CN113949955B - Noise reduction processing method, device, electronic equipment, earphone and storage medium - Google Patents

Abstract

The application discloses a noise reduction processing method, a device, electronic equipment, an earphone and a storage medium, and relates to the technical field of noise reduction, wherein the method comprises the following steps: acquiring environmental sounds acquired by an audio acquisition device, wherein the environmental sounds comprise noise signals; preprocessing the environmental sound to obtain a noise signal to be analyzed, wherein the noise signal to be analyzed corresponds to a plurality of frequency bands; acquiring sound energy values of the noise signals to be analyzed in a plurality of frequency bands; determining corresponding target noise reduction parameters according to the proportional relation among the sound energy values of the frequency bands; and carrying out noise reduction processing on the environmental sound based on the target noise reduction parameter. According to the method and the device, the target noise reduction parameters are determined according to the proportional relation among the sound energy values of the noise signals in the plurality of frequency bands through identification, so that targeted noise reduction processing is performed based on the target noise reduction parameters, and therefore a better active noise reduction effect can be achieved for the noise signals in various frequency bands encountered daily.

Description

Noise reduction processing method, device, electronic equipment, earphone and storage medium

Technical Field

The present application relates to the technical field of noise reduction, and more specifically, to a noise reduction processing method, device, electronic device, earphone and storage medium.

Background technique

Currently, active noise reduction usually uses a method of fixed filter parameters. However, due to the different noise compositions in different environments, when the noise in the surrounding environment changes significantly, the noise reduction effect of active noise reduction is not stable enough. For example, for active noise reduction headphones with a peak noise reduction performance of 80 to 200 Hz, the active noise reduction performance will drop significantly between 400 and 2000 Hz, and it cannot effectively eliminate part of the noise in the daily environment at 400 to 2000 Hz. That is, the current active noise reduction effect of active noise reduction headphones is not good.

Summary of the invention

The embodiments of the present application provide a noise reduction processing method, device, electronic device, earphone and storage medium, which can improve the active noise reduction effect.

In a first aspect, an embodiment of the present application provides a noise reduction processing method, the method comprising: obtaining ambient sound collected by an audio collection device, the ambient sound containing a noise signal; preprocessing the ambient sound to obtain a noise signal to be analyzed, the noise signal to be analyzed corresponding to multiple frequency bands; obtaining sound energy values of the noise signal to be analyzed in multiple frequency bands; determining corresponding target noise reduction parameters based on a proportional relationship between the sound energy values of the multiple frequency bands; and performing noise reduction processing on the ambient sound based on the target noise reduction parameters.

In a second aspect, an embodiment of the present application provides a noise reduction processing device, which includes: an audio acquisition module, used to acquire ambient sound acquired by an audio acquisition device, wherein the ambient sound includes a noise signal; a preprocessing module, used to preprocess the ambient sound to obtain a noise signal to be analyzed, wherein the noise signal to be analyzed corresponds to multiple frequency bands; an energy acquisition module, used to acquire sound energy values of the noise signal to be analyzed in multiple frequency bands; a parameter determination module, used to determine corresponding target noise reduction parameters according to a proportional relationship between the sound energy values of the multiple frequency bands; and a noise reduction processing module, used to perform noise reduction processing on the ambient sound based on the target noise reduction parameters.

In a third aspect, an embodiment of the present application provides an electronic device, comprising: a memory; one or more processors coupled to the memory; one or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the one or more processors, and the one or more applications are configured to execute the noise reduction processing method provided in the first aspect above. In a fourth aspect, an embodiment of the present application provides a headset, comprising an audio acquisition device, an audio output device, and an audio signal processing circuit, wherein: the audio acquisition device is used to acquire ambient sound; the audio signal processing circuit is used to acquire the ambient sound acquired by the audio acquisition device; the ambient sound is preprocessed to obtain a noise signal to be analyzed, and the noise signal to be analyzed corresponds to multiple frequency bands; the sound energy values of the noise signal to be analyzed in multiple frequency bands are acquired; the corresponding target noise reduction parameters are determined according to the proportional relationship between the sound energy values of the multiple frequency bands; the audio output device is used to perform noise reduction processing on the ambient sound based on the target noise reduction parameters.

In a fifth aspect, an embodiment of the present application provides a computer-readable storage medium, in which a program code is stored. The program code can be called by a processor to execute the noise reduction processing method provided in the first aspect above.

The embodiments of the present application provide a noise reduction processing method, device, electronic device, headset and storage medium, which obtains ambient sound collected by an audio collection device, wherein the ambient sound includes a noise signal, and then pre-processes the ambient sound to obtain a noise signal to be analyzed corresponding to multiple frequency bands, then obtains the sound energy value of the noise signal to be analyzed in multiple frequency bands, and determines the corresponding target noise reduction parameters according to the proportional relationship between the sound energy values of the multiple frequency bands, and finally performs noise reduction processing on the ambient sound based on the noise reduction parameters. Thus, by identifying the sound energy values of the noise signal in multiple frequency bands, determining the target noise reduction parameters according to the proportional relationship between them, and performing targeted noise reduction processing based on the target noise reduction parameters, a better active noise reduction effect can be achieved for noise signals of various frequency bands encountered in daily life.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

FIG. 1 shows a schematic diagram of an active noise reduction principle.

FIG. 2 shows a schematic diagram of an application environment suitable for an embodiment of the present application.

FIG3 is a schematic flow chart of a noise reduction method provided in an embodiment of the present application.

FIG4 is a schematic flow chart of a noise reduction method provided in another embodiment of the present application.

FIG. 5 shows a schematic flow chart of step S260 in FIG. 4 provided by an exemplary embodiment of the present application.

FIG. 6 shows a frequency spectrum characteristic diagram of a type of noise signal provided by an exemplary embodiment of the present application.

FIG. 7 shows a frequency spectrum characteristic diagram of another type of noise signal provided by an exemplary embodiment of the present application.

FIG8 is a schematic flow chart of a noise reduction method provided in yet another embodiment of the present application.

FIG. 9 is a schematic flow chart of step S370 in FIG. 8 provided by an exemplary embodiment of the present application.

FIG. 10 shows a module block diagram of a noise reduction processing device provided in an embodiment of the present application.

FIG. 11 shows a structural block diagram of an electronic device provided in an embodiment of the present application.

FIG12 shows a structural block diagram of the earphone provided in an embodiment of the present application.

FIG13 shows a storage unit provided in an embodiment of the present application for storing or carrying a program code for implementing a noise reduction processing method according to an embodiment of the present application.

Detailed ways

In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

At present, most of the active noise cancellation (ANC) headphones are designed with fixed active noise cancellation parameters to eliminate external environmental noise based on a fixed active noise cancellation curve (Active Noise Cancellation Level Curve). The technical principle is shown in Figure 1. After the microphone collects the external environmental noise, it is processed by the noise reduction circuit with fixed noise reduction filter parameters to generate an inverted signal that is broadcast through the speaker to offset the external environmental noise. After using fixed noise reduction filter parameters, the noise reduction performance curve does not change with the external noise environment during the operation of the noise reduction circuit.

Definition of Terms

Active Noise Cancellation (ANC): is a noise reduction method. Its principle is to use a speaker to play sound based on the noise sound waves picked up at a specified location, and generate a sound wave with the same amplitude and opposite phase to the original sound wave at the specified location. Since sound waves are mechanical vibrations, when two sound waves meet in space, they will be linearly superimposed. The secondary sound field generated by the speaker and the original noise field meet at the specified location and are linearly superimposed. The two sound waves with the same amplitude and opposite phase will cancel each other out after superposition, thereby weakening or even eliminating the noise, allowing the listener to have a quieter listening experience.

Active noise reduction curve: The active noise reduction curve is a curve showing the change of the active noise reduction amount of the noise reduction device with frequency, which is used to reflect the noise reduction ability of the device at different sound frequencies. It is specifically reflected in a curve with the active noise reduction amount on the vertical axis and the frequency on the horizontal axis. The noise reduction amount of the noise reduction device indicates the degree to which the audible sound waves are reduced before reaching the human eardrum. The noise reduction amount of active noise reduction for sound waves of different frequencies is not the same. Compared with high-frequency signals, the noise reduction effect of active noise reduction on low-frequency sound waves is more obvious. The noise reduction amount of the noise reduction device at each frequency point is measured using standardized professional instruments. The curve formed by connecting the noise reduction values at each frequency point is called the noise reduction curve, which accurately describes the noise reduction ability at different frequency points.

