CN116312606A - High-frequency noise suppression method and device, terminal equipment and storage medium - Google Patents
High-frequency noise suppression method and device, terminal equipment and storage medium Download PDFInfo
- Publication number
- CN116312606A CN116312606A CN202310238047.0A CN202310238047A CN116312606A CN 116312606 A CN116312606 A CN 116312606A CN 202310238047 A CN202310238047 A CN 202310238047A CN 116312606 A CN116312606 A CN 116312606A
- Authority
- CN
- China
- Prior art keywords
- noise
- audio signal
- target
- curve
- frequency band
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 230000001629 suppression Effects 0.000 title claims abstract description 110
- 238000000034 method Methods 0.000 title claims abstract description 57
- 230000005236 sound signal Effects 0.000 claims abstract description 203
- 238000012545 processing Methods 0.000 claims abstract description 35
- 238000012549 training Methods 0.000 claims description 44
- 238000001228 spectrum Methods 0.000 claims description 36
- 230000005764 inhibitory process Effects 0.000 claims description 18
- 238000010183 spectrum analysis Methods 0.000 claims description 12
- 238000000605 extraction Methods 0.000 claims description 7
- 238000010586 diagram Methods 0.000 description 9
- 230000003595 spectral effect Effects 0.000 description 9
- 238000004891 communication Methods 0.000 description 7
- 230000011218 segmentation Effects 0.000 description 4
- 238000004590 computer program Methods 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
- G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
- G10L21/0208—Noise filtering
- G10L21/0216—Noise filtering characterised by the method used for estimating noise
- G10L21/0232—Processing in the frequency domain
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
- G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
- G10L21/0208—Noise filtering
- G10L21/0216—Noise filtering characterised by the method used for estimating noise
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Quality & Reliability (AREA)
- Signal Processing (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Noise Elimination (AREA)
Abstract
The invention discloses a method, a device, terminal equipment and a computer readable storage medium for suppressing high-frequency noise, wherein the method obtains a target audio signal, inputs the target audio signal into a target noise model for classification, and obtains the noise type of the target audio signal; and when the noise type is burst noise, performing multi-section suppression processing on the target audio signal based on a noise multi-section suppression curve to obtain a noise reduction audio signal, wherein the noise multi-section suppression curve is obtained by performing sectional processing on a preset noise suppression default curve. The invention provides a high-frequency noise suppression scheme for performing sectional suppression on high-frequency burst noise so as to effectively suppress the high-frequency burst noise and ensure the voice call quality.
Description
Technical Field
The present invention relates to the field of audio processing technologies, and in particular, to a method and apparatus for suppressing high frequency noise, a terminal device, and a computer readable storage medium.
Background
With the continuous development of technology, people's life can hardly leave the electronic equipment.
At present, people have the requirements and habits of voice communication in the occasions such as at home, on roads, on subways, or in companies, but sudden noise, such as car whistling sounds on roads, broadcasting sounds on subways, door closing warning sounds and the like, inevitably exists in the occasions, and the sudden noise affects the voice quality of a user when the user uses electronic equipment to perform voice communication, generally, the sudden noise or harmonics thereof are mainly concentrated in a middle-high frequency part, and the main characteristic information of human voice is mainly concentrated in a middle-low frequency part, so that the voice quality is guaranteed while the high-frequency sudden noise is effectively restrained, and the technical problem to be solved in the technical field of audio processing is urgent.
Disclosure of Invention
The main object of the present invention is to provide a method, an apparatus, a terminal device and a computer readable storage medium for suppressing high frequency noise. The high-frequency burst noise suppression scheme aims to provide a high-frequency noise suppression scheme for performing sectional suppression on high-frequency burst noise so as to effectively suppress the high-frequency burst noise and ensure the voice call quality.
In order to achieve the above object, the present invention provides a method of suppressing high frequency noise, the method of suppressing high frequency noise including:
acquiring a target audio signal, inputting the target audio signal into a target noise model for classification, and obtaining the noise type of the target audio signal;
and when the noise type is burst noise, performing multi-section suppression processing on the target audio signal based on a noise multi-section suppression curve to obtain a noise reduction audio signal, wherein the noise multi-section suppression curve is obtained by performing sectional processing on a preset noise suppression default curve.
Optionally, the method further comprises:
segmenting a preset noise suppression default curve according to each preset frequency band to obtain each frequency band curve;
determining each target frequency band curve based on the frequency band curves and the noise reduction coefficients corresponding to the preset frequency bands;
and combining the target frequency band curves according to the frequency bands corresponding to the target frequency band curves to obtain the noise multi-section inhibition target curve.
Optionally, the step of determining each target frequency band curve based on the frequency band curves and the noise reduction coefficients corresponding to the preset frequency bands includes:
multiplying the frequency band curve corresponding to the preset frequency band by the noise reduction coefficient corresponding to the preset frequency band to obtain a target frequency band curve.
Optionally, the higher the preset frequency band is, the smaller the noise reduction coefficient corresponding to the preset frequency band is.
