CN105931648B - Audio signal solution reverberation method and device - Google Patents
Audio signal solution reverberation method and device Download PDFInfo
- Publication number
- CN105931648B CN105931648B CN201610474006.1A CN201610474006A CN105931648B CN 105931648 B CN105931648 B CN 105931648B CN 201610474006 A CN201610474006 A CN 201610474006A CN 105931648 B CN105931648 B CN 105931648B
- Authority
- CN
- China
- Prior art keywords
- filter
- signal
- sub
- audio signal
- channel audio
- 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.)
- Active
Links
- 230000005236 sound signal Effects 0.000 title claims abstract description 129
- 238000000034 method Methods 0.000 title claims abstract description 44
- 238000001914 filtration Methods 0.000 claims abstract description 11
- 230000003595 spectral effect Effects 0.000 claims description 24
- 230000017105 transposition Effects 0.000 claims description 20
- 239000011159 matrix material Substances 0.000 claims description 10
- 238000005516 engineering process Methods 0.000 claims description 9
- 230000000694 effects Effects 0.000 claims description 8
- 238000001514 detection method Methods 0.000 claims description 7
- 238000012545 processing Methods 0.000 description 13
- 230000006854 communication Effects 0.000 description 7
- 230000008569 process Effects 0.000 description 7
- 238000004422 calculation algorithm Methods 0.000 description 6
- 238000004891 communication Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 6
- 230000006870 function Effects 0.000 description 6
- 238000004590 computer program Methods 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 230000006835 compression Effects 0.000 description 2
- 238000007906 compression Methods 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 241000209140 Triticum Species 0.000 description 1
- 235000021307 Triticum Nutrition 0.000 description 1
- 230000005856 abnormality Effects 0.000 description 1
- 239000011358 absorbing material Substances 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 230000002427 irreversible effect Effects 0.000 description 1
- 210000003127 knee Anatomy 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 239000004065 semiconductor Substances 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
- G10L19/00—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
- G10L19/02—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis using spectral analysis, e.g. transform vocoders or subband vocoders
- G10L19/0204—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis using spectral analysis, e.g. transform vocoders or subband vocoders using subband decomposition
-
- 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)
- Physics & Mathematics (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- Signal Processing (AREA)
- Health & Medical Sciences (AREA)
- Human Computer Interaction (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Quality & Reliability (AREA)
- Stereophonic System (AREA)
- Cable Transmission Systems, Equalization Of Radio And Reduction Of Echo (AREA)
Abstract
This application discloses a kind of audio signal solution reverberation method and devices.The method includes: acquisition channel audio signal, and channel audio signal includes early stage reverb signal and late reverberation signal；Judge whether channel audio signal is voice signal；If, then update the variance of the joint probability density distribution of early stage reverb signal, and the variance based on the distribution of the joint probability density of early stage reverb signal updates the filter coefficient of sub-filter, wherein sub-filter is for filtering out the late reverberation signal for including in channel audio signal；And the channel audio signal of solution reverberation is determined based on updated filter coefficient.The scheme of the application, can be by the late reverberation target signal filter in the audio signal of input, to improve the accuracy rate of subsequent speech recognition.
Description
Technical field
This application involves field of computer technology, and in particular to Audio Signal Processing field more particularly to audio signal solution
Reverberation method and device.
Background technique
During audio especially Speech processing, if the acquisition device for acquiring audio signal is (for example, wheat
Gram wind) position apart from sound source farther out, the audio signal that acquisition device receives will be inevitably by the influence of reverberation.It is mixed
Loud presence can not only reduce the sense of hearing quality of audio signal, but also under will lead to the precision of existing voice identifying system sharply
Drop.
Reverberation can be decomposed into early stage reverberation and late reverberation, wherein on audio quality and identifying system precision influence compared with
Big is late reverberation, therefore the main target for solving reverberation is how to reduce late reverberation.
In the prior art, more existing for filtering out the late reverberation in the collected voice signal of acquisition device
Algorithm.However, these algorithms usually all have the following problems:
1) when solving the filter coefficient of the filter for filtering out late reverberation, need to obtain whole audio numbers
According to, it is lower so as to cause the real-time rate of algorithm, and then the delay that will lead to solution reverberation algorithm is relatively high.And in voice communication and language
It is higher to the requirement of real-time of solution reverberation algorithm in sound identification field.
2), when solving the filter coefficient of the filter for filtering out late reverberation, it will usually be related to matrix inversion
It calculates.Once and matrix is irreversible during matrix inversion, then the filter coefficient acquired just inaccuracy, and then it is mixed to influence solution
Loud performance.In addition, the operand of matrix inversion is larger, the real-time for also resulting in understanding reverberation algorithm from another point of view is poor.
Summary of the invention
The purpose of the application is to propose a kind of improved audio signal solution reverberation method and device, to solve background above
The technical issues of technology segment is mentioned.
In a first aspect, this application provides a kind of audio signal solution reverberation methods, comprising: channel audio signal is obtained,
Channel audio signal includes early stage reverb signal and late reverberation signal；Judge whether channel audio signal is voice letter
Number；If so, updating the variance of the joint probability density distribution of early stage reverb signal, and the joint based on early stage reverb signal is general
The variance of rate Density Distribution updates the filter coefficient of sub-filter, wherein sub-filter is for filtering out single channel audio
The late reverberation signal for including in signal；And the audio signal after solution reverberation is determined based on updated filter coefficient.
