Noise reduction method, device and system for double-frequency-response earphone and electronic equipment thereof

Abstract

The embodiment of the invention provides a noise reduction method, device and system for a dual-frequency-response earphone, and the method comprises the steps: at least obtaining an audio signal of a main device, the audio signal comprising an environment audio signal and a return audio signal; performing frequency band noise reduction processing on the environment audio signal and the return audio signal to obtain a corresponding noise offset signal; and outputting the noise cancellation signal through the MEMS loudspeaker. Active noise reduction processing is carried out on an environment audio signal where a main device is located and a return audio signal sent by the main device to obtain a noise offset signal corresponding to current noise, and the noise offset signal is amplified and sent out through an MEMS loudspeaker, so that the noise offset signal can be amplified according to the current environment or scene where the earphone is located. And active noise reduction is carried out on the specific noise in the specific frequency band, so that the noise reduction effect of the earphone on active noise reduction is improved.

Description

Technical Field

[0001] The present invention relates to the field of smart headphones, and in particular to a dual-frequency headphone noise reduction method, device, system, electronic device and storage medium thereof. Background Art

[0002] The active noise reduction technology of modern headphones uses signal processing algorithms to analyze the sound entering and emitted by the headphones to create reverse sound waves corresponding to the noise, thereby canceling out background noise based on the reverse sound waves.

[0003] Since the current active noise reduction technology of headphones still has limitations in efficiently processing wide-band noise, and it is even more difficult to capture the corresponding noise characteristic frequency bands in dynamic and changing noise environments, the existing headphone noise reduction methods are unable to analyze the characteristic frequency bands of different characteristics of noise in different scenarios, and thus cannot generate reverse sound waves based on the analysis results to perform noise reduction, resulting in poor active noise reduction effect of the headphones. Summary of the invention

[0004] The embodiment of the present invention provides a dual-frequency response headphone noise reduction method to solve the problem that the existing headphone noise reduction method is unable to analyze the characteristic frequency bands of different characteristic noises in different scenarios, thereby generating reverse sound waves according to the analysis results to perform noise reduction, resulting in poor active noise reduction effect of the headphone.

[0005] In a first aspect, an embodiment of the present invention provides a dual-frequency response headphone noise reduction method, the method comprising the following steps: At least obtaining an audio signal of a master device, wherein the audio signal includes an ambient audio signal and a return audio signal; Performing frequency band noise reduction processing on the ambient audio signal and the returned audio signal to obtain corresponding noise cancellation signals; The noise cancellation signal is outputted through a MEMS speaker.

[0006] Optionally, the at least obtaining an audio signal of a master device includes: The external noise of the main device is received and processed by a feedforward microphone arranged outside the main device to obtain an ambient noise data audio signal of the main device; The sound emitted by the main device is received and processed by a feedback microphone arranged inside the main device to obtain a return audio signal of the main device.

[0007] Optionally, performing frequency band noise reduction processing on the ambient audio signal and the returned audio signal to obtain a corresponding noise cancellation signal includes: By using a preset adaptive filtering algorithm, the reverse waveforms of the ambient audio signal and the returned audio signal in a special frequency band are extracted to obtain a corresponding anti-noise waveform digital signal; The anti-noise waveform digital signal is converted into a digital-to-analog converter to obtain a corresponding noise cancellation signal.

[0008] Optionally, the performing frequency band noise reduction processing on the ambient audio signal and the returned audio signal to obtain a corresponding noise cancellation signal, the method further includes: Get the usage scenario of the current main device; Based on the usage scenario, determining a noise frequency band of the specific noise; Based on the noise frequency band, the reverse waveform in the noise frequency band is extracted by a preset adaptive filtering algorithm to obtain a corresponding anti-noise waveform digital signal.

[0009] Optionally, the determining a noise frequency band of the specific noise based on the usage scenario includes: The specific noise in the time domain is converted into the specific noise signal in the frequency domain through the trained frequency domain analysis model, and the frequency spectrum recognition processing is performed on the specific noise signal to determine the noise frequency band of the specific noise.

[0010] Optionally, before converting the specific noise in the time domain into the specific noise signal in the frequency domain through the trained frequency domain analysis model, and performing frequency spectrum recognition processing on the specific noise signal to determine the noise frequency band of the specific noise, the method further includes: Acquire a frequency domain analysis model to be trained and a training noise sample set, wherein the training noise sample set includes original audio, noise audio corresponding to the original audio, and a waveform label corresponding to the noise audio; Based on the original audio, the noise audio corresponding to the original audio, and the waveform label corresponding to the noise audio, the frequency domain analysis model to be trained is iteratively trained, and a trained frequency domain analysis model is obtained after the iterative training is completed.

[0011] In a second aspect, an embodiment of the present invention further provides a dual-frequency headphone noise reduction device, the dual-frequency headphone noise reduction device comprising: A first acquisition module, used to acquire an audio signal of a master device, wherein the audio signal includes an ambient audio signal and a return audio signal; A second acquisition module is used to perform frequency band noise reduction processing on the ambient audio signal and the returned audio signal to obtain a corresponding noise cancellation signal; The output module is used to output the noise cancellation signal through the MEMS speaker.

[0012] In a third aspect, an embodiment of the present invention further provides a dual-frequency headphone noise reduction system, which includes: a dual-frequency headphone noise reduction device, a server, and a smart terminal.

[0013] In a fourth aspect, an embodiment of the present invention provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps in the dual-frequency response headphone noise reduction method provided in an embodiment of the present invention are implemented.

[0014] In a fifth aspect, an embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps in the dual-frequency response headphone noise reduction method provided in the embodiment of the invention are implemented.

[0015] In an embodiment of the present invention, an audio signal of a main device is obtained, and the audio signal includes an ambient audio signal and a return audio signal; frequency band noise reduction processing is performed on the ambient audio signal and the return audio signal to obtain a corresponding noise cancellation signal; and the noise cancellation signal is output through a MEMS speaker. By performing active noise reduction processing on the ambient audio signal in which the main device is located and the return audio signal emitted by the main device, a noise cancellation signal corresponding to the current noise is obtained, and the noise cancellation signal is amplified and emitted through a MEMS speaker, so that active noise reduction can be performed on specific noise in a specific frequency band according to the current environment or scene in which the headset is located, thereby improving the noise reduction effect of the headset in active noise reduction. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 It is an architecture diagram of a dual-frequency headphone noise reduction system provided by an embodiment of the present invention; Figure 2 is a flow chart of a dual-frequency response headphone noise reduction method provided by an embodiment of the present invention; Figure 3 yes Figure 2 Specific flow chart of step S201; Figure 4 yes Figure 2 Specific flow chart of step S202; Figure 5 yes Figure 2Another specific flow chart of step S202; Figure 6 yes Figure 5 The specific flow chart of step S205; Figure 7 yes Figure 6 The specific flow chart before step S2051; Figure 8 is a structural schematic diagram of a dual-frequency headphone noise reduction device provided in an embodiment of the present invention; Fig. 9 It is a structural schematic diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] like Figure 1 As shown, Figure 1 It is an architecture diagram 100 of a dual-frequency headphone noise reduction system provided by an embodiment of the present invention, and the dual-frequency headphone noise reduction system includes: a dual-frequency headphone noise reduction device 800, a server 101, and a smart terminal 102. The dual-frequency headphone noise reduction device 800 also includes a first acquisition module, which can be used to acquire the audio signal of the main device; a second acquisition module, which can be used to perform frequency band noise reduction processing on the ambient audio signal and the return audio signal to obtain a corresponding noise cancellation signal; and an output module, which can be used to output the noise cancellation signal through a MEMS speaker.

[0020] The above-mentioned smart terminal may include but is not limited to smart headphones in any wearable form, including but not limited to wireless and wired headphones.

[0021] Among them, the above-mentioned main device can be a headset for sending and receiving audio signals. It can be understood that the headset can carry an auxiliary MEMS speaker, that is, the main device can carry two sound speakers, one is the native sound speaker, and the other is the auxiliary MEMS speaker.

[0022] The above-mentioned audio signal may include but is not limited to an ambient audio signal and a return audio signal. Specifically, the above-mentioned ambient audio signal may be obtained by analyzing and processing the frequency of the ambient sound to obtain the frequency corresponding to the ambient audio, so that the acquired ambient audio signal may be determined based on the known frequency of the ambient audio. The above-mentioned return audio signal may be an audio signal sent by the sound-emitting speaker of the above-mentioned main device and received by the microphone of the main device after return transmission. Generally speaking, the return audio signal may be captured by a microphone attached to the main device or a microphone arranged near the MEMS speaker.

[0023] The above-mentioned frequency band noise reduction processing can be performed by using a preset adaptive filtering algorithm to perform reverse wave calculation on the above-mentioned return audio signal and the above-mentioned ambient audio signal, thereby reducing the noise of the above-mentioned return audio signal and the ambient audio signal by emitting corresponding reverse waves.

[0024] The above-mentioned reverse wave can be understood as a waveform corresponding to an audio signal that can effectively resist noise from the above-mentioned return audio signal and the above-mentioned ambient audio signal. Generally speaking, after extracting the waveform of the above-mentioned reverse wave, the audio signal corresponding to the waveform can be output through a sound-emitting device through methods such as digital-to-analog conversion, thereby canceling the audio of the above-mentioned return audio signal and the above-mentioned ambient audio signal to achieve a noise reduction effect.

[0025] The above-mentioned preset adaptive filtering algorithm can be used to perform noise reduction processing on the above-mentioned audio signal in real time, so as to adjust the noise resistance capability of the above-mentioned dual-frequency response headphone noise reduction system for different audio signals. Specifically, the above-mentioned preset adaptive filtering algorithm may include but is not limited to the least mean square error algorithm (LMS), the normalized least mean square error algorithm (NLMS), the recursive least squares algorithm (RLS) and the affine projection algorithm (APA) and other algorithms that can calculate the reverse wave of the above-mentioned audio signal.

[0026] The above-mentioned noise cancellation signal may be an audio signal obtained by outputting the above-mentioned reverse wave through the above-mentioned digital-to-analog conversion and other methods, and the noise cancellation signal may be used to resist noise reduction for the above-mentioned return audio signal and the above-mentioned ambient audio signal. It is understandable that the noise cancellation signal may be adjusted according to the strength of the collected return audio signal or the strength of the ambient audio signal, that is, the resistance of the generated noise inefficiency signal to the return audio signal and the resistance to the ambient audio signal may be adjusted according to the strength of the corresponding audio signal or the noise strength collected specifically.

[0027] In a possible embodiment, the above-mentioned noise cancellation signal can be output by a conventional speaker or a MEMS speaker. Generally speaking, the noise cancellation signal corresponding to low-frequency noise, such as wind noise, can be output through a conventional speaker. Another method is to output the noise cancellation signal corresponding to medium and high frequency noise, such as a baby crying, through a MEMS speaker.

[0028] In another possible embodiment, the dual-frequency headphone noise reduction system extracts the audio signal of the main device, performs reverse wave extraction on the noise in the audio signal through a preset adaptive filtering algorithm, and outputs a corresponding anti-noise audio signal according to the waveform of the reverse wave, thereby achieving the effect of anti-noise on the audio signal in the main device.

[0029] like Figure 2 As shown, Figure 2 : is a flow chart of a dual-frequency response headphone noise reduction method provided by an embodiment of the present invention, and the dual-frequency response headphone noise reduction method comprises the steps of: S201. At least obtain an audio signal of a master device.

[0030] In an embodiment of the present invention, the dual-frequency response headphone noise reduction method can be applied to a dual-frequency response headphone noise reduction system. The dual-frequency response headphone noise reduction system has functions such as audio signal processing, audio signal transmission and reception, and audio signal storage, and can be constructed based on a server or a server cluster. The server or server cluster can be an electronic device with audio signal processing capabilities.

[0031] The above-mentioned main device can be a headset for sending and receiving audio signals. It can be understood that the headset can carry an auxiliary MEMS speaker, that is, the main device can carry two sound speakers, one is the native sound speaker, and the other is the auxiliary MEMS speaker. The above-mentioned audio signals may include but are not limited to ambient audio signals and return audio signals.

[0032] The above audio signal may include but is not limited to an ambient audio signal and a return audio signal. Specifically, the above ambient audio signal may be obtained by analyzing and processing the frequency of the ambient sound to obtain the frequency corresponding to the ambient audio, so that the acquired ambient audio signal may be determined based on the known frequency of the ambient audio. The above return audio signal may be an audio signal sent by the sound-emitting speaker of the above main device and received by the microphone of the main device after return.

[0033] In a possible embodiment, the dual-frequency response headphone noise reduction system captures the returned audio signal and the ambient audio signal through a microphone attached to the main device or a microphone arranged near the MEMS speaker.

[0034] S202: Perform frequency band noise reduction processing on the ambient audio signal and the returned audio signal to obtain corresponding noise cancellation signals.

[0035] In an embodiment of the present invention, the above-mentioned ambient audio signal may be obtained by analyzing and processing the frequency of the ambient sound to obtain the frequency corresponding to the ambient audio, so that the acquired ambient audio signal may be determined based on the frequency of the known ambient audio to determine what ambient audio it is. The above-mentioned return audio signal may be an audio signal sent by the sound-emitting speaker of the above-mentioned main device and received by the microphone of the main device after return transmission. Generally speaking, the return audio signal may be captured by a microphone attached to the main device or a microphone arranged near the MEMS speaker.

[0036] The above-mentioned frequency band noise reduction processing can be performed by using a preset adaptive filtering algorithm to perform reverse wave calculation on the above-mentioned return audio signal and the above-mentioned ambient audio signal, thereby reducing the noise of the above-mentioned return audio signal and the ambient audio signal by emitting corresponding reverse waves.

[0037] The above-mentioned noise cancellation signal may be an audio signal obtained by outputting the above-mentioned reverse wave through the above-mentioned digital-to-analog conversion and other methods, and the noise cancellation signal may be used to resist noise reduction for the above-mentioned return audio signal and the above-mentioned ambient audio signal. It is understandable that the noise cancellation signal may be adjusted according to the strength of the collected return audio signal or the strength of the ambient audio signal, that is, the resistance of the generated noise inefficiency signal to the return audio signal and the resistance to the ambient audio signal may be adjusted according to the strength of the corresponding audio signal or the noise strength collected specifically.

[0038] In a possible embodiment, the dual-frequency response headphone noise reduction system performs frequency band noise reduction processing on the collected audio signal through a preset adaptive filtering algorithm to obtain a noise cancellation signal with corresponding frequency band characteristics.

[0039] S203: Output the noise cancellation signal through the MEMS speaker.

[0040] In a possible embodiment, the dual-frequency headphone noise reduction system controls the MEMS speaker to output the noise cancellation signal with the corresponding frequency band characteristics, thereby achieving the effect of anti-noise reduction.

[0041] In an embodiment of the present invention, an audio signal of a main device is obtained, and the audio signal includes an ambient audio signal and a return audio signal; a frequency band noise reduction process is performed on the ambient audio signal and the return audio signal to obtain a corresponding noise cancellation signal; and the noise cancellation signal is output through a MEMS speaker. By performing active noise reduction processing on the ambient audio signal in which the main device is located and the return audio signal emitted by the main device, a noise cancellation signal corresponding to the current noise is obtained, and the noise cancellation signal is amplified and emitted through a MEMS speaker, so that active noise reduction can be performed on specific noise in a specific frequency band according to the current environment or scene in which the headset is located, thereby improving the noise reduction effect of the headset in active noise reduction.

[0042] Optional, such as Figure 3 As shown, in step S201, at least the step of acquiring the audio signal of the master device further includes steps S2011-S2012, wherein: S2011, receiving and processing the external noise of the main device through a feedforward microphone disposed outside the main device to obtain an ambient noise data audio signal of the main device; S2012: Receive and process the sound emitted by the main device through a feedback microphone disposed inside the main device to obtain a return audio signal of the main device.

[0043] In an embodiment of the present invention, the dual-frequency headphone noise reduction system captures the ambient noise outside the headphone, such as the sound of vehicles on the street, the noise of the crowd, etc., through a feedforward microphone arranged outside the main device, and converts these audio signals into ambient noise data audio signals. It also detects the sound generated by the headphone itself, such as the music being played or the sound of the call by the user, through a feedback microphone arranged inside the main device, and then generates a return audio signal.

[0044] It can be understood that the audio signals obtained through the microphone above can be processed for noise reduction by the digital signal processor attached to the headset and the corresponding digital signal algorithm, such as the preset adaptive filtering algorithm, and the processed audio signals can be output through the MEMS speaker.

[0045] By using the above method to reduce the noise of the audio signal, not only can the purity of the earphone sound quality and the level of detail of the sound be significantly improved, but also the clarity of calls and music playback can be maintained in various noisy environments. In this embodiment, the above dual-frequency response earphone noise reduction system uses the advantages of the feedforward microphone and the feedback microphone to accurately perform noise reduction processing on the sound, thereby improving the noise reduction capability of the dual-frequency response earphone for the audio signal.

[0046] Optional, such as Figure 4As shown, in step S202, the step of performing frequency band noise reduction processing on the ambient audio signal and the returned audio signal to obtain the corresponding noise cancellation signal also includes steps S2021-S2022, wherein: S2021. Extract the reverse waveforms of the ambient audio signal and the return audio signal in a special frequency band through a preset adaptive filtering algorithm to obtain a corresponding anti-noise waveform digital signal.

[0047] In an embodiment of the present invention, the preset adaptive filtering algorithm can be used to perform noise reduction processing on the audio signal in real time so as to adjust the noise resistance of the dual-frequency headphone noise reduction system for different audio signals. Specifically, the preset adaptive filtering algorithm can include but is not limited to the least mean square error algorithm (LMS), the normalized least mean square error algorithm (NLMS), the recursive least squares algorithm (RLS) and the affine projection algorithm (APA) and other algorithms that can calculate the reverse wave of the audio signal.

[0048] The above-mentioned special frequency band may be the special waveform features of the above-mentioned environmental audio signal and the returned audio signal on the waveform diagram, such as the high-rise and fall waveform that may exist in the noisy environmental audio signal, and the high-frequency audio waveform that may exist in the baby crying audio signal. Generally speaking, since the waveform diagrams of different audio signals collected in different environments may be different, and have special waveform features in a certain period of time, the audio signals corresponding to the special waveform features can be collected and integrated to obtain a data set of audio data corresponding to the special waveform features, so as to determine the corresponding noise audio signal based on the data set.

[0049] The above-mentioned reverse wave can be understood as a waveform corresponding to an audio signal that can effectively resist noise from the above-mentioned return audio signal and the above-mentioned ambient audio signal. Generally speaking, after extracting the waveform of the above-mentioned reverse wave, the audio signal corresponding to the waveform can be output through a sound-emitting device through methods such as digital-to-analog conversion, thereby canceling the audio of the above-mentioned return audio signal and the above-mentioned ambient audio signal to achieve a noise reduction effect.

[0050] The above-mentioned anti-noise waveform digital signal can be understood as the carrier of the above-mentioned reverse wave in the digital signal field, that is, the above-mentioned reverse wave is displayed in the form of a digital signal.

[0051] In a possible embodiment, the dual-frequency headphone noise reduction system extracts the waveforms of the collected ambient audio signals and the reverse waves of the returned audio signals in the corresponding special frequency bands through the preset adaptive filtering algorithm, and generates corresponding anti-noise waveform digital signals based on the extracted waveforms.

[0052] S2022. Perform digital-to-analog conversion on the anti-noise waveform digital signal through a digital-to-analog converter to obtain a corresponding noise cancellation signal.

[0053] In an embodiment of the present invention, the dual-frequency headphone noise reduction system extracts the reverse wave waveform corresponding to the collected noise audio signal through a preset adaptive filtering algorithm, and converts the anti-noise waveform digital signal corresponding to the reverse wave waveform into a digital-to-analog converter to obtain a corresponding noise cancellation signal, and the noise inefficiency signal can be output through the MEMS speaker.

[0054] Through the above method, sound waves with opposite phases to the ambient noise can be generated in real time, effectively canceling out the noise, thereby greatly improving the purity of the listening environment and the user's auditory experience.

[0055] Optional, such as Figure 5 As shown, in step S202, the ambient audio signal and the returned audio signal are subjected to frequency band noise reduction processing to obtain a corresponding noise cancellation signal, and steps S204-S206 are also included, wherein: S204: Obtain the usage scenario of the current main device.

[0056] In the embodiment of the present invention, the above-mentioned usage scenario may be the environment where the current main device is located or the environment where it may be put into use, for example, it may be an office area, a noisy street area, or a quiet reading area, etc. It is understandable that different usage scenarios have different noise reduction requirements for the main device. For example, the noise reduction in a quiet area is mainly for the sound emitted by the main device itself, that is, the noise reduction of the return audio signal, while in a noisy environment, the noise reduction is for the ambient audio signal. Due to the noise reduction requirements in different environments, the objects for the corresponding reverse wave extraction may also be different.

[0057] Specifically, the current geographic location of the main device can be obtained based on the positioning device of the main device, or features can be extracted based on audio data in the current environment. The audio features in the current environment can be extracted through an audio feature model to determine the current environment of the main device.

[0058] S205: Determine a noise frequency band of the specific noise based on the usage scenario.

[0059] In this embodiment, the noise frequency band of the above-mentioned specific noise can be the frequency range generated by a certain noise source. Since each noise source, such as a car engine, industrial machinery, or a crowd chatting, has its own unique frequency characteristics, these frequencies determine the tone and characteristics of the noise. Therefore, the noise frequency band corresponding to the specific noise can be determined by extracting the features of the audio data in different scenarios.

[0060] S206 , based on the noise frequency band, extract the reverse waveform in the noise frequency band by using a preset adaptive filtering algorithm to obtain a corresponding anti-noise waveform digital signal.

[0061] In a possible embodiment, the dual-frequency response headphone noise reduction system analyzes the noise frequency bands of the user's environment, and then uses a preset adaptive filtering algorithm to perform reverse waveform extraction on these noise frequency bands. For example, if the user is in a high traffic noise environment, the dual-frequency response headphone noise reduction system will identify and process the main low-frequency noise bands, and then adjust the preset adaptive filter to dynamically adjust its parameters according to the corresponding low-frequency noise bands, and generate reverse waveform digital signals that can accurately offset the noise in these frequency bands.

[0062] Optional, such as Figure 6 As shown, in step S205, the step of determining the noise frequency band of the specific noise based on the usage scenario further includes step S2051, wherein: S2051. Convert the specific noise in the time domain into a specific noise signal in the frequency domain through the trained frequency domain analysis model, and perform frequency spectrum recognition processing on the specific noise signal to determine the noise frequency band of the specific noise.

[0063] In an embodiment of the present invention, the above-mentioned trained frequency domain analysis model can be any deep learning model that can convert time domain data in specific noise into a specific noise signal in the frequency domain. Specifically, the above-mentioned trained frequency domain analysis model can decompose the sine waves and cosine waves of different frequency domains of the above-mentioned specific noise signal through mathematical methods such as Fourier transform, thereby completing the conversion from time domain to frequency domain.

[0064] By identifying the frequency spectrum corresponding to a specific noise signal in the frequency domain, the noise frequency band of the specific noise signal can be obtained. For example, when the frequency domain analysis shows that the specific noise is mainly concentrated in the mid-high frequency area, the frequency reverse wave corresponding to the mid-high frequency can be used as the filtering parameter of the filter, so that noise suppression can be performed on the mid-high frequency band.

[0065] Optional, such as Figure 7 As shown, before the step S2051, the specific noise in the time domain is converted into a specific noise signal in the frequency domain through the trained frequency domain analysis model, and the frequency spectrum recognition processing is performed on the specific noise signal to determine the noise frequency band of the specific noise, steps S207-S208 are also included, wherein: S207: Obtain a frequency domain analysis model to be trained and a training noise sample set.

[0066] In an embodiment of the present invention, the above-mentioned frequency domain analysis model to be trained may include but is not limited to BERT (Bidirectional Encoder Representations from Transformers), CLIP (Contrastive Language–Image Pre-training) and MMBT (Multimodal Bitransformer) and other deep learning models that cannot accurately convert time domain data in specific noise into specific noise signals in the frequency domain, that is, when specific noise is input into the above-mentioned frequency domain analysis model to be trained, the corresponding specific noise signal cannot be accurately output, and after frequency spectrum recognition of the specific noise signal, the corresponding noise frequency band cannot be accurately obtained.

[0067] The above-mentioned training noise sample set may include but is not limited to original audio, noise audio corresponding to the original audio, and waveform labels corresponding to the noise audio.

[0068] The original audio may be the audio collected by the feedforward microphone and the feedback microphone, including but not limited to the ambient audio signal and the return audio signal.

[0069] The noise audio corresponding to the original audio may be all noise signals other than the required original audio signal, that is, this embodiment needs to remove or perform noise reduction processing on the noise audio corresponding to the original audio.

[0070] The waveform label corresponding to the above-mentioned noise audio can be a sound wave audio waveform collected according to the corresponding noise audio, that is, each noise audio can display the waveform corresponding to the noise through a waveform through digital-to-analog conversion and other methods.

[0071] S208. Based on the original audio, the noise audio corresponding to the original audio, and the waveform label corresponding to the noise audio, the frequency domain analysis model to be trained is iteratively trained, and a trained frequency domain analysis model is obtained after the iterative training is completed.

[0072] In a possible embodiment, the dual-frequency response headphone noise reduction system iteratively trains the frequency domain analysis model to be trained through the original audio, the noise audio corresponding to the original audio, and the waveform label corresponding to the noise audio. Specifically, the original audio and the noise audio corresponding to the original audio are input into the frequency domain analysis model to be trained, and the output waveform is compared with the waveform label corresponding to the noise audio. According to the comparison result, the parameters of the frequency domain analysis model to be trained are adjusted, and the adjusted frequency domain analysis model to be trained continues to perform the above operation of inputting the original audio and the noise audio corresponding to the original audio into the frequency domain analysis model to be trained, until the similarity between the output waveform and the waveform label corresponding to the noise audio is greater than or equal to the preset similarity threshold, the training of the frequency domain analysis model to be trained is completed, and a trained frequency domain analysis model is obtained.

[0073] like Figure 8 As shown, the embodiment of the present invention further provides a dual-frequency headphone noise reduction device 800, the dual-frequency headphone noise reduction device comprising: A first acquisition module 801 is used to acquire an audio signal of a master device, where the audio signal includes an ambient audio signal and a return audio signal; The second acquisition module 802 is used to perform frequency band noise reduction processing on the ambient audio signal and the returned audio signal to obtain a corresponding noise cancellation signal; The output module 803 is used to output the noise cancellation signal through the MEMS speaker.

[0074] Optionally, the first acquisition module 801 further includes: A first acquisition submodule is used to receive and process the external noise of the main device through a feedforward microphone arranged outside the main device to obtain an ambient noise data audio signal of the main device; The second acquisition submodule is used to receive and process the sound emitted by the main device through a feedback microphone arranged inside the main device to obtain a return audio signal of the main device.

[0075] Optionally, the second acquisition module 802 includes: The third acquisition submodule is used to extract the reverse waveforms of the ambient audio signal and the returned audio signal in a special frequency band through a preset adaptive filtering algorithm to obtain a corresponding anti-noise waveform digital signal; The fourth acquisition submodule is used to perform digital-to-analog conversion on the anti-noise waveform digital signal through a digital-to-analog converter to obtain a corresponding noise cancellation signal.

[0076] Optionally, the above device further includes: The third acquisition module is used to obtain the usage scenario of the current main device; A first determination module, configured to determine a noise frequency band of specific noise based on the usage scenario; The fourth acquisition module is used to extract the reverse waveform in the noise frequency band through a preset adaptive filtering algorithm based on the noise frequency band to obtain a corresponding anti-noise waveform digital signal.

[0077] Optionally, the first determining module includes: The training unit is used to convert the specific noise in the time domain into the specific noise signal in the frequency domain through a trained frequency domain analysis model, and perform frequency spectrum recognition processing on the specific noise signal to determine the noise frequency band of the specific noise.

[0078] Optionally, the above device further includes: An audio acquisition module, used to acquire a frequency domain analysis model to be trained and a training noise sample set, wherein the training noise sample set includes an original audio, a noise audio corresponding to the original audio, and a waveform label corresponding to the noise audio; The training module is used to iteratively train the frequency domain analysis model to be trained based on the original audio, the noise audio corresponding to the original audio, and the waveform label corresponding to the noise audio, and obtain the trained frequency domain analysis model after the iterative training is completed.

[0079] like Fig. 9 As shown, an embodiment of the present invention further provides an electronic device 900, including a processor, and the processor can execute any one of the above dual-frequency response headphone noise reduction methods.

[0080] Specifically, it includes a processor 901 and a memory 902, and a computer program for executing the dual-frequency response headphone noise reduction method stored in the memory 902 and capable of running on the processor 901, wherein: The processor 901 runs the computer program of the dual-frequency response headphone noise reduction method stored in the memory 902, and performs the following steps: At least obtaining an audio signal of a master device, wherein the audio signal includes an ambient audio signal and a return audio signal; Performing frequency band noise reduction processing on the ambient audio signal and the returned audio signal to obtain corresponding noise cancellation signals; The noise cancellation signal is outputted through a MEMS speaker.

[0081] Optionally, the processor 901 executes the step of at least acquiring the audio signal of the master device, including: The external noise of the main device is received and processed by a feedforward microphone arranged outside the main device to obtain an ambient noise data audio signal of the main device; The sound emitted by the main device is received and processed by a feedback microphone arranged inside the main device to obtain a return audio signal of the main device.

[0082] Optionally, the processor 901 performs frequency band noise reduction processing on the ambient audio signal and the returned audio signal to obtain a corresponding noise cancellation signal, including: By using a preset adaptive filtering algorithm, the reverse waveforms of the ambient audio signal and the returned audio signal in a special frequency band are extracted to obtain a corresponding anti-noise waveform digital signal; The anti-noise waveform digital signal is converted into a digital-to-analog converter to obtain a corresponding noise cancellation signal.

[0083] Optionally, the processor 901 performs frequency band noise reduction processing on the ambient audio signal and the returned audio signal to obtain a corresponding noise cancellation signal, and the method further includes: Get the usage scenario of the current main device; Based on the usage scenario, determining a noise frequency band of the specific noise; Based on the noise frequency band, the reverse waveform in the noise frequency band is extracted by a preset adaptive filtering algorithm to obtain a corresponding anti-noise waveform digital signal.

[0084] Optionally, the processor 901 executes the step of determining a noise frequency band of specific noise based on the usage scenario, including: The specific noise in the time domain is converted into the specific noise signal in the frequency domain through the trained frequency domain analysis model, and the frequency spectrum recognition processing is performed on the specific noise signal to determine the noise frequency band of the specific noise.

[0085] Optionally, the processor 901 further executes the trained frequency domain analysis model to convert the specific noise in the time domain into the specific noise signal in the frequency domain, and performs frequency spectrum recognition processing on the specific noise signal to determine the noise frequency band of the specific noise. The method also includes: Acquire a frequency domain analysis model to be trained and a training noise sample set, wherein the training noise sample set includes original audio, noise audio corresponding to the original audio, and a waveform label corresponding to the noise audio; Based on the original audio, the noise audio corresponding to the original audio, and the waveform label corresponding to the noise audio, the frequency domain analysis model to be trained is iteratively trained, and a trained frequency domain analysis model is obtained after the iterative training is completed.

[0086] An embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, each process of the dual-frequency response headphone noise reduction method or the application-side dual-frequency response headphone noise reduction method provided in the embodiment of the present invention is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be described here.

[0087] Those skilled in the art can understand that the implementation of all or part of the processes in the above-mentioned embodiments is a program that can be completed by instructing related hardware through a computer program, and can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM).

[0088] The above disclosure is only the preferred embodiment of the present invention, which certainly cannot be used to limit the scope of the present invention. Therefore, equivalent changes made according to the claims of the present invention are still within the scope of the present invention.

Claims

1. A dual-frequency headphone noise reduction method, characterized in that: The headset at least comprises a main device and an auxiliary device, wherein the auxiliary device comprises a MEMS speaker, including: At least obtaining an audio signal of a master device, wherein the audio signal includes an ambient audio signal and a return audio signal; Performing frequency band noise reduction processing on the ambient audio signal and the returned audio signal to obtain corresponding noise cancellation signals; The noise cancellation signal is outputted through a MEMS speaker.

2. The dual-frequency headphone noise reduction method according to claim 1, wherein: The at least obtaining the audio signal of the master device comprises: The external noise of the main device is received and processed by a feedforward microphone arranged outside the main device to obtain an ambient noise data audio signal of the main device; The sound emitted by the main device is received and processed by a feedback microphone arranged inside the main device to obtain a return audio signal of the main device.

3. The dual-frequency headphone noise reduction method according to claim 1, wherein: The performing frequency band noise reduction processing on the ambient audio signal and the returned audio signal to obtain a corresponding noise cancellation signal includes: By using a preset adaptive filtering algorithm, the reverse waveforms of the ambient audio signal and the returned audio signal in a special frequency band are extracted to obtain a corresponding anti-noise waveform digital signal; The anti-noise waveform digital signal is converted into a digital-to-analog converter to obtain a corresponding noise cancellation signal.

4. The dual-frequency headphone noise reduction method according to claim 1, characterized in that: The frequency band noise reduction processing is performed on the ambient audio signal and the returned audio signal to obtain a corresponding noise cancellation signal, and the method further includes: Get the usage scenario of the current main device; Based on the usage scenario, determining a noise frequency band of the specific noise; Based on the noise frequency band, the reverse waveform in the noise frequency band is extracted by a preset adaptive filtering algorithm to obtain a corresponding anti-noise waveform digital signal.

5. The dual-frequency headphone noise reduction method according to claim 4, characterized in that: The determining the noise frequency band of the specific noise based on the usage scenario includes: The specific noise in the time domain is converted into the specific noise signal in the frequency domain through the trained frequency domain analysis model, and the frequency spectrum recognition processing is performed on the specific noise signal to determine the noise frequency band of the specific noise.

6. The dual-frequency headphone noise reduction method according to claim 5, characterized in that: Before converting the specific noise in the time domain into the specific noise signal in the frequency domain by using the trained frequency domain analysis model and performing frequency spectrum recognition processing on the specific noise signal to determine the noise frequency band of the specific noise, the method further includes: Acquire a frequency domain analysis model to be trained and a training noise sample set, wherein the training noise sample set includes original audio, noise audio corresponding to the original audio, and a waveform label corresponding to the noise audio; Based on the original audio, the noise audio corresponding to the original audio, and the waveform label corresponding to the noise audio, the frequency domain analysis model to be trained is iteratively trained, and a trained frequency domain analysis model is obtained after the iterative training is completed.

7. A dual-frequency headphone noise reduction device, characterized in that: include: A first acquisition module, used to acquire an audio signal of a master device, wherein the audio signal includes an ambient audio signal and a return audio signal; A second acquisition module is used to perform frequency band noise reduction processing on the ambient audio signal and the returned audio signal to obtain a corresponding noise cancellation signal; The output module is used to output the noise cancellation signal through the MEMS speaker.

8. A dual-frequency headphone noise reduction system, characterized in that: The dual-frequency headphone noise reduction system comprises: a dual-frequency headphone noise reduction device; The dual-frequency response headphone noise reduction device is determined by a dual-frequency response headphone noise reduction method as shown in claims 1-6.

9. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps in the dual-frequency response headphone noise reduction method according to any one of claims 1 to 6 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by the processor, the steps in the dual-frequency response headphone noise reduction method according to any one of claims 1 to 6 are implemented.

Images