Optionally, the step of performing multi-segment suppression processing on the target audio signal based on the noise multi-segment suppression curve to obtain a noise reduction audio signal includes:
and multiplying the noise multi-section inhibition curve by the target audio signal to obtain a noise reduction audio signal.
Optionally, the method further comprises:
and establishing an audio signal training set based on a plurality of audio signals with known noise types, and performing model training on a pre-constructed initial noise model based on the audio signal training set to obtain the target noise model.
Optionally, the step of performing model training on the pre-constructed initial noise model based on the audio signal training set to obtain the target noise model includes:
inputting a target audio signal in the audio signal training set into a pre-constructed initial noise model;
extracting voice characteristics of the target audio signal through a characteristic extraction module of the initial noise model, and separating out noise signals in the target audio signal;
performing spectrum analysis on the noise signal through a spectrum analysis module of the initial noise model, determining a spectrum range of the noise signal, and determining a noise type of the noise signal based on the spectrum range;
and adjusting model parameters of the initial noise model according to the noise type to obtain a target noise model.
In addition, in order to achieve the above object, the present invention also provides a high-frequency noise suppressing apparatus including:
the noise type module is used for acquiring a target audio signal, inputting the target audio signal into a target noise model for classification, and obtaining the noise type of the target audio signal;
and the multi-section suppression module is used for performing multi-section suppression processing on the target audio signal based on a noise multi-section suppression curve when the noise type is burst noise to obtain a noise reduction audio signal, wherein the noise multi-section suppression curve is obtained by performing sectional processing on a preset noise suppression default curve.
In addition, to achieve the above object, the present invention also provides a terminal device including: the method comprises the steps of a memory, a processor and a high-frequency noise suppression program stored in the memory and capable of running on the processor, wherein the high-frequency noise suppression program of the terminal equipment is executed by the processor to realize the high-frequency noise suppression method.
In addition, in order to achieve the above object, the present invention also provides a computer-readable storage medium having stored thereon a high-frequency noise suppressing program which, when executed by a processor, implements the steps of the high-frequency noise suppressing method as described above.
The embodiment of the invention provides a method, a device, terminal equipment and a computer readable storage medium for suppressing high-frequency noise, wherein the method obtains a noise type of a target audio signal by acquiring the target audio signal, inputting the target audio signal into a target noise model for classification; and when the noise type is burst noise, performing multi-section suppression processing on the target audio signal based on a noise multi-section suppression curve to obtain a noise reduction audio signal, wherein the noise multi-section suppression curve is obtained by performing sectional processing on a preset noise suppression default curve.
According to the embodiment of the invention, the target audio signal recorded by the earphone microphone is detected, the target audio signal is input into the target noise model for classification, the noise type of the target audio signal is obtained, then when the noise type is burst noise, the target audio signal is subjected to multi-section inhibition processing based on a noise multi-section inhibition curve, and the noise reduction audio signal is obtained, wherein the noise multi-section inhibition curve is obtained by carrying out sectional processing on a preset noise inhibition default curve. Thus, the invention provides a high-frequency noise suppression scheme for performing sectional suppression on high-frequency burst noise so as to effectively suppress the high-frequency burst noise and ensure the voice call quality.
Drawings
FIG. 1 is a schematic device architecture diagram of a hardware operating environment of a terminal device according to an embodiment of the present invention;
FIG. 2 is a flowchart illustrating a high frequency noise suppression method according to a first embodiment of the present invention;
FIG. 3 is a flowchart of a frequency domain noise multi-segment suppression algorithm according to an embodiment of the high frequency noise suppression method of the present invention;
FIG. 4 is a schematic diagram of a noise environment spectrum according to an embodiment of the high frequency noise suppression method of the present invention;
FIG. 5 is a schematic diagram of a noise reduction spectrum according to an embodiment of the high frequency noise suppression method of the present invention;
fig. 6 is a schematic functional block diagram of an embodiment of a high-frequency noise suppression device according to the present invention.
The achievement of the objects, functional features and advantages of the present invention will be further described with reference to the accompanying drawings, in conjunction with the embodiments.
Detailed Description
It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
Referring to fig. 1, fig. 1 is a schematic device structure diagram of a hardware running environment of a terminal device according to an embodiment of the present invention.
It should be noted that, the terminal device in the embodiment of the present invention may be a left and right earphone, or a terminal device integrated with a system composed of left and right earphones, which is applied to the technical field of earphone audio. Specifically, the terminal device may be a smart phone, a PC (PerSonal Computer ), a tablet computer, a portable computer, or the like. There is no particular limitation herein.
As shown in fig. 1, the terminal device may include: a processor 1001, such as a CPU, a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Wherein the communication bus 1002 is used to enable connected communication between these components. The user interface 1003 may include a DiSplay (diselay), an input unit such as a Keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface, a wireless interface. The network interface 1004 may optionally include a standard wired interface, a wireless interface (e.g., wi-Fi interface). The memory 1005 may be a high-speed RAM memory or a stable memory (non-volatile memory), such as a disk memory. The memory 1005 may also optionally be a storage device separate from the processor 1001 described above.
