CN106572419B - A kind of stereo audio enhancing system - Google Patents
A kind of stereo audio enhancing system Download PDFInfo
- Publication number
- CN106572419B CN106572419B CN201510645398.9A CN201510645398A CN106572419B CN 106572419 B CN106572419 B CN 106572419B CN 201510645398 A CN201510645398 A CN 201510645398A CN 106572419 B CN106572419 B CN 106572419B
- Authority
- CN
- China
- Prior art keywords
- signal
- unit
- frequency
- module
- submodule
- 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
Landscapes
- Stereophonic System (AREA)
Abstract
The present invention relates to a kind of stereo audios to enhance system, which includes：Signal separation module, dialogue enhance module, decorrelation processing module, signal synthesizing module, environmental sound module, virtual bass boost module and dynamic range control module；The system is for output after being pre-processed the stereo signal received to playback equipment to obtain desired audio.
Description
Technical field
The present invention relates to Underwater Acoustic channels technical field, more particularly to a kind of stereo audio enhances system.
Background technology
Currently, multichannel ambiophonic system has been widely used, for example, the 5.1/7.1 formats of Doby and DTS and
22.2 formats of NHK etc..But in these terminal devices such as laptop, smart television, PAD, wireless sound box and mobile phone
In, the system of two channel stereo formats is still widely used, and can also long-term existence.These above-mentioned electronic products
Playback equipment is usually small size loud speaker.But there are many defects, such as miniature loudspeaker for such small size loud speaker
Due to being limited by physical conditions such as unit sizes, low frequency signal can not be effectively reset；The high frequency loss of many loud speakers
Also than more serious.In short, the Frequency Response of this kind of loud speaker is bad, non-linear distortion is than more serious.In many practical applications
In, the distance between two loud speakers are close, this makes stereosonic sound field narrow, and audience can not experience the ring really shaken
Around sound effective value.In addition, user often complains that not clear enough the problem of dialogue, user are desirable to hear improved sound in occasion in this way
Sound.
To solve the above-mentioned problems, there are a variety of audio effect processing technologies.HRTF is a kind of to be used for by extensive discussions
The technology for manufacturing virtual surround sound has corresponding processing method in the pretreatment of earphone or loud speaker.But HRTF is high
Degree personalization, everyone has a set of HRTF data of oneself, and there are prodigious othernesses by the HRTF of different people.It is another
Aspect, HRTF processing can so that tone color changes and sound is unnatural.These defects make the practical application of HRTF by very big
Limitation.Another, which is so-called stereo enhancement technology, such as SRS, Dolby and WAVES etc., has corresponding business to solve
Certainly scheme.These technologies all achieve certain success in electronics field, but the perception due to different people to sound
With very strong subjectivity, it also is difficult to that there are one specific conclusions to the evaluation of these technologies.Patent of the present invention is above-mentioned from solving
There are problems to set out, and provides a kind of audio Enhancement Method of suitable consumer electronics field application.
This patent provides a kind of method extending sound field width using decorrelation technique.Decorrelation technique is a kind of classical
Acoustic image method for widening, the very strong signal decomposition of correlation is that the signal with low correlation is reset by it, to reduce ears
Sense of hearing cross-correlation coefficient (Inter-Aural Cross Correlation, IACC), to obtain the acoustic image of broadening, contributes to
Improve the problem that sound field is narrow during general audio frequency apparatus plays.The physical basis of this method is：One broad sound source can be with
Several point sources being spatially separating are decomposed into, if there is this group of point source high correlation, hearer can only perceive positioned at this group of point source
One narrow acoustic image of center of gravity；If there is this group of point source low correlation, hearer can experience close with original broad sound source
Broad acoustic image.Decorrelating method is not to create physically correct acoustic image, but psychologic acoustics is used to act on to obtain one
A diffusion, broad acoustic image.Therefore, this method significantly reduces computation complexity, has very strong actual application value.
Foregoing miniature loudspeaker is limited by physical conditions such as unit sizes, can not effectively reset low frequency letter
Number.Traditionally use the method for balanced device or frequency divider that system energy consumption can be brought to increase, efficiency reduces, not readily portable etc. many
Problem.Therefore, the virtual bass (virtual of psychologic acoustics missing fundamental (Missing Fundmental) principle is utilized
Bass) enhancing technology can effectively enhance miniature loudspeaker replaying bass quality.Existing virtual bass boost technology is mainly led to
It crosses different harmonic waves and generates the harmonic components that unit generates low frequency signal, be broadly divided into Time-Domain algorithm and frequency domain algorithm two major classes.
Time harmonic generates unit：Half-wave full-wave rectifier and unlimited feedback multiplier etc..Due to the presence of nonlinear interaction, make
These technologies also will produce serious intermodulation distortion while generating harmonic wave, influence the sound quality of audio playback.Frequency domain harmonic produces
Raw unit utilizes phase vocoder to generate harmonic wave and is limited to time-frequency although this method can reduce non-linear distortion mostly
Resolution ratio will produce transient state and obscure and the distortions such as reverberation effect.These defects constrain virtual bass boost to a certain extent
The promotion and application of technology.
Invention content
It is an object of the present invention to solve in existing technology, the sound field actually presented is narrow so that surround sound
Ineffective, being submerged in ambient sound to white signal causes dialogue not clear enough, and replaying bass is ineffective and loud speaker dynamic
The small problem of range, the present invention provides a kind of stereo audios to enhance system.
The present invention provides a kind of stereo audios to enhance system, including：Signal separation module, dialogue enhance module, solution
Related process module, signal synthesizing module, environmental sound module, virtual bass boost module and dynamic range control module；It should
System is for output after being pre-processed the stereo signal received to playback equipment to obtain desired audio.
Concrete implementation method is as follows：The left channel signals x that the signal separation module receives systemLinThe right side and
Sound channel signal xRinIt detaches and obtains through acoustical signal xsWith circular acoustical signal xd；Wherein, the through acoustical signal xsIt is input to
The dialogue enhances module, and the dialogue enhances module to above-mentioned through acoustical signal xsIt is handled, is exported
xsout；And around acoustical signal xdIt is input to the decorrelation processing module, the decorrelation processing module is by above-mentioned surround sound
Signal xdTwo output signal x with very low correlation are obtained by a pair of of de-correlation filter groupL1And xR1；Described
xsout、xL1And xR1Three signals fully enter the signal synthesizing module, and the signal synthesizing module is obtained above-mentioned
Xsout、xL1And xR1Three signals are weighted combination and obtain new left channel signals xL2With right-channel signals xR2；Then, institute
The environmental sound module stated is by xL2And xR2Respectively with the binaural room impulse response h in left and right channelLAnd hRIt is obtained after carrying out convolution
Export xL3And xR3；Then, the virtual bass boost module receives input signal xL3And xR3It is obtained after progress bass boost defeated
Go out xL4And xR4；The dynamic range control module, according to the adaptive adjusting x of the dynamic range of output equipmentL4And xR4Width
Degree, obtains the output signal x of systemLoutAnd xRout。
As a further improvement of the above technical scheme, dialogue enhancing module is an IIR or FIR filtering
Device, the amplitude spectrum of the filter are to be obtained to contour of equal loudness against after being modified.
As a further improvement of the above technical scheme, the decorrelation processing module is to pass through the input of the module
A pair of of de-correlation filter hLDAnd hRDAfter export xL1And xR1, the filter hLDAnd hRDAmplitude near flat is rung
It answers, and its phase sound is then completely uncorrelated or weak relevant.
As a further improvement of the above technical scheme, the signal synthesizing module is by the de-correlation block
Export xL1And xR1It is multiplied by gain factor k respectively, then adds the output x of the dialogue enhancing modulesoutWith gain factor ρ's
The left channel signals x that product is exportedL2With right-channel signals xR2, the wherein range of k is 0≤k≤10, and the range of ρ is 0≤ρ
≤10。
As a further improvement of the above technical scheme, the binaural room impulse response is first by image method mould
Intend specified acoustics scene or measure acquisition binaural room impulse response in actual room, then to its early reflection
What part point carried out obtaining after simplifying.
As a further improvement of the above technical scheme, the virtual bass boost module includes：
First high pass filter unit, first high pass filter unit is to input signal xL3High-pass filtering is carried out, it is described
The first high pass filter unit cutoff frequency be system speaker low-frequency cut-off frequency；
Second high pass filter unit, second high pass filter unit is to input signal xR3High-pass filtering is carried out, it is described
The second high pass filter unit cutoff frequency be system speaker low-frequency cut-off frequency；
First summing unit, first summing unit is by left channel signals xL3With right-channel signals xR3It is added；
Low-pass filter unit, the low-pass filter unit extracts low frequency signal from first summing unit, described
The cutoff frequency of low-pass filter unit is the low-frequency cut-off frequency of system speaker；
Down-sampled unit, the down-sampled unit carry out the low frequency signal that the low-pass filter unit extracts down-sampled
Processing, the down-sampled factor are determined by the low-frequency cut-off frequency of system speaker and the sample rate of input audio signal；
Preanalysis and separative element, the preanalysis and separative element are for the low frequency obtained by the down-sampled unit
Signal carries out fundamental frequency estimation and to high frequency obtained by first high pass filter unit and second high pass filter unit
Signal carries out transient state Induction Peried detection；
Harmonic wave generates unit, and the harmonic wave generates unit and handles fundamental component and remaining point respectively using different methods
Amount；
Sampling unit is risen, described liter of sampling unit generates the signal that unit generates to the harmonic wave and carries out a liter sampling processing,
It is equal with the down-sampled factor of down-sampled unit to rise decimation factor；
First delay unit, first delay unit to by signal obtained by first high pass filter unit into line delay
Processing, delay duration delay duration caused by the preanalysis and separative element determine；
Second delay unit, second delay unit to by signal obtained by second high pass filter unit into line delay
Processing, delay duration delay duration caused by the preanalysis and separative element determine；
Second summing unit carries out signal obtained by signal obtained by first delay unit and the bandpass filtering unit
It is added, obtains new left channel signals xL4；
Third summing unit carries out signal obtained by signal obtained by second delay unit and the bandpass filtering unit
It is added, obtains new right-channel signals xR4。
Bandpass filtering unit, the bandpass filtering unit by the design of low-frequency cut-off frequency and high-frequency cut-off frequency,
Bandpass filtering treatment is carried out to described liter of sampling unit, filters out low frequency part and high-frequency harmonic distorted portion.
As a further improvement of the above technical scheme, the preanalysis and separative element further comprise：
Audio mixing synthesizes submodule, and audio mixing synthesis submodule will pass through the first, the second high-pass filtering list
First and generation audio signal xLhpAnd xRhpIt is synthesized, generates audio signal xMhp；
Transient state Induction Peried detection sub-module, the transient state Induction Peried detection sub-module is to input signal time domain energy
Envelope be detected.In the starting of oscillation part of note, transient components account for it is leading, so as to judge whether music signal is in
Transient state starting of oscillation state；
Adaptive fundamental frequency tracks submodule, and adaptive fundamental frequency tracking submodule uses two constrained based on zero pole point
Rank lattice trapper is tracked to do adaptive base spectrum line, updates trap parameter using lattice recursive least squares；
Filter parameter designs submodule, and the filter parameter designs submodule and utilizes the transient state Induction Peried
The testing result T of detection sub-moduleAEThe bandwidth of trapper is controlled, when being judged as the transient state starting of oscillation moment, it is believed that audio is believed at this time
Number based on transient state broadband signal, harmonic components are weaker, and improved sigmoid functions is utilized to change the bandwidth factor α of trapper
Reduce bandwidth；Smoothing processing is done for the fundamental frequency estimation result F of the adaptive fundamental frequency tracking submodule simultaneously, is reduced adaptive
The influence of oscillation；
Dynamic cascading trap submodule, the dynamic cascading trap submodule are cascaded by the adjustable trapper of multiple bandwidth
It forms, the bandwidth of trapper is controlled by the transient state Induction Peried detection sub-module, when being judged as the transient state starting of oscillation moment, is reduced
Bandwidth；The centre frequency of trapper is determined by the adaptive fundamental frequency tracking submodule, need to track submodule to adaptive fundamental frequency
The fundamental frequency estimation result of block does smoothing processing.
As a further improvement of the above technical scheme, the harmonic wave generates unit and further comprises：
Residual components harmonic generation submodule, the residual components harmonic generation submodule utilize special non-linear letter
Number handles the residual components of the dynamic cascading trap submodule output, generates the harmonic signal of residual components；
Fundamental component harmonic generation submodule, the fundamental component harmonic generation submodule utilize specific harmonic frequency
Calculation formula is calculated by the instantaneous of 2 to 6 harmonics of the adaptive frequency tracking submodule estimation gained fundamental frequency
Frequency；
Harmonic energy adjusts submodule, and the harmonic energy adjusts submodule and uses the fundamental component harmonic generation
The instantaneous frequency for the harmonics that submodule estimates adjusts the amplitude of each harmonic according to psychoacoustic principle；
Harmonic and reactive detection submodule, the harmonic and reactive detection submodule are produced using the residual components harmonic generation submodule
Raw harmonic component xvb_resiThe harmonic component x generated with the harmonic energy adjustment submodulevb_harmWeighting synthesis is done, is obtained
To virtual bass harmonic component xvb。
The advantage of the invention is that：Non-linear distortion is reduced, the Auditory Perception of bass is improved, effectively increases dialogue
Clarity so that the dialogue in film or in music is more clear variable, and surrounding sound effect is obviously improved, computation complexity drop
It is low, it not will produce that transient state is fuzzy and reverberation effect.
Description of the drawings
Fig. 1 is the schematic block diagram of the stereo audio enhancing system of the present invention；
Fig. 2 is the dialogue enhancing mould filter in the block of the present invention；
Fig. 3 is the phase response of the de-correlation filter group of the present invention；
Fig. 4 is the binaural room impulse response Functions in Time Domain waveform of the present invention；
Fig. 5 is the binaural room impulse response Functions in Time Domain waveform that the simplification of the present invention obtains；
Fig. 6 is the environmental sound specific implementation block diagram of the present invention；
Fig. 7 is the frequency response of certain TV loudspeaker of the present invention；
Fig. 8 is the structural schematic diagram of the virtual supper bass enhancing system of the present invention；
Fig. 9 is the preanalysis of the present invention and the structural schematic diagram of separative element；
Figure 10 is that the virtual bass harmonic wave of the present invention generates the structural schematic diagram of unit module；
Figure 11 is the dynamic range control unit of the present invention；
Figure 12 is the dynamic range control input-output mappings relational graph of the present invention.
102 signal separation module, 104 dialogue enhances module
106 decorrelation processing module, 108 signal synthesizing module
110 environmental sound module, 112 virtual bass boost module
114 dynamic range control module, 801 low-pass filter unit
802 down-sampled unit, 803 preanalysis and separative element
804 harmonic waves generate 805 liters of sampling units of unit
806 bandpass filtering unit the first high pass filter units of 807a
807b the second high pass filter unit the first delay units of 808a
809 first summing unit of the second delay units of 808b
810a the second summing unit 810b third summing units
901 transient state Induction Peried detection sub-module, 902 adaptive fundamental frequency tracks submodule
903 filter parameters design 904 dynamic cascading trap submodule of submodule
1001 residual components harmonic generation submodule, 1002 fundamental component harmonic generation submodule
1004 harmonic energies adjust 1005 harmonic and reactive detection submodule of submodule
The 4th delay units of 1102a third delay units 1102b
1104 envelope detected unit, 1106 gain calculating unit
1108a the first multiplier the second multipliers of 1108b
Specific implementation mode
In conjunction with attached drawing, the invention will be further described.
As shown in Figure 1, the present invention provides a kind of stereo audios to enhance system, including：Signal separation module 102 is right
White enhancing module 104, decorrelation processing module 106, signal synthesizing module 108, environmental sound module 110, virtual bass boost
Module 112 and dynamic range control module 114；The system for output after being pre-processed the stereo signal received to
Playback equipment is to obtain desired audio.
The dialogue enhancing module 104 is an IIR or FIR filter, and the amplitude-frequency response of the filter is reference
The inverse of the loudness contour of human ear is designed；
The decorrelation processing module 106 is that the input of the module is passed through a pair of of de-correlation filter hLDAnd hRDAfterwards
Export xL1And xR1, the filter hLDAnd hRDAmplitude response near flat and its phase response are completely uncorrelated
Or it is weak relevant；
The signal synthesizing module 108 is by the output x of the de-correlation block 106L1And xR1It is multiplied by gain respectively
Then factor k adds the output x of the dialogue enhancing module 104soutThe left sound exported with the product of gain factor ρ
Road signal xL2With right-channel signals xR2, the wherein range of k is 0≤k≤10, and the range of ρ is 0≤ρ≤10.
The environmental sound module 110 is will to input xL2And xR2Respectively by the binaural room impulse response of setting come
Enhance the acoustic efficiency of specific environment, the binaural room impulse response is first by the specified acoustic field of image method simulation
Scape measures acquisition binaural room impulse response in actual room, then simplifies to its reflection part
It obtains afterwards；
As shown in figure 8, the virtual bass boost module 112 further comprises：
First high pass filter unit 807a, the first high pass filter unit 807a is to input signal xL3Carry out high pass filter
Wave, the cutoff frequency of the first high pass filter unit 807a are the low-frequency cut-off frequency of system speaker；
Second high pass filter unit 807b, the second high pass filter unit 807b is to input signal xR3Carry out high pass filter
Wave, the cutoff frequency of the second high pass filter unit 807b are the low-frequency cut-off frequency of system speaker；
First summing unit 809, first summing unit 809 is by left channel signals xL3With right-channel signals xR3It carries out
It is added；
Low-pass filter unit 801, the low-pass filter unit 801 extract low frequency from first summing unit 809
Signal, the cutoff frequency of the low-pass filter unit 801 are the low-frequency cut-off frequency of system speaker；
Down-sampled unit 802, the low frequency signal that the down-sampled unit 802 extracts the low-pass filter unit 801 into
The down-sampled processing of row, the down-sampled factor determine by the low-frequency cut-off frequency of system speaker and the sample rate of input audio signal,
It is set as 16 times of down-sampled units；
Preanalysis and separative element 803, the preanalysis and separative element 803 are for the down-sampled unit
The low frequency signal of 802 gained carries out fundamental frequency estimation and to the first high pass filter unit 807a and second high pass
High-frequency signal obtained by filter unit 807b carries out transient state Induction Peried detection；
Harmonic wave generates unit 804, and the harmonic wave generates unit 804 and handles base respectively using different virtual bass technologies
Frequency component and residual components；
Sampling unit 805 is risen, described liter of sampling unit 805 generates the signal that unit 804 generates to the harmonic wave and rise
Sampling unit processing, it is equal with the down-sampled factor of the down-sampled unit 802 to rise decimation factor；
First delay unit 808a, the first delay unit 808a is to obtained by the first high pass filter unit 807a
Signal carries out delay process, and delay duration delay duration caused by the preanalysis and separative element 803 determines；
Second delay unit 808b, the second delay unit 808b is to obtained by the second high pass filter unit 807b
Signal carries out delay process, and delay duration delay duration caused by the preanalysis and separative element 803 determines；
Second summing unit 810a, by signal and the bandpass filtering unit 806 obtained by the first delay unit 808a
Gained signal is added, and new left channel signals x is obtainedL4；
Third summing unit 810b, by signal and the bandpass filtering unit 806 obtained by the second delay unit 808b
Gained signal is added, and new right-channel signals x is obtainedR4；
Bandpass filtering unit 806, the bandpass filtering unit 806 pass through low-frequency cut-off frequency and high-frequency cut-off frequency
Design, filters out the low frequency part and high-frequency harmonic distorted portion that tone playing equipment can not restore.Utilize the bandpass filtering list
806 finishing of member rises the signal after sampling unit, obtains xmp。
As shown in figure 9, the preanalysis and separative element 803 further comprise：
Audio mixing synthesizes submodule 905, and audio mixing synthesis submodule 905 will pass through the first, second high pass
Filter unit 807a and 807b and the audio signal x generatedLhpAnd xRhpIt is synthesized, generates audio signal xMhp；
Transient state Induction Peried detection sub-module 901, when the transient state Induction Peried detection sub-module 901 is to input signal
The envelope of domain energy is detected.In the starting of oscillation part of note, transient components account for it is leading, so as to judge that music signal is
It is no to be in transient state starting of oscillation state；
Adaptive fundamental frequency tracks submodule 902, and the adaptive fundamental frequency is tracked submodule 902 and used based on zero pole point about
The second order lattice trapper of beam is tracked to do adaptive base spectrum line, using lattice recursive least squares update trapper ginseng
Number；
Filter parameter designs submodule 903, and the filter parameter is designed submodule 903 and risen using the transient state
The testing result T for time detection sub-module 901 of shakingAEThe bandwidth of trapper is controlled, when being judged as the transient state starting of oscillation moment, can be recognized
It is audio signal at this time based on transient state broadband signal, harmonic components are weaker, and improved sigmoid functions is utilized to change trapper
Bandwidth factor α reduce bandwidth.It is done smoothly for the fundamental frequency estimation result F of the adaptive fundamental frequency tracking submodule 902 simultaneously
Processing, reduces the influence of adaptive oscillation.
Dynamic cascading trap submodule 904, the dynamic cascading trap submodule 904 is by the adjustable trap of multiple bandwidth
Device cascades, and the bandwidth of trapper is controlled by the transient state Induction Peried detection sub-module 901, when being judged as transient state starting of oscillation
Moment reduces bandwidth.The centre frequency of trapper is tracked submodule 902 by the adaptive fundamental frequency and is determined, need to be to described
The fundamental frequency estimation result of adaptive fundamental frequency tracking submodule 902 does smoothing processing.
As shown in Figure 10, the harmonic wave generates unit 804 and further comprises：
Residual components harmonic generation submodule 1001, it is special that the residual components harmonic generation submodule 1001 utilizes
Nonlinear function handles the residual components that the dynamic cascading trap submodule 904 exports, and manufactures the harmonic signal of residual components；
Fundamental component harmonic generation submodule 1002, the fundamental component harmonic generation submodule 1002 is using specifically
Harmonic frequency calculation formula calculates the wink of the fundamental frequency fundamental frequency exported by the adaptive fundamental frequency tracking submodule 902
When frequency；
Harmonic energy adjusts submodule 1004, and the harmonic energy adjusts submodule 1004 and uses the fundamental component
The instantaneous frequency for the fundamental frequency that harmonic generation submodule 1002 estimates adjusts the amplitude of each harmonic according to psychoacoustic principle.
Harmonic and reactive detection submodule 1005, the harmonic and reactive detection submodule 1005 utilize the residual components harmonic generation
The harmonic component x that submodule 1001 generatesvb_resiThe harmonic component generated with the harmonic energy adjustment submodule 1004
xvb_harmWeighting synthesis is done, virtual bass harmonic component x is obtainedvb。
Concrete implementation method is as follows：
Step 1, as shown in Figure 1, the left channel signals x that the signal separation module 102 receives systemLinThe right side and
Sound channel signal xRinIsolate through acoustical signal xsWith circular acoustical signal xd.Currently, document has many methods that direct sound wave may be implemented
With the extraction around acoustical signal, a kind of possible realization method is only provided here.In the signal separation module 102, adopt
With Signal separator is realized based on the method for adaptive Panning, iterated to calculate out corresponding to left and right acoustic channels using following equation
The weight w of signalLAnd wR：
xs=wLxLin+wRxRin (1)
Wherein μ is iteration step length, determines that convergence speed of the algorithm, value range are 0≤μ≤1, the bigger convergence of the value
Speed is faster.It can be computed accordingly around acoustical signal：xd=wRxLin-wLxRin。
Step 2, the dialogue enhance module 104 to the through acoustical signal xsIt is handled to obtain output xsoutWith
Enhancing be included in it is stereo in white signal, improve speech intelligibility；The decorrelation processing module 106 is surround above-mentioned
Acoustical signal xdTwo output signal x with very low correlation are obtained by a pair of of de-correlation filter groupL1And xR1.It is specific next
It says, it is by input signal x that the dialogue, which enhances module 104,SProcessing output is carried out by the filter of a special designing
xsout.The amplitude-frequency response of the filter is that the inverse of the loudness contour with reference to human ear designs, it is therefore an objective to as small as possible
Dialogue clarity is improved to the greatest extent under the premise of increasing signal amplitude so that the dialogue in film or in music clearly may be used
It distinguishes.Loudness contour with reference to human ear is known that human ear to the intermediate frequency information susceptibility higher of sound and low frequency and high frequency is believed
Breath is then relatively not sensitive enough, and carrying out promotion in the frequency range of human ear sensitivity helps to make one to perceive the volume of bigger, to be promoted
The clarity of voice.A dialogue based on the design of this principle enhances filter, as shown in Fig. 2, should be kept away in 1000Hz or less
Exempt to enhancing white signal, since 1000Hz enhancing amplitude be gradually increased to 5000Hz reach maximum value (value according to
Specific application settings, are not higher than 18dB generally), the amplitude then enhanced since 10000Hz is gradually reduced, can be effective
Promotion voice clarity.
Meanwhile using a kind of new acoustic image extended method to the circular acoustical signal x that isolatesdIt is handled, according to psychology
Principles of Acoustics, ears cross-correlation coefficient have substantial connection with the sound source width that audience perceives, are defined as：
If ears cross-correlation coefficient is close to 1, audience perceives a very narrow dimension；If ears cross-correlation coefficient
Close to zero, then audience perceives a very wide dimension.Dialogue sound should there are one specific directions to position, but surround
As soon as sound should not come from a fixed orientation but should be full of entire space, this, which needs to reduce ears to the greatest extent, receives
The degree of correlation around acoustical signal.Therefore, the decorrelation processing module 106 utilizes a pair of of de-correlation filter hLDAnd hRD
It is respectively acting on around acoustical signal xdTo feed back to the new circular acoustical signal x of left and right acoustic channelsL1And xR1, receive reducing ears
Around the correlation of acoustical signal.hLDAnd hRDIt can be FIR or iir filter, but must assure that they have approximate straight width
Frequency response is answered, and its phase should be weak related or completely unrelated.As shown in figure 3, the phase response of two filters is
It generates at random, amplitude response is all 1 in all frequencies, and the pulse that time domain is then obtained using inverse Fourier transform is rung
It answers.In order to reduce computation complexity, the de-correlation filter group of IIR forms can also be designed.In other embodiments, also according to
Need the de-correlation filter group of design time-varying.
Step 3, the signal synthesizing module 108 is by x obtained abovesout、xL1And xR1Three signals are weighted group
Conjunction obtains new left channel signals xL2With right-channel signals xR2.More exactly, described by executing following operation
Signal synthesizing module 108 synthesizes new left channel signals xL2With right-channel signals xR2：
xL2=ρ xsout+kxL1 (5)
xR2=ρ xsout+kxR1
Wherein gain factor ρ and k is respectively intended to adjust the ratio to white signal and between acoustical signal.Larger k can
Preferably to protrude background sound and obtain better spatial impression；And smaller k values can then suppress ambient sound.ρ values are bigger, dialogue
The ratio that sound accounts for is more, and dialogue clarity is more preferable.Different parameters should be arranged to different application scenarios, such as at some scenes
In the music program of recording, background sound causes voice to be flooded by background sound than more prominent, at this moment can tune low k-value appropriate with
Prominent voice；In film, it is sometimes desirable to which at this moment the scene effect for building shock can increase k values.The suggestion value of k and ρ
Ranging from (0≤k≤10) and (0≤ρ≤10).
Step 4, the environmental sound module 110 is by xL2And xR2Respectively with the binaural room impulse response in left and right channel
hLAnd hROutput x is obtained after carrying out convolutionL3And xR3.As shown in fig. 6, the environmental sound module 110 receives the signal
The output signal x of synthesis module 108L2And xR2, it is somebody's turn to do respectively with after the aforementioned binaural room impulse response convolution calculated
The output signal x of moduleL3And xR3.Specifically left channel signals xL2Respectively pass through N number of delay unit (602a in corresponding diagram,
604a, 606a etc.), amount of delay is respectively Li (1≤i≤N), the output of each delay unit respectively with corresponding gain factor
For gLi(1≤i≤N) is multiplied (corresponding multiplier is respectively 608a, 610a, 612a etc.), and results added is obtained L channel
Output xL3.Similar, right-channel signals xR2Pass through M delay unit (602b in corresponding diagram, 604b, 606b etc.) respectively,
Its amount of delay is respectively Ri (1≤i≤M), and the output of each delay unit is respectively g with corresponding gain factorRi(1≤i≤M)
It is multiplied (corresponding multiplier is respectively 608b, 610b, 612b etc.), and results added is obtained the output x of right channelR3。
Step 5, the virtual bass boost module 112 receive input signal xL3And xR3It is defeated after progress bass boost afterwards
Go out xL4And xR4.As shown in fig. 7, being the amplitude-frequency response figure of a TV loudspeaker.The low frequency performance of the loud speaker is bad, no
Can effectively replay signal 180Hz frequency contents below.Based on this, the present embodiment provides a kind of reduction loud speaker cutoff frequencies
The following voice frequency signalling system of rate.The system is based on adaptive notch filter fundamental frequency tracer technique and energy envelope starting of oscillation detection technique,
It can be reliably separated and be generated each harmonic of audio Signal Pitch, the defect of traditional scheme is compensated for, improve supper bass
The effect of playback.Specifically, as shown in figure 8, first, the three-dimensional vocal input of the virtual bass boost module 112 is believed
Number xL3And xR3By first summing unit 809, stereo mix audio x is obtainedmix.Utilize described first, second
High pass filter unit 807a and 807b and the low-pass filter unit 801 are separated into audio signal to be ended higher than loud speaker
The high-frequency signal x of frequencyLhpAnd xRhpAnd the low frequency signal x less than loud speaker cutoff frequencylp, and utilize down-sampled unit 802
16 times of down-sampled processing are carried out to low frequency signal, reduce the computing cost of subsequent processing.Then, to the low frequency signal x of acquisitionlpWith
High-frequency signal xLhpAnd xRhp, fundamental frequency estimation and the detection of transient state Induction Peried are carried out successively, utilize the result of estimation and detection
Dynamic notch device is instructed more accurately to detach fundamental frequency and its harmonic component in low frequency signal.It then uses different types of virtual
Bass technology handles fundamental component F respectivelyharmWith residual components xresi, obtain more accurate bass chromatic rendition.Finally, such as Fig. 8
It is shown, first, to obtained virtual bass enhancement signal xvbBy the liter sampling unit 805, restore crude sampling rate；So
Afterwards, the signal after sampling unit is risen using the finishing of bandpass filtering unit 806, obtains xmp；Meanwhile by high-frequency signal xLhp
And xRhpInto line delay, the time difference of the preanalysis and the generation of separative element 803 is compensated；Then, it is asked using described second
With device 810a and the third summing unit 810b, to after obtained delay high-frequency signal and virtual bass enhancement signal
xmpAudio mixing output respectively, obtains the output x of virtual bass boost unitL4And xR4。
Step 6, the dynamic range control module 114, according to the adaptive adjusting x of the dynamic range of output equipmentL4
And xR4Amplitude, obtain the output signal x of systemLoutAnd xRout.As shown in figure 11, left-channel signal xL4By the third
Then signal D sampled point of delay is multiplied by gain factor g at the first multiplier 1108a and is exported by delay unit 1102a
xLout；Likewise, right channel signal xR4By the 4th delay unit 1102b by signal D sampled point of delay, then exist
It is multiplied by gain factor g at second multiplier 1108b and obtains output xRout。
In the environmental sound module 110, a kind of method of simplification is used to realize the simulation of particular room, it should
Method adds environmental effect with smaller calculation amount to stereo signal.As shown in figure 4, first, utilizing in-site measurement or sound
The method for learning modeling obtains sound source to binaural room impulse response function.Then, it is simplified, according to the resource feelings of system
Condition, it is all zero that the impulse response after simplifying has effective numerical value, other sampled points in about 10-80 sampled point, in this way
Computational complexity can be greatly reduced, convenient for carrying out real-time implementation in the effective platform of resource.This method is specific as follows：It is first
First, the impulse response of left and right ear is respectively processed, that is, intercepts first 80 milliseconds and is used as valid data section, remaining data house
It abandons；10 subsegments are averagely divided into for valid data section, the maximum sampled point of 1 to 8 absolute figure is selected to each subsegment,
It should ensure that the distribution of these sampled points is relatively average as far as possible.Secondly, arteries and veins between the left and right side room by above-mentioned steps simplification
Punching response also needs to further carry out decorrelative transformation, that is, the transmission function of ears is made to have correlation as small as possible
Property, such as there are one values at sampled point 120 for left ear transmission function, then should ensure that the impulse response of auris dextra exists as far as possible
Numerical value near sampled point 120 is all zero.Finally, the transmission function that above-mentioned steps simplify should be examined by subjective experiment and be imitated
Fruit.It should be ensured that lamprophonia, Sensurround are strong after ideal binaural transfer function and sound-source signal convolution.Fig. 5 be to Fig. 4 into
Row obtains simple version after simplifying, practical audiometry show the transmission function using Fig. 5 instead of maintaining preferable effect after Fig. 4 and
Computation complexity is significantly reduced again.
In the virtual bass boost module 112, the low-pass filter unit 801 carries out input audio signal
Low-pass filtering treatment, the cutoff frequency f of filterLIt is determined by system speaker.By the resume module, input sound can be obtained
The low frequency part of frequency signal.
In the virtual bass boost module 112, the down-sampled unit 802 carries out 16 times of down-sampled processing,
Reduce the sample rate of the input signal after the low-pass filter unit 801.The down-sampled unit 802 and the liter
805 main function of sampling unit is the sample rate of shift input signal, facilitate preanalysis described below and separative element 803 with
And the harmonic wave generates the processing of unit 804.Because before the signal input down-sampled unit 802, by described
The processing of low-pass filter unit 801, therefore the down-sampled unit 802 mainly realizes the function of an extraction, i.e., from input
In discrete digital signal, a point is extracted every Q point, Q is down-sampled unit multiple, need to meet relationship (6)：
Wherein fsIndicate the sample frequency of input signal, fLFor the cutoff frequency of low-pass filter unit 801, Q values are here
16。
In the virtual bass boost module 112, the first, second high pass filter unit 807a and 807b,
Main function is the high frequency section for extracting input audio signal, and on the one hand this part signal is used for later stage recovering signal, another
Aspect is used by starting of oscillation detection module.
In the preanalysis and separative element 803, the preanalysis and separative element 803 are to low after down-sampled
Frequency signal does Analysis of Fundamental Frequencies and extraction, and analysis and extraction have used adaptive notch filter technology.Simultaneously using by first, second
The energy envelope analysis of signal high frequency section, transient state oscillation starting points detection is carried out to signal after high pass filter unit 807a and 807b.
The specific implementation mode of the preanalysis and separative element 803 is as shown in Figure 9.The preanalysis and separative element 803 wrap
It includes：Audio mixing synthesizes submodule 905, transient state Induction Peried detection sub-module 901, and adaptive fundamental frequency tracks submodule 902, filter
Wave device parameter designing submodule 903 and dynamic cascading trap submodule 904.
In the preanalysis and separative element 803, the audio mixing synthesize submodule 905, will pass through first,
Second high pass filter unit 807a and 807b and the audio signal x generatedLhpAnd xRhpIt is synthesized, generates audio signal xMhp。
In the preanalysis and separative element 803, the transient state Induction Peried detection sub-module 901, to input
The envelope of signal time domain energy is detected, be then based on each note starting of oscillation part transient components account for leading role this often
Know, judges whether to reach transient state.And in transient portion thereof, frequency content is abundant, and changes rapidly, is easy to bring to fundamental frequency separation tired
Difficulty identifies them, can facilitate follow-up work.Induction Peried point can be obtained by following two relational expressions：
P (n)=max { y (n), P (n-1) × kDECAY}. (8)
Wherein xMhp(k) it is to synthesize submodule 905 by the audio mixing to generate, y (n) is local maximum, and P (n) is
Peak value following device, kDECAYIt is delay factor, n indicates current time, L1It is that detection window is long.When the condition that meets (9), can obtain
Audio signal starting of oscillation moment TAE, i.e. the value of n in this case：
Wherein MD,AEMark starting of oscillation starts threshold value.
In the preanalysis and separative element 803, the adaptive fundamental frequency tracks submodule 902 and uses two bases
It is cascaded in series in the second order lattice trapper of zero pole point constraint to realize the adaptive real-time tracing of two fundamental frequency spectral lines, first
The transmission function of filter is：
Wherein α is zero pole point convergence factor, for controlling the bandwidth of adaptive notch filter；Be first it is adaptive more
New coefficient.With first of trapper center frequency point flRelationship meet following relational expression：
Using the coefficient of lattice recurrence least square (RLSL) algorithm adaptive updates trapper, detailed process for example following three
A relational expression：
Cl(n)=λ Cl(n-1)+(1-λ)xlp,l(n-1)[xlp,l(n)+xlp,l(n-2)], (13)
Dl(n)=λ Dl(n-1)+2(1-λ)xlp,l(n-1)2 (14)
Wherein λ is the forgetting factor of RLSL algorithms, for controlling adaptive convergence rate.It isAfter constraint is smooth
As a result, the two meets relationship：
Wherein smoothing factor η=0.5.xlp,1(n) the as output x of the down-sampled unit 802lp(n), xlp,2(n)
It is the output of first lattice trapper, is represented by：
The form that final adaptive fundamental frequency tracking submodule 902 is cascaded in series for using two trappers, will predict simultaneously
The frequency of the strongest fundamental component of ingredient
In the preanalysis and separative element 803, filter parameter design submodule 903 utilizes described
The testing result T of transient state Induction Peried detection sub-module 901AEThe bandwidth of trapper is controlled, when being judged as the transient state starting of oscillation moment,
It is considered that for audio signal based on transient state broadband signal, harmonic components are weaker at this time, therefore utilize improved sigmoid functions
The bandwidth factor α for changing trapper reduces bandwidth.Specific implementation mode is such as shown in (18)：
Wherein d=n-TAE, i.e. nearest with it current time n starting of oscillation moment TAETime difference.Simultaneously for described adaptive
It answers the fundamental frequency estimation result F of fundamental frequency tracking submodule 902 to do smoothing processing, reduces the influence of adaptive oscillation.
In the preanalysis and separative element 803, the dynamic cascading trap submodule 904 be a parameter by
The trapper unit that the filter parameter design submodule 903 controls, transmission function is by shown in (19)：
WhereinWithIt can be calculated by formula (15), be determined by (17) with the relationship of F.
In the virtual bass boost module 112, the harmonic wave generates unit 804 and is distinguished by different means
The harmonic signal of fundamental frequency part and residual components is generated, and two parts of signals is reasonably combined, embodiments thereof is as schemed
Shown in 10.Specifically harmonic wave generation unit 804 includes：Residual components harmonic generation submodule 1001, fundamental component
Harmonic generation submodule 1002, harmonic energy adjust submodule 1004 and harmonic wave synthon module 1005.
In the harmonic wave generates unit 804, the residual components harmonic generation submodule 1001 is using form
(20) nonlinear function handles the residual components x that the dynamic cascading trap submodule 904 generatesresi, generate remaining point
The harmonic signal of amount.Below this function be proved to virtual bass perception and non-linear distortion control aspect general performance compared with
It is good：
In the harmonic wave generates unit 804, the fundamental component harmonic generation submodule 1002 is using form
(21) harmonic frequency calculation formula, calculates the parameters frequency of harmonic signal, wherein fl(1, n) it is the filter parameter control
The frequency for first of fundamental frequency that system module 903 is estimated, l={ 1,2 }.fl(i, n) is fundamental frequency flThe i subfrequencies of (1, n), meter
Calculation method is：
And the harmonic energy adjustment submodule 1004 is estimated using the fundamental component harmonic generation submodule 1002
The harmonic frequency f of meterl(i, n) adjusts the amplitude of each harmonic according to psychoacoustic principle, obtains best hearing effect, specifically
Shown in calculating process such as formula (22)：
Wherein AiAmplitude is weighted for each harmonic.
In the harmonic wave generates unit 804, the harmonic and reactive detection submodule 1005 utilizes the residual components
The harmonic component x that harmonic generation submodule 1001 generatesvb_resiIt is generated with the harmonic energy adjustment submodule 1004 humorous
Wave component xvb_harmWeighting synthesis is done, virtual bass harmonic component x is obtainedvb。
In the virtual bass boost module 112,805 main function of liter sampling unit is to input signal
Into row interpolation, it is to rise sampling unit multiple that Q-1 zero, Q are inserted into behind each input point, and value is 16 here.Generally into
It has gone after interpolation processing, has needed frequency overlapped-resistable filter to prevent spectral aliasing, anti-aliasing filter function can be by described here
Bandpass filtering unit 806 is completed.
In the virtual bass boost module 112, the bandpass filtering unit 806 by low-frequency cut-off frequency and
The design of high-frequency cut-off frequency filters out low frequency part and high-frequency harmonic distorted portion that tone playing equipment can not restore.Described
Bandpass filtering unit 806 is mainly there are two effect, and one is anti-aliasing filter, the signal after smooth interpolation, another is pair
Virtual bass harmonic signal carries out tone color adjustment, filter out low-frequency cut-off frequency hereinafter, and especially high-frequency signal, reduce and lose
Very.General bandpass filter band connection frequency fBP_lowAnd fBP_highValue meet following two formula：
fBP_low≥flow, (23)
Wherein flowIndicate the playback lower-frequency limit of tone playing equipment.
In the virtual bass boost module 112, the first, second delay unit 808a and 808b is for more
Mend low frequency processing generate time lag, when signal being made to restore high and low frequency part will not generation time it is poor.When software realization,
The spending that memory is read is saved using Cushion Technology.
In dynamic range control module 114, as shown in figure 11, the calculating of gain factor is by the envelope detected
What unit 1104 and the gain calculating unit 1106 obtained.The envelope detected unit 1104 passes through a low-pass filtering
Unit detects the envelope value in left and right channel.The gain calculating unit 1106 is obtained according to the envelope value of input by tabling look-up
Corresponding gain factor g, dynamic range control have different applications in different occasions, so should be set according to specific application
Set different gains.As shown in figure 12, a mapping relations of input/output relation are given, the input signal more than 0dB is all
It is compressed in 0dB, plays the role of wire pressing device, prevents distorted signals；The signal inputted before -20~-10dB is exported -10
Between~0dB, the dynamic range of this segment signal is compressed；The dynamic range of signals inputted between -80~-20dB does not become
Change, output is more whole big 10dB than input, that is, this segment signal is linearly expanded；The dynamic of input signal less than -80dB
Range is then expanded, that is, small amplitude signal is taken as noise and is suppressed.In practical applications, other shapes can also be taken
The signal input/output relation of formula maps, and needs to be configured according to specific circumstances.
It should be noted that stereo audio enhancing system described in the invention can be realized with various ways, such as
The combination of hardware, software either hardware and software.Hardware platform can be FPGA, PLD or other application-specific integrated circuit ASICs.
Software platform includes DSP, ARM or other microprocessors.
It should be noted last that above example is only used to illustrate the technical scheme of the present invention and unrestricted.Although ginseng
It is described the invention in detail according to embodiment, it will be understood by those of ordinary skill in the art that, to the technical side of the present invention
Case is modified or replaced equivalently, and without departure from the spirit and scope of technical solution of the present invention, should all be covered in the present invention
Right in.
Claims (7)
1. a kind of stereo audio enhances system, which is characterized in that the system includes：Signal separation module, dialogue enhance module,
Decorrelation processing module, signal synthesizing module, environmental sound module, virtual bass boost module and dynamic range control module；
The left channel signals x that the signal separation module receives systemLinWith right-channel signals xRinIt detaches and obtains direct sound wave
Signal xsWith circular acoustical signal xd；Wherein, the through acoustical signal xsIt is input to the dialogue enhancing module, pair
White enhancing module is to above-mentioned through acoustical signal xsIt is handled, the x exportedsout；And around acoustical signal xdIt is input to described
Decorrelation processing module, the decorrelation processing module is by above-mentioned circular acoustical signal xdPass through a pair of of de-correlation filter group
Obtain two output signal x with very low correlationL1And xR1；The xsout、xL1And xR1Three signals fully enter
The signal synthesizing module, the signal synthesizing module is by x obtained abovesout、xL1And xR1Three signals are weighted
Combination obtains new left channel signals xL2With right-channel signals xR2；Then, the environmental sound module is by xL2And xR2Respectively with
The binaural room impulse response h in left and right channelLAnd hROutput x is obtained after carrying out convolutionL3And xR3；Then, the virtual bass
Enhance module and receives input signal xL3And xR3X is exported after carrying out bass boostL4And xR4；The dynamic range control module, root
According to the adaptive adjusting x of the dynamic range of output equipmentL4And xR4Amplitude, obtain the output signal x of systemLoutAnd xRout；
The virtual bass boost module includes：
First high pass filter unit, first high pass filter unit is to input signal xL3Carry out high-pass filtering, described the
The cutoff frequency of one high pass filter unit is the low-frequency cut-off frequency of system speaker；
Second high pass filter unit, second high pass filter unit is to input signal xR3Carry out high-pass filtering, described the
The cutoff frequency of two high pass filter units is the low-frequency cut-off frequency of system speaker；
First summing unit, first summing unit is by left channel signals xL3With right-channel signals xR3It is added；
Low-pass filter unit, the low-pass filter unit extract low frequency signal, the low pass from first summing unit
The cutoff frequency of filter unit is the low-frequency cut-off frequency of system speaker；
Down-sampled unit, the down-sampled unit carry out down-sampled place to the low frequency signal that the low-pass filter unit extracts
Reason, the down-sampled factor are determined by the low-frequency cut-off frequency of system speaker and the sample rate of input audio signal；
Preanalysis and separative element, the preanalysis and separative element are for the low frequency signal obtained by the down-sampled unit
Carry out fundamental frequency estimation and to high-frequency signal obtained by first high pass filter unit and second high pass filter unit
Carry out transient state Induction Peried detection；
Harmonic wave generates unit, and the harmonic wave generates the higher harmonic component that unit is used for generating fundamental component and residual components；
Sampling unit is risen, the signal that described liter of sampling unit generates the harmonic wave unit output carries out a liter sampling processing, and liter is adopted
Like factor is equal with the down-sampled factor of down-sampled unit；
Bandpass filtering unit, the bandpass filtering unit carry out bandpass filtering to the signal that described liter of sampling unit exports, are used for
Filter out the low frequency part and high-frequency harmonic distorted portion of signal；
First delay unit, first delay unit to by signal obtained by first high pass filter unit into line delay
Reason, delay duration delay duration caused by the preanalysis and separative element determine；
Second delay unit, second delay unit to by signal obtained by second high pass filter unit into line delay
Reason, delay duration delay duration caused by the preanalysis and separative element determine；
Signal obtained by signal obtained by first delay unit and the bandpass filtering unit is carried out phase by the second summing unit
Add, obtains new left channel signals xL4；
Signal obtained by signal obtained by second delay unit and the bandpass filtering unit is carried out phase by third summing unit
Add, obtains new right-channel signals xR4。
2. stereo audio according to claim 1 enhances system, which is characterized in that the dialogue enhancing module is one
A IIR or FIR filter, the amplitude spectrum of the filter are to be obtained to contour of equal loudness against after being modified.
3. stereo audio according to claim 1 enhances system, which is characterized in that the decorrelation processing module is
The input of the module is passed through into a pair of of de-correlation filter hLDAnd hRDAfter export xL1And xR1, the de-correlation filter hLDWith
hRDAmplitude response near flat, and its phase response is completely uncorrelated or weak relevant.
4. stereo audio according to claim 1 enhances system, which is characterized in that the signal synthesizing module be by
The output x of the de-correlation blockL1And xR1It is multiplied by gain factor k respectively, then adds the defeated of the dialogue enhancing module
Go out xsoutThe left channel signals x exported with the product of gain factor ρL2With right-channel signals xR2, the wherein range of k is 0≤k
The range of≤10, ρ are 0≤ρ≤10.
5. stereo audio according to claim 1 enhances system, which is characterized in that the binaural room impulse response
Be by image method simulation setting acoustics scene or actual room measure obtain binaural room impulse response after it is right
Its reflection part obtained after simplifying.
6. stereo audio according to claim 1 enhances system, which is characterized in that the preanalysis and separative element into
One step includes：
Audio mixing synthesize submodule, the audio mixing synthesize submodule will by the first, second high pass filter unit and
The audio signal x of generationLhpAnd xRhpIt is synthesized, generates audio signal xMhp；
Transient state Induction Peried detection sub-module, packet of the transient state Induction Peried detection sub-module to input signal time domain energy
Network is detected, and judges whether music signal is in transient state starting of oscillation state；
Adaptive fundamental frequency tracks submodule, and adaptive base rate tracking submodule uses the second order lattice constrained based on zero pole point
Type trapper is tracked to do adaptive base spectrum line, updates trap parameter using lattice recursive least squares；
Filter parameter designs submodule, and the filter parameter is designed submodule and detected using the transient state Induction Peried
The testing result T of submoduleAEThe bandwidth of trapper is controlled, when being judged as the transient state starting of oscillation moment, it is believed that at this time audio signal with
Based on transient state broadband signal, harmonic components are weaker, and the bandwidth factor α for being changed trapper using improved sigmoid functions is reduced
Bandwidth；Smoothing processing is done for the fundamental frequency estimation result F of the adaptive fundamental frequency tracking submodule simultaneously, reduces adaptive oscillation
Influence；
Dynamic cascading trap submodule, the dynamic cascading trap submodule by the cascade of multiple bandwidth adjustable trapper and
At the bandwidth of trapper is controlled by the transient state Induction Peried detection sub-module, when being judged as the transient state starting of oscillation moment, reduces band
It is wide；The centre frequency of trapper is determined by the adaptive fundamental frequency tracking submodule, need to track submodule to adaptive fundamental frequency
Fundamental frequency estimation result do smoothing processing.
7. stereo audio according to claim 1 enhances system, which is characterized in that the harmonic wave generates unit into one
Step includes：
Residual components harmonic generation submodule, the residual components harmonic generation submodule is using at special nonlinear function
The residual components for managing the output of dynamic cascading trap submodule, generate the harmonic signal of residual components；
Fundamental component harmonic generation submodule, the fundamental component harmonic generation submodule are calculated using specific harmonic frequency
Method calculates the instantaneous frequency of 2 to 6 harmonics by adaptive fundamental frequency tracking submodule estimation gained fundamental frequency；
Harmonic energy adjusts submodule, and the harmonic energy adjusts submodule and uses the fundamental component harmonic generation submodule
The instantaneous frequency for the harmonics that block estimates adjusts the amplitude of each harmonic according to psychoacoustic principle；
Harmonic and reactive detection submodule, the harmonic and reactive detection submodule are generated using the residual components harmonic generation submodule
The harmonic component that harmonic component and harmonic energy adjustment submodule generate does weighting synthesis, obtains virtual bass harmonic wave point
Amount.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510645398.9A CN106572419B (en) | 2015-10-08 | 2015-10-08 | A kind of stereo audio enhancing system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510645398.9A CN106572419B (en) | 2015-10-08 | 2015-10-08 | A kind of stereo audio enhancing system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106572419A CN106572419A (en) | 2017-04-19 |
| CN106572419B true CN106572419B (en) | 2018-08-03 |
Family
ID=58507110
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510645398.9A Active CN106572419B (en) | 2015-10-08 | 2015-10-08 | A kind of stereo audio enhancing system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106572419B (en) |
Families Citing this family (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10153744B1 (en) * | 2017-08-02 | 2018-12-11 | 2236008 Ontario Inc. | Automatically tuning an audio compressor to prevent distortion |
| US10499153B1 (en) * | 2017-11-29 | 2019-12-03 | Boomcloud 360, Inc. | Enhanced virtual stereo reproduction for unmatched transaural loudspeaker systems |
| JP6699677B2 (en) * | 2018-02-06 | 2020-05-27 | ヤマハ株式会社 | Information processing method, information processing apparatus, and program |
| CN110166893A (en) * | 2018-02-13 | 2019-08-23 | 晶豪科技股份有限公司 | Virtual bass generation circuit, loudspeaker and method |
| CN109947385A (en) * | 2019-03-08 | 2019-06-28 | 广东小天才科技有限公司 | Dynamic tuning method, apparatus, wearable device and storage medium |
| CN109920446B (en) * | 2019-03-12 | 2021-03-26 | 腾讯音乐娱乐科技（深圳）有限公司 | Audio data processing method and device and computer storage medium |
| CN110021304A (en) * | 2019-05-10 | 2019-07-16 | 腾讯音乐娱乐科技（深圳）有限公司 | A kind of audio-frequency processing method, device, terminal and storage medium |
| CN110267163B (en) * | 2019-06-18 | 2021-07-20 | 重庆清文科技有限公司 | Method, system, medium and device for enhancing directional sound virtual low frequency |
| CN110972053B (en) * | 2019-11-25 | 2021-06-25 | 腾讯音乐娱乐科技（深圳）有限公司 | Method and related apparatus for constructing a listening scene |
| CN111405420A (en) * | 2020-02-28 | 2020-07-10 | 浙江吉利新能源商用车集团有限公司 | Vehicle sound system, control method and vehicle |
| CN112351379B (en) * | 2020-10-28 | 2021-07-30 | 歌尔光学科技有限公司 | Control method of audio component and intelligent head-mounted device |
| CN114363793B (en) * | 2022-01-12 | 2024-06-11 | 厦门市思芯微科技有限公司 | System and method for converting double-channel audio into virtual surrounding 5.1-channel audio |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101902674A (en) * | 2010-08-13 | 2010-12-01 | 西安交通大学 | Self-excitation eliminating method of high gain public address system based on space counteracting |
| CN103634726A (en) * | 2013-08-30 | 2014-03-12 | 苏州上声电子有限公司 | Automatic loudspeaker equalization method |
| CN104936088A (en) * | 2015-04-21 | 2015-09-23 | 上海大学 | Mixed virtual bass enhancing method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP5802753B2 (en) * | 2010-09-06 | 2015-11-04 | ドルビー・インターナショナル・アクチボラゲットＤｏｌｂｙ Ｉｎｔｅｒｎａｔｉｏｎａｌ Ａｂ | Upmixing method and system for multi-channel audio playback |
-
2015
- 2015-10-08 CN CN201510645398.9A patent/CN106572419B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101902674A (en) * | 2010-08-13 | 2010-12-01 | 西安交通大学 | Self-excitation eliminating method of high gain public address system based on space counteracting |
| CN103634726A (en) * | 2013-08-30 | 2014-03-12 | 苏州上声电子有限公司 | Automatic loudspeaker equalization method |
| CN104936088A (en) * | 2015-04-21 | 2015-09-23 | 上海大学 | Mixed virtual bass enhancing method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106572419A (en) | 2017-04-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106572419B (en) | A kind of stereo audio enhancing system | |
| KR101333031B1 (en) | Method of and device for generating and processing parameters representing HRTFs | |
| US9729969B2 (en) | System and method for bass enhancement | |
| RU2658872C2 (en) | Equalisation filter coefficient determinator, apparatus, equalisation filter coefficient processor, system and methods | |
| CN102395098B (en) | Method of and device for generating 3D sound | |
| EP2064699B1 (en) | Method and apparatus for extracting and changing the reverberant content of an input signal | |
| CN107770718B (en) | Generating binaural audio by using at least one feedback delay network in response to multi-channel audio | |
| US20030216907A1 (en) | Enhancing the aural perception of speech | |
| JP3537674B2 (en) | Audio system | |
| US10701505B2 (en) | System, method, and apparatus for generating and digitally processing a head related audio transfer function | |
| US11611828B2 (en) | Systems and methods for improving audio virtualization | |
| CN108886650A (en) | It is eliminated for the subband spatial of audio reproduction and crosstalk | |
| JP2023517720A (en) | Reverb rendering | |
| AU2017210021A1 (en) | Synthesis of signals for immersive audio playback | |
| CN112956210B (en) | Audio signal processing method and device based on equalization filter | |
| Romoli et al. | A mixed decorrelation approach for stereo acoustic echo cancellation based on the estimation of the fundamental frequency | |
| KR102194515B1 (en) | Subband spatial processing and crosstalk cancellation system for conferences | |
| Romoli et al. | A novel decorrelation approach for multichannel system identification | |
| CN108040317A (en) | A kind of hybrid sense of hearing sound field broadening method | |
| EP4032321A1 (en) | Enhancement of audio from remote audio sources | |
| WO2017121245A1 (en) | Method for achieving surround sound, electronic device, and storage medium | |
| US20030202665A1 (en) | Implementation method of 3D audio | |
| Rämö et al. | Graphic delay equalizer | |
| Czyżewski et al. | Adaptive personal tuning of sound in mobile computers | |
| Weller et al. | Application of a circular 2D hard-sphere microphone array for higher-order Ambisonics auralization |
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 |