Currently, ANC headphones basically eliminate surrounding noise through a fixed active noise reduction performance and a fixed active noise reduction curve. The effective range of active noise reduction of headphones is generally between 20 and 2000Hz, and the peak is generally between 80 and 250Hz.

However, if a fixed active noise reduction curve is used to eliminate surrounding noise, the noise reduction experience of ANC headphones will be unstable when the noise in the surrounding environment changes significantly. For example, an ANC headphone with a noise reduction performance peak at 80-250Hz will have a good active noise reduction experience when the surrounding noise is mainly low-frequency noise of 80-250Hz. However, when the user wears the headphones and enters an environment where the surrounding noise is mainly medium- and low-frequency noise of 250-400Hz, the active noise reduction experience will deteriorate significantly. Ultimately, when users use general ANC headphones, the active noise reduction experience in different environments is significantly different.

In addition, a small number of ANC headphones will reduce noise in certain frequency bands based on the characteristics of the surrounding environment. For example, when this type of ANC headphones is used in conjunction with its mobile phone APP, the transparency mode (that is, amplifying the external environmental noise to the ears, so that the headphone wearer can hear the external environmental noise more clearly. This function is similar to processing in the opposite direction of noise reduction) can be set to "about voice" mode. In the voice-related mode, the headphones will have a fixed amount of noise reduction for noise signals below about 300Hz, and at the same time amplify signals above about 300Hz (the frequency components of voice signals are mainly above 300Hz) to the ears of the headphone wearer. However, this processing still uses a fixed active noise reduction curve to perform the same processing on all different environmental noises, and the above-mentioned problems will still exist.

In addition, some ANC headphones switch between noise reduction modes with different noise reduction intensity levels according to the overall strength of the ambient noise, but whether the switching is automatic or manual, different frequency components in the noise cannot be flexibly processed differently.

Based on the above problems, the embodiments of the present application provide a noise reduction processing method, device, electronic device and computer-readable storage medium, which processes and analyzes the acquired environmental noise signal, identifies the spectrum characteristics of the environmental noise, and adjusts the noise reduction parameters according to the spectrum characteristics of the noise, so as to obtain the best noise reduction effect for the noise. In order to facilitate a better understanding of the embodiments of the present application, the application environment applicable to the embodiments of the present application is described below.

Please refer to Fig. 2, which shows a schematic diagram of an application environment applicable to the embodiment of the present application. The noise reduction processing method provided in the embodiment of the present application can be applied to the noise reduction processing system 10 shown in Fig. 2. The noise reduction processing system 10 includes a terminal 100 and a headset 200.

The terminal 100 may be, but is not limited to, a mobile phone, a tablet computer, an MP3 player (Moving Picture Experts Group Audio Layer III, dynamic image compression standard audio layer 3), an MP4 player (Moving Picture Experts Group Audio Layer IV, dynamic image compression standard audio layer 4), a notebook computer, an e-book or a wearable electronic device, etc. The embodiment of the present application does not limit the specific device type of the terminal 100.

In some embodiments, an application capable of controlling the noise reduction mode of the headset may be installed in the terminal 100. The terminal 100 may send the noise reduction mode to be implemented to the headset 200, and the headset 200 plays audio, which may include anti-phase sound waves, music signals, masking sounds, etc. The terminal 100 may include an audio acquisition device for collecting ambient sound, and may be provided with a processor for executing the noise reduction processing method provided in the embodiment of the present application, and then play the audio correspondingly through the headset 200 to achieve noise reduction.

The terminal 100 and the headset 200 may be connected by wire or wirelessly. Optionally, if they are connected wirelessly, the terminal 100 and the headset 200 may be connected wirelessly based on Bluetooth, 2.4G wireless communication technology, infrared transmission technology or wireless network to achieve data transmission. For example, after the headset 200 is connected wirelessly to the terminal 100, the audio source data may be obtained through the terminal 100 for playback. Optionally, the wireless network may be a mobile communication network or a wireless fidelity network (WiFi).

Among them, the earphone 200 can be a wired earphone or a wireless earphone. Optionally, the earphone 200 can also be a true wireless earphone. The following is an example of a completely cable-free true wireless earphone, but those skilled in the art should understand that the completely cable-free true wireless earphone is only an exemplary description. In actual use, those skilled in the art can refer to the solution of the embodiment of the present application and select other types of earphones to implement this solution, including but not limited to wired earphones and wireless earphones with cables between the two earphones.

In addition, in some embodiments, the noise reduction processing system 10 may also include only headphones 200, that is, the noise reduction processing method provided in the embodiments of the present application can be implemented without a terminal. For example, in a scenario where the headphones 200 do not play music but only perform noise reduction processing, the headphones 200 may not be connected to the terminal 100, and the noise reduction processing method provided in the embodiments of the present application can be implemented independently.

In addition, only one earphone is shown in the figure. In actual applications, those skilled in the art may refer to the solution of the embodiment of the present application and select a pair of earphones to implement the solution. It should be noted that the noise reduction processing of multiple earphones may be independent of each other or not, and each earphone in a pair of earphones may be connected to the terminal 100 respectively, and the earphones may also be connected to each other, which is not limited in the embodiment of the present application.

In some embodiments, each earphone 200 may include an audio acquisition device, an audio output device, and an audio signal processing circuit, specifically, may include at least one speaker, at least one microphone that can pick up ambient noise, at least one audio signal processing circuit that can run an algorithm, and the earphone 200 may also include at least one power supply circuit.

The speaker is used to play audio and ANC anti-phase noise, so as to realize the music playing and ANC noise reduction functions of the earphone 200. When the earphone 200 is a mono earphone or a true wireless earphone, each earphone 200 has at least one speaker. When the earphone 200 is a dual-channel earphone or a multi-channel earphone, each earphone 200 has at least two speakers.

The microphone is located at a position on the structure of the earphone 200 that can pick up ambient sound, and the picked up ambient sound is used for at least the following two purposes: the original noise signal required for ANC noise reduction, which is used as the input signal required for the noise reduction circuit to output the reverse noise; and the ambient noise signal required for noise detection and analysis. The above two purposes can be achieved simultaneously by one microphone, saving hardware costs, or multiple microphones can be used to achieve the above two purposes separately.

Among them, the audio signal processing circuit can be used for the following two purposes: ANC noise reduction processing function, used to determine the noise reduction parameters used for noise reduction processing and send them to the speaker output corresponding inverted sound waves to achieve noise reduction processing; noise detection and analysis function, used to detect and analyze noise signals in audio signals.

The power supply circuit can provide power for other hardware components, and the power supply source can be a battery built into the headset 200, can be power input from the outside, or can be a power generation device built into the headset 200.

The noise reduction processing method, device, electronic device, headset and storage medium provided in the embodiments of the present application will be described in detail below through specific embodiments.

Please refer to FIG3, which shows a schematic flow chart of a noise reduction processing method provided in an embodiment of the present application, which can be applied to an electronic device, and the electronic device can be the above-mentioned terminal or headset. The following will be described in detail with respect to the schematic flow chart shown in FIG3. The noise reduction processing method may include the following steps:

Step S110: Acquire the ambient sound collected by the audio collection device.

The ambient sound is a sound signal of the environment collected by the audio collection device based on the current position, and the ambient sound includes a noise signal. The audio collection device can be set in the terminal or in the headset, and the ambient sound is collected based on the audio collection device. In some embodiments, the audio collection device can be a device such as a microphone that can be used to collect sound signals, which is not limited here.

Taking the example of an audio collection device being set in a headset, the ambient sound can be collected based on the audio collection device of the headset, such as a microphone. When it is set in the headset, cost can be saved, and the external ambient sound signal picked up by the feedforward microphone can be reused, which can be used for feedforward noise reduction design and can also be used to pick up user input voice without incurring additional hardware costs.

In some embodiments, the audio collection device can collect ambient sound in real time. In other embodiments, the audio collection device can also collect ambient sound based on a collection instruction, that is, the electronic device collects ambient sound based on the audio collection device according to the collection instruction when receiving the collection instruction. For example, the user can trigger the collection instruction through the terminal or the operation of the headset, so that the audio collection device can receive the collection instruction to collect ambient sound.

In some embodiments, the electronic device obtains the ambient sound collected by the audio collection device. If the ambient sound contains a noise signal, the active noise reduction function can be activated to execute step S120 and subsequent steps.

In other embodiments, after obtaining the ambient sound collected by the audio collection device, the electronic device may also activate the active noise reduction function and execute step S120 and subsequent steps to perform noise reduction processing on the noise if the sound energy value of the noise signal in the ambient sound exceeds the preset energy value. If the sound energy value of the noise signal in the ambient sound does not exceed the preset energy value, step S120 and subsequent steps may not be executed, for example, the ambient sound may continue to be monitored or the monitoring may be ended, so that active noise reduction processing may not be performed when the noise signal in the ambient sound is weak and has little impact on the user, thereby reducing device power consumption and saving resources.

Step S120: pre-processing the collected ambient sound to obtain a noise signal to be analyzed.

In some embodiments, preprocessing the collected ambient sound may include performing analog-to-digital conversion on the collected ambient sound to obtain a digital signal, and performing pre-emphasis, framing, windowing, Mel-Frequency Cepstral Coefficients (MFCC) extraction and other processing on the digital signal to obtain a noise signal to be analyzed. In other embodiments, preprocessing may include more or fewer processing steps than the aforementioned, which are not limited here.

In this embodiment, the collected ambient sound is preprocessed, and the noise signal to be analyzed can be divided into multiple frequency bands. Specifically, the frequency band refers to the frequency range of the signal, and the unit is generally Hertz (Hz). For example, the frequency band can be 400Hz to 600Hz, and the noise signal to be analyzed can correspond to 100Hz to 200Hz, 200Hz to 400Hz, and 400Hz to 600Hz.

Step S130: Acquire sound energy values of the noise signal to be analyzed in multiple frequency bands.

After obtaining the noise signal to be analyzed, the sound energy values of the noise signal to be analyzed in multiple frequency bands can be obtained. In one embodiment, a noise signal corresponding to multiple frequency bands is obtained, and the sound energy values of the noise signal to be analyzed in multiple frequency bands can be obtained based on the sound energy value of the noise signal in each frequency band.

The unit of the sound energy value may be dB. In some examples, the spectrum energy may also be referred to as energy, amplitude, or sound pressure level, i.e., how many dB is used to represent a sound signal in an environment. As a method, the sound energy values of the noise signal in multiple frequency bands may be determined based on the distribution of the noise signal to be analyzed on the spectrum graph, where the horizontal axis of the spectrum graph may be the frequency and the vertical axis may be the sound energy value.

Step 140: Determine corresponding target noise reduction parameters according to the proportional relationship between the sound energy values of multiple frequency bands.

When performing noise reduction processing on ambient sound, the corresponding target noise reduction parameters can be determined according to the proportional relationship between the sound energy values of multiple frequency bands of the noise signal, so as to perform noise reduction processing on the noise signal in the ambient sound based on the target noise reduction parameters. By obtaining the proportional relationship between the sound energy values of multiple frequency bands, the frequency bands where the sound energy values are more concentrated can be more accurately determined to more accurately distinguish the noise types, which is conducive to determining more accurate noise reduction parameters and achieving better noise reduction effects.

Among them, the target noise reduction parameter can be generated in real time according to the sound energy value of the collected noise signal, or it can be pre-set. For example, in some embodiments, multiple groups of preset noise reduction parameters for various noise signals can be pre-constructed, and a mapping relationship between each preset noise reduction parameter and the proportional relationship can be constructed, so that according to the proportional relationship between the sound energy values of multiple frequency bands, the corresponding preset noise reduction parameter is determined as the target noise reduction parameter, and the ambient sound is noise-reduced based on the target noise reduction parameter, thereby reducing the amount of calculation. If this embodiment is applied to headphones, the power consumption of the headphones can be reduced and the battery life can be increased.

In some embodiments, based on the trained neural network model, the noise type of the noise signal to be analyzed can be determined according to the proportional relationship of the energy values of multiple frequency bands, and then the noise reduction parameters corresponding to the noise type can be determined as the target noise reduction parameters. The neural network model can use the collected noise signals of various frequency bands as training samples, and mark the noise type of each training sample. The neural network model can be trained in this way to obtain a trained neural network model that can be used to implement step S140. The details can be seen in the embodiments described later and will not be repeated here.

In other embodiments, the sound energy values of multiple frequency bands may be specifically compared, and the noise reduction parameter corresponding to the frequency band with the highest sound energy value may be determined as the target noise reduction parameter. In some implementations, the proportional relationship between the sound energy values of multiple frequency bands may be obtained by first determining the frequency band with the highest sound energy value, and then determining the proportional relationship between the frequency band and other frequency bands. The details can also be seen in the embodiments described below, and will not be repeated here.

In other embodiments, the proportional relationship between the sound energy values of multiple frequency bands can also be directly determined by the ratio between the sound energy values of multiple frequency bands. Specifically, if the ratio between the sound energy values of multiple frequency bands matches the preset ratio, the frequency band with the highest sound energy value can be determined as a candidate frequency band from the multiple frequency bands, and the noise reduction parameter corresponding to the candidate frequency band can be determined as the target noise reduction parameter. Among them, the preset ratio can be determined according to actual needs. For example, if the number of multiple frequency bands is 3, the preset ratio can be 1:1:2, 1:1:3, 1:2:4, etc., which are not limited here. In addition, matching with the preset ratio can be an exact match or an approximate match. For example, the ratio between the sound energy values of the three frequency bands is 1.1:1:2, and the preset ratio is 1:1:2. At this time, it can also be determined that the ratio of the three frequency bands matches the preset ratio, thereby allowing a certain error, and the error tolerance can also be determined according to actual needs. In addition, when the ratio of multiple frequency bands matches the preset ratio, the order of the frequency bands can be unrestricted, that is, as long as the ratio of the multiple frequency bands can match the preset ratio.

In a specific example, if the ratio between the sound energy values of frequency band A, frequency band B, and frequency band C is 1:2:1, and the preset ratio is 1:1:2, it can be determined that the ratio between the sound energy values of frequency band A, frequency band B, and frequency band C matches the preset ratio, and the noise reduction parameter corresponding to frequency band B can be determined as the target noise reduction parameter.

Step 150: Perform noise reduction processing on the ambient sound based on the target noise reduction parameter.

The determined target noise reduction parameter may be an active noise reduction curve or a noise reduction parameter corresponding to the active noise reduction curve. Based on the target noise reduction parameter, the earphone may output a corresponding anti-phase sound wave to perform noise reduction on the ambient sound.

The noise reduction processing method provided in the embodiment of the present application collects ambient sound based on an audio collection device, wherein the ambient sound may contain a noise signal, and then pre-processes the collected ambient sound to obtain a noise signal to be analyzed corresponding to multiple frequency bands, then obtains the sound energy value of the noise signal to be analyzed in multiple frequency bands, and determines the corresponding target noise reduction parameters based on the proportional relationship between the sound energy values of the multiple frequency bands, and finally performs noise reduction processing on the ambient sound based on the noise reduction parameters. Thus, by identifying the sound energy values of the noise signal in multiple frequency bands, determining the target noise reduction parameters based on the proportional relationship between them, and performing targeted noise reduction processing based on the target noise reduction parameters, a better active noise reduction effect can be achieved for noise signals of various frequency bands encountered in daily life.

Please refer to FIG. 4 , which shows a schematic flow chart of a noise reduction processing method provided by another embodiment of the present application. Specifically, the method may include:

Step S210: Collecting ambient sound based on the audio collection device.

Step S220: Based on the preset frequency range, the preset frequency range is divided into a plurality of frequency bands to be analyzed according to octaves.

Since the resolution of the human ear hearing system for sound frequency is not fixed, the higher the frequency, the lower the frequency resolution of the human ear hearing system, and the frequency interval that the human ear hearing system can distinguish is approximately logarithmically proportional to the center frequency of the octave, when a noise signal to be analyzed corresponding to multiple frequency bands is obtained, the frequency bandwidth of the noise signal to be analyzed can be set to a logarithmic frequency band, such as an octave. In some embodiments, if you want to identify the noise signal in a more fine-grained manner, you can also set the analysis frequency bandwidth to 1/2 octave or even 1/3 octave, which is not limited in this embodiment.

In some embodiments, the preset frequency range can be determined according to actual needs. For example, considering that most ANC headphones have a weak noise reduction effect in the 1-2kHz frequency band, only low-frequency noise components below 1kHz can be analyzed, and the preset frequency range can be a frequency range below 1kHz. Of course, a wider or narrower frequency range can also be processed and analyzed, and this embodiment does not limit the preset frequency range.

In some implementations, the division of frequency bands can also be determined according to actual needs. In practical applications, in order to prevent the ANC headset from entering the nonlinear working area and causing abnormal sounds in the presence of strong low-frequency vibrations, the noise reduction effect of the ANC headset below 100Hz is generally poor. In view of this, the first frequency band to be analyzed can be set to 100-200Hz. At the same time, combined with the frequency resolution performance of the human auditory system, the noise signal can be subjected to logarithmic frequency analysis, and the second frequency band to be analyzed can be set to 200-400Hz. In addition, as an implementation method, according to the logarithmic octave characteristics, the third frequency band to be analyzed can be set to 400-800Hz.

As another implementation, considering that most ANC headphones have a weak noise reduction effect in the 1-2kHz frequency band, only low-frequency noise components below 1kHz can be analyzed, and only the spectrum components below 1kHz can be considered, and the third frequency band to be analyzed can be set to 400-1000Hz. In this way, the frequency characteristics of active noise reduction and the characteristics of the human hearing system can be combined to determine the frequency ranges of different frequency bands to be analyzed.

It should be noted that different ANC headphones have different noise reduction performances. According to the noise reduction performance of the ANC headphones in different frequency bands, the third frequency band to be analyzed can be flexibly set. For example, the highest frequency at which the noise reduction performance begins to deteriorate is set as the upper limit frequency value of the third frequency band to be analyzed. In one example, if the noise reduction effect of the ANC headphones in the frequency band above 1.2kHz begins to deteriorate or is lower than the set threshold, the third frequency band to be analyzed can also be set to 400-1200Hz, and the preset frequency range can be the frequency range below 1.2kHz.

It should be noted that the frequency band to be analyzed may also be set to be narrower or wider, the center frequency may be moved to other frequencies, the frequency band may be split into multiple finer frequency bands, etc., and this embodiment does not limit this. In addition, the logarithmic frequency band analysis may be based on octaves, 1/2 octaves, or 1/3 octaves, etc., and this embodiment does not limit this either.

Considering that the human ear hearing system has a logarithmic characteristic in its ability to distinguish sound frequencies, that is, the resolution of the low-frequency part is high, but the resolution of the high-frequency part is low. Therefore, compared with the noise reduction method that performs Fourier transform on the noise signal picked up by the audio acquisition device and then analyzes the noise spectrum difference to call different noise reduction modes, this embodiment sets the frequency band to be analyzed and then identifies the noise in combination with the noise spectrum characteristics within the frequency band, which can more fully consider the subjective listening characteristics of the human ear hearing system, reduce the consideration of excessive noise spectrum details, and make the noise signal more robust when processing and analyzing.

Step S230: Determine a corresponding bandpass filter according to the upper limit frequency value and the lower limit frequency value of each frequency band.

According to the upper limit frequency value and the lower limit frequency value of each frequency band, the frequency range of each frequency band can be obtained, and the bandpass filter corresponding to each frequency range can be generated.

Step S240: performing bandpass filtering on the noise signal of each frequency band based on the bandpass filter corresponding to each frequency band in the time domain, and obtaining a filtered noise signal as the noise signal to be analyzed.

In the time domain, based on the bandpass filter corresponding to each frequency band, the noise signal of each frequency band is bandpass filtered to filter out the signals in different frequency bands to be analyzed, so as to facilitate the subsequent identification of the spectrum characteristics of the environmental noise signal. In this filtering process, unnecessary low and high frequencies are also filtered out, which not only ensures the integrity of the data information, but also reduces the amount of data processed by the signal. And because the filtering is performed directly in the time domain, there is no need to perform Fourier transform to the spectrum, only filtering in the time domain, so the processing is simpler and more in line with the characteristics of noise reduction.

Among them, the bandpass filtering process can retain the frequency components within a certain frequency range in the signal, while attenuating the frequency components outside the frequency range to a lower level. In this embodiment, the bandpass filtering process can be used to intercept the sound signals in different frequency ranges, and then the sound signal energy in the corresponding frequency range is calculated, that is, the sound energy value, and then different noise types are distinguished according to the distribution difference of the sound energy value in different frequency bands, and then the corresponding noise reduction parameters are determined as the target noise reduction parameters.

In some other embodiments, low-pass filtering may be performed instead of band-pass filtering, which is not limited here.

In some embodiments, preprocessing may also include downsampling processing. For example, before the noise signal of each frequency band is band-pass filtered in the time domain based on the band-pass filter corresponding to each frequency band, the collected ambient sound may be downsampled to obtain the downsampled ambient sound; and the noise signal to be analyzed is obtained based on the downsampled ambient sound. Specifically, in some embodiments, the downsampled ambient sound is obtained, and the downsampled ambient sound may be preprocessed to obtain the noise signal to be analyzed, wherein the preprocessing step may refer to the aforementioned step S120, which will not be described in detail here. Of course, in other embodiments, when preprocessing the downsampled ambient sound, the noise signal of the ambient sound may also be filtered, and the specific implementation method may refer to the aforementioned steps S220 to S240, which will not be described in detail here.

Since active noise reduction processing has high requirements for system latency, for example, in some scenarios, the system's hardware latency needs to be within 20 microseconds. Therefore, in the active noise reduction processing path, the sampling rate of the digital signal is relatively high, basically above 192kHz, and some ANC headphones even use a high sampling rate of 768kHz. However, when identifying, analyzing, and classifying the noise spectrum, its sensitivity to system latency is much lower, and considering that the higher the sampling rate, the greater the amount of calculation, the high sampling rate digital signal can be downsampled before performing noise spectrum identification and analysis.

In some embodiments, since active noise reduction is mainly effective for frequency components below 2kHz, the noise signals within this frequency range can be identified and analyzed. In order to cover the frequency range below 2kHz, the microphone signal sampling rate only needs to be no less than 4kHz. At the same time, the lower the frequency, the more computing power can be saved, so downsampling processing can be used; if the computing power of the electronic device is sufficient, such as the computing power of the audio signal processing circuit, a higher sampling rate can also be used. As one way, a 16kHz sampling rate can be used, and the collected ambient sound can be downsampled based on the sampling rate of 16kHz; as another way, different sampling rates can also be considered. As long as the sampling rate is not less than 4kHz, the accuracy of the noise signal can still be guaranteed while reducing the amount of computing to a certain extent.

Step S250: Acquire sound energy values of the noise signal to be analyzed in multiple frequency bands.

By bandpass filtering, the noise components of the noise signal in different frequency bands to be analyzed can be obtained, and at this time, the sound energy values of the noise signal to be analyzed in different frequency bands can be calculated.

In some embodiments, after obtaining the sound energy values of the noise signal to be analyzed in multiple frequency bands, the sound energy values of the noise signal to be analyzed in multiple frequency bands can be smoothed. Since the noise signal in the environment always changes over time, in actual use, in order to prevent the active noise reduction effect from switching frequently and resulting in a poor user experience, the sound energy values of the noise signals in different frequency bands are smoothed, wherein the specific smoothing speed can be adjusted according to actual needs, and can also be preset by the program or customized by the user. Energy smoothing can slow down the tracking speed of noise signal changes, eliminate the impact of some short-term transient changes in the environmental noise signal on the noise reduction effect, and improve the user experience.

Step S260: Determine corresponding target noise reduction parameters according to the proportional relationship between the sound energy values of multiple frequency bands.

In some other embodiments, step S260 may be implemented based on a trained neural network model, and step S260 may include steps S261 to S262. Specifically, please refer to FIG. 5, which shows a flow chart of step S260 in FIG. 4 provided by an exemplary embodiment of the present application. In this embodiment, step S260 may include:

Step S261: Based on the trained deep learning model, determine the noise type according to the proportional relationship of the sound energy values of multiple frequency bands.

Before training the deep learning model, a noise database can be collected first. For example, based on different noise environments, noise signals in different environments can be collected to obtain corresponding noise signals. In order to improve the recognition accuracy of the deep learning model, as many noise signals as possible can be collected.

After the noise signal is collected, the collected noise signal can be segmented based on the preset time length to obtain multiple noise signals with a duration of the preset time length, wherein the preset time length can be determined according to actual needs, or can be preset by the program or customized by the user, and this embodiment does not limit this. In an example, the preset time length can be set to range from several seconds to tens of seconds.

After the collected noise signal is segmented, the sound energy values of each noise signal in different frequency bands can be marked. In some embodiments, only the sound energy value of the noise signal in each frequency band can be marked; in other embodiments, the proportional relationship of the sound energy values of the noise signal in multiple frequency bands can be directly marked; in still other embodiments, the noise type of the noise signal can also be marked, which is not limited in this embodiment and can be determined specifically according to the input and output definitions when the deep learning model is constructed.

The noise types can be divided according to various dimensions. In some embodiments, the noise types can be divided according to the frequency band with the highest sound energy value; in other embodiments, the noise types can be divided according to different noise generation scenes; in still other embodiments, the noise types can be divided according to different noise generation methods, such as snoring, air conditioning, and talking. This embodiment does not limit the method of dividing the noise types.

In one embodiment, since the sound energy value distributions of different noises in different frequency bands are different, the frequency bands with the highest sound energy values of different noises will also be different. The noise types can also be divided according to the frequency bands with the highest sound energy values. For example, the frequency band with the highest sound energy value of noise type 1 is below 200Hz, and the frequency band with the highest sound energy value of noise type 2 is 500Hz~600Hz. If the sound energy value of the noise signal is mainly concentrated in 500Hz~600Hz, its noise type can be marked as noise type 2.

In addition, in some embodiments, when determining the frequency band with the highest sound energy value from multiple frequency bands, it can also be determined whether the ratios of the frequency band with the highest sound energy value to the sound energy values of other frequency bands all exceed a preset ratio. If they do, the frequency band with the highest sound energy value is marked as a frequency band corresponding to a noise type. If they do not, the frequency band with the highest sound energy value can be widened, and other frequency bands whose ratios do not exceed the preset ratio can be merged with the candidate frequency bands into a target frequency band, so that the merged target frequency band can be marked as a frequency band corresponding to a noise type, and for the merged target frequency band, corresponding noise reduction parameters can be generated as noise reduction parameters corresponding to the noise type, so as to achieve more accurate targeted noise reduction. The specific implementation method can be seen in the embodiments described later, and will not be repeated here.

In another embodiment, since the spectral characteristics of noise in different scenes are different, the sound energy values in different frequency bands are also different, and thus the noise type can be divided according to the scene. For example, the noise type may include subway noise in a subway environment, office noise in an office environment, etc., then noise signals of various noise types such as subway noise based on a subway environment and office noise based on an office environment can be collected in advance, and the corresponding noise types can be marked. When training a deep learning model, the spectral characteristics of the noise signal can be analyzed, such as the proportional relationship of the noise signal in multiple frequency bands as the input of the deep learning model, and the noise type corresponding to the noise signal is used as the expected output. Thus, in one example, if the spectral characteristics of the collected noise signal match the spectral characteristics of the subway noise collected based on the subway environment, for example, the proportional relationship of the sound energy values of the noise signal in multiple frequency bands matches the proportional relationship of the sound energy values of the subway noise in the corresponding frequency bands, then it can be determined that the noise type of the noise signal is subway noise, wherein the judgment of whether the spectral characteristics match can be realized by the trained deep learning model, and the noise type is determined by the trained deep learning model.

In another embodiment, since the spectral characteristics of noise generated in different ways are quite different, for example, the sound energy value of snoring is mainly concentrated in 250-800Hz, and the general noise reduction curve of active noise reduction headphones can only have a good noise reduction effect on noise signals in the range of 80-200Hz, so the noise reduction effect on snoring is not strong. In order to improve the targeted noise reduction effect, the noise types can be divided according to the different noise generation methods, such as snoring, air conditioning, speaking, etc., then noise signals of various noise types such as snoring, air conditioning, speaking, etc. can be collected in advance, and the corresponding noise types can be marked. When training the deep learning model, the above description can be referred to. In an example, if the proportional relationship of the sound energy values of the collected noise signal in multiple frequency bands matches the proportional relationship of the sound energy values of snoring in the corresponding frequency bands, it can be determined that the noise type of the noise signal is snoring.

A deep learning model is established based on a neural network, and based on the above-mentioned labeled noise signal, the deep learning model is trained to train the parameters of the deep learning model, including the number of network layers, activation functions, etc., so as to obtain the desirable range of the training parameters, and then determine whether the training can be stopped based on the loss function curve obtained by training and testing, and when it can be stopped, a deep learning model that can identify different types of noise can be obtained. Among them, the neural network can be a convolutional neural network (CNN), a deep neural network (DNN), a recurrent neural network (RNN), etc., which are not limited here. Therefore, based on the trained deep learning model, the noise type of the noise signal to be analyzed can be determined according to the proportional relationship of the sound energy values of multiple frequency bands.

In some embodiments, the trained deep learning model can be transplanted into the audio signal processing circuit of the headset, and the trained deep learning model can be run based on the headset. In other embodiments, it can also be deployed on the terminal to run the trained deep learning model based on the terminal, thereby reducing the computing burden of the headset and reducing the power consumption of the headset. In particular, when the headset is a true wireless headset, its battery life can be improved.

Step S262: Determine the noise reduction parameter corresponding to the noise type as the target noise reduction parameter.

In some embodiments, a mapping relationship between each noise type and a noise reduction parameter may be pre-stored, so that the corresponding noise reduction parameter may be determined as a target noise reduction parameter according to the noise type. In other embodiments, the corresponding noise reduction parameter may be generated in real time according to the noise type as a target noise reduction parameter, so as to implement adaptive noise reduction processing for the current noise type, so as to achieve a better noise reduction effect.

In some implementations, at least three sets of noise reduction parameters may be stored inside the earphone, which may correspond to three sets of active noise reduction curves, respectively, to match noise reduction processing of different noise types.

Step S270: performing noise reduction processing on the ambient sound based on the target noise reduction parameters.

After determining the noise type, the corresponding target noise reduction parameters can be determined according to the noise reduction parameters corresponding to the noise type. For example, if the energy of the noise signal in the current environment is concentrated in the frequency band below 200Hz, the active noise reduction curve of the ANC headset can be adjusted to an active noise reduction curve with noise reduction performance concentrated in the frequency band below 200Hz, thereby obtaining the best noise reduction effect; and when the environment changes, the energy in the spectrum of the noise signal is mainly concentrated in the range of 400 to 600Hz, then the active noise reduction curve of the ANC headset can be adjusted to a noise reduction performance concentrated in the frequency band of 400 to 600Hz, thereby continuing to obtain a better noise reduction effect.

In a specific example, taking Figures 6 and 7 as examples, Figures 6 and 7 are spectrum feature diagrams of two types of noise signals, respectively. The energy of the noise signal in Figure 6 is concentrated in the low-frequency band below 200Hz. When the spectrum feature of the noise signal is identified as shown in Figure 6, the noise reduction parameter whose noise reduction performance is mainly concentrated below 200Hz can be determined as the target noise reduction parameter, and the ambient sound can be processed by this, such as adjusting the active noise reduction curve to concentrate the noise reduction performance below 200Hz. The energy of the noise signal in Figure 7 is mostly distributed around 500Hz to 600Hz. When the spectrum feature of the noise signal is identified as shown in Figure 7, the noise reduction parameter whose noise reduction performance is mainly concentrated between 500Hz and 600Hz can be determined as the target noise reduction parameter, and the ambient sound can be processed by this, such as adjusting the active noise reduction curve to concentrate the noise reduction performance around 500Hz to 600Hz, and the headphones can perform noise reduction based on the active noise reduction curve.

It should be noted that the parts not described in detail in this embodiment can be referred to the previous embodiments and will not be described again here.

The noise reduction processing method provided in this embodiment collects ambient sound based on an audio collection device, then generates a corresponding bandpass filter based on multiple frequency bands to be analyzed divided by octaves, and directly performs bandpass filtering in the time domain to obtain the noise signal to be analyzed, so that the processing is simpler and more in line with the characteristics of noise reduction. Then, according to the proportional relationship between the sound energy values of the noise signal to be analyzed in multiple frequency bands, the spectrum characteristics of the noise signal are identified, and then the noise reduction parameters are adjusted accordingly according to the spectrum characteristics, so as to obtain a better noise reduction effect under different noises. In addition, the collected ambient sound can be downsampled before the bandpass filtering, which can take into account the amount of calculation of the headset and reduce its power consumption. And by using octaves that are more in line with the human ear hearing system to divide multiple frequency bands to be analyzed, it is possible to achieve robust and more in line with the subjective perception characteristics of the human ear hearing system to identify the noise type without considering too many noise spectrum details, and obtain a more in line with the subjective hearing and noise reduction effect of the human ear, thereby improving the subjective experience of the listener. In addition, due to the noise reduction processing method provided in this embodiment, a pair of ANC headphones can achieve noise reduction processing for various noises encountered in daily life, saving the number of ANC headphones used and purchased by users.

Please refer to FIG8 , which shows a schematic flow chart of a noise reduction processing method provided by another embodiment of the present application. Specifically, the method may include:

Step S310: Collecting ambient sound based on the audio collection device.

Step S320: Based on the preset frequency range, the preset frequency range is divided into a plurality of frequency bands to be analyzed according to octaves.

Step S330: Determine a corresponding bandpass filter according to the upper limit frequency value and the lower limit frequency value of each frequency band.

Step S340: performing bandpass filtering on the noise signal of each frequency band based on the bandpass filter corresponding to each frequency band in the time domain, and obtaining a filtered noise signal as the noise signal to be analyzed.

Step S350: Acquire sound energy values of the noise signal to be analyzed in multiple frequency bands.

Step S360: Determine a frequency band with the highest sound energy value from among the multiple frequency bands as a candidate frequency band.

Based on the obtained sound energy values of the noise signal in different frequency bands, the spectral characteristics of the noise signal can be further identified through the magnitude relationship between the different energy values. In this embodiment, the frequency band with the highest sound energy value can be determined as a candidate frequency band from multiple frequency bands. Since the noise signal is generally concentrated in the frequency band with the highest sound energy value, it can be determined whether the candidate frequency band is sufficient to determine the target noise reduction parameter.

In one embodiment, after the bandpass filtering, the noise signal may be smoothed. In one example, if the noise energy value of the smoothed noise signal in the first frequency band of 100 Hz to 200 Hz is A, the noise energy value in the second frequency band of 200 Hz to 400 Hz is B, and the noise energy value in the third frequency band of 400 Hz to 1000 Hz is C, then if A>B>C, the first frequency band of 100 Hz to 200 Hz corresponding to A may be determined as a candidate frequency band, and A is the corresponding candidate sound energy value.

Step S370: Determine whether the ratios of the candidate sound energy value corresponding to the candidate frequency band to the sound energy values of other frequency bands exceed a preset ratio.

In some embodiments, the preset ratio includes a first preset ratio and a second preset ratio, and step S370 may include steps S371 to S373. Specifically, please refer to FIG. 9, which shows a flow chart of step S370 in FIG. 8 provided by an exemplary embodiment of the present application. In this embodiment, step S370 may include:

Step S371: Determine a first ratio of the sound energy value of the candidate frequency band to the sound energy value of the first frequency band.

Step S372: Determine a second ratio of the sound energy value of the candidate frequency band to the sound energy value of the second frequency band.

Among them, other frequency bands may include a continuous first frequency band and a second frequency band. It should be noted that the continuous first frequency band and the second frequency band refer to that the first frequency band, the second frequency band and the candidate frequency band are continuous in frequency. For example, if the candidate frequency band is 100Hz to 200Hz, the first frequency band may be 200Hz to 400Hz, and the second frequency band may be 400Hz to 1000Hz. For another example, if the candidate frequency band is 200Hz to 400Hz, the first frequency band may be 100Hz to 200Hz, and the second frequency band may be 400Hz to 1000Hz. Therefore, by comparing the sound energy value of the candidate frequency band with the sound energy values of the first frequency band and the second frequency band, the degree of difference between the candidate sound energy value of the candidate frequency band and the sound energy values of other frequency bands can be determined.

In some possible embodiments, the first frequency band, the second frequency band and the candidate frequency band may be spaced apart and continuous in frequency, for example, the candidate frequency band is 100 Hz to 200 Hz, the first frequency band may be 250 Hz to 450 Hz, and the second frequency band may be 500 Hz to 1000 Hz.

In some implementations, the first preset ratio and the second preset ratio can be determined according to actual needs, or can be preset by a program, or can be user-defined, which is not limited in this embodiment. In addition, the first preset ratio and the second preset ratio can be the same or different.

As an implementation mode, if the first preset ratio is the same as the second preset ratio, it can be determined by determining whether the ratio of the candidate sound energy value corresponding to the candidate frequency band to the sound energy value of each frequency band in other frequency bands exceeds the first preset ratio or the second preset ratio. Only when both exceed, can it be determined that the ratio of the candidate sound energy value corresponding to the candidate frequency band to the sound energy values of other frequency bands exceeds the preset ratio.

As another implementation, the first preset ratio and the second preset ratio may be different. For example, different first preset ratios and second preset ratios may be set according to the frequency difference between the first frequency band and the second frequency band and the candidate frequency band, and in one example, the larger the frequency difference, the larger the preset ratio. In another example, the larger the frequency difference, the smaller the preset ratio. In this case, the target noise reduction parameter can be determined as long as the difference between the sound energy value of the candidate frequency band and the frequency band closest to it is large enough.

The frequency difference may be the difference between the center frequency of the first frequency band, the second frequency band and the candidate frequency band, or the difference between the first frequency band, the second frequency band and the upper limit frequency or the lower limit frequency of the candidate frequency band, which is not limited here. For example, if the candidate frequency band is 100Hz to 200Hz, the first frequency band may be 200Hz to 400Hz, and the second frequency band may be 400Hz to 1000Hz, then the frequency difference between the first frequency band and the candidate frequency band is the difference between the center frequencies, i.e., 300Hz-150Hz=150Hz, and the frequency difference between the second frequency band and the candidate frequency band is the difference between the center frequencies, i.e., 700Hz-150Hz=550Hz, then the frequency difference corresponding to the second frequency band is greater than the frequency difference corresponding to the first frequency band.

In some implementations, a mapping relationship between different frequency difference intervals and preset ratios may be preset, and a corresponding preset ratio may be determined according to the frequency difference to serve as the first preset ratio or the second preset ratio.

Step S373: If the first ratio exceeds the first preset ratio and the second ratio exceeds the second preset ratio, it is determined that the ratios of the candidate sound energy value corresponding to the candidate frequency band and the sound energy values of other frequency bands all exceed the preset ratios.

If the first ratio exceeds the first preset ratio and the second ratio exceeds the second preset ratio, it is determined that the ratios of the candidate sound energy value corresponding to the candidate frequency band and the sound energy values of other frequency bands both exceed the preset ratios. At this time, the difference between the candidate frequency band and the other frequency bands is large enough, and the noise signal to be analyzed can be determined as a noise type.

In some embodiments, if the ratio of the candidate sound energy value corresponding to the candidate frequency band to the sound energy value of other frequency bands does not exceed the preset ratio, there may be a mixture of multiple noise signals. In this case, the frequency band with the highest sound energy value may not be included in the pre-divided frequency band, so that the difference in sound energy values between the frequency bands is not large enough. At this time, the frequency band with the highest sound energy value can be widened, and other frequency bands whose ratios do not exceed the preset ratio can be merged with the candidate frequency band into a target frequency band. Thus, corresponding noise reduction parameters can be generated as target noise reduction parameters for the merged target frequency band. In this way, targeted noise reduction processing can be implemented for scenes with a mixture of multiple high noise signals to achieve better noise reduction effects.

Step S380: If both are exceeded, the noise reduction parameter corresponding to the candidate frequency band is determined as the target noise reduction parameter.

For example, if the sound energy value of the noise signal to be analyzed in the first frequency band of 100Hz to 200Hz is A, the sound energy value in the second frequency band of 200Hz to 400Hz is B, and the sound energy value in the third frequency band of 400Hz to 1000Hz is C, if A is greater than a certain multiple of B, and A is greater than a certain multiple of C, that is, the first frequency band 100Hz to 200Hz corresponding to A is the candidate frequency band, the second frequency band 200Hz to 400Hz can be recorded as the first frequency band, and the third frequency band 400Hz to 1000Hz can be recorded as the second frequency band, and the first ratio of A to B is greater than the first preset ratio, and the second ratio of A to C is greater than the second preset ratio, then the noise signal to be analyzed can be classified as a type of noise. At this time, the noise reduction parameter can be adjusted to a noise reduction parameter with the best noise reduction performance for the candidate frequency band 100Hz to 200Hz, so as to perform deeper and more targeted noise reduction processing.

Step S390: performing noise reduction processing on the ambient sound based on the target noise reduction parameters.

It should be noted that the parts not described in detail in this embodiment can be referred to the previous embodiments and will not be described again here.

The noise reduction processing method provided in the present embodiment is based on the above-mentioned embodiment. By determining whether the ratios of the candidate sound energy value corresponding to the candidate frequency band and the sound energy values of other frequency bands all exceed the preset ratios, and only when both exceed the ratios, the noise reduction parameters corresponding to the candidate frequency band are determined as the target noise reduction parameters. It can be determined whether the difference between the candidate sound energy value and the sound energy values of other frequency bands is large enough, so that when both exceed the preset ratios, it is considered that the difference is large enough. Then, when the difference is large enough, the noise reduction parameters corresponding to the candidate frequency band, that is, the noise reduction parameters whose noise reduction performance is mainly concentrated in the candidate frequency band, are determined as the target noise reduction parameters, thereby achieving more accurate noise reduction for various noise signals and improving the noise reduction effect.

Please refer to FIG. 10, which shows a structural block diagram of a noise reduction processing device 1000 provided in an embodiment of the present application, which can be applied to an electronic device, and the electronic device can be the above-mentioned terminal or headset. Specifically, the noise reduction processing device 1000 may include: an audio acquisition module 1010, a preprocessing module 1020, an energy acquisition module 1030, a parameter determination module 1040 and a noise reduction processing module 1050, specifically:

An audio collection module 1010 is used to collect ambient sound based on an audio collection device, where the ambient sound includes a noise signal;

A preprocessing module 1020 is used to preprocess the collected ambient sound to obtain a noise signal to be analyzed, where the noise signal to be analyzed corresponds to multiple frequency bands;

The energy acquisition module 1030 is used to acquire the sound energy values of the noise signal to be analyzed in multiple frequency bands;

A parameter determination module 1040, configured to determine corresponding target noise reduction parameters according to a proportional relationship between the sound energy values of the multiple frequency bands;

The noise reduction processing module 1050 is used to perform noise reduction processing on the ambient sound based on the target noise reduction parameter.

Further, the parameter determination module 1040 may include: a first candidate determination submodule, a first candidate comparison submodule and a first candidate denoising submodule, wherein:

A first candidate determination submodule, configured to determine a frequency band with the highest sound energy value from the multiple frequency bands as a candidate frequency band;

A first candidate comparison submodule, used to determine whether the ratios of the candidate sound energy values corresponding to the candidate frequency band and the sound energy values of other frequency bands all exceed a preset ratio;

The first candidate denoising submodule is configured to determine the denoising parameter corresponding to the candidate frequency band as the target denoising parameter if both exceed the target denoising parameter.

Furthermore, the noise reduction processing device 1000 may also include:

a target frequency band determination module, configured to combine other frequency bands whose ratios do not exceed the preset ratio with the candidate frequency band as the target frequency band if both are not exceeded;

The target noise reduction module is used to determine the noise reduction parameter corresponding to the target frequency band as the target noise reduction parameter.

Further, the other frequency bands include a continuous first frequency band and a second frequency band, the preset ratio includes a first preset ratio and a second preset ratio, and the candidate comparison submodule includes: a first ratio determination unit, a second ratio determination unit and a ratio comparison unit, wherein:

A first ratio determination unit, configured to determine a first ratio of a sound energy value of a candidate frequency band to a sound energy value of a first frequency band;

A second ratio determination unit, used to determine a second ratio of the sound energy value of the candidate frequency band to the sound energy value of the second frequency band;

A ratio comparison unit is used to determine that the ratios of the candidate sound energy value corresponding to the candidate frequency band to the sound energy values of other frequency bands exceed the preset ratios if the first ratio exceeds the first preset ratio and the second ratio exceeds the second preset ratio.

Further, the parameter determination module 1040 may include: a second candidate determination submodule and a second candidate denoising submodule, wherein:

A second candidate determination submodule, configured to determine a frequency band with the highest sound energy value from among the multiple frequency bands as a candidate frequency band if a ratio between the sound energy values of the multiple frequency bands matches a preset ratio;

The second candidate noise reduction submodule is used to determine the noise reduction parameters corresponding to the candidate frequency band as target noise reduction parameters.

Furthermore, the preprocessing module 1020 may include: a frequency band division submodule, a filter determination submodule and a time domain filtering submodule, wherein:

A frequency band division submodule, configured to divide the preset frequency range into a plurality of frequency bands to be analyzed according to octaves based on the preset frequency range;

The filter determination submodule is used to determine the corresponding bandpass filter according to the upper limit frequency value and the lower limit frequency value of each frequency band;

The time domain filtering submodule is used to perform bandpass filtering on the noise signal of each frequency band in the time domain based on the bandpass filter corresponding to each frequency band, and obtain the filtered noise signal as the noise signal to be analyzed.

Furthermore, the preprocessing module 1020 may include: a downsampling submodule and a noise acquisition submodule, wherein:

A downsampling submodule, used to downsample the collected ambient sound to obtain the downsampled ambient sound;

The noise acquisition submodule is used to obtain the noise signal to be analyzed according to the downsampled ambient sound.

Furthermore, the preprocessing module 1020 may include: a model determination submodule and a parameter determination submodule, wherein:

A model determination submodule, configured to determine the noise type of the noise signal to be analyzed based on the trained deep learning model and the proportional relationship of the sound energy values of the multiple frequency bands;

The parameter determination submodule is used to determine the noise reduction parameter corresponding to the noise type as the target noise reduction parameter.

Furthermore, the noise reduction processing device 1000 further includes: a smoothing processing module, wherein:

The smoothing processing module is used to perform smoothing processing on the sound energy values of the noise signal to be analyzed in multiple frequency bands.

Further, the pre-processing module 1020 may include: a noise reduction starter module, wherein:

The noise reduction starter module is used to pre-process the ambient sound to obtain the noise signal to be analyzed if the sound energy value of the noise signal in the ambient sound exceeds a preset energy value.

The noise reduction processing device provided in the embodiment of the present application is used to implement the corresponding noise reduction processing method in the aforementioned method embodiment, and has the beneficial effects of the corresponding method embodiment, which will not be repeated here.

In several embodiments provided in the present application, the coupling between modules may be electrical, mechanical or other forms of coupling.

In addition, each functional module in each embodiment of the present application can be integrated into a processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The above integrated modules can be implemented in the form of hardware or software functional modules.

Please refer to Figure 11, which shows a structural block diagram of an electronic device provided in an embodiment of the present application. The electronic device 1100 can be a terminal capable of running applications, such as headphones or a smart phone, a tablet computer, an MP3 player, an MP4 player, an e-book, a laptop computer, a personal computer, a wearable electronic device, etc. The electronic device 1100 in the present application may include one or more of the following components: a processor 1110, a memory 1120, and one or more applications, wherein one or more applications may be stored in the memory 1120 and configured to be executed by one or more processors 1110, and one or more programs are configured to execute the method described in the aforementioned method embodiment.

The processor 1110 may include one or more processing cores. The processor 1110 uses various interfaces and lines to connect various parts of the entire electronic device 1100, and executes various functions and processes data of the electronic device 1100 by running or executing instructions, programs, code sets or instruction sets stored in the memory 1120, and calling data stored in the memory 1120. Optionally, the processor 1110 can be implemented in at least one hardware form of digital signal processing (DSP), field-programmable gate array (FPGA), and programmable logic array (PLA). The processor 1110 can integrate one or a combination of a central processing unit (CPU), a graphics processing unit (GPU), and a modem. Among them, the CPU mainly processes the operating system, user interface, and application programs; the GPU is responsible for rendering and drawing display content; and the modem is used to process wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor 1110, but may be implemented separately through a communication chip.

The memory 1120 may include a random access memory (RAM) or a read-only memory (ROM). The memory 1120 may be used to store instructions, programs, codes, code sets or instruction sets. The memory 1120 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for implementing at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the following various method embodiments, etc. The data storage area may also store data (such as a phone book, audio and video data, chat record data) created by the electronic device 1100 during use.

Please refer to FIG. 12 , which shows a block diagram of the structure of the earphone provided in an embodiment of the present application. The earphone 1200 may include an audio acquisition device 1210 , an audio output device 1220 , and an audio signal processing circuit 1230 . Among them:

The audio collection device 1210 is used to collect ambient sound. In some implementations, the audio collection device 1210 may be a microphone or other device capable of collecting audio signals, and is used to collect audio signals and transmit them to the audio signal processing circuit 1220 .

The audio signal processing circuit 1220 is used to obtain the ambient sound collected by the audio collection device; pre-process the ambient sound to obtain a noise signal to be analyzed, wherein the noise signal to be analyzed corresponds to multiple frequency bands; obtain the sound energy values of the noise signal to be analyzed in multiple frequency bands; and determine the corresponding target noise reduction parameters according to the proportional relationship between the sound energy values of the multiple frequency bands.

The audio output device 1230 is used to output an audio signal. As an implementation, the audio output device 1230 can perform noise reduction processing on the ambient sound based on the target noise reduction parameter. In some implementations, the audio output device 1230 can be a speaker or other device that can output an audio signal.

In addition, in some embodiments, the earphone 1200 may also include a power supply circuit, which can provide power to other hardware components. The power supply source may be a battery built into the earphone 1200, power input from an external source, or a power generating device built into the earphone 1200.

The earphone 1200 provided in the embodiment of the present application is used to implement the corresponding noise reduction processing method in the aforementioned method embodiment, and has the beneficial effects of the corresponding method embodiment, which will not be repeated here.

Please refer to Figure 13, which shows a block diagram of a computer-readable storage medium provided in an embodiment of the present application. The computer-readable storage medium 1300 stores program codes, which can be called by a processor to execute the method described in the above embodiment.

The computer readable storage medium 1300 may be an electronic memory such as a flash memory, an EEPROM (electrically erasable programmable read-only memory), an EPROM, a hard disk, or a ROM. Optionally, the computer readable storage medium 1300 includes a non-transitory computer-readable storage medium. The computer readable storage medium 1300 has storage space for program code 1310 that performs any method step of the above method. These program codes can be read from or written to one or more computer program products. The program code 1310 can be compressed, for example, in an appropriate form.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims (10)

1. A noise reduction processing method, the method comprising:

acquiring environmental sounds acquired by an audio acquisition device, wherein the environmental sounds comprise noise signals;

preprocessing the environmental sound to obtain a noise signal to be analyzed, wherein the noise signal to be analyzed corresponds to a plurality of frequency bands;

acquiring sound energy values of the noise signals to be analyzed in a plurality of frequency bands;

determining a frequency band having the highest sound energy value from the plurality of frequency bands as a candidate frequency band;

determining whether the ratio of the sound energy values of the candidate sound energy values corresponding to the candidate frequency bands to the sound energy values of other frequency bands exceeds a preset ratio, wherein partial frequency bands which are not overlapped with the candidate frequency bands exist;

if the noise reduction parameters are all exceeded, determining the noise reduction parameters corresponding to the candidate frequency bands as target noise reduction parameters;

if the ratio is not exceeded, combining other frequency bands with the ratio which is not exceeded by the preset ratio with the candidate frequency band to serve as a target frequency band;

determining the noise reduction parameters corresponding to the target frequency bands as target noise reduction parameters;

and carrying out noise reduction processing on the environmental sound based on the target noise reduction parameter.

2. The method of claim 1, wherein the other frequency bands include a first frequency band and a second frequency band that are consecutive, the preset ratio includes a first preset ratio and a second preset ratio, and the determining whether the ratio of the candidate sound energy value corresponding to the candidate frequency band to the sound energy value of the other frequency band exceeds the preset ratio includes:

Determining a first ratio of sound energy values of the candidate frequency bands to sound energy values of the first frequency band;

determining a second ratio of the sound energy value of the candidate frequency band to the sound energy value of the second frequency band;

if the first ratio exceeds a first preset ratio and the second ratio exceeds a second preset ratio, judging that the ratio of the candidate sound energy value corresponding to the candidate frequency band to the sound energy values of other frequency bands exceeds the preset ratio.

3. The method according to claim 1 or 2, wherein the preprocessing of the ambient sound to obtain a noise signal to be analyzed comprises:

dividing a preset frequency range into a plurality of frequency bands to be analyzed according to octaves based on the preset frequency range;

determining a corresponding band-pass filter according to the upper limit frequency value and the lower limit frequency value of each frequency band;

and carrying out band-pass filtering processing on the noise signal of each frequency band in the time domain based on the band-pass filter corresponding to each frequency band, and obtaining the noise signal after filtering as the noise signal to be analyzed.

4. The method of claim 1, wherein the preprocessing the ambient sound to obtain a noise signal to be analyzed comprises:

Performing downsampling processing on the acquired environmental sound to obtain downsampled environmental sound;

and obtaining a noise signal to be analyzed according to the down-sampled environmental sound.

5. The method of claim 1, wherein the acquiring the noise signal to be analyzed is followed by sound energy values for a plurality of frequency bands, the method further comprising:

and smoothing the sound energy values of the noise signal to be analyzed in a plurality of frequency bands.

6. The method of claim 1, wherein the preprocessing the ambient sound to obtain a noise signal to be analyzed comprises:

if the sound energy value of the noise signal in the environmental sound exceeds the preset energy value, preprocessing the environmental sound to obtain the noise signal to be analyzed.

7. A noise reduction processing apparatus, characterized in that the apparatus comprises:

the audio acquisition module is used for acquiring the environmental sound acquired by the audio acquisition device, wherein the environmental sound comprises a noise signal;

the preprocessing module is used for preprocessing the environmental sound to obtain a noise signal to be analyzed, and the noise signal to be analyzed corresponds to a plurality of frequency bands;

The energy acquisition module is used for acquiring sound energy values of the noise signal to be analyzed in a plurality of frequency bands;

a parameter determination module configured to determine a frequency band having a highest sound energy value from the plurality of frequency bands as a candidate frequency band; determining whether the ratio of the sound energy values of the candidate sound energy values corresponding to the candidate frequency bands to the sound energy values of other frequency bands exceeds a preset ratio, wherein partial frequency bands which are not overlapped with the candidate frequency bands exist; if the noise reduction parameters are all exceeded, determining the noise reduction parameters corresponding to the candidate frequency bands as target noise reduction parameters; if the ratio is not exceeded, combining other frequency bands with the ratio which is not exceeded by the preset ratio with the candidate frequency band to serve as a target frequency band; determining the noise reduction parameters corresponding to the target frequency bands as target noise reduction parameters;

and the noise reduction processing module is used for carrying out noise reduction processing on the environmental sound based on the target noise reduction parameters.

8. An earphone, characterized by including audio acquisition device, audio output device and audio signal processing circuit, wherein:

the audio acquisition device is used for acquiring environmental sounds;

the audio signal processing circuit is used for acquiring the environmental sound acquired by the audio acquisition device; preprocessing the environmental sound to obtain a noise signal to be analyzed, wherein the noise signal to be analyzed corresponds to a plurality of frequency bands; acquiring sound energy values of the noise signals to be analyzed in a plurality of frequency bands; determining a frequency band having the highest sound energy value from the plurality of frequency bands as a candidate frequency band; determining whether the ratio of the sound energy values of the candidate sound energy values corresponding to the candidate frequency bands to the sound energy values of other frequency bands exceeds a preset ratio, wherein partial frequency bands which are not overlapped with the candidate frequency bands exist; if the noise reduction parameters are all exceeded, determining the noise reduction parameters corresponding to the candidate frequency bands as target noise reduction parameters; if the ratio is not exceeded, combining other frequency bands with the ratio which is not exceeded by the preset ratio with the candidate frequency band to serve as a target frequency band; determining the noise reduction parameters corresponding to the target frequency bands as target noise reduction parameters;

The audio output device is used for carrying out noise reduction processing on the environmental sound based on the target noise reduction parameter.

9. An electronic device, comprising:

one or more processors;

a memory;

one or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the one or more processors, the one or more applications configured to perform the method of any of claims 1-6.

10. A computer readable storage medium, characterized in that the computer readable storage medium has stored therein a program code, which is callable by a processor for executing the method of any one of the claims 1-6.