It will be appreciated by those skilled in the art that the terminal device structure shown in fig. 1 is not limiting of the terminal device and may include more or fewer components than shown, or may combine certain components, or a different arrangement of components.
As shown in fig. 1, an operating system, a network communication module, a user interface module, and a high-frequency noise suppressing program may be included in a memory 1005 as one type of computer storage medium.
In the terminal shown in fig. 1, the network interface 1004 is mainly used for connecting to a background server and performing data communication with the background server; the user interface 1003 is mainly used for connecting a client and communicating data with the client; and the processor 1001 may be configured to call a high-frequency noise suppression program stored in the memory 1005, and perform the following operations:
acquiring a target audio signal, inputting the target audio signal into a target noise model for classification, and obtaining the noise type of the target audio signal;
and when the noise type is burst noise, performing multi-section suppression processing on the target audio signal based on a noise multi-section suppression curve to obtain a noise reduction audio signal, wherein the noise multi-section suppression curve is obtained by performing sectional processing on a preset noise suppression default curve.
Further, the processor 1001 may be further configured to call a high-frequency noise suppression program stored in the memory 1005 to perform the following operations:
segmenting a preset noise suppression default curve according to each preset frequency band to obtain each frequency band curve;
determining each target frequency band curve based on the frequency band curves and the noise reduction coefficients corresponding to the preset frequency bands;
and combining the target frequency band curves according to the frequency bands corresponding to the target frequency band curves to obtain the noise multi-section inhibition target curve.
Further, the determining the target frequency band curves based on the frequency band curves and the noise reduction coefficients corresponding to the preset frequency bands respectively includes:
multiplying the frequency band curve corresponding to the preset frequency band by the noise reduction coefficient corresponding to the preset frequency band to obtain a target frequency band curve.
Further, the higher the preset frequency band is, the smaller the noise reduction coefficient corresponding to the preset frequency band is.
Further, the performing multi-segment suppression processing on the target audio signal based on the noise multi-segment suppression curve to obtain a noise reduction audio signal includes:
and multiplying the noise multi-section inhibition curve by the target audio signal to obtain a noise reduction audio signal.
Further, the processor 1001 may be further configured to call a high-frequency noise suppression program stored in the memory 1005 to perform the following operations:
and establishing an audio signal training set based on a plurality of audio signals with known noise types, and performing model training on a pre-constructed initial noise model based on the audio signal training set to obtain the target noise model.
Further, the performing model training on the pre-constructed initial noise model based on the audio signal training set, and the obtaining the target noise model includes:
inputting a target audio signal in the audio signal training set into a pre-constructed initial noise model;
extracting voice characteristics of the target audio signal through a characteristic extraction module of the initial noise model, and separating out noise signals in the target audio signal;
performing spectrum analysis on the noise signal through a spectrum analysis module of the initial noise model, determining a spectrum range of the noise signal, and determining a noise type of the noise signal based on the spectrum range;
and adjusting model parameters of the initial noise model according to the noise type to obtain a target noise model.
Based on the above-described structure, various embodiments of a method of suppressing high-frequency noise are proposed.
Referring to fig. 2, fig. 2 is a flowchart illustrating a method for suppressing high frequency noise according to a first embodiment of the present invention. It should be noted that although a logic sequence is shown in the flowchart, the high-frequency noise suppressing method of the present invention may, of course, also perform the steps shown or described in a different order than that here in some cases. In this embodiment, the execution body of the high-frequency noise suppression method may be devices such as an earphone, a personal computer, and a smart phone, and is not limited in this embodiment, and for convenience of description, the execution body is omitted from description of each embodiment. In this embodiment, the method for suppressing high-frequency noise includes:
step S10, obtaining a target audio signal, inputting the target audio signal into a target noise model for classification, and obtaining the noise type of the target audio signal;
the method comprises the steps of acquiring real-time audio signals (hereinafter referred to as target audio signals for distinguishing) received by an earphone microphone, inputting the target audio signals into a pre-trained noise model (hereinafter referred to as target noise model for distinguishing) for classification, and obtaining the noise type of the target audio signals based on the target noise model.
In a possible implementation manner, based on the detected target audio signal acquired by the earphone microphone, the target audio signal is input into a target noise model, and the target noise model outputs the noise type of the target audio signal according to the input target audio signal, wherein the noise type of the audio signal can be but not limited to wind noise, road noise and burst noise, and the burst noise can be but not limited to door closing alarm prompt sound on a subway or the like.
And step S20, when the noise type is burst noise, performing multi-section suppression processing on the target audio signal based on a noise multi-section suppression curve to obtain a noise reduction audio signal, wherein the noise multi-section suppression curve is obtained by performing segmentation processing on a preset noise suppression default curve.
When the noise type of the target audio signal is detected as burst noise, the target audio signal is subjected to multi-segment suppression processing, and a processed audio signal (hereinafter referred to as a noise reduction audio signal to show a distinction) is generated, wherein a noise multi-segment suppression curve is obtained by performing segment processing on a preset suppression curve (hereinafter referred to as a noise suppression default curve to show a distinction).
In a possible implementation manner, when the noise type output by the target noise model is detected to be burst noise, it is determined that the target audio signal detected at the moment contains the burst noise, and then fourier transformation is performed on the target audio signal to obtain a frequency domain audio signal. Performing multi-section suppression processing on a frequency domain audio signal corresponding to the target audio signal based on a preset noise suppression target curve, generating a processed frequency domain noise reduction signal, converting the frequency domain noise reduction signal into a time domain noise reduction signal according to an inverse Fourier transform principle, and playing the time domain noise reduction signal through an earphone loudspeaker after the time domain noise reduction signal is obtained, namely replacing the target audio signal with the noise type of burst noise by the time domain noise reduction signal in real time. The time domain noise reduction signal is a noise reduction audio signal.
It should be noted that the target audio signal belongs to a time-domain signal, and the frequency-domain audio signal belongs to a frequency-domain signal, that is, the target audio signal may be a time-domain audio curve, and the frequency-domain audio signal may be a frequency-domain audio curve. Fourier transforms are mainly used to transform a real time domain signal into a virtual one of the mathematical structures, i.e. the frequency domain.
In a possible embodiment, as shown in fig. 4, the spectrum diagram of the noise environment is a spectrum diagram when the noise reduction processing is not performed in the noise environment, based on the upper half of fig. 4, it can be seen that noise is relatively high except that obvious human voice exists, based on the lower half of fig. 4, noise is present from 0Hz to 15KHz, as shown in fig. 5, the spectrum diagram of the noise reduction effect is shown, based on the upper half of fig. 5, three segments of signals are shown in fig. 5, three words spoken and/or heard by a user during a voice call can be understood, the interval between the three segments is blank when no voice exists, the noise has been reduced to be noiseless, based on the lower half of fig. 5, the audio signal is obviously weakened compared with the lower half of fig. 4, and the noise reduction effect is realized.
Further, in a possible embodiment, in the step S20, the step of performing the multi-segment suppression processing on the target audio signal based on the noise multi-segment suppression curve to obtain the noise reduction audio signal includes:
step S201, multiplying the noise multi-section suppression curve by the target audio signal to obtain a noise reduction audio signal.
And performing curve fitting on a preset noise multi-section suppression curve and a target audio signal to obtain a noise reduction audio signal after fitting.
It should be noted that, in this embodiment, the amplitude range of the default noise suppression curve is between 0 and 1.
In one possible implementation, the noise multi-segment suppression curve is multiplied by the target audio signal, thereby acting to attenuate the frequency domain audio signal and obtaining a noise-reduced audio signal after noise reduction.
Further, in a possible embodiment, the method for suppressing high-frequency noise of the present invention further includes:
step A10, segmenting a preset noise suppression default curve according to each preset frequency band to obtain each frequency band curve;
the preset noise curves (hereinafter referred to as noise suppression default curves for distinguishing) are subjected to frequency band division according to preset frequency bands (hereinafter referred to as preset frequency bands for distinguishing), and a plurality of divided noise curves (hereinafter referred to as frequency band curves for distinguishing) are obtained.
In one possible implementation, the preset frequency band may be 0Hz to 2KHz, 2KHz to 4KHz, and 4KHz to 8KHz, and the noise suppression default curve is determined according to factory parameters of the earphone, so that the preset noise suppression default curve is divided according to the preset frequency band, and frequency band curves of three frequency bands can be obtained.
It should be noted that, the present solution does not limit the manner of obtaining the default noise suppression curve.
Step A20, determining each target frequency band curve based on each frequency band curve and the noise reduction coefficient corresponding to each preset frequency band;
and determining the noise reduction coefficient corresponding to each frequency band curve according to each frequency band curve, and calculating to obtain each target frequency band curve based on each frequency band curve and each noise reduction coefficient.
In a possible implementation manner, the noise reduction coefficient corresponding to the frequency band curve with the frequency band of 0Hz to 2KHz is set to be 1, the noise reduction coefficient corresponding to the frequency band curve with the frequency band of 2KHz to 4KHz is set to be 0.005, the noise reduction coefficient corresponding to the frequency band curve with the frequency band of 4KHz to 8KHz is set to be 0.002, and then each target frequency band curve is obtained according to the noise reduction coefficient corresponding to each preset frequency band curve. It should be understood that the frequency band of 0Hz to 2KHz is mainly distributed with human voice, so it is not preferable to suppress noise too much to cause voice damage, so the noise reduction coefficient of the frequency band is set to 1, and the higher the frequency of the frequency band is, the smaller the noise reduction coefficient is to achieve a greater noise reduction intensity for noise of higher frequency.
It should be noted that, the method for suppressing high-frequency noise of the present invention is not limited to the preset frequency band and the number of frequency bands, and it should be understood that, based on different design requirements of practical applications, in different feasible embodiments, the preset frequency band may be any range meeting practical requirements, and the number of frequency bands may be any number meeting practical requirements.
And step A30, combining the target frequency band curves according to the corresponding frequency bands to obtain the noise multi-section inhibition target curve.
And combining the target frequency band curves according to the frequency bands corresponding to the target frequency band curves, and obtaining the finished noise multi-section inhibition target curve after the combination is finished.
In a possible implementation manner, the three frequency band standard curves are connected in sequence according to respective frequency bands, specifically, the right end point of the first frequency band standard curve is connected with the left end point of the second frequency band standard curve, and the right end point of the second frequency band standard curve is connected with the left end point of the third frequency band standard curve, so that a complete noise multi-segment suppression target curve containing three frequency bands of 0Hz to 2KHz, 2KHz to 4KHz and 4KHz to 8KHz is obtained.
In a possible implementation manner, as shown in fig. 3, a flow chart of a frequency domain noise multi-section suppression algorithm is shown, firstly, a target audio signal received by a headset microphone is detected, then, a noise model is utilized to perform type estimation on the audio signal, a noise type of the target audio signal is determined, then, when the noise type is burst noise, FFT (Fast Fourier transform ) is performed on the target audio signal, a frequency domain audio signal for calculation is obtained after fourier transform, a preset noise multi-section suppression target curve is utilized to fit the frequency domain audio signal, namely, two curves are multiplied to obtain the frequency domain audio signal, finally, IFFT (Inverse Fast Fourier Transform ) is performed on the frequency domain audio signal to convert the frequency domain audio signal into a time domain audio signal, and the time domain audio signal is played through a headset speaker, so that noise reduction of the burst noise is completed.
Further, in a possible embodiment, the step a20 includes:
and step A201, multiplying a frequency band curve corresponding to the preset frequency band by a noise reduction coefficient corresponding to the preset frequency band to obtain a target frequency band curve.
Multiplying each frequency band curve by the noise reduction coefficient corresponding to each frequency band curve to obtain each target frequency band curve.
In one possible implementation, a frequency band curve with a frequency band of 0Hz to 2KHz is multiplied by a noise reduction coefficient 1 to obtain a first item of standard frequency band curve, a frequency band curve with a frequency band of 2KHz to 4KHz is multiplied by a noise reduction coefficient 0.005 to obtain a second item of standard frequency band curve, and a frequency band curve with a frequency band of 4KHz to 8KHz is multiplied by a noise reduction coefficient 0.002 to obtain a third item of standard frequency band curve.
Further, the higher the frequency band is, the smaller the noise reduction coefficient corresponding to the preset frequency band is.
In one possible embodiment, the higher the frequency band of the audio signal, the smaller the preset noise reduction coefficient of the frequency band, and the greater the noise reduction strength.
In the embodiment, the method for suppressing high-frequency noise acquires a real-time target audio signal received by an earphone microphone, inputs the target audio signal into a pre-trained target noise model for classification, and obtains the noise type of the target audio signal based on the target noise model; when the noise type of the target audio signal is detected to be burst noise, performing multi-section inhibition processing on the target audio signal to generate a processed noise reduction audio signal, wherein a noise multi-section inhibition curve is obtained by performing sectional processing on a preset noise inhibition default curve; performing curve fitting on a preset noise multi-section suppression curve and a target audio signal to obtain a noise reduction audio signal after fitting; frequency band segmentation is carried out on a preset noise suppression default curve according to a preset frequency band, so that a plurality of segmented frequency band curves are obtained; determining the noise reduction coefficient corresponding to each frequency band curve according to each frequency band curve, and calculating to obtain each target frequency band curve based on each frequency band curve and each noise reduction coefficient; combining the target frequency band curves according to the frequency bands corresponding to the target frequency band curves, and obtaining a finished noise multi-section inhibition target curve after the combination is finished; multiplying each frequency band curve by the noise reduction coefficient corresponding to each frequency band curve to obtain each target frequency band curve.
In this way, the embodiment of the invention detects the target audio signal recorded by the earphone microphone, inputs the target audio signal into the target noise model for classification to obtain the noise type of the target audio signal, and then carries out multi-section suppression processing on the target audio signal based on the noise multi-section suppression curve when the noise type is burst noise to obtain the noise-reduction audio signal, wherein the noise multi-section suppression curve is obtained by carrying out sectional processing on the preset noise suppression default curve. Thus, the invention provides a high-frequency noise suppression scheme for performing sectional suppression on high-frequency burst noise so as to effectively suppress the high-frequency burst noise and ensure the voice call quality.
Further, based on the first embodiment of the high-frequency noise suppressing method of the present invention described above, a second embodiment of the high-frequency noise suppressing method of the present invention is proposed.
In this embodiment, the method for suppressing high-frequency noise of the present invention further includes:
and step B10, an audio signal training set is established based on a plurality of audio signals with known noise types, and model training is carried out on a pre-established initial noise model based on the audio signal training set to obtain the target noise model.
And establishing an audio signal training set, and carrying out model training on a pre-constructed noise model (hereinafter referred to as an initial noise model to show and distinguish) according to the audio signal training set to obtain a trained target noise model.
In a possible implementation manner, the audio signal training set is composed of a plurality of audio signals, wherein each audio signal includes but is not limited to a wind noise type, a road noise type, a burst noise type and other audio signals, and model training is performed on a pre-built initial noise model based on the audio signal training set, wherein the initial noise model is an untrained model structure, and after model training is finished, a target noise model is obtained, wherein the target noise model is used for outputting the noise type of the audio signal according to an input audio signal.
Further, in a possible embodiment, in the step B10, the step of "model training the pre-constructed initial noise model based on the audio signal training set to obtain the target noise model" includes:
step B101, inputting a target audio signal in the audio signal training set into a pre-constructed initial noise model;
audio signals in the audio signal training set (hereinafter referred to as target audio signals to show distinction) are input to the initial noise model constructed in advance.
It should be noted that, the method for suppressing high-frequency noise of the present invention does not limit how to determine the spectral energy of the audio signal.
Step B102, extracting the voice characteristics of the target audio signal through a characteristic extraction module of the initial noise model, and separating out the noise signal in the target audio signal;
and extracting the voice characteristics of the target audio signal through a characteristic extraction module of the initial noise model, and separating out the noise signal of the target audio signal.
In a possible embodiment, the speech signal in the target audio signal is determined by extracting speech features in the target audio signal, so that noise signals in the target audio signal other than the speech signal are separated.
Step B103, performing spectrum analysis on the noise signal through a spectrum analysis module of the initial noise model, determining a spectrum range of the noise signal, and determining a noise type of the noise signal based on the spectrum range;
determining a spectrum energy range of the spectrum energy of the target audio signal through a spectrum analysis module of the initial noise model, detecting whether the spectrum energy range corresponding to the spectrum energy of the target audio signal is a preset spectrum energy range (hereinafter referred to as a preset spectrum energy range for distinguishing), and determining that the noise type of the target audio signal is burst noise when the spectrum energy range corresponding to the spectrum energy of the target audio signal is the preset spectrum energy range.
In one possible implementation, each spectral energy range is preset in the initial noise model, the spectral energy range in which the spectral energy of the high-frequency audio signal is located in each spectral energy range is taken as the preset spectral energy range, and the high-frequency audio signal is the burst audio signal, and then when the spectral energy range corresponding to the spectral energy of the target audio signal is the preset spectral energy range, the noise type of the target audio signal is determined to be the burst noise.
And step B104, adjusting model parameters of the initial noise model according to the noise type to obtain a target noise model.
And adjusting model parameters of the initial noise model according to the noise type obtained by model training to obtain a target noise model.
In the embodiment, the high-frequency noise suppression method comprises the steps of establishing an audio signal training set, and carrying out model training on a pre-established initial noise model according to the audio signal training set to obtain a trained target noise model; inputting a target audio signal in an audio signal training set into a pre-constructed initial noise model; extracting the voice characteristics of the target audio signal through a characteristic extraction module of the initial noise model, and separating out the noise signal of the target audio signal; determining a spectrum energy range of the spectrum energy of the target audio signal through a spectrum analysis module of the initial noise model, detecting whether the spectrum energy range corresponding to the spectrum energy of the target audio signal is a preset spectrum energy range, and determining that the noise type of the target audio signal is burst noise when the spectrum energy range corresponding to the spectrum energy of the target audio signal is the preset spectrum energy range; and adjusting model parameters of the initial noise model according to the noise type obtained by model training to obtain a target noise model.
Thus, whether the target audio signal acquired by the earphone microphone needs to be subjected to multi-stage noise suppression processing is detected by utilizing a pre-constructed noise model, so that key judgment conditions of the multi-stage noise suppression processing are established.
In addition, the embodiment of the invention also provides a device for suppressing high-frequency noise.
Referring to fig. 6, fig. 6 is a functional block diagram of an embodiment of a high-frequency noise suppression device according to the present invention, and as shown in fig. 6, the high-frequency noise suppression device according to the present invention includes:
the noise type module 10 is configured to obtain a target audio signal, input the target audio signal to a target noise model, and classify the target audio signal to obtain a noise type of the target audio signal;
and the multi-section suppression module 20 is configured to perform multi-section suppression processing on the target audio signal based on a multi-section suppression curve of noise when the noise type is burst noise, so as to obtain a noise reduction audio signal, where the multi-section suppression curve of noise is obtained by performing segmentation processing on a preset default noise suppression curve.
Further, the high-frequency noise suppression device of the present invention further includes:
the segmentation module is used for segmenting a preset noise suppression default curve according to each preset frequency band to obtain each frequency band curve;
the target frequency band curve module is used for determining each target frequency band curve based on each frequency band curve and each noise reduction coefficient corresponding to each preset frequency band;
and the combination module is used for combining the target frequency band curves according to the corresponding frequency bands to obtain the noise multi-section inhibition target curve.
Further, the target frequency band curve module includes:
and the target frequency band curve unit is used for multiplying the frequency band curve corresponding to the preset frequency band by the noise reduction coefficient corresponding to the preset frequency band to obtain a target frequency band curve.
Further, the higher the preset frequency band is, the smaller the noise reduction coefficient corresponding to the preset frequency band is
Further, the multi-segment suppression module 20 further includes:
and the noise reduction audio signal module is used for multiplying the noise multi-section inhibition curve with the target audio signal to obtain a noise reduction audio signal.
Further, the high-frequency noise suppression device of the present invention further includes:
the model training module is used for establishing an audio signal training set based on a plurality of audio signals with known noise types, and carrying out model training on a pre-constructed initial noise model based on the audio signal training set to obtain the target noise model.
Further, the model training module is further used for inputting the target audio signals in the audio signal training set into a pre-constructed initial noise model; extracting voice characteristics of the target audio signal through a characteristic extraction module of the initial noise model, and separating out noise signals in the target audio signal; performing spectrum analysis on the noise signal through a spectrum analysis module of the initial noise model, determining a spectrum range of the noise signal, and determining a noise type of the noise signal based on the spectrum range; and adjusting model parameters of the initial noise model according to the noise type to obtain a target noise model.
Further, the target noise model unit is further configured to detect whether the spectrum energy range is a preset spectrum energy range; and when the spectrum energy range is the preset spectrum energy range, determining that the noise type of the target audio signal is burst noise.
The present invention also provides a computer storage medium having stored thereon a high-frequency noise suppression program which, when executed by a processor, implements the steps of the high-frequency noise suppression program method according to any one of the above embodiments.
The specific embodiments of the computer storage medium of the present invention are substantially the same as the embodiments of the method for suppressing high frequency noise of the present invention described above, and will not be described herein.
The present invention also provides a computer program product, which comprises a computer program, wherein the computer program, when executed by a processor, implements the steps of the high frequency noise suppression method according to any one of the above embodiments, which is not described herein.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or system. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or system that comprises the element.
The foregoing embodiment numbers of the present invention are merely for the purpose of description, and do not represent the advantages or disadvantages of the embodiments.
From the above description of the embodiments, it will be clear to those skilled in the art that the above-described embodiment method may be implemented by means of software plus a necessary general hardware platform, but of course may also be implemented by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (e.g. ROM/RAM, magnetic disk, optical disk) as described above, comprising instructions for causing a terminal device (which may be a TWS headset or the like) to perform the method according to the embodiments of the present invention.
The foregoing description is only of the preferred embodiments of the present invention, and is not intended to limit the scope of the invention, but rather is intended to cover any equivalents of the structures or equivalent processes disclosed herein or in the alternative, which may be employed directly or indirectly in other related arts.
Claims (10)
1. A method of suppressing high frequency noise, characterized in that the method of suppressing high frequency noise comprises the steps of:
acquiring a target audio signal, inputting the target audio signal into a target noise model for classification, and obtaining the noise type of the target audio signal;
and when the noise type is burst noise, performing multi-section suppression processing on the target audio signal based on a noise multi-section suppression curve to obtain a noise reduction audio signal, wherein the noise multi-section suppression curve is obtained by performing sectional processing on a preset noise suppression default curve.
2. The method of suppressing high frequency noise according to claim 1, wherein the method further comprises:
segmenting a preset noise suppression default curve according to each preset frequency band to obtain each frequency band curve;
determining each target frequency band curve based on the frequency band curves and the noise reduction coefficients corresponding to the preset frequency bands;
and combining the target frequency band curves according to the frequency bands corresponding to the target frequency band curves to obtain the noise multi-section inhibition target curve.
3. The method of suppressing high frequency noise as defined in claim 2, wherein the step of determining each target frequency band curve based on each of the frequency band curves and the noise reduction coefficients corresponding to each of the preset frequency bands comprises:
multiplying the frequency band curve corresponding to the preset frequency band by the noise reduction coefficient corresponding to the preset frequency band to obtain a target frequency band curve.
4. A method of suppressing high frequency noise as defined in claim 3, wherein the higher the preset frequency band is, the smaller the noise reduction coefficient corresponding to the preset frequency band is.
5. The method of suppressing high frequency noise according to claim 1, wherein the step of performing a multi-stage suppression process on the target audio signal based on a noise multi-stage suppression curve to obtain a noise reduction audio signal comprises:
and multiplying the noise multi-section inhibition curve by the target audio signal to obtain a noise reduction audio signal.
6. The high-frequency noise suppressing method according to any one of claims 1 to 5, characterized in that the method further comprises:
and establishing an audio signal training set based on a plurality of audio signals with known noise types, and performing model training on a pre-constructed initial noise model based on the audio signal training set to obtain the target noise model.
7. The method for suppressing high frequency noise as defined in claim 6, wherein said step of model training a pre-constructed initial noise model based on said audio signal training set to obtain a target noise model comprises:
inputting a target audio signal in the audio signal training set into a pre-constructed initial noise model;
extracting voice characteristics of the target audio signal through a characteristic extraction module of the initial noise model, and separating out noise signals in the target audio signal;
performing spectrum analysis on the noise signal through a spectrum analysis module of the initial noise model, determining a spectrum range of the noise signal, and determining a noise type of the noise signal based on the spectrum range;
and adjusting model parameters of the initial noise model according to the noise type to obtain a target noise model.
8. A high-frequency noise suppressing apparatus, characterized by comprising:
the noise type module is used for acquiring a target audio signal, inputting the target audio signal into a target noise model for classification, and obtaining the noise type of the target audio signal;
and the multi-section suppression module is used for performing multi-section suppression processing on the target audio signal based on a noise multi-section suppression curve when the noise type is burst noise to obtain a noise reduction audio signal, wherein the noise multi-section suppression curve is obtained by performing sectional processing on a preset noise suppression default curve.
9. A terminal device, characterized in that the terminal device comprises: a memory, a processor, and a high-frequency noise suppressing program stored on the memory and executable on the processor, which when executed by the processor, realizes the steps of the high-frequency noise suppressing method according to any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored thereon a high-frequency noise suppressing program which, when executed by a processor, implements the steps of the high-frequency noise suppressing method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310238047.0A CN116312606A (en) | 2023-03-13 | 2023-03-13 | High-frequency noise suppression method and device, terminal equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310238047.0A CN116312606A (en) | 2023-03-13 | 2023-03-13 | High-frequency noise suppression method and device, terminal equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116312606A true CN116312606A (en) | 2023-06-23 |
Family
ID=86782950
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310238047.0A Pending CN116312606A (en) | 2023-03-13 | 2023-03-13 | High-frequency noise suppression method and device, terminal equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116312606A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN119052692A (en) * | 2024-08-15 | 2024-11-29 | 恩平市艺星电子有限公司 | Noise suppression method and system for microphone, microphone and medium |
| CN119485105A (en) * | 2024-11-15 | 2025-02-18 | 广州雷萌科技有限公司 | A digital audio processing method and system for audio |
-
2023
- 2023-03-13 CN CN202310238047.0A patent/CN116312606A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN119052692A (en) * | 2024-08-15 | 2024-11-29 | 恩平市艺星电子有限公司 | Noise suppression method and system for microphone, microphone and medium |
| CN119485105A (en) * | 2024-11-15 | 2025-02-18 | 广州雷萌科技有限公司 | A digital audio processing method and system for audio |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3698360B1 (en) | Noise reduction using machine learning | |
| CN110197670B (en) | Audio noise reduction method and device and electronic equipment | |
| US10373609B2 (en) | Voice recognition method and apparatus | |
| CN103236263B (en) | Method, system and mobile terminal for improving call quality | |
| US9640194B1 (en) | Noise suppression for speech processing based on machine-learning mask estimation | |
| Latif et al. | Adversarial machine learning and speech emotion recognition: Utilizing generative adversarial networks for robustness | |
| JP6764923B2 (en) | Speech processing methods, devices, devices and storage media | |
| CN107995360B (en) | Call processing method and related products | |
| CN116312606A (en) | High-frequency noise suppression method and device, terminal equipment and storage medium | |
| CN112951259B (en) | Audio noise reduction method and device, electronic equipment and computer readable storage medium | |
| US9749741B1 (en) | Systems and methods for reducing intermodulation distortion | |
| CN113949955B (en) | Noise reduction processing method, device, electronic equipment, earphone and storage medium | |
| CN104021798A (en) | Method for soundproofing an audio signal by an algorithm with a variable spectral gain and a dynamically modulatable hardness | |
| JP6439682B2 (en) | Signal processing apparatus, signal processing method, and signal processing program | |
| CN108600893A (en) | Military environments audio classification system, method and military noise cancelling headphone | |
| CN112750451A (en) | Noise reduction method for improving voice listening feeling | |
| CN112992190B (en) | Audio signal processing method, device, electronic device and storage medium | |
| WO2024002896A1 (en) | Audio signal processing method and system for enhancing a bone-conducted audio signal using a machine learning model | |
| CN115348507A (en) | Impulse noise suppression method, system, readable storage medium and computer equipment | |
| CN114333874B (en) | Method for processing audio signal | |
| CN115720317A (en) | Audio signal squeaking detection and suppression method and device | |
| JP2012063614A (en) | Masking sound generation device | |
| JP2012181561A (en) | Signal processing apparatus | |
| CN116980804B (en) | Volume adjustment method, device, equipment and readable storage medium | |
| US11308970B2 (en) | Voice correction apparatus and voice correction method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |