CN110289009A - Sound signal processing method and device and interactive intelligent equipment - Google Patents
Sound signal processing method and device and interactive intelligent equipment Download PDFInfo
- Publication number
- CN110289009A CN110289009A CN201910616082.5A CN201910616082A CN110289009A CN 110289009 A CN110289009 A CN 110289009A CN 201910616082 A CN201910616082 A CN 201910616082A CN 110289009 A CN110289009 A CN 110289009A
- Authority
- CN
- China
- Prior art keywords
- filter
- signal
- coefficient
- processing
- linear
- 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.)
- Granted
Links
- 230000005236 sound signal Effects 0.000 title claims abstract description 108
- 230000002452 interceptive effect Effects 0.000 title claims abstract description 10
- 238000003672 processing method Methods 0.000 title claims description 19
- 238000012545 processing Methods 0.000 claims abstract description 168
- 230000003044 adaptive effect Effects 0.000 claims abstract description 115
- 238000001914 filtration Methods 0.000 claims abstract description 80
- 238000000034 method Methods 0.000 claims abstract description 60
- 230000008569 process Effects 0.000 claims description 29
- 230000017105 transposition Effects 0.000 claims description 15
- 238000001228 spectrum Methods 0.000 claims description 14
- 239000011159 matrix material Substances 0.000 claims description 12
- 230000008901 benefit Effects 0.000 claims description 4
- 230000021615 conjugation Effects 0.000 claims description 2
- 230000001629 suppression Effects 0.000 abstract 5
- 230000000694 effects Effects 0.000 description 12
- 230000002401 inhibitory effect Effects 0.000 description 11
- 230000008859 change Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000006872 improvement Effects 0.000 description 3
- 238000007781 pre-processing Methods 0.000 description 3
- 230000003321 amplification Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 235000013399 edible fruits Nutrition 0.000 description 2
- 230000005611 electricity Effects 0.000 description 2
- 230000005764 inhibitory process Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000003199 nucleic acid amplification method Methods 0.000 description 2
- 230000000644 propagated effect Effects 0.000 description 2
- 206010019133 Hangover Diseases 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000002146 bilateral effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 230000001902 propagating effect Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
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
-
- 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
- G10L2021/02082—Noise filtering the noise being echo, reverberation of the speech
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)
- Circuit For Audible Band Transducer (AREA)
Abstract
The invention discloses a method and a device for processing a sound signal and interactive intelligent equipment, which are applied to the field of signal processing. Wherein, the method comprises the following steps: receiving an original sound signal; performing first filtering processing on an original sound signal based on a nonlinear filter to obtain a first processing result, wherein the first filtering processing is used for performing reverberation suppression on the original sound signal; updating the coefficient of the linear adaptive filter according to the first processing result; and performing second filtering processing on the first processing result based on the linear adaptive filter after the coefficient is updated, and outputting a final processing result of the original sound signal, wherein the second filtering processing is used for performing reverberation suppression on the first processing result. The invention solves the technical problem that when the reverberation suppression is carried out on the sound signal in the prior art, the better suppression effect is difficult to obtain and the lower algorithm complexity is kept, and achieves the purpose of reducing the execution complexity while ensuring the suppression effect.
Description
Technical field
The present invention relates to field of signal processing, in particular to a kind of processing method of voice signal, device and interaction
Smart machine.
Background technique
With the development of multimedia equipment, the requirement propagated acoustic information is higher and higher.For example, teleconference, long-range
In the scenes such as teaching, while propagating image information, it is also necessary to also clearly be propagated acoustic information.But sound is passing
During broadcasting, reverberation can be generated by multiple reflections, can include reverberation in the acoustic information received so as to cause receiving end
Signal.And reverb signal can damage the quality of original signal, therefore also need to inhibit reverb signal, to improve sound
Quality.
The receiving end of voice signal can carry out Reverberation Rejection, but current Reverberation Rejection to the voice signal received at present
Mode haves the defects that complexity is high or inhibitory effect is poor, it is difficult to which the two gets both.
When for carrying out Reverberation Rejection to voice signal in the prior art, it is difficult to while obtaining preferable inhibitory effect
The problem of keeping lower algorithm complexity, currently no effective solution has been proposed.
Summary of the invention
The embodiment of the invention provides a kind of processing method of voice signal, device and interactive intelligence equipment, at least to solve
When certainly carrying out Reverberation Rejection to voice signal in the prior art, it is difficult to keep lower while obtaining preferable inhibitory effect
The technical issues of algorithm complexity.
According to an aspect of an embodiment of the present invention, a kind of processing method of voice signal is provided, comprising: receive primary sound
Sound signal;The first filtering processing is carried out to primary sound sound signal based on nonlinear filter, obtains the first processing result, wherein the
One filtering processing is for carrying out Reverberation Rejection to primary sound sound signal;Linear adaptive filter is updated according to the first processing result
Coefficient;The second filtering processing is carried out to the first processing result based on the linear adaptive filter after update coefficient, exports primary sound
The final process result of sound signal, wherein the second filtering processing is for carrying out Reverberation Rejection to the first processing result.
Further, before receiving primary sound sound signal, initialization process is carried out to the coefficient of linear adaptive filter,
The coefficient that initial time linear adaptive filter is arranged is the sequence of preset length, and the value of each element in sequence is 0.
Further, the signal-to-noise ratio of primary sound sound signal is obtained, wherein signal-to-noise ratio is that primary sound sound signal mid-term hopes signal power
With the ratio of reverb signal power;Nonlinear filter is determined according to signal-to-noise ratio and preset compensation factor;Based on nonlinear filtering
Wave device carries out the first filtering processing to primary sound sound signal.
Further, it is determined that the type of nonlinear filter, comprising: determine that signal-to-noise ratio is less than snr threshold, determine non-
Linear filter is that power spectrum subtracts filter;Or determine that signal-to-noise ratio is greater than or equal to snr threshold, determine nonlinear filter
Subtract filter for amplitude spectrum.
Further, it is determined that the distortion factor of demand is greater than distortion factor threshold value, determine nonlinear filter for wiener filter
Wave device.
Further, the more new gain of linear adaptive filter is obtained;According to linear adaptive filter in upper a period of time
The current coefficient that the coefficient at quarter, more new gain and the first processing result filter linear adaption is updated.
Further, linear adaptive filter is least square method of recursion filter, obtains linear adaptive filter
Current input signal vector, current input signal vector conjugate transposition, preset forgetting factor, current first processing result
Power and linear adaptive filter coefficient covariance matrix;Obtain the of the power of forgetting factor and the first processing result
One product;It obtains the conjugate transposition of current input signal vector, the covariance matrix of linear adaptive filter coefficient and works as
Second product of preceding input signal vector;Obtain linear adaptive filter coefficient covariance matrix and current input signal to
The third product of amount;Obtain first the second product of sum of products and value;Determine third product with and value ratio be more new gain.
Further, the variable quantity of the current coefficient of linear adaptive filter and the coefficient of last moment is obtained, wherein
Variable quantity is the 4th product of the conjugate transposition of more new gain and the first processing result;Determine linear adaptive filter upper one
The sum of the coefficient and variable quantity at moment are that linear adaption filters current coefficient.
Further, according to the linear adaptive filter after update coefficient to the input signal of linear adaptive filter
It is handled, obtains the estimated value of reverb signal in the first processing result;Determine the estimation of the first processing result and reverb signal
The difference of value is final process result.
According to another aspect of an embodiment of the present invention, a kind of processing unit of voice signal is additionally provided, comprising: receive mould
Block, for receiving primary sound sound signal;First filter module, for carrying out the first filter to primary sound sound signal based on nonlinear filter
Wave processing, obtains the first processing result, wherein the first filtering processing is for carrying out Reverberation Rejection to primary sound sound signal;Update mould
Block, for updating the coefficient of linear adaptive filter according to the first processing result;Second filter module, for based on update system
Linear adaptive filter after number carries out the second filtering processing to the first processing result, exports the final process of primary sound sound signal
As a result, wherein the second filtering processing is for carrying out Reverberation Rejection to the first processing result.
Further, the first filter module includes: the first acquisition submodule, for obtaining the signal-to-noise ratio of primary sound sound signal,
Wherein, signal-to-noise ratio is the ratio that primary sound sound signal mid-term hopes signal power and reverb signal power;It determines submodule, is used for basis
Signal-to-noise ratio and preset compensation factor determine nonlinear filter;Submodule is filtered, for being based on nonlinear filter to primary sound
Sound signal carries out the first filtering processing.
Further, update module includes: the second acquisition submodule, for obtaining the more newly-increased of linear adaptive filter
Benefit;Update submodule, for according to linear adaptive filter in the coefficient of last moment, more new gain and the first processing result
The current coefficient of linear adaption filtering is updated.
According to another aspect of an embodiment of the present invention, a kind of storage medium is additionally provided, storage medium includes the journey of storage
Sequence, wherein equipment where control storage medium executes the processing method of above-mentioned voice signal in program operation.
According to another aspect of an embodiment of the present invention, a kind of processor is additionally provided, processor is used to run program,
In, program executes the processing method of above-mentioned voice signal when running.
According to another aspect of an embodiment of the present invention, a kind of interactive intelligence equipment is additionally provided, comprising: voice pickup dress
It sets, for receiving primary sound sound signal;Nonlinear filter is connected with sound pickup device, for carrying out the to primary sound sound signal
One filtering processing, obtains the first processing result, wherein the first filtering processing is for carrying out Reverberation Rejection to primary sound sound signal;Line
Property sef-adapting filter, be connected with nonlinear filter, for updating linear adaptive filter according to first processing result
Coefficient, and the second filtering processing is carried out to the first processing result based on updated coefficient, export the final place of primary sound sound signal
Manage result, wherein the second filtering processing is for carrying out Reverberation Rejection to the first processing result;Sound play device, for playing
The final process result of primary sound sound signal.
In embodiments of the present invention, audio signal is pre-processed first with nonlinear filtering, then with pretreatment
Obtained result goes to update the coefficient of linear adaptive filter.Pretreatment can inhibit the reverb signal of a part, so that input
Reverberation power into linear filter is smaller.Meanwhile the signal after pre-processing for original received signal more
Accurately, but also the related coefficient of linear filter is more accurate.The above two o'clock receive linear filtering can at faster speed
It holds back, to reduce the complexity of algorithm, promotes the inhibitory effect to reverb signal, and then solve in the prior art to sound
When signal carries out Reverberation Rejection, it is difficult to lower algorithm complexity technology be kept to ask while obtaining preferable inhibitory effect
Topic.
Detailed description of the invention
The drawings described herein are used to provide a further understanding of the present invention, constitutes part of this application, this hair
Bright illustrative embodiments and their description are used to explain the present invention, and are not constituted improper limitations of the present invention.In the accompanying drawings:
Fig. 1 is the flow chart of the processing method of voice signal according to an embodiment of the present invention;
Fig. 2 is a kind of schematic diagram of the processing method of optional voice signal according to an embodiment of the present invention;
Fig. 3 a, which is shown, individually uses linear filter to inhibit linear filter coefficients convergent during reverberation;
Fig. 3 b is linear during showing according to the joint nonlinear and linear filter inhibition reverberation of the scheme of the application
Filter coefficient convergent;
Fig. 4 is a kind of schematic diagram of the processing unit of voice signal according to an embodiment of the present invention.
Specific embodiment
In order to enable those skilled in the art to better understand the solution of the present invention, below in conjunction in the embodiment of the present invention
Attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is only
The embodiment of a part of the invention, instead of all the embodiments.Based on the embodiments of the present invention, ordinary skill people
The model that the present invention protects all should belong in member's every other embodiment obtained without making creative work
It encloses.
It should be noted that description and claims of this specification and term " first " in above-mentioned attached drawing, "
Two " etc. be to be used to distinguish similar objects, without being used to describe a particular order or precedence order.It should be understood that using in this way
Data be interchangeable under appropriate circumstances, so as to the embodiment of the present invention described herein can in addition to illustrating herein or
Sequence other than those of description is implemented.In addition, term " includes " and " having " and their any deformation, it is intended that cover
Cover it is non-exclusive include, for example, the process, method, system, product or equipment for containing a series of steps or units are not necessarily limited to
Step or unit those of is clearly listed, but may include be not clearly listed or for these process, methods, product
Or other step or units that equipment is intrinsic.
Embodiment 1
According to embodiments of the present invention, a kind of embodiment of the processing method of voice signal is provided, it should be noted that
The step of process of attached drawing illustrates can execute in a computer system such as a set of computer executable instructions, also,
It, in some cases, can be to be different from shown in sequence execution herein although logical order is shown in flow charts
The step of out or describing.
Fig. 1 is the flow chart of the processing method of voice signal according to an embodiment of the present invention, as shown in Figure 1, this method packet
Include following steps:
Step S102 receives primary sound sound signal.
Specifically, above-mentioned primary sound sound signal can be sound transmitting terminal in the sound letter for collecting sound generation object generation
After number, by transmission of sound signals to the signal of sound receiving end, i.e. the voice signal that directly receives of equipment.
In an alternative embodiment, by taking the scene of teleconference as an example, sound transmitting terminal can be meeting scene
Intelligent terminal (such as: interactive intelligent tablet computer, tablet computer, mobile terminal etc.), during meeting carries out, the intelligence at meeting scene
The acoustic information of terminal collection site, and acoustic information is transmitted to live terminal of remotely attending a meeting, the terminal at the scene of attending a meeting is i.e.
Receive above-mentioned primary sound sound signal.
Step S104 carries out the first filtering processing to primary sound sound signal based on nonlinear filter, obtains the first processing knot
Fruit, wherein the first filtering processing is for carrying out Reverberation Rejection to primary sound sound signal.
Specifically, above-mentioned nonlinear filter can be realized by sound receiving end.Nonlinear filtering has structure simple, easy
In realize the advantages that.
In an alternative embodiment, above-mentioned nonlinear filter can pass through median filtering, mean filter, bilateral filter
A variety of nonlinear filtering algorithms such as wave, Kalman filtering carry out the first filtering processing to primary sound sound signal, obtain the first processing knot
Fruit.
Above-mentioned first processing result inhibits part reverberation to believe via nonlinear filter compared with primary sound sound signal
Number, quality is higher compared with primary sound sound signal.But since the effect that nonlinear filter inhibits reverb signal is limited, and can also
New noise is introduced, therefore after being filtered using nonlinear filter to primary sound sound signal, it is also necessary to enter step
S106 continues Reverberation Rejection processing to the first processing result.
Step S106 updates the coefficient of linear adaptive filter according to the first processing result.
Specifically, linear adaptive filter is the change according to environment, change filter using adaptive algorithm
The linear filter of coefficient, that is, the coefficient of above-mentioned linear adaptive filter is updated based on preset adaptive algorithm
Time-varying coefficient.By constantly adjusting the coefficient of filter, filter is enable to reach its optimum performance.Different linear filterings
Device has its corresponding more new algorithm.
The above-mentioned linear adaptive filter of the application updates filter coefficient according to the first processing result, so as to
To with the matched optimum linear filter of the first processing result.
Step S108 carries out the second filtering processing to the first processing result based on updated linear adaptive filter,
Export the final process result of primary sound sound signal, wherein the second filtering processing is for carrying out Reverberation Rejection to the first processing result.
In an alternative embodiment, it is still illustrated by taking above-mentioned teleconference as an example, receives sound in receiving end
After message breath, the first filtering processing of Reverberation Rejection is carried out to primary sound sound signal by nonlinear filter first, based on non-thread
Property filter processing result update linear adaptive filter filter coefficient, and using update coefficient after it is linear adaptive
It answers filter that the first processing result is further processed, so as to complete the Reverberation Rejection carried out to original sound information, obtains
To the higher acoustic information of quality.After receiving end as above handle to original sound information, then play out, so that with
Family can hear the higher acoustic information of quality.
It should be noted that Reverberation Rejection mainly includes nonlinear filtering and two class of linear filtering.Nonlinear filtering is to mixed
Loud inhibitory effect is poor, and can introduce new noise, requires so that such method is difficult to apply to sound signal quality
In higher scene.Linear filter method has preferable effect in inhibiting reverberation, but such method is generally required with high
Computation complexity be that cost can just exchange ideal Reverberation Rejection effect for, and the convergence rate of filter also will affect
Reverberation Rejection effect.So that high inhibitory effect and low algorithm complexity are difficult to when carrying out Reverberation Rejection to voice signal
It gets both.
And the above embodiments of the present application pre-process audio signal first with nonlinear filtering, then with pretreatment
The part coefficient of obtained result deinitialization linear adaptive filter.Pretreatment can inhibit the reverb signal of a part, make
The reverberation power that must be input in linear filter is smaller.Meanwhile the signal after pre-processing comes relative to original received signal
Say it is more accurate, but also the related coefficient of linear filter is more accurate.The above two o'clock makes linear filtering can be with faster
Speed convergence, to promote the inhibitory effect to reverb signal;Further, the present invention program utilizes low non-of computation complexity
Linear filtering pre-processes reverb signal.Pretreatment can inhibit the reverb signal of a part, so that being input to linear filtering
Reverberation power in device is smaller.Linear filter can disinthibite remaining reverb signal using lower filter order as cost,
Thus greatly reduce the execution complexity of this method.
As a kind of optional embodiment, before receiving primary sound sound signal, the above method further include: to linear adaption
The coefficient of filter carries out initialization process, and the coefficient of setting initial time linear adaptive filter is the sequence of preset length
It arranges, the value of each element in sequence is 0.
Specifically, since linear adaptive filter needs to update the filter system of its own according to the first processing result
Number, therefore in the initial stage, linear adaptive filter is initialized.In an alternative embodiment, line is initialized
Property filter length L, can be set zero moment linear filter coefficients value be 0 namely GL×1=(0,0 ..., 0)Τ, wherein
(·)TIndicate transposition operation.
As a kind of optional embodiment, the first filtering processing is carried out to primary sound sound signal based on nonlinear filter, is obtained
To the first processing result, comprising: obtain the signal-to-noise ratio of primary sound sound signal, wherein signal-to-noise ratio is that primary sound sound signal mid-term hopes signal
The ratio of power and reverb signal power;Nonlinear filter is determined according to signal-to-noise ratio and preset compensation factor;Based on non-thread
Property filter to primary sound sound signal carry out first filtering processing.
Specifically, the signal-to-noise ratio (SNR, Signal-Noise Ratio) of primary sound sound signal can be electronic equipment or electricity
The ratio of signal and noise in subsystem.In the scheme of the application, signal-to-noise ratio is for indicating in primary sound sound signal, normal sound
The ratio of information and reverberation information.
In an alternative embodiment, received acoustic information can be expressed as Y=S+N, wherein S is for indicating
Desired signal, N need repressed reverb signal for indicating, Y is for indicating the original sound information that receiving end receives.It is above-mentioned
The signal-to-noise ratio of original sound information can be the ratio of the signal power of the signal power and reverb signal of desired signal.It can with formula
To indicate are as follows:Wherein, E [S2] for indicating the signal power of desired signal, E [N2] for indicating reverb signal
Signal power.
Above-mentioned preset compensation factor can pass through following formula in an alternative embodiment for empirical value
Indicate above-mentioned nonlinear filter:Wherein, μ is for indicating that preset compensation factor, SNR are used
In the signal-to-noise ratio for indicating primary sound sound signal, α and β are for indicating the type of nonlinear filter, wherein as α=0.5, β=1,
W indicates that power spectrum subtracts filter;As α=1, β=0.5, W indicates that amplitude spectrum subtracts filter;As α=1, β=1, W indicates dimension
Receive filter.
After obtaining nonlinear filter W, nonlinear filter W can be used, primary sound sound signal is carried out at the first filtering
Reason, obtains the first processing result.First processing result is the estimation output of nonlinear filter W.
Still in the above-described embodiments, the estimation output of nonlinear filter can be with are as follows:Wherein, W is for indicating
In the numerical value of each frequency point of each frequency domain, Y is used to indicate the value of each frequency point of primary sound sound signal above-mentioned nonlinear filter,
For indicate nonlinear filter to desired signal S frequency domain estimated value.
In the above embodiments of the present application and the coefficient of primary sound sound signal itself (signal-to-noise ratio of primary sound sound signal) determine it is non-
Linear filter, and using determining nonlinear filter, the first filtering processing is carried out to primary sound sound signal, and then obtained non-
Linear filter is to the estimated value of desired signal, i.e. the first processing result.First processing result has pressed down compared with primary sound sound signal
Part reverberation signal is made, but inhibitory effect is limited, and introduced new noise, it is therefore desirable to linear sef-adapting filter pair
The output of nonlinear filter continues the second filtering processing,
As a kind of optional embodiment, the above method further include: determine the type of nonlinear filter, comprising: determine
Signal-to-noise ratio is less than snr threshold, determines that nonlinear filter is that power spectrum subtracts filter;Or determine that signal-to-noise ratio is greater than or equal to
Snr threshold determines that nonlinear filter is that amplitude spectrum subtracts filter.
Specifically, in order to determine the nonlinear filter for carrying out the first filtering processing to primary sound sound signal, it is also true
Determine the type of nonlinear filter.In an alternative embodiment, the type of nonlinear filter, can be by adjusting above-mentioned
α and β determines the type of nonlinear filter.
When noise is bigger, amplitude spectrum subtracts filter with preferable Reverberation Rejection effect;When noise is smaller, power
Spectrum subtracts filter with preferable Reverberation Rejection effect, therefore in above scheme, by the signal-to-noise ratio of primary sound sound signal, and it is pre-
If snr threshold determine the type of nonlinear filter.
In an alternative embodiment, the signal-to-noise ratio of primary sound sound signal is compared with preset snr threshold,
When the signal-to-noise ratio of primary sound sound signal is less than snr threshold, α=0.5, β=1 can be set, so that W indicates that power spectrum subtracts filter
Wave device;In the case that the noise of primary sound sound signal is greater than or equal to noise than threshold value, α=1, β=0.5, so that W can be set
Indicate that amplitude spectrum subtracts filter.
The above embodiments of the present application pass through type of the noise than determining nonlinear filter of primary sound sound signal as a result, and
The type of nonlinear filter is adjusted by adjusting the coefficient of filter, so that nonlinear filter and primary sound sound signal
Feature more match.
As a kind of optional embodiment, the step of determining the type of nonlinear filter further include: determine the mistake of demand
True coefficient is greater than distortion factor threshold value, determines that nonlinear filter is Wiener filter.
Specifically, above-mentioned distortion factor can be the distortion factor.The distortion factor is using one without the letter before amplifier amplification
It number makes comparisons with by the amplified signal of amplifier, the difference of the signal and original signal that are amplified is the distortion factor.The distortion factor
Value it is bigger, then by amplifier amplification or signal and original signal difference it is bigger.
In a scenario, the fidelity of sound is required to increase, it is thus determined that the type of nonlinear filter is wiener filter
Wave device, so as to when carrying out Reverberation Rejection processing to primary sound sound signal, guarantee that primary sound sound signal only spends distortion.
As a kind of optional embodiment, the first processing result is carried out at the second filtering based on linear adaptive filter
Reason exports the final process result of primary sound sound signal, comprising: obtain the more new gain of linear adaptive filter;According to linear
The current coefficient that sef-adapting filter filters linear adaption in the coefficient of last moment, more new gain and the first processing result
It is updated.
It should be noted that different types of linear adaptive filter is with corresponding for updating filter coefficient
Algorithm criterion.
In the above scheme, after getting the more new gain of linear adaptive filter, linear adaptive filter root
According to the first processing result, linear adaptive filter is in the coefficient of last moment and the current more new gain of acquisition, to upper
The coefficient at one moment is updated, and obtains linear adaptive filter in the coefficient at current time.Through the above scheme, so that line
Property sef-adapting filter coefficient it is associated with processing result of the nonlinear filter to primary sound sound signal so that it is linear from
Adaptive filter can adjust its coefficient at any time, reach and the most matched optimum state of the first processing result.
As a kind of optional embodiment, linear adaptive filter is least square method of recursion filter, is obtained linear
The more new gain of sef-adapting filter, comprising: obtain current input signal vector, the current input letter of linear adaptive filter
The conjugate transposition of number vector, preset forgetting factor, current first processing result power and linear adaptive filter coefficient
Covariance matrix;Obtain the first product of the power of forgetting factor and the first processing result;Obtain current input signal vector
Conjugate transposition, the covariance matrix of linear adaptive filter coefficient and the second product of current input signal vector;It obtains
The covariance matrix of line taking adaptive filter coefficient and the third product of current input signal vector;Obtain the first sum of products
Second product and value;Determine third product with and value ratio be more new gain.
Specifically, above-mentioned linear adaptive filter can be filtered for LMS (Least Mean Square, Minimum Mean Square Error)
Wave device, RLS filter or Kalman filter.Replacement criteria of the different types of linear adaptive filter according to corresponding to it
It is updated gain.
The current input signal vector of above-mentioned linear adaptive filter include current time original sound signal vector, on
The original sound signal vector ... at one moment and the original sound signal vector at preceding (L+1) moment, wherein L is linear adaption
The length of filter.Following formula, which can be used, indicates the current input signal vector of linear adaptive filter: X (n)=(Y
(n),Y(n-1),...,Y(n-L+1))T, wherein X (n) is the current input signal vector of linear adaptive filter, Y (n)
For original sound signal vector.
First processing result is the estimated value of the desired audio signal of nonlinear filter output, therefore at current first
The power of reason result is the power of current desired signal estimated value.
Above-mentioned least square method of recursion filter is to be based on least square method of recursion (Recursive Least
Square, RLS) filter.In an alternative embodiment, it can indicate to obtain recursive least-squares by following formula
The mode of the more new gain of method filter:
Wherein, K ∈ CL×1For indicating the update gain vector of RLS filter;XHIndicate the conjugate transposition of X;Φ∈CL×L
Indicate linear adaptive filter in the covariance matrix of the coefficient G (n-1) of last moment;X is for indicating current input signal
Vector, wherein the filter input at the n-th moment is X (n)=(Y (n), Y (n-1) ..., Y (n-L+1))T;λ indicate forget because
Son.Indicate the power of current expectation audio signal.In the examples described above, λ δ2For the first product, XΗΦ X is the
Two products, Φ X are third product.
As a kind of optional embodiment, according to linear adaptive filter the coefficient of last moment, more new gain and
The current coefficient that first processing result filters linear adaption is updated, comprising: obtains working as linear adaptive filter
The variable quantity of preceding coefficient and the coefficient of last moment, wherein variable quantity is the conjugate transposition of more new gain and the first processing result
The 4th product;Determine that linear adaptive filter is worked as in the sum of coefficient and variable quantity of last moment for linear adaption filtering
Preceding coefficient.
In the above scheme, the change of the coefficient of linear adaptive filter is determined by more new gain and the first processing result
Change amount determines linear adaption on the basis of coefficient of the linear adaptive filter in last moment further according to the variable quantity
The current coefficient of filter.
In an alternative embodiment, it can be carried out more by coefficient of the following formula to linear adaptive filter
It is new:
Wherein, G (n) indicates that the current coefficient of linear adaptive filter, G (n-1) indicate on linear adaptive filter for the moment
The coefficient at quarter, K indicate update gain vector,Indicate the conjugate transposition of the first processing result, i.e. nonlinear filter believes expectation
Number estimated value conjugate transposition.Specifically, i-th of element in G (n) can be expressed as
As a kind of optional embodiment, based on update the linear adaptive filter after coefficient to the first processing result into
Row second is filtered, comprising: the input according to the linear adaptive filter after update coefficient to linear adaptive filter
Signal is handled, and the estimated value of reverb signal in the first processing result is obtained;Determine the first processing result and reverb signal
The difference of estimated value is final process result.
Specifically, the sef-adapting filter after above-mentioned update coefficient, the estimation using updated coefficient to reverb signal
Value carries out operation.After the estimated value for obtaining the reverb signal in the first processing result, this is subtracted using the first processing result and is estimated
Final process result can be obtained in evaluation.
In an alternative embodiment, final process result can be obtained by following formula:
Wherein, SlIndicate final process result,Indicate the first processing result, what G indicated linear adaptive filter is
Number vector, X indicate the input signal vector of linear adaptive filter, and X (n)=(Y (n), Y (n-1) ..., Y (n-L+1))T。
Specifically, GHThe calculating of X can indicate are as follows:
Wherein G*(i) conjugation of i-th of element in G is indicated.X (i) indicates i-th yuan of X
Element.
Fig. 2 is a kind of schematic diagram of the processing method of optional voice signal according to an embodiment of the present invention, below with reference to
Fig. 2 is illustrated a kind of optional embodiment of the processing method of the voice signal in the application.
Firstly, receiving end receives the primary sound sound signal Y (n) of transmitting terminal transmission, by nonlinear filter to primary sound message
Number Y (n) carries out the first filtering processing, obtains the first processing result
It is then based on the first processing resultWith input signal X (n)=(Y (n), Y (n- of linear adaptive filter
1),...,Y(n-L+1))TDetermine the more new gain K of linear adaptive filter.It is based on the first processing result againIt is more newly-increased
The beneficial K and coefficient G of linear adaptive filter last moment (n-1), is updated the coefficient of linear adaptive filter, obtains
The coefficient G (n) current to linear adaptive filter.
Finally based on the input signal X (n) of linear adaptive filter, the current coefficient G of linear adaptive filter
(n) and the first processing resultObtain final process result Sl。
The effect that can reach to the application below is illustrated.By provided by the present application by nonlinear filter and line
Property sef-adapting filter be combined to voice signal carry out inhibit reverberation scheme so that the hangover feelings of signal on a timeline
Condition is improved, and its improvement carries out Reverberation Rejection processing to voice signal better than exclusive use nonlinear filter
Scheme.
Fig. 3 a, which is shown, individually uses linear filter to inhibit linear filter coefficients convergent during reverberation.Figure
Linear filter coefficients during 3b is shown according to the joint nonlinear and linear filter inhibition reverberation of the scheme of the application
Convergent.Horizontal axis indicates the number of sampled point in figure, and the longitudinal axis indicates the average value of linear filter coefficients.Pass through Fig. 3 a and figure
3b can speed up the convergence rate of linear filter it is found that using the structure for combining nonlinear and linear filter in the application,
To promote the performance of linear filter.
Further, computational complexity is primarily present in the calculating process of linear filter, computation complexity and filter
Wave device length it is square directly proportional.Through comparison 3a and Fig. 3 b it is found that in the case where linear filter length remains unchanged, connection
The convergence rate for closing nonlinear and linear filter is significantly larger than linear filter, that is, joint nonlinear and linear filter
Linear filter is better than to Reverberation Rejection effect.So, in the case where reaching same Reverberation Rejection effect, Federated filter
The length of linear filter can be reduced suitably in device, to reduce whole computational complexity.
Embodiment 2
According to embodiments of the present invention, a kind of embodiment of the processing unit of voice signal is provided, Fig. 4 is according to the present invention
A kind of schematic diagram of the processing unit of voice signal of embodiment, as shown in connection with fig. 4, which includes:
Receiving module 40, for receiving primary sound sound signal.
First filter module 42 is obtained for carrying out the first filtering processing to primary sound sound signal based on nonlinear filter
First processing result, wherein the first filtering processing is for carrying out Reverberation Rejection to primary sound sound signal.
Update module 44, for updating the coefficient of linear adaptive filter according to the first processing result.
Second filter module 46, for being carried out based on the linear adaptive filter after update coefficient to the first processing result
Second filtering processing, exports the final process result of primary sound sound signal, wherein the second filtering processing is for the first processing result
Carry out Reverberation Rejection.
As a kind of optional embodiment, above-mentioned apparatus further include: initialization module, for receive primary sound sound signal it
Before, initialization process is carried out to the coefficient of linear adaptive filter, the coefficient of initial time linear adaptive filter is set
For the sequence of preset length, the value of each element in sequence is 0.
As a kind of optional embodiment, the first filter module includes: the first acquisition submodule, for obtaining primary sound message
Number signal-to-noise ratio, wherein signal-to-noise ratio be primary sound sound signal mid-term hope signal power and reverb signal power ratio;Determine submodule
Block, for determining nonlinear filter according to signal-to-noise ratio and preset compensation factor;Submodule is filtered, for being based on nonlinear filtering
Wave device carries out the first filtering processing to primary sound sound signal.
As a kind of optional embodiment, the above method further include: determining module, for determining the class of nonlinear filter
Type, determining module includes: the first determining submodule, for determining that signal-to-noise ratio less than snr threshold, determines nonlinear filter
Subtract filter for power spectrum;Or second determine submodule, for determine signal-to-noise ratio be greater than or equal to snr threshold, determine non-thread
Property filter be amplitude spectrum subtract filter.
As a kind of optional embodiment, determining module further include: third determines submodule, for determining the distortion of demand
Coefficient is greater than distortion factor threshold value, determines that nonlinear filter is Wiener filter.
As a kind of optional embodiment, update module includes: the second acquisition submodule, for obtaining linear adaption filter
The more new gain of wave device;Update submodule, for according to linear adaptive filter the coefficient of last moment, more new gain and
The current coefficient that first processing result filters linear adaption is updated.
As a kind of optional embodiment, linear adaptive filter is least square method of recursion filter, and second obtains
Submodule includes: first acquisition unit, for obtaining current input signal vector, the current input letter of linear adaptive filter
The conjugate transposition of number vector, preset forgetting factor, current first processing result power and linear adaptive filter coefficient
Covariance matrix;Second acquisition unit, the first product of the power for obtaining forgetting factor and the first processing result;Third
Acquiring unit, for obtaining the conjugate transposition of current input signal vector, the covariance matrix of linear adaptive filter coefficient
And the second product of current input signal vector;4th acquiring unit, for obtaining the association of linear adaptive filter coefficient
The third product of variance matrix and current input signal vector;5th acquiring unit, for obtaining first the second product of sum of products
And value;First determination unit, for determine third product with and value ratio be more new gain.
As a kind of optional embodiment, updating submodule includes: the 6th acquiring unit, for obtaining linear adaption filter
The variable quantity of the coefficient of the current coefficient and last moment of wave device, wherein variable quantity is more new gain and the first processing result
4th product of conjugate transposition;Second determination unit, for determining coefficient and change of the linear adaptive filter in last moment
The sum of change amount is that linear adaption filters current coefficient.
As a kind of optional embodiment, the second filter module includes: processing submodule, after according to coefficient is updated
Linear adaptive filter handles the input signal of linear adaptive filter, obtains reverberation in the first processing result and believes
Number estimated value;4th determines submodule, for determining that the difference of estimated value of the first processing result and reverb signal is final place
Manage result.
It should be noted that the processing unit of the voice signal in the present embodiment has function same as Example 1, and
It can achieve the effect that same as Example 1.
Embodiment 3
According to embodiments of the present invention, a kind of storage medium is provided, storage medium includes the program of storage, wherein in institute
State the processing method that equipment where controlling the storage medium when program operation executes voice signal described in embodiment 1.
It should be noted that the program that the storage medium in the present embodiment is stored has and 1 phase of embodiment at runtime
Same function, and can achieve the effect that same as Example 1.
Embodiment 4
According to embodiments of the present invention, a kind of processor is provided, processor is for running program, wherein described program fortune
The processing method of voice signal described in embodiment 1 is executed when row.
It should be noted that when the processor in the present embodiment executes the processing method of the voice signal in embodiment 1, tool
Standby function same as Example 1, and can achieve the effect that same as Example 1.
Embodiment 5
According to embodiments of the present invention, a kind of interactive intelligence equipment is provided, comprising:
Sound pickup device, for receiving primary sound sound signal;
Nonlinear filter is connected with sound pickup device, for carrying out the first filtering processing to primary sound sound signal, obtains
First processing result, wherein the first filtering processing is for carrying out Reverberation Rejection to primary sound sound signal;
Linear adaptive filter is connected with nonlinear filter, linear adaptive for being updated according to the first processing result
The coefficient of filter is answered, and the second filtering processing is carried out to the first processing result based on updated coefficient, exports primary sound message
Number final process result, wherein second filtering processing for the first processing result carry out Reverberation Rejection;
Sound play device, for playing the final process result of primary sound sound signal.
The interactive intelligence equipment that the above embodiments of the present application propose can pass through filter after receiving primary sound sound signal
Audio signal is pre-processed first with nonlinear filtering, the result deinitialization then obtained with pretreatment is linearly adaptive
Answer the part coefficient of filter.Pretreatment can inhibit the reverb signal of a part, so that being input to the reverberation in linear filter
Power is smaller.Meanwhile the signal after pre-processing is more accurate for original received signal, but also linear filter
Related coefficient it is more accurate.The above two o'clock restrain linear filtering can at faster speed, to be promoted to reverb signal
Inhibitory effect.Further, the above scheme of the present embodiment is using the low nonlinear filtering of computation complexity to reverb signal
It is pre-processed.Pretreatment can inhibit the reverb signal of a part, so that the reverberation power being input in linear filter is smaller.
Linear filter can disinthibite remaining reverb signal using lower filter order as cost, thus greatly reduce interactive intelligence
The complexity that energy equipment is filtered primary sound sound signal.
It should be noted that the nonlinear filter and linear adaptive filter in the present embodiment may each be and pass through electricity
The filter circuit that sub- component is constituted is also possible to execute the processor of response of step.
It should also be noted that, filter executes the visible embodiment 1 of specific embodiment of above-mentioned steps, and in the present embodiment
Nonlinear filter and linear adaptive filter can also execute other steps in embodiment 1, details are not described herein again.
The serial number of the above embodiments of the invention is only for description, does not represent the advantages or disadvantages of the embodiments.
In the above embodiment of the invention, it all emphasizes particularly on different fields to the description of each embodiment, does not have in some embodiment
The part of detailed description, reference can be made to the related descriptions of other embodiments.
In several embodiments provided herein, it should be understood that disclosed technology contents can pass through others
Mode is realized.Wherein, the apparatus embodiments described above are merely exemplary, such as the division of the unit, Ke Yiwei
A kind of logical function partition, there may be another division manner in actual implementation, for example, multiple units or components can combine or
Person is desirably integrated into another system, or some features can be ignored or not executed.Another point, shown or discussed is mutual
Between coupling, direct-coupling or communication connection can be through some interfaces, the INDIRECT COUPLING or communication link of unit or module
It connects, can be electrical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
On unit.It can some or all of the units may be selected to achieve the purpose of the solution of this embodiment according to the actual needs.
It, can also be in addition, the functional units in various embodiments of the present invention may be integrated into one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.Above-mentioned integrated list
Member both can take the form of hardware realization, can also realize in the form of software functional units.
If the integrated unit is realized in the form of SFU software functional unit and sells or use as independent product
When, it can store in a computer readable storage medium.Based on this understanding, technical solution of the present invention is substantially
The all or part of the part that contributes to existing technology or the technical solution can be in the form of software products in other words
It embodies, which is stored in a storage medium, including some instructions are used so that a computer
Equipment (can for personal computer, server or network equipment etc.) execute each embodiment the method for the present invention whole or
Part steps.And storage medium above-mentioned includes: that USB flash disk, read-only memory (ROM, Read-Only Memory), arbitrary access are deposited
Reservoir (RAM, Random Access Memory), mobile hard disk, magnetic or disk etc. be various to can store program code
Medium.
The above is only a preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For member, various improvements and modifications may be made without departing from the principle of the present invention, these improvements and modifications are also answered
It is considered as protection scope of the present invention.
Claims (15)
1. a kind of processing method of voice signal characterized by comprising
Receive primary sound sound signal;
The first filtering processing is carried out to the primary sound sound signal based on nonlinear filter, obtains the first processing result, wherein institute
The first filtering processing is stated for carrying out Reverberation Rejection to the primary sound sound signal;
The coefficient of linear adaptive filter is updated according to first processing result;
The second filtering processing is carried out to first processing result based on the linear adaptive filter after update coefficient, exports institute
State the final process result of primary sound sound signal, wherein second filtering processing is for mixing first processing result
It rings and inhibits.
2. the method according to claim 1, wherein before receiving primary sound sound signal, the method also includes:
Initialization process is carried out to the coefficient of the linear adaptive filter, the filtering of linear adaption described in initial time is set
The coefficient of device is the sequence of preset length, and the value of each element in the sequence is 0.
3. the method according to claim 1, wherein described carry out primary sound sound signal based on nonlinear filter
First filtering processing, obtains the first processing result, comprising:
Obtain the signal-to-noise ratio of the primary sound sound signal, wherein the signal-to-noise ratio is that the primary sound sound signal mid-term hopes signal power
With the ratio of reverb signal power;
Nonlinear filter is determined according to the signal-to-noise ratio and preset compensation factor;
First filtering processing is carried out to the primary sound sound signal based on the nonlinear filter.
4. according to the method described in claim 3, it is characterized in that, the method also includes: determine the nonlinear filter
Type, comprising:
It determines that the signal-to-noise ratio is less than snr threshold, determines that the nonlinear filter is that power spectrum subtracts filter;Or
It determines that the signal-to-noise ratio is greater than or equal to the snr threshold, determines that the nonlinear filter is that amplitude spectrum subtracts filtering
Device.
5. according to the method described in claim 4, it is characterized in that, the step of determining the type of the nonlinear filter is also wrapped
It includes:
It determines that the distortion factor of demand is greater than distortion factor threshold value, determines that the nonlinear filter is Wiener filter.
6. the method according to claim 1, wherein updating linear adaption filter according to first processing result
The coefficient of wave device, comprising:
Obtain the more new gain of the linear adaptive filter;
According to the linear adaptive filter in the coefficient of last moment, the more new gain and first processing result pair
The current coefficient of the linear adaption filtering is updated.
7. according to the method described in claim 6, it is characterized in that, the linear adaptive filter is least square method of recursion
Filter obtains the more new gain of the linear adaptive filter, comprising:
The conjugation of the current input signal vector, the current input signal vector that obtain the linear adaptive filter turns
It sets, the covariance square of the power of preset forgetting factor, current first processing result and the linear adaptive filter coefficient
Battle array;
Obtain the first product of the power of the forgetting factor and first processing result;
Obtain the conjugate transposition of the current input signal vector, the linear adaptive filter coefficient covariance matrix with
And the second product of the current input signal vector;
Obtain the covariance matrix of the linear adaptive filter coefficient and the third product of the current input signal vector;
Obtain second product of the first sum of products and value;
It determines the third product and described and value the ratio is the more new gain.
8. according to the method described in claim 6, it is characterized in that, according to the linear adaptive filter in last moment
The current coefficient that coefficient, the more new gain and first processing result filter the linear adaption is updated, packet
It includes:
Obtain the variable quantity of the current coefficient of the linear adaptive filter and the coefficient of last moment, wherein the variation
4th product of the conjugate transposition of amount more new gain for described in and first processing result;
Determine the linear adaptive filter the sum of the coefficient of last moment and the variable quantity be the linear adaption
Filter current coefficient.
9. the method according to claim 1, wherein based on the linear adaptive filter after update coefficient to institute
It states the first processing result and carries out the second filtering processing, comprising:
The input signal of the linear adaptive filter is handled according to the linear adaptive filter after update coefficient,
Obtain the estimated value of reverb signal in first processing result;
The difference for determining the estimated value of first processing result and the reverb signal is the final process result.
10. a kind of processing unit of voice signal characterized by comprising
Receiving module, for receiving primary sound sound signal;
First filter module obtains at first for carrying out the first filtering processing to primary sound sound signal based on nonlinear filter
Manage result, wherein first filtering processing is for carrying out Reverberation Rejection to the primary sound sound signal;
Update module, for updating the coefficient of linear adaptive filter according to first processing result;
Second filter module, for carrying out the to first processing result based on updating the linear adaptive filter after coefficient
Two filtering processings, export the final process result of the primary sound sound signal, wherein second filtering processing is for described the
One processing result carries out Reverberation Rejection.
11. device according to claim 10, which is characterized in that first filter module includes:
First acquisition submodule, for obtaining the signal-to-noise ratio of the primary sound sound signal, wherein the signal-to-noise ratio is the original sound
The ratio of desired signal power and reverb signal power in signal;
Submodule is determined, for determining nonlinear filter according to the signal-to-noise ratio and preset compensation factor;
Submodule is filtered, for carrying out first filtering processing to the primary sound sound signal based on the nonlinear filter.
12. device according to claim 10, which is characterized in that the update module includes:
Second acquisition submodule, for obtaining the more new gain of the linear adaptive filter;
Update submodule, for according to the linear adaptive filter in the coefficient of last moment, the more new gain and institute
The current coefficient that the first processing result filters the linear adaption is stated to be updated.
13. a kind of storage medium, which is characterized in that the storage medium includes the program of storage, wherein run in described program
When control the storage medium where equipment perform claim require any one of 1 to 9 described in voice signal processing method.
14. a kind of processor, which is characterized in that the processor is for running program, wherein right of execution when described program is run
Benefit require any one of 1 to 9 described in voice signal processing method.
15. a kind of interactive intelligence equipment, it is characterised in that, include:
Sound pickup device, for receiving primary sound sound signal;
Nonlinear filter is connected with the sound pickup device, for carrying out the first filtering processing to the primary sound sound signal,
Obtain the first processing result, wherein first filtering processing is for carrying out Reverberation Rejection to the primary sound sound signal;
Linear adaptive filter is connected with the nonlinear filter, linear for being updated according to first processing result
The coefficient of sef-adapting filter, and the second filtering processing, output are carried out to first processing result based on updated coefficient
The final process result of the primary sound sound signal, wherein second filtering processing is for carrying out first processing result
Reverberation Rejection;
Sound play device, for playing the final process result of the primary sound sound signal.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910616082.5A CN110289009B (en) | 2019-07-09 | 2019-07-09 | Sound signal processing method and device and interactive intelligent equipment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910616082.5A CN110289009B (en) | 2019-07-09 | 2019-07-09 | Sound signal processing method and device and interactive intelligent equipment |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110289009A true CN110289009A (en) | 2019-09-27 |
CN110289009B CN110289009B (en) | 2021-06-15 |
Family
ID=68022017
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910616082.5A Active CN110289009B (en) | 2019-07-09 | 2019-07-09 | Sound signal processing method and device and interactive intelligent equipment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110289009B (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111598806A (en) * | 2020-05-15 | 2020-08-28 | Oppo(重庆)智能科技有限公司 | Image filtering method and device, electronic equipment and storage medium |
CN112201267A (en) * | 2020-09-07 | 2021-01-08 | 北京达佳互联信息技术有限公司 | Audio processing method and device, electronic equipment and storage medium |
CN113473342A (en) * | 2021-05-20 | 2021-10-01 | 中国科学院声学研究所 | Signal processing method and device for hearing aid, hearing aid and computer storage medium |
CN113470676A (en) * | 2021-06-30 | 2021-10-01 | 北京小米移动软件有限公司 | Sound processing method, sound processing device, electronic equipment and storage medium |
CN114333880A (en) * | 2022-03-04 | 2022-04-12 | 南京大鱼半导体有限公司 | Signal processing method, device, equipment and storage medium |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102347028A (en) * | 2011-07-14 | 2012-02-08 | 瑞声声学科技(深圳)有限公司 | Double-microphone speech enhancer and speech enhancement method thereof |
CN103440869A (en) * | 2013-09-03 | 2013-12-11 | 大连理工大学 | Audio-reverberation inhibiting device and inhibiting method thereof |
CN105324982A (en) * | 2013-05-06 | 2016-02-10 | 波音频有限公司 | A method and apparatus for suppression of unwanted audio signals |
CN107424623A (en) * | 2016-05-24 | 2017-12-01 | 展讯通信(上海)有限公司 | Audio signal processing method and device |
US20180182411A1 (en) * | 2016-12-23 | 2018-06-28 | Synaptics Incorporated | Multiple input multiple output (mimo) audio signal processing for speech de-reverberation |
JP2018165834A (en) * | 2014-04-08 | 2018-10-25 | 華為技術有限公司Huawei Technologies Co.,Ltd. | Noise signal processing method, noise signal generating method, encoder, decoder, and encoding and decoding system |
-
2019
- 2019-07-09 CN CN201910616082.5A patent/CN110289009B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102347028A (en) * | 2011-07-14 | 2012-02-08 | 瑞声声学科技(深圳)有限公司 | Double-microphone speech enhancer and speech enhancement method thereof |
CN105324982A (en) * | 2013-05-06 | 2016-02-10 | 波音频有限公司 | A method and apparatus for suppression of unwanted audio signals |
CN103440869A (en) * | 2013-09-03 | 2013-12-11 | 大连理工大学 | Audio-reverberation inhibiting device and inhibiting method thereof |
JP2018165834A (en) * | 2014-04-08 | 2018-10-25 | 華為技術有限公司Huawei Technologies Co.,Ltd. | Noise signal processing method, noise signal generating method, encoder, decoder, and encoding and decoding system |
CN107424623A (en) * | 2016-05-24 | 2017-12-01 | 展讯通信(上海)有限公司 | Audio signal processing method and device |
US20180182411A1 (en) * | 2016-12-23 | 2018-06-28 | Synaptics Incorporated | Multiple input multiple output (mimo) audio signal processing for speech de-reverberation |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111598806A (en) * | 2020-05-15 | 2020-08-28 | Oppo(重庆)智能科技有限公司 | Image filtering method and device, electronic equipment and storage medium |
CN111598806B (en) * | 2020-05-15 | 2023-06-13 | Oppo(重庆)智能科技有限公司 | Image filtering method and device, electronic equipment and storage medium |
CN112201267A (en) * | 2020-09-07 | 2021-01-08 | 北京达佳互联信息技术有限公司 | Audio processing method and device, electronic equipment and storage medium |
CN113473342A (en) * | 2021-05-20 | 2021-10-01 | 中国科学院声学研究所 | Signal processing method and device for hearing aid, hearing aid and computer storage medium |
CN113470676A (en) * | 2021-06-30 | 2021-10-01 | 北京小米移动软件有限公司 | Sound processing method, sound processing device, electronic equipment and storage medium |
CN114333880A (en) * | 2022-03-04 | 2022-04-12 | 南京大鱼半导体有限公司 | Signal processing method, device, equipment and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN110289009B (en) | 2021-06-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110289009A (en) | Sound signal processing method and device and interactive intelligent equipment | |
CN109754813B (en) | Variable step size echo cancellation method based on rapid convergence characteristic | |
CN111885275B (en) | Echo cancellation method and device for voice signal, storage medium and electronic device | |
CN102204276B (en) | Sound emission and collection device, and sound emission and collection method | |
Krusienski et al. | Design and performance of adaptive systems based on structured stochastic optimization strategies | |
JP5102371B2 (en) | Apparatus and method for calculating control information of echo suppression filter, and apparatus and method for calculating delay value | |
CN107123430A (en) | Echo cancellation method, device, conference tablet and computer storage medium | |
CN111768796B (en) | Acoustic echo cancellation and dereverberation method and device | |
CN110992923B (en) | Echo cancellation method, electronic device, and storage device | |
DE3141500A1 (en) | FILTER COEFFICIENT ADJUSTMENT UNIT FOR THE DIGITAL FILTER OF AN AUTOMATIC DIGITAL RECURSIVE EQUALIZER | |
CN109087662A (en) | A kind of echo cancel method and device | |
Gil-Cacho et al. | Nonlinear acoustic echo cancellation based on a parallel-cascade kernel affine projection algorithm | |
JP2004349806A (en) | Multichannel acoustic echo canceling method, apparatus thereof, program thereof, and recording medium thereof | |
CN117789744A (en) | Voice noise reduction method and device based on model fusion and storage medium | |
CN111370016B (en) | Echo cancellation method and electronic equipment | |
Naylor et al. | Affine projection and recursive least squares adaptive filters employing partial updates | |
CN110199528A (en) | Far field voice capturing | |
JP3673727B2 (en) | Reverberation elimination method, apparatus thereof, program thereof, and recording medium thereof | |
CN115620737A (en) | Voice signal processing device, method, electronic equipment and sound amplification system | |
Zeller et al. | Coefficient pruning for higher-order diagonals of Volterra filters representing Wiener-Hammerstein models | |
CN112802487A (en) | Echo processing method, device and system | |
JP3180739B2 (en) | Method and apparatus for identifying unknown system by adaptive filter | |
CN111883155A (en) | Echo cancellation method, device and storage medium | |
CN117496989A (en) | Voice processing method and device of intercom terminal, terminal equipment and storage medium | |
Tummarello et al. | Stepsize control in NLMS acoustic echo cancellation using a neural network |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |