind noise signal are respectively divided into multiple overlapping signal segments; Select a window function, multiply the window function by each signal segment, and then perform windowing processing on each signal segment; Fourier transform is performed on the windowed signal segments to obtain the power spectral density corresponding to each signal segment.

3. The wind noise reduction method for OWS clip-on headphones according to claim 2, characterized in that: The formula for calculating the power spectral density X(k) is as follows: Where x[n] is the nth signal segment after windowing; N is the total number of signal segments; k is the frequency point, ranging from 0 to N-1; j is the imaginary unit, satisfying j²=-1.

4. The wind noise reduction method for OWS clip-on headphones according to claim 1, characterized in that: The process of enhancing the signal segment to be converted to obtain the target wind noise signal segment includes: Calculate the frequency difference and phase difference between the first specified wind noise signal and the second specified wind noise signal to obtain the reference frequency difference and reference phase difference; The signal to be converted is enhanced based on the reference frequency difference and reference phase difference to obtain the target wind noise signal segment to be determined; The target wind noise signal segment to be determined is compared with the second specified wind noise signal. If the similarity between the two is less than or equal to the preset similarity threshold, the target wind noise signal segment to be determined is taken as the final target wind noise signal segment. If the similarity between the two is greater than the preset similarity threshold, the target wind noise signal segment to be determined is enhanced again until the enhanced target wind noise signal segment to be determined is compared with the second specified wind noise signal. If the similarity between the two is less than or equal to the preset similarity threshold, the enhanced target wind noise signal segment to be determined is taken as the final target wind noise signal segment.

5. The wind noise reduction method for OWS clip-on headphones according to claim 1, characterized in that: The OWS clip-on earphone has an air duct on its shell, and the auxiliary microphone corresponds to the air duct.

6. The wind noise reduction method for OWS clip-on headphones according to claim 1, characterized in that: The feedforward microphone is mounted on the side of the housing of the OWS clip-on headphones away from the ear.

7. The wind noise reduction method for OWS clip-on headphones according to claim 1, characterized in that: The call microphone is installed on the side of the housing of the OWS clip-on earphone near the mouth.

8. A computer-readable storage medium, characterized in that, The system stores multiple instructions adapted for loading by a processor and executing the steps of the wind noise reduction method for the OWS clip-on headphones according to any one of claims 1 to 7.

9. An electronic device, characterized in that, It includes a memory, a processor, and program instructions stored in the memory and executable on the processor, wherein when the program instructions are executed by the processor, they implement the wind noise reduction method of the OWS clip-on headphones as described in any one of claims 1 to 7.

Images