In some embodiments, judge that channel audio signal whether be voice signal includes: to pass through voice activity detection
Technology judges whether channel audio signal is voice signal.
In some embodiments, the variance of the joint probability density distribution of early stage reverb signal is the sub-band filter before updating
The product of the input signal vector of the transposed matrix and sub-filter of the filter coefficient of device and the single-pass received in t moment
The spectral coefficient x of audio channel signalt,fAbsolute value of the difference square；Updated filter coefficient gfIt (t+1) is the filter before update
The sum of wave device coefficient and update variable quantity；Wherein, updating variable quantity is that the first update running parameter and second update running parameter
The ratio between；First update running parameter be the iteration step length of sub-filter, the reality output of sub-filter and desired output it
Between error and sub-filter input signal vector product；Second updates running parameter for the defeated of sub-filter
Enter the product of the transposition of signal vector and the input signal vector of sub-filter；The reality output of sub-filter and expectation are defeated
Error between out is equal to the spectral coefficient of the channel audio signal received in t moment and the early stage reverb signal of t moment
The ratio between the variance of joint probability density distribution subtracts the transposition and sub-band filter of the filter coefficient of the sub-filter before updating
The product of the input signal vector of device.
In some embodiments, the audio signal d after reverberation is solvedt,fEqual to the channel audio signal received in t moment
Spectral coefficient subtract t+1 moment sub-filter filter coefficient transposition and sub-filter input signal vector
Product.In some embodiments, judging whether channel audio signal is method after voice signal further include: if it is not, then
The filter system of sub-filter before variance and update that the joint probability density of early stage reverb signal before update is distributed
The filtering of variance and updated sub-filter that number is distributed as the joint probability density of updated early stage reverb signal
Device coefficient.
In some embodiments, method further include: judge the mistake between the reality output of sub-filter and desired output
Whether difference meets the absolute value of the spectral coefficient for square being greater than the channel audio signal received in t moment of Error Absolute Value
Product square with preset threshold K；If so, filter coefficient is set to null vector；Wherein, K > 1.
Second aspect, this application provides a kind of audio signal solution reverberation units, comprising: obtains module, is configured to obtain
Channel audio signal is taken, channel audio signal includes early stage reverb signal and late reverberation signal；Judgment module, configuration are used
In judging whether channel audio signal is voice signal；First update module, if being configured to channel audio signal is language
Sound signal then updates the variance of the joint probability density distribution of early stage reverb signal, and the joint based on early stage reverb signal is general
The variance of rate Density Distribution updates the filter coefficient of sub-filter, wherein sub-filter is for filtering out single channel audio
The late reverberation signal for including in signal；And determining module, it is configured to determine that solution is mixed based on updated filter coefficient
Audio signal after sound.
In some embodiments, judgment module is further configured to: judging single channel by voice activity detection technology
Whether audio signal is voice signal.
In some embodiments, the variance of the joint probability density distribution of early stage reverb signal are as follows: the subband filter before update
The product of the input signal vector of the transposed matrix and sub-filter of the filter coefficient of wave device and the list received in t moment
The spectral coefficient x of channel audio signalt,fAbsolute value of the difference square；Updated filter coefficient gf(t+1) before to update
The sum of filter coefficient and update variable quantity；Wherein, updating variable quantity is the first update running parameter and the second more new change ginseng
The ratio between number；First updates the reality output and desired output that running parameter is the iteration step length of sub-filter, sub-filter
Between error and sub-filter input signal vector product；Second updates running parameter for sub-filter
The product of the input signal vector of the transposition and sub-filter of input signal vector；The reality output and expectation of sub-filter
Error between output is equal to the spectral coefficient of the channel audio signal received in t moment and the early stage reverb signal of t moment
Joint probability density distribution the ratio between variance subtract the sub-filter before updating filter coefficient transposition and subband filter
The product of the input signal vector of wave device.
In some embodiments, the audio signal d after reverberation is solvedt,fEqual to the channel audio signal received in t moment
Spectral coefficient subtract t+1 moment sub-filter filter coefficient transposition and sub-filter input signal vector
Product.
In some embodiments, device further includes the second update module；If the second update module is configured to single channel sound
Frequency signal is not voice signal, then before the variance and update that are distributed the joint probability density of the early stage reverb signal before update
The variance and update that the filter coefficient of sub-filter is distributed as the joint probability density of updated early stage reverb signal
The filter coefficient of sub-filter afterwards.
In some embodiments, device further includes zero setting module；Zero setting module is configured to judge the reality of sub-filter
What whether the error between border output and desired output met Error Absolute Value square is greater than the single channel sound that receives in t moment
The product square with preset threshold K of the absolute value of the spectral coefficient of frequency signal；And if so, by filter coefficient be set to zero to
Amount；Wherein, K > 1.
Audio signal solution reverberation method and device provided by the present application, are continuously updated by the voice signal based on input
The variance of the joint probability density distribution of early stage reverb signal and the parameter of sub-filter, so that sub-filter filters energy
Enough by the late reverberation target signal filter in the channel audio signal of input, to improve the accuracy rate of subsequent speech recognition.
In addition, the audio signal solution reverberation method and device of the application, the joint probability density distribution of early stage reverb signal
Variance update and sub-filter parameter update needed for calculation amount it is smaller, and renewal process only with the list in a period of time
Channel audio signal is related, real-time with higher.
Detailed description of the invention
By reading a detailed description of non-restrictive embodiments in the light of the attached drawings below, the application's is other
Feature, objects and advantages will become more apparent upon:
Fig. 1 is that this application can be applied to exemplary system architecture figures therein；
Fig. 2 is the flow chart according to one embodiment of the audio signal solution reverberation method of the application；
Fig. 3 is the flow chart according to another embodiment of the audio signal solution reverberation method of the application；
Fig. 4 is the structural schematic diagram according to one embodiment of the audio signal solution reverberation unit of the application；
Fig. 5 is adapted for the structural representation of the computer system for the terminal device or server of realizing the embodiment of the present application
Figure.
Specific embodiment
The application is described in further detail with reference to the accompanying drawings and examples.It is understood that this place is retouched
The specific embodiment stated is used only for explaining related invention, rather than the restriction to the invention.It also should be noted that in order to
Convenient for description, part relevant to related invention is illustrated only in attached drawing.
It should be noted that in the absence of conflict, the features in the embodiments and the embodiments of the present application can phase
Mutually combination.The application is described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.
Fig. 1 is shown can be using the implementation of the audio signal solution reverberation method or audio signal solution reverberation unit of the application
The exemplary system architecture 100 of example.
As shown in Figure 1, system architecture 100 may include terminal device 101,102,103, network 104 and server 105.
Network 104 between terminal device 101,102,103 and server 105 to provide the medium of communication link.Network 104 can be with
Including various connection types, such as wired, wireless communication link or fiber optic cables etc..
User can be used terminal device 101,102,103 and be interacted by network 104 with server 105, to receive or send out
Send message etc..Various telecommunication customer end applications can be installed, such as web browser is answered on terminal device 101,102,103
With, shopping class application, searching class application, instant messaging tools, mailbox client, social platform software etc..
Terminal device 101,102,103 can be with acquisition audio signal ability various electronic equipments, including but
It is not limited to smart phone, tablet computer, MP3 player (Moving Picture Experts Group Audio Layer
III, dynamic image expert's compression standard audio level 3), MP4 (Moving Picture Experts Group Audio
Layer IV, dynamic image expert's compression standard audio level 4) player, pocket computer on knee and desktop computer etc.
Deng.
Server 105 can be to provide the server of various services, such as to the sound that terminal device 101,102,103 acquires
The audio processing service device that frequency signal is handled.Audio processing service device can carry out the data such as the audio signal received
The processing such as analysis, and processing result (such as audio data through reverberation removal processing) is fed back into terminal device.
It should be noted that audio signal solution reverberation method both can be by terminal device provided by the embodiment of the present application
101, it 102,103 executes, can also be executed, can be held by terminal device 101,102,103 with a part of step by server 105
Row and another part step is executed by server 105.Correspondingly, audio signal solution reverberation unit both can be set in terminal device
101, in 102,103, it also can be set in server 105 or a part of module be set to terminal device 101,102,103
In and another part module is set in server.
It should be understood that the number of terminal device, network and server in Fig. 1 is only schematical.According to realization need
It wants, can have any number of terminal device, network and server.
With continued reference to Fig. 2, the process of one embodiment of the audio signal solution reverberation method according to the application is shown
200.The audio signal solution reverberation method, comprising the following steps:
Step 210, channel audio signal is obtained, wherein channel audio signal includes early stage reverb signal and advanced stage
Reverb signal.
In the present embodiment, electronic equipment (such as the service shown in FIG. 1 of audio signal solution reverberation method operation thereon
Device) it can be received from user using its terminal for carrying out audio signal sample by wired connection mode or radio connection
Channel audio signal.It should be pointed out that above-mentioned radio connection can include but is not limited to 3G/4G connection, WiFi connects
Connect, bluetooth connection, WiMAX connection, Zigbee connection, UWB (ultra wideband) connection and other it is currently known or will
Come the radio connection developed.
In general, when, there are when a certain distance, the equipment for acquiring sound is adopted between the equipment and sound source of acquisition sound
The audio signal collected is by the influence by reverberation.Reverberation usually (namely is directly conveyed to acquisition sound from sound source according to direct sound wave
The audio signal of the equipment of sound) and be transferred to acquisition sound equipment reverberation between time difference be early stage reverberation and advanced stage
Reverberation.For example, the reverberation of the equipment of acquisition sound can will be reached in 30ms after direct sound wave (millisecond) as early stage reverberation, and incite somebody to action
The reverberation for the equipment for acquiring sound is reached as late reverberation more than 30ms.Early stage reverberation is to voice amplitude, phase delay, resonance
Peak influence is smaller, and late reverberation then can be bigger on voice amplitude, phase delay, formant influence, and will lead to language
The phase inter-masking of syllable, these all reduce speech intelligibility, and very big difficulty is brought to speech recognition.
Step 220, judge whether channel audio signal is voice signal.
The audio signal solution reverberation method of the present embodiment, it is intended to solution reverberation processing is carried out to collected voice signal, into
And enables that treated voice signal more realistically reflects the practical voice said of user.And then promote subsequent speech recognition
The accuracy rate of equal signal processings.
In addition, therefore voice signal and other audio signals are obtained in the presence of more significant difference by judging in step
Whether the channel audio signal got is voice signal, can be carried out in subsequent processing step only for voice signal
The operations such as corresponding filtering, to improve the solution reverberation treatment effeciency and real-time of voice signal.
In some optional implementations, for example, VAD (Voice Activity Detection, voice can be passed through
Activity detection) technology carries out the identification of voice signal, so that the channel audio signal that judgement is got in step 210 is
No is voice signal.
Step 230, if so, updating the variance of the joint probability density distribution of early stage reverb signal, and it is mixed based on early stage
The variance for ringing the joint probability density distribution of signal updates the filter coefficient of sub-filter.
The variance of the joint probability density distribution of early stage reverb signal can be the filter of the sub-filter before updating
The product of the input signal vector of the transposed matrix and sub-filter of coefficient and the channel audio signal received in t moment
Spectral coefficient xt,fAbsolute value of the difference square.
Illustratively, the variance of the joint probability density distribution of early stage reverb signal can satisfy:
Wherein, xt,fSpectral coefficient for the channel audio signal received in t moment,For the sub-band filter before update
The transposition of the filter coefficient of device,For the input signal vector of sub-filter.
Updated filter coefficient gf(t+1) for the filter coefficient before updating and the sum of variable quantity is updated.Wherein, more
New change amount is that the first update running parameter and second update the ratio between running parameter.First updates running parameter for sub-filter
Iteration step length, sub-filter reality output and desired output between error and sub-filter input signal
The product of vector.Second updates the input of the transposition and sub-filter of the input signal vector that running parameter is sub-filter
The product of signal vector.Error between the reality output and desired output of sub-filter is equal to the list received in t moment
The ratio between the variance of the joint probability density distribution of the early stage reverb signal of the spectral coefficient and t moment of channel audio signal subtracts update
The product of the input signal vector of the transposition and sub-filter of the filter coefficient of preceding sub-filter.
Illustratively, updated filter coefficient gf(t+1) it can satisfy:
Wherein, μ is the iteration step length of sub-filter, and e (t) is between the reality output and desired output of sub-filter
Error e (t) meet:
Herein, the initial value g of the filter coefficient of sub-filterf(0) it can be single order number is N complete null vector.
Wherein, N is the number of taps of sub-filter, and is met:
N=L-D+1.
L is the frame time length ratio of reverberation time and one frame of channel audio signal, and D is the reverberation time of early stage reverberation
With the frame time length ratio of one frame of channel audio signal.And the reverberation time can for example be defined as indoor sound and reach stable
State, remnant voice is absorbed through sound-absorbing material repeatedly in the room after sound source stops sounding, and average sound pressure level is decayed needed for 60dB
Time.
In addition, the input signal vector of sub-filterIt can have the following form of expression:
In other words,Can be understood as sub-filter from the t-D moment to t-L+1 reception to input signal
The input signal vector of formation.
In addition, in formula (3):
Correspondingly, in formula (2) and formula (3):
By above-mentioned formula (1)~(3) as can be seen that in the side of the joint probability density distribution to early stage reverb signal
When difference is updated, the filter coefficient used is filter coefficient (the i.e. g of t momentfOr the g in formula (2)f(t)), exist
It completes the update of the variance of the joint probability density distribution of early stage reverb signal and then the coefficient of filter is updated
(the g i.e. in formula (2)f(t+1))。
Step 240, the audio signal after solution reverberation is determined based on updated filter coefficient.
Using formula as described above (1)~(3) to the variance of the joint probability density of early stage reverb signal distribution and
Audio signal d after the filter coefficient of sub-filter is updated, after solving reverberationt,fEqual to the single-pass received in t moment
The spectral coefficient of audio channel signal subtracts the transposition and the sub-filter of the filter coefficient of the sub-filter at t+1 moment
The product of input signal vector.
Illustratively, the audio signal d after reverberation is solvedT, fIt can satisfy:
In addition, the audio signal solution reverberation method of the present embodiment can also include such as in some optional implementations
Under step:
Judge whether the error between the reality output of sub-filter and desired output meets square of Error Absolute Value
Greater than the product square with preset threshold K of the absolute value of the spectral coefficient of the channel audio signal received in t moment.Namely
It is to say, judges whether the error e (t) between the reality output of sub-filter and desired output meets | e (t) |2> K × | xt,f
|2；
If so, filter coefficient is set to null vector.
Wherein, K is pre-set threshold value, and meets K > 1.
In addition, working as | e (t) |2> K × | xt,f|2When, it is believed that sub-filter diverging believes early stage reverberation in next step
Number joint probability density distribution variance when being updated, the g of formula (1)fFor null vector.
Using the audio signal solution reverberation method of the present embodiment, calculation amount needed for solving reverberation process is smaller, and operation
Journey only need t-D~t-L+1 moment between the channel audio signal that inputs and the channel audio signal of present frame input,
So that the real-time of the solution reverberation process of the present embodiment is stronger.
With further reference to Fig. 3, it illustrates the processes 300 of another embodiment of audio signal solution reverberation method.The sound
The process 300 of frequency signal solution reverberation method.
Can have in the present embodiment step 310 corresponding with step 210~step 240 in embodiment illustrated in fig. 2~
Step 340.
Unlike embodiment illustrated in fig. 2, the present embodiment be may further comprise:
Step 350, if judging result in step 320 be it is no, the joint of the early stage reverb signal before update is general
The connection of the variance of rate Density Distribution and the filter coefficient of the sub-filter before update as updated early stage reverb signal
Close the variance of probability density distribution and the filter coefficient of updated sub-filter.
In other words, if judging the channel audio signal got in step 310 in step 320 not is voice letter
Number, then the variance of the joint probability density distribution of early stage reverb signal and the filter coefficient of sub-filter are not updated, and
The variance of joint probability density distribution based on the early stage reverb signal before update and the filter coefficient pair of sub-filter
The channel audio signal got in step 310 carries out solution reverberation processing (step 340).
Compared with embodiment shown in Fig. 2, the audio signal solution reverberation method of the present embodiment can be sentenced to avoid in step 320
The output abnormality that the error of disconnected result may cause further promotes the method through the present embodiment treated the evening of audio signal
Phase reverberation filter effect.
With further reference to Fig. 4, as the realization to method shown in above-mentioned each figure, this application provides a kind of audio signal solutions
One embodiment of reverberation unit, the Installation practice is corresponding with embodiment of the method shown in Fig. 2, which can specifically answer
For in various electronic equipments.
As shown in figure 4, audio signal solution reverberation unit 400 described in the present embodiment includes obtaining module 410, judgment module
420, the first update module 430 and determining module 440:
Wherein:
It obtains module 410 to be configurable to obtain channel audio signal, channel audio signal includes early stage reverberation letter
Number and late reverberation signal.
Judgment module 420 is configurable to judge whether channel audio signal is voice signal.
If it is voice signal that the first update module 430, which is configurable to channel audio signal, early stage reverberation letter is updated
Number joint probability density distribution variance, and based on the joint probability density of early stage reverb signal distribution variance update subband
The filter coefficient of filter, wherein sub-filter is for filtering out the late reverberation signal for including in channel audio signal.
Determining module 440 is configurable to determine the audio signal after solution reverberation based on updated filter coefficient.
In some optional implementations, judgment module 420, which can be further configured, to be used for: by voice activity detection skill
Art judges whether channel audio signal is voice signal.
In some optional implementations,
The variance of the joint probability density distribution of early stage reverb signal can be the filter of the sub-filter before updating
The product of the input signal vector of the transposed matrix and sub-filter of coefficient and the channel audio signal received in t moment
Spectral coefficient xt,fAbsolute value of the difference square.
Illustratively, the variance of the joint probability density distribution of early stage reverb signal can satisfy:
Wherein, xt,fSpectral coefficient for the channel audio signal received in t moment,For the sub-band filter before update
The transposition of the filter coefficient of device,For the input signal vector of the sub-filter；
Updated filter coefficient gf(t+1) for the filter coefficient before updating and the sum of variable quantity is updated.Wherein, more
New change amount is that the first update running parameter and second update the ratio between running parameter.First updates running parameter for sub-filter
Iteration step length, sub-filter reality output and desired output between error and sub-filter input signal
The product of vector.Second updates the input of the transposition and sub-filter of the input signal vector that running parameter is sub-filter
The product of signal vector.Error between the reality output and desired output of sub-filter is equal to the list received in t moment
The ratio between the variance of the joint probability density distribution of the early stage reverb signal of the spectral coefficient and t moment of channel audio signal subtracts update
The product of the input signal vector of the transposition and sub-filter of the filter coefficient of preceding sub-filter.
Illustratively, updated filter coefficient gf(t+1) it can satisfy:
Wherein, μ is the iteration step length of sub-filter, and e (t) is between the reality output and desired output of sub-filter
Error e (t) meet:
Audio signal d in some optional implementations, after solving reverberationt,fEqual to the single channel received in t moment
The spectral coefficient of audio signal subtracts the defeated of the transposition of the filter coefficient of the sub-filter at t+1 moment and the sub-filter
Enter the product of signal vector.
Illustratively, the audio signal d after reverberation is solvedt,fIt can satisfy:
In some optional implementations, the audio signal solution reverberation unit of the present embodiment can further include
Two update module (not shown)s.
If it is not voice signal that the second update module, which is configurable to channel audio signal, the early stage before update is mixed
The filter coefficient of sub-filter before ringing the variance and update of the joint probability density distribution of signal is as updated morning
The variance of the joint probability density distribution of phase reverb signal and the filter coefficient of updated sub-filter.
In some optional implementations, the audio signal solution reverberation unit of the present embodiment, which can further include, to be set
Zero module (not shown).
Zero setting module is configurable to judge whether the error between the reality output of sub-filter and desired output is full
The absolute value of the spectral coefficient for square being greater than the channel audio signal received in t moment of sufficient Error Absolute Value square with it is pre-
If the product of threshold k.In other words, judge whether the error e (t) between the reality output of sub-filter and desired output is full
Foot | e (t) |2> K × | xt,f|2；And if so, filter coefficient is set to null vector；Wherein, K is pre-set threshold value,
And meet K > 1.
It will be understood by those skilled in the art that above-mentioned audio signal solution reverberation unit 400 can also include some other public affairs
Know structure, such as processor, memory etc., in order to unnecessarily obscure embodiment of the disclosure, these well known structures are in Fig. 4
In be not shown.
Below with reference to Fig. 5, it illustrates the calculating of the terminal device or server that are suitable for being used to realize the embodiment of the present application
The structural schematic diagram of machine system 500.
As shown in figure 5, computer system 500 includes central processing unit (CPU) 501, it can be read-only according to being stored in
Program in memory (ROM) 502 or be loaded into the program in random access storage device (RAM) 503 from storage section 508 and
Execute various movements appropriate and processing.In RAM 503, also it is stored with system 500 and operates required various programs and data.
CPU 501, ROM 502 and RAM 503 are connected with each other by bus 504.Input/output (I/O) interface 505 is also connected to always
Line 504.
I/O interface 505 is connected to lower component: the importation 506 including keyboard, mouse etc.；It is penetrated including such as cathode
The output par, c 507 of spool (CRT), liquid crystal display (LCD) etc. and loudspeaker etc.；Storage section 508 including hard disk etc.；
And the communications portion 509 of the network interface card including LAN card, modem etc..Communications portion 509 via such as because
The network of spy's net executes communication process.Driver 510 is also connected to I/O interface 505 as needed.Detachable media 511, such as
Disk, CD, magneto-optic disk, semiconductor memory etc. are mounted on as needed on driver 510, in order to read from thereon
Computer program be mounted into storage section 508 as needed.
Particularly, in accordance with an embodiment of the present disclosure, it may be implemented as computer above with reference to the process of flow chart description
Software program.For example, embodiment of the disclosure includes a kind of computer program product comprising be tangibly embodied in machine readable
Computer program on medium, the computer program include the program code for method shown in execution flow chart.At this
In the embodiment of sample, which can be downloaded and installed from network by communications portion 509, and/or from removable
Medium 511 is unloaded to be mounted.
Flow chart and block diagram in attached drawing are illustrated according to the system of the various embodiments of the application, method and computer journey
The architecture, function and operation in the cards of sequence product.In this regard, each box in flowchart or block diagram can generation
A part of one module, program segment or code of table, a part of the module, program segment or code include one or more
Executable instruction for implementing the specified logical function.It should also be noted that in some implementations as replacements, institute in box
The function of mark can also occur in a different order than that indicated in the drawings.For example, two boxes succeedingly indicated are practical
On can be basically executed in parallel, they can also be executed in the opposite order sometimes, and this depends on the function involved.Also it wants
It is noted that the combination of each box in block diagram and or flow chart and the box in block diagram and or flow chart, Ke Yiyong
The dedicated hardware based system of defined functions or operations is executed to realize, or can be referred to specialized hardware and computer
The combination of order is realized.
Being described in module involved in the embodiment of the present application can be realized by way of software, can also be by hard
The mode of part is realized.Described module also can be set in the processor, for example, can be described as: a kind of processor packet
It includes and obtains module, judgment module, the first update module and determining module.Wherein, the title of these modules is under certain conditions simultaneously
The restriction to the module itself is not constituted, is also described as " obtaining the mould of channel audio signal for example, obtaining module
Block ".
As on the other hand, present invention also provides a kind of nonvolatile computer storage media, the non-volatile calculating
Machine storage medium can be nonvolatile computer storage media included in device described in above-described embodiment；It is also possible to
Individualism, without the nonvolatile computer storage media in supplying terminal.Above-mentioned nonvolatile computer storage media is deposited
One or more program is contained, when one or more program is executed by an equipment, so that equipment: obtaining single channel sound
Frequency signal, channel audio signal include early stage reverb signal and late reverberation signal；Judge channel audio signal whether be
Voice signal；If so, updating the variance of the joint probability density distribution of early stage reverb signal, and based on early stage reverb signal
The variance of joint probability density distribution updates the filter coefficient of sub-filter, wherein sub-filter is for filtering out single-pass
The late reverberation signal for including in audio channel signal；And the single channel sound of solution reverberation is determined based on updated filter coefficient
Frequency signal.
Above description is only the preferred embodiment of the application and the explanation to institute's application technology principle.Those skilled in the art
Member is it should be appreciated that invention scope involved in the application, however it is not limited to technology made of the specific combination of above-mentioned technical characteristic
Scheme, while should also cover in the case where not departing from the inventive concept, it is carried out by above-mentioned technical characteristic or its equivalent feature
Any combination and the other technical solutions formed.Such as features described above has similar function with (but being not limited to) disclosed herein
Can technical characteristic replaced mutually and the technical solution that is formed.
Claims (10)
1. a kind of audio signal solution reverberation method characterized by comprising
Channel audio signal is obtained, the channel audio signal includes early stage reverb signal and late reverberation signal；
Judge whether the channel audio signal is voice signal；
If so, updating the variance of the joint probability density distribution of the early stage reverb signal, and believed based on the early stage reverberation
Number joint probability density distribution variance update sub-filter filter coefficient, wherein the sub-filter is used for
Filter out the late reverberation signal for including in the channel audio signal；And
The channel audio signal of solution reverberation is determined based on the updated filter coefficient；The judgement single channel sound
Whether frequency signal is that voice signal includes:
Judge whether the channel audio signal is voice signal by voice activity detection technology.
2. according to the method described in claim 1, it is characterized by:
The variance of the joint probability density distribution of the early stage reverb signal is the filter of the sub-filter before updating
The product of the input signal vector of the transposed matrix of coefficient and the sub-filter and the single channel audio received in t moment
The spectral coefficient x of signalt,fAbsolute value of the difference square；
Updated filter coefficient gf(t+1) for the filter coefficient before updating and the sum of variable quantity is updated；
Wherein, the update variable quantity is that the first update running parameter and second update the ratio between running parameter；
Described first update running parameter be the iteration step length of the sub-filter, the sub-filter reality output and
The product of the input signal vector of error and the sub-filter between desired output；
Described second updates the transposition and the sub-filter of the input signal vector that running parameter is the sub-filter
Input signal vector product；
Error between the reality output and desired output of the sub-filter is equal to the single channel audio received in t moment
The ratio between the variance of the joint probability density distribution of the early stage reverb signal of the spectral coefficient and t moment of signal subtracts described before updating
The product of the input signal vector of the transposition of the filter coefficient of sub-filter and the sub-filter.
3. the method according to claim 1, wherein the audio signal d after the solution reverberationt,fEqual in t moment
The spectral coefficient of the channel audio signal received subtract the transposition of the filter coefficient of the sub-filter at t+1 moment with it is described
The product of the input signal vector of sub-filter.
4. method according to claim 1 to 3, which is characterized in that in the judgement single channel audio letter
After number whether being voice signal, the method also includes:
If it is not, the subband before the variance and update that are then distributed the joint probability density of the early stage reverb signal before update is filtered
After variance and update that the filter coefficient of wave device is distributed as the joint probability density of the updated early stage reverb signal
Sub-filter filter coefficient.
5. method according to claim 1 to 3, which is characterized in that the method also includes:
Judge whether the error between the reality output and desired output of the sub-filter meets square of Error Absolute Value
Greater than the product square with preset threshold K of the absolute value of the spectral coefficient of the channel audio signal received in t moment；
If so, the filter coefficient is set to null vector；
Wherein, K > 1.
6. a kind of audio signal solution reverberation unit characterized by comprising
Obtain module, be configured to obtain channel audio signal, the channel audio signal include early stage reverb signal and
Late reverberation signal；
Judgment module is configured to judge whether the channel audio signal is voice signal；
First update module updates the early stage reverberation letter if being configured to the channel audio signal is voice signal
Number joint probability density distribution variance, and based on the joint probability density of the early stage reverb signal distribution variance update
The filter coefficient of sub-filter, wherein the sub-filter includes for filtering out in the channel audio signal
Late reverberation signal；And
Determining module is configured to determine the channel audio signal of solution reverberation based on the updated filter coefficient；Institute
Judgment module is stated further to be configured to:
Judge whether the channel audio signal is voice signal by voice activity detection technology.
7. device according to claim 6, it is characterised in that:
The variance of the joint probability density distribution of the early stage reverb signal are as follows:
The input signal of the transposed matrix of the filter coefficient of the sub-filter before update and the sub-filter to
The spectral coefficient x of the product of amount and the channel audio signal received in t momentt,fAbsolute value of the difference square；
Updated filter coefficient gf(t+1) for the filter coefficient before updating and the sum of variable quantity is updated；
Wherein, the update variable quantity is that the first update running parameter and second update the ratio between running parameter；
Described first update running parameter be the iteration step length of the sub-filter, the sub-filter reality output and
The product of the input signal vector of error and the sub-filter between desired output；
Described second updates the transposition and the sub-filter of the input signal vector that running parameter is the sub-filter
Input signal vector product；
Error between the reality output and desired output of the sub-filter is equal to the single channel audio received in t moment
The ratio between the variance of the joint probability density distribution of the early stage reverb signal of the spectral coefficient and t moment of signal subtracts described before updating
The product of the input signal vector of the transposition of the filter coefficient of sub-filter and the sub-filter.
8. device according to claim 6, which is characterized in that the audio signal d after the solution reverberationt,fEqual in t moment
The spectral coefficient of the channel audio signal received subtract the transposition of the filter coefficient of the sub-filter at t+1 moment with it is described
The product of the input signal vector of sub-filter.
9. according to device described in claim 6-8 any one, which is characterized in that described device further includes the second update mould
Block；
If it is not voice signal that second update module, which is configured to the channel audio signal, described in front of update
Described in the filter coefficient of sub-filter before the variance and update of the joint probability density distribution of early stage reverb signal is used as
The variance of the joint probability density distribution of updated early stage reverb signal and the filter coefficient of updated sub-filter.
10. according to device described in claim 6-8 any one, which is characterized in that described device further includes zero setting module；
Whether the zero setting module is configured to judge the error between the reality output and desired output of the sub-filter
Meet the absolute value of the spectral coefficient for square being greater than the channel audio signal received in t moment of Error Absolute Value square with
The product of preset threshold K；And
If so, the filter coefficient is set to null vector；
Wherein, K > 1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610474006.1A CN105931648B (en) | 2016-06-24 | 2016-06-24 | Audio signal solution reverberation method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610474006.1A CN105931648B (en) | 2016-06-24 | 2016-06-24 | Audio signal solution reverberation method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105931648A CN105931648A (en) | 2016-09-07 |
| CN105931648B true CN105931648B (en) | 2019-05-03 |
Family
ID=56829221
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610474006.1A Active CN105931648B (en) | 2016-06-24 | 2016-06-24 | Audio signal solution reverberation method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105931648B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107328410B (en) * | 2017-06-30 | 2020-07-28 | 百度在线网络技术（北京）有限公司 | Method for locating an autonomous vehicle and vehicle computer |
| CN107328411B (en) * | 2017-06-30 | 2020-07-28 | 百度在线网络技术（北京）有限公司 | Vehicle-mounted positioning system and automatic driving vehicle |
| CN111489760B (en) * | 2020-04-01 | 2023-05-16 | 腾讯科技（深圳）有限公司 | Speech signal dereverberation processing method, device, computer equipment and storage medium |
| CN113223543B (en) * | 2021-06-10 | 2023-04-28 | 北京小米移动软件有限公司 | Speech enhancement method, device and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2009535674A (en) * | 2006-05-01 | 2009-10-01 | 日本電信電話株式会社 | Method and apparatus for speech dereverberation based on stochastic model of sound source and room acoustics |
| CN103033815A (en) * | 2012-12-19 | 2013-04-10 | 中国科学院声学研究所 | Detection Method and detection device of distance expansion target based on reverberation covariance matrix |
| CN104995676A (en) * | 2013-02-14 | 2015-10-21 | 杜比实验室特许公司 | Signal decorrelation in an audio processing system |
| WO2016014254A1 (en) * | 2014-07-23 | 2016-01-28 | Pcms Holdings, Inc. | System and method for determining audio context in augmented-reality applications |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8886526B2 (en) * | 2012-05-04 | 2014-11-11 | Sony Computer Entertainment Inc. | Source separation using independent component analysis with mixed multi-variate probability density function |
| TWI618051B (en) * | 2013-02-14 | 2018-03-11 | 杜比實驗室特許公司 | Audio signal processing method and apparatus for audio signal enhancement using estimated spatial parameters |
-
2016
- 2016-06-24 CN CN201610474006.1A patent/CN105931648B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2009535674A (en) * | 2006-05-01 | 2009-10-01 | 日本電信電話株式会社 | Method and apparatus for speech dereverberation based on stochastic model of sound source and room acoustics |
| CN103033815A (en) * | 2012-12-19 | 2013-04-10 | 中国科学院声学研究所 | Detection Method and detection device of distance expansion target based on reverberation covariance matrix |
| CN104995676A (en) * | 2013-02-14 | 2015-10-21 | 杜比实验室特许公司 | Signal decorrelation in an audio processing system |
| WO2016014254A1 (en) * | 2014-07-23 | 2016-01-28 | Pcms Holdings, Inc. | System and method for determining audio context in augmented-reality applications |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105931648A (en) | 2016-09-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105931648B (en) | Audio signal solution reverberation method and device | |
| CN109121057B (en) | Intelligent hearing aid method and system | |
| CN110428842A (en) | Speech model training method, device, equipment and computer readable storage medium | |
| CN108196820A (en) | For adjusting the method and apparatus of play parameter | |
| CN110444202B (en) | Composite voice recognition method, device, equipment and computer readable storage medium | |
| CN106165015B (en) | Apparatus and method for facilitating watermarking-based echo management | |
| CN111986691B (en) | Audio processing method, device, computer equipment and storage medium | |
| CN111863015A (en) | Audio processing method and device, electronic equipment and readable storage medium | |
| CN109410918A (en) | For obtaining the method and device of information | |
| CN109817222A (en) | A kind of age recognition methods, device and terminal device | |
| CN112562648A (en) | Adaptive speech recognition method, apparatus, device and medium based on meta learning | |
| CN112259116A (en) | Method and device for reducing noise of audio data, electronic equipment and storage medium | |
| CN110428835A (en) | A kind of adjusting method of speech ciphering equipment, device, storage medium and speech ciphering equipment | |
| EP4040764A2 (en) | Method and apparatus for in-vehicle call, device, computer readable medium and product | |
| CN114898762A (en) | Real-time voice noise reduction method and device based on target person and electronic equipment | |
| CN113555032A (en) | Multi-speaker scene recognition and network training method and device | |
| CN105701686A (en) | Voiceprint advertisement implementation method and device | |
| CN112992190B (en) | Audio signal processing method and device, electronic equipment and storage medium | |
| CN110169082A (en) | Combining audio signals output | |
| CN106340310B (en) | Speech detection method and device | |
| CN111508500B (en) | Voice emotion recognition method, system, device and storage medium | |
| CN115295024A (en) | Signal processing method, signal processing device, electronic apparatus, and medium | |
| CN107798556A (en) | For updating method, equipment and the storage medium of situation record | |
| CN113823312A (en) | Speech enhancement model generation method and device and speech enhancement method and device | |
| Liu et al. | Speech enhancement with stacked frames and deep neural network for VoIP applications |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |