CN101977349A - Decoding optimizing and improving method of Ambisonic voice repeating system - Google Patents

Abstract

The invention relates to a decoding optimizing and improving method of an Ambisonic voice repeating system, which comprises the steps of: respectively establishing a direction optimizing target function, a loudness optimizing target function and a definition optimizing target function through calculating velocity vectors, sound energy flow vectors and loudness of a virtual sound image according to a sound image positioning principle, and introducing an amplitude optimizing target function of directivity of a virtual microphone into a comprehensive optimizing target function; determining the comprehensive optimizing target function with optimal fitness value through an iteration operation by adopting an optimizing algorithm, and obtaining an optimal decoding result; controlling the introduction of the amplitude optimizing target function of directivity of the virtual microphone to obtain a decoding result the same phase with each speaker, and furthest reducing the mixing of repeated voice in the space direction, wherein the optimizing convergence speed is increased and the operation time is shortened. The invention is suitable for a decoding process of the Ambisonic voice repeating system in any order, any speaker quantity and any arrangement mode.

Description

The optimization of Ambisonic sound reproduction system decodes is improved one's methods

Technical field

The present invention relates to the improvement of Ambisonic multipath three dimensional sound repeater system decoding technique, specifically be meant and utilize synthetic virtual microphone and directive property control technology thereof, Ambisonic sound reproduction system decodes method based on the vector composition principle is optimized improvement, improve the synthetic precision of virtual sound image, improve the sound reproduction quality of Ambisonic system.

Background technology

Plane wave can be decomposed and reconstruct by a series of orthogonal sphere hamonic functions.Use spheric harmonic function

It is the theoretical foundation of Ambisonic sound reproduction system that plane wave is decomposed with reconstruct.The exponent number of spheric harmonic function is high more, and the precision of decomposition or reconstruct is just high more.Use spheric harmonic function to s signal plane ripple S (θ _S, φ _S) sample, can obtain original Ambisonic code signal

$A_{m,n}^{\mathrm{\σ}}=Y_{m,n}^{\mathrm{\σ}}({\mathrm{\θ}}_S,{\mathrm{\φ}}_S)\·s---\left(1\right)$

In the formula, m is the spheric harmonic function exponent number, m 〉=0,0≤n≤m, σ=± 1, θ _SAnd φ _SGlancing incidence angle and vertical incidence angle for plane wave s.Utilize the loud speaker of some to form specific array, simultaneously Ambisonic code signal branch is equipped with different weights and reconfigures, flow to the loud speaker of diverse location, can reconstruct plane wave S (θ in the loudspeaker array center _S, φ _S).Flow to orientation angles and be (θ _j, φ _j) the recurrent signal g of j loud speaker _jCan obtain by formula (2), wherein

It is the code signal of corresponding j loud speaker

The weight coefficient group, and

$g_j=\underset{0\≤m\≤M,0\≤n\≤m,\mathrm{\σ}=\±1}{\mathrm{\Σ}}{c}_{m,n,j}^{\mathrm{\σ}}\·A_{m,n}^{\mathrm{\σ}}---\left(2\right)$

Wherein, M is that the spheric harmonic function that is used to calculate is ended exponent number.How according to the specific arrangements form of loudspeaker array, it is suitable to obtain

Make the virtual sound image that reconstructs at the array center place near original plane ripple signal S (θ _S, φ _S) process, be exactly the decode procedure of Ambisonic sound reproduction system.

Gerzon serves as the vector composition principle that the basis proposes with the psychologic acoustics research that the human auditory system synthesizes virtual sound image to many sound sources, is applied to Ambisonic sound reproduction system decodes process.The original acoustical signal of different directions incident can produce the true acoustic image of different directions in people's auditory system.Gerzon thinks, as long as this virtual sound image is with true acoustic image direction unanimity, loudness is identical and have acoustic image clearly, can think the accurate reconstruct to original signal.Therefore the key of Ambisonic sound reproduction system is the weight coefficient group

Determine.For this reason, at first need at specific loudspeaker array, respectively according to the psychoacoustic principle of the synthetic virtual sound image of lower frequency region and high-frequency domain, calculate the data such as direction, loudness and definition of virtual sound image, and then utilize the system optimization theory, set up the optimization aim function, come the different weight coefficient groups of comparison by the computer optimization program

Fitness value, the final decoded result optimized of obtaining.

Craven and Wiggins are once based on this method and use 4 rank spheric harmonic functions respectively the horizontal multi-loudspeaker repeater system of ITU form to be optimized, the former uses the conjugate gradient optimization method, the latter as initial value, uses the TABU search method to be optimized computing with the former optimization result.The decoded result of Craven shows, the scope of the quality of virtual sound image between three loud speakers of the place ahead comparatively stable and accurately, the less stable of side direction and rear virtual sound image, the direction of virtual sound image are also not too accurate.Comparatively speaking, the decoded result of Wiggins, all better aspect the stability of virtual sound image and accuracy.But all only at the horizontal plane sound reproduction system of IUT form, the system to other types does not discuss in both research.David Moore and P.W.M Tsang have also announced them to how improving searching algorithm improving the research of 1 rank Ambisonic system optimization, but do not see to have and be higher than 1 rank systematic research report.And it is to the research of searching algorithm itself, do not relate to sound control, and can not effectively solve in optimizing process, because the nonlinear optimization that introducing caused of high-frequency domain acoustic energy flow vector is difficult to obtain globally optimal solution, and the problem of calculating length consuming time.

Summary of the invention

The objective of the invention is to overcome the deficiency that has now based on the Ambisonic sound reproduction system decodes method of virtual sound image vector composition principle, providing a kind of controls based on sound, utilize the optimization of the Ambisonic sound reproduction system decodes of virtual directivity of microphone control technology to improve one's methods, improve the synthetic precision of virtual sound image, improve the sound reproduction quality of Ambisonic system.。

The present invention realizes above-mentioned purpose by the following technical solutions:

The optimization of this Ambisonic sound reproduction system decodes is improved one's methods, and may further comprise the steps:

Step 1:, obtain the velocity r of decision low-frequency range and high band virtual sound image alignment quality respectively according to the acoustic image positioning principle _VWith acoustic energy flow vector r _E

Step 2:, calculate the loudness V of virtual sound image in low-frequency range according to the acoustic image positioning principle _LiLoudness V with high band _Hi

Step 3: ideally, the virtual sound image that is produced during Ambisonic system sound reproduction, should with the counterparty to true acoustic image direction unanimity, loudness identical, acoustic image is clear simultaneously.Set up direction optimization aim function Afit, loudness optimization aim function Lfit and resolution optimization target function Mfit thus respectively;

Step 4: the optimization aim function C fit that sets up virtual directivity of microphone control;

Step 5: synthetic complex optimum target function Tfit;

Step 6: in conjunction with the position of loud speaker in the Ambisonic sound reproduction system, adopt optimized Algorithm, obtain the complex optimum target function Tfit fitness value with optimum, its pairing spheric harmonic function weight coefficient group by interative computation Be optimum decoded result.

Compared with prior art, the present invention has following advantage and beneficial effect:

1) introduces the amplitude optimization aim function C fit that is used to control the virtual directivity of microphone.Virtual microphone with the single flap directive property feature that does not have secondary lobe, can suppress from the sound of corresponding repeating transmission loud speaker different directions, farthest reduce and retransmit sound obscuring in the direction in space sense;

2) decoded result can make each loudspeaker signal homophase (in phase), and the natural tone color and the maximum range of audibility that this has at utmost guaranteed the sound reproduction system are particularly suitable for many people and listen to, or head has the sound reproduction of rotation situation to use.

3) behind the introducing Cfit, optimization aim is more clear and definite, and the convergence rate of optimization aim function is accelerated, and shorten operation time greatly.

4) new optimization is improved one's methods all suitable to the Ambisonic system of any exponent number, any number of loudspeakers and arrangement.Wherein, for the Ambisonic system of asymmetric loudspeaker arrangement mode, can be by the independent directive property of adjusting the synthetic virtual microphone of the pairing difference of repeating transmission loud speaker of asymmetric distribution, the guiding optimizer obtains decoded result with fast speed.

5) proposed by the invention by the synthetic virtual directivity of microphone control method of spheric harmonic function, only relevant with spheric harmonic function itself, have nothing to do with adopting which kind of acoustic image orientation criterion.Therefore, the synthetic virtual directivity of microphone control method that the present invention proposes also can with other any acoustic image orientation criterion theories, as combining, the Ambisonic system is decoded based on the acoustic image positioning principle of number of people transfer function etc.

6) the invention solves prior art and be difficult to obtain the problem of globally optimal solution in the nonlinear optimization of using the vector composition principle to carry out occurring in the Ambisonic decode procedure; Though the overall merit target function value is better when having overcome computation optimization, the synthetic virtual directivity of microphone is undesirable, thus the physical property error result that acoustic image is obscured before and after when causing sound reproduction.

Description of drawings

Fig. 1 is the directive property parameter schematic diagram of synthetic virtual microphone;

Fig. 2 is a flow chart of the present invention;

Fig. 3 optimizes flow chart.

Embodiment

The present invention is described in further detail below in conjunction with embodiment and accompanying drawing, but embodiments of the present invention are not limited thereto.

Embodiment

Spheric harmonic function can be counted as the virtual microphone with particular orientation.At this moment, the cataloged procedure of Ambisonic system can be regarded the pick up process of the virtual microphone of a plurality of spheric harmonic functions to original sound wave of using as.During sound reproduction, the recurrent signal of each loud speaker of feeding can be by signal that these virtual microphones picked up by weight coefficient

Stack obtains, and the stack back produces the synthetic virtual microphone SVM with new directive property, and anti-phase back is corresponding to a loud speaker of retransmitting in the loudspeaker array.

If the directive property of a certain synthetic virtual microphone SVM is undesirable, then when sound reproduction, the sound of all directions of being picked up all will be applied and retransmit the mistake that all causes the sound fore-and-aft direction to obscure by a certain loud speaker on physical significance and subjective sensation.If virtual microphone SVM is to other direction, particularly the inhibition of rear sound is very capable, the problem that the sound fore-and-aft direction that occurs easily in the time of can solving sound reproduction to a great extent during sound reproduction is obscured.Therefore, the synthetic virtual microphone SVM that utilizes the virtual microphone array of spheric harmonic function to produce, whether its directive property is desirable, with having a strong impact on original sound field carried out the quality that pick up in the space, and the subjectivity when influencing sound reproduction is simultaneously listened to effect.

The present invention has controlled the directive property of synthetic virtual microphone SVM effectively by setting up virtual directivity of microphone controlled target function C fit.

As Fig. 1, suppose that the azimuth of virtual microphone orientation of its axis is (θ _Aim, φ _Aim), its level is Δ θ and Δ φ with vertical effective angle of coverage, the start angle that calculates cone of coverage is respectively θ ₁, θ ₂, and φ ₁, φ ₂, and θ ₁＜θ ₂, φ ₁＜φ ₂Then under perfect condition, this virtual microphone should have good restraining ability to the outer sound of its effective coverage range.Microphone is picked up the power of ability by amplitude r to certain direction sound _M(θ, φ) expression, r _M(θ, φ) 〉=0.The directive property of virtual microphone should be subjected to effective control aspect following three:

(1) orientation of its axis of virtual microphone, i.e. azimuth (θ _Aim, φ _Aim);

(2) effective angle of coverage of virtual microphone, i.e. level and vertical angle of coverage Δ θ and Δ φ;

(3) the outer sound ability to accept of virtual microphone effective coverage range should be able to be subjected to the inhibition of maximum program.

For reaching above-mentioned controlled target, use the amplitude comparison method, set up relatively optimization aim function C fit of amplitude, the amplitude sum that the microphone effective coverage range is outer, divided by the amplitude sum in the effective coverage range, consider θ and φ in definition at the spherical coordinates of Ambisonic system, promptly-180 °≤θ≤180 °,-90 °≤φ≤90 °, have:

(1) at θ ₁＜θ _Aim＜θ ₂And under ° situation of Δ φ≤180:

$\mathrm{Cfit}=\frac{\underset{-180\≤\mathrm{\θ}\≤180}{\mathrm{\Σ}}\left(\underset{-90\≤\mathrm{\φ}\≤90}{\mathrm{\Σ}}{r}_M(\mathrm{\θ},\mathrm{\φ})\right)}{\underset{{\mathrm{\θ}}_1\≤\mathrm{\θ}\≤{\mathrm{\θ}}_2}{\mathrm{\Σ}}\left(\underset{{\mathrm{\φ}}_1\≤\mathrm{\φ}\≤{\mathrm{\φ}}_2}{\mathrm{\Σ}}{r}_M(\mathrm{\θ},\mathrm{\φ})\right)}-1---\left(3\right)$

(2) at θ ₁＜θ _Aim＜θ ₂And under ° situation of Δ φ≤180:

$\mathrm{Cfit}=\frac{\underset{-180\&le;\mathrm{\&theta;}\&le;180}{\mathrm{\&Sigma;}}\left(\underset{-90\&le;\mathrm{\&phi;}\&le;90}{\mathrm{\&Sigma;}}{r}_M(\mathrm{\&theta;},\mathrm{\&phi;})\right)}{\underset{{\mathrm{\&theta;}}_1\&le;\mathrm{\&theta;}\&le;{\mathrm{\&theta;}}_2}{\mathrm{\&Sigma;}}\left(\underset{-90\&le;\mathrm{\&phi;}\&le;90}{\mathrm{\&Sigma;}}{r}_M(\mathrm{\&theta;},\mathrm{\&phi;})\right)\text{+}\underset{{\mathrm{\&theta;}}_2<\mathrm{\&theta;}\&le;180}{\underset{-180\&le;\mathrm{\&theta;}<{\mathrm{\&theta;}}_1,}{\mathrm{\&Sigma;}}}\left(\underset{{\mathrm{\&phi;}}_2\&le;\mathrm{\&phi;}<90}{\underset{-90<\mathrm{\&phi;}\&le;{\mathrm{\&phi;}}_1,}{\mathrm{\&Sigma;}}}{r}_M(\mathrm{\&theta;},\mathrm{\&phi;})\right)}-1---\left(4\right)$

(3) at θ _Aim＜θ ₁Or θ _Aim＞θ ₂And under ° situation of Δ φ≤180:

$\mathrm{Cfit}=\frac{\underset{-180\&le;\mathrm{\&theta;}\&le;180}{\mathrm{\&Sigma;}}\left(\underset{-90\&le;\mathrm{\&phi;}\&le;90}{\mathrm{\&Sigma;}}{r}_M(\mathrm{\&theta;},\mathrm{\&phi;})\right)}{\underset{{\mathrm{\&theta;}}_2<\mathrm{\&theta;}\&le;180}{\underset{-180\&le;\mathrm{\&theta;}<{\mathrm{\&theta;}}_1,}{\mathrm{\&Sigma;}}}\left(\underset{{\mathrm{\&phi;}}_1\&le;\mathrm{\&phi;}\&le;{\mathrm{\&phi;}}_2}{\mathrm{\&Sigma;}}{r}_M(\mathrm{\&theta;},\mathrm{\&phi;})\right)}-1---\left(5\right)$

(4) at θ _Aim＜θ ₁Or θ _Aim＞θ ₂And under ° situation of Δ φ≤180:

$\mathrm{Cfit}=\frac{\underset{-180\&le;\mathrm{\&theta;}\&le;180}{\mathrm{\&Sigma;}}\left(\underset{-90\&le;\mathrm{\&phi;}\&le;90}{\mathrm{\&Sigma;}}{r}_M(\mathrm{\&theta;},\mathrm{\&phi;})\right)}{\underset{{\mathrm{\&theta;}}_2<{\mathrm{\&theta;}}_i\&le;180}{\underset{-180\&le;\mathrm{\&theta;}<{\mathrm{\&theta;}}_1,}{\mathrm{\&Sigma;}}}\left(\underset{-90\&le;\mathrm{\&phi;}\&le;90}{\mathrm{\&Sigma;}}{r}_M(\mathrm{\&theta;},\mathrm{\&phi;})\right)+\underset{{\mathrm{\&theta;}}_1\&le;\mathrm{\&theta;}<{\mathrm{\&theta;}}_2}{\mathrm{\&Sigma;}}\left(\underset{{\mathrm{\&phi;}}_2\&le;\mathrm{\&phi;}<90}{\underset{-90<\mathrm{\&phi;}\&le;{\mathrm{\&phi;}}_1,}{\mathrm{\&Sigma;}}}{r}_M(\mathrm{\&theta;},\mathrm{\&phi;})\right)}-1---\left(6\right)$

r _M(θ, φ) be equivalent to all loud speakers at the azimuth for (θ, the absolute value of the signal level value of closing on direction φ), as shown in the formula:

$r_M(\mathrm{\&theta;},\mathrm{\&phi;})=\left|\underset{j=1}{\overset{k}{\mathrm{\&Sigma;}}}{g}_j\right|=\left|\underset{j=1}{\overset{k}{\mathrm{\&Sigma;}}}(\underset{0\&le;m\&le;M,0\&le;n\&le;m,\mathrm{\&sigma;}=\&PlusMinus;1}{\mathrm{\&Sigma;}}{c}_{m,n,j}^{\mathrm{\&sigma;}}\&CenterDot;Y_{m,n}^{\mathrm{\&sigma;}}(\mathrm{\&theta;},\mathrm{\&phi;}))\right|---\left(7\right)$

In the formula (7), k represents the sum of loud speaker in the Ambisonic system, the recurrent signal g of loud speaker _jCan obtain function by formula (1) and (2)

Be spheric harmonic function,

Be spheric harmonic function

The weight coefficient group.Obviously, carry out the m rank Ambisonic system of encoding and decoding to adopting any m rank spheric harmonic function, formula (3) all is suitable for to formula (7).

Under any circumstance, must make that all Cfit has minimum value (under the perfect condition, Cfit=0), promptly when weakening microphone as far as possible sound pick up ability outside to effective coverage range, will be to the maximized of picking up of the sound in the effective coverage range, reach the purpose that the directive property of virtual microphone is controlled, can obtain comparatively desirable virtual microphone directional property this moment.

Performing step of the present invention and concrete scheme following (seeing accompanying drawing 2):

Step 1:, obtain the velocity r of decision low-frequency range and high band virtual sound image alignment quality respectively according to the acoustic image positioning principle _VWith acoustic energy flow vector r _E, so that carry out virtual sound image and the truly contrast of acoustic image.

r _VAnd r _EDirection, i.e. the deflection of virtual sound image is by θ _ViAnd φ _Vi, θ _EiAnd φ _EiExpression.Wherein, θ _ViAnd φ _ViBe the horizontal angle and the vertical angle of low-frequency range virtual sound image, θ _EiAnd φ _EiThen be the horizontal angle and the vertical angle of high band virtual sound image.r _VAnd r _EAmplitude can represent the definition of virtual sound image in the physical sense.Vector synthetic sound with Gerzon is an example as positioning principle, and these angles can be calculated by formula (8)～(11):

$\tan {\mathrm{\&theta;}}_{\mathrm{Vi}}=\frac{\underset{j=1}{\overset{k}{\mathrm{\&Sigma;}}}{g}_j\cos {\mathrm{\&phi;}}_j\sin {\mathrm{\&theta;}}_j}{\underset{j=1}{\overset{k}{\mathrm{\&Sigma;}}}{g}_j\cos {\mathrm{\&phi;}}_j\cos {\mathrm{\&theta;}}_j}---\left(8\right)$

$\tan {\mathrm{\&phi;}}_{\mathrm{Vi}}=\frac{\underset{j=1}{\overset{k}{\mathrm{\&Sigma;}}}\left(g_j\sin {\mathrm{\&phi;}}_j\right)}{\sqrt{{\left(\underset{j=1}{\overset{k}{\mathrm{\&Sigma;}}}{g}_j\cos {\mathrm{\&phi;}}_j\cos {\mathrm{\&theta;}}_j\right)}^2+{\left(\underset{j=1}{\overset{k}{\mathrm{\&Sigma;}}}{g}_j\cos {\mathrm{\&phi;}}_j\sin {\mathrm{\&theta;}}_j\right)}^2}}---\left(9\right)$

$\tan {\mathrm{\&theta;}}_{\mathrm{Ei}}=\frac{\underset{j=1}{\overset{k}{\mathrm{\&Sigma;}}}{g}_j^2\cos {\mathrm{\&phi;}}_j\sin {\mathrm{\&theta;}}_j}{\underset{j=1}{\overset{k}{\mathrm{\&Sigma;}}}{g}_j^2\cos {\mathrm{\&phi;}}_j\cos {\mathrm{\&theta;}}_j}---\left(10\right)$

$\tan {\mathrm{\&phi;}}_{\mathrm{Ei}}=\frac{\underset{j=1}{\overset{k}{\mathrm{\&Sigma;}}}\left(g_j^2\sin {\mathrm{\&phi;}}_j\right)}{\sqrt{{\left(\underset{j=1}{\overset{k}{\mathrm{\&Sigma;}}}{g}_j^2\cos {\mathrm{\&phi;}}_j\cos {\mathrm{\&theta;}}_j\right)}^2+{\left(\underset{j=1}{\overset{k}{\mathrm{\&Sigma;}}}{g}_j^2\cos {\mathrm{\&phi;}}_j\sin {\mathrm{\&theta;}}_j\right)}^2}}---\left(11\right)$

With corresponding low-frequency range of true sound source and the high band virtual sound image definition evaluation of estimate r on a certain direction in space _ViAnd r _EiTry to achieve by formula (12)～(13)

$r_{\mathrm{Vi}}=\frac{\sqrt{{\left(\underset{j=1}{\overset{k}{\mathrm{\&Sigma;}}}{g}_j\cos {\mathrm{\&phi;}}_j\cos {\mathrm{\&theta;}}_j\right)}^2+{\left(\underset{j=1}{\overset{k}{\mathrm{\&Sigma;}}}{g}_j\cos {\mathrm{\&phi;}}_j\sin {\mathrm{\&theta;}}_j\right)}^2+{\left(\underset{j=1}{\overset{k}{\mathrm{\&Sigma;}}}{g}_j\sin {\mathrm{\&phi;}}_j\right)}^2}}{\underset{j=1}{\overset{k}{\mathrm{\&Sigma;}}}{g}_j}---\left(12\right)$

$r_{\mathrm{Ei}}=\frac{\sqrt{{\left(\underset{j=1}{\overset{k}{\mathrm{\&Sigma;}}}{g}_j^2\cos {\mathrm{\&phi;}}_j\cos {\mathrm{\&theta;}}_j\right)}^2+{\left(\underset{j=1}{\overset{k}{\mathrm{\&Sigma;}}}{g}_j^2\cos {\mathrm{\&phi;}}_j\sin {\mathrm{\&theta;}}_j\right)}^2+{\left(\underset{j=1}{\overset{k}{\mathrm{\&Sigma;}}}{g}_j^2\sin {\mathrm{\&phi;}}_j\right)}^2}}{\underset{j=1}{\overset{k}{\mathrm{\&Sigma;}}}{g}_j^2}---\left(13\right)$

Step 2:, calculate the loudness parameter V of virtual sound image respectively at low-frequency range and high band according to the acoustic image positioning principle _LiAnd V _HiVector synthetic sound with Gerzon is an example as positioning principle, V _LiAnd V _HiCan calculate by formula (14)～(15):

$V_{\mathrm{Li}}=\underset{j=1}{\overset{k}{\mathrm{\&Sigma;}}}{g}_j---\left(14\right)$

$V_{\mathrm{Hi}}=\underset{j=1}{\overset{k}{\mathrm{\&Sigma;}}}{g}_j^2---\left(15\right)$

During with Ambisonic system sound reproduction, can be by the size of each loud speaker input signal of adjusted in concert Ambisonic system, the sound level of control virtual sound image.

Step 3: ideally, the virtual sound image that is produced during Ambisonic system sound reproduction, with the counterparty to true acoustic image identical on direction, loudness and definition, set up direction optimization aim function Afit, loudness optimization aim function Lfit and resolution optimization target function Mfit thus respectively.

The level angle and the vertical angle of the true sound source on the definition space direction are respectively θ _OiAnd φ _Oi, be example with the vector synthetic sound of Gerzon as positioning principle, then have:

r _Vi＝r _Ei＝1 (16)

θ _Vi＝θ _Ei＝θ _Oi (17)

φ _Vi＝φ _Ei＝φ _Oi (18)

In addition, when same sound when different direction in spaces occurs, with the loudness of the virtual sound image of its corresponding all directions all should be identical, have thus:

V _Li＝V _L0 (19)

V _Hi＝V _H0 (20)

To (23), the coupling system optimum theory can be set up relevant direction optimization aim function Afit, loudness optimization aim function Lfit and resolution optimization target function Mfit according to formula (19).

$\mathrm{Afit}=\underset{k=1}{\overset{6}{\mathrm{\&Sigma;}}}\left(\sqrt{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}\frac{{({\mathrm{\&alpha;}}_{\mathrm{ki}}-{\mathrm{\&beta;}}_{\mathrm{ki}})}^2}{n}}\right)---\left(21\right)$

$\mathrm{Lfit}=\sqrt{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}\frac{{(1-V_{\mathrm{Li}}/V_{L0})}^2}{n}}+\sqrt{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}\frac{{(1-V_{\mathrm{Hi}}/V_{H0})}^2}{n}}---\left(22\right)$

$\mathrm{Mfit}=\sqrt{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}\frac{{(1-r_{\mathrm{Vi}})}^2}{n}}+\sqrt{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}\frac{{(1-r_{\mathrm{Ei}})}^2}{n}}---\left(23\right)$

Wherein, n represents that virtual sound image is at space different directions hits.。

Step 4: the optimization aim function C fit that sets up the virtual directivity of microphone control of being calculated by formula (3) to (7);

Step 5: synthetic complex optimum target function Tfit;

Complex optimum target function Tfit is:

Tfit＝b ₁·Afit+b ₂·Lfit+b ₃·Mfit+b ₄·Cfit (24)

In the formula, b _j(j=1,2,3,4) are optimization aim function shared weight in total optimization aim function, and 0≤b _j≤ 1.When Tfit has minimum value, show that virtual acoustic and actual sound that the Ambisonic system produced are the most approaching on subjective sensation, this moment, system reached optimum;

Step 6: retransmit loudspeaker array in conjunction with complex optimum target function Tfit and concrete Ambisonic system, adopt optimized Algorithm, calculate the fitness value of optimum optimization aim function T fit, its pairing spheric harmonic function weight coefficient group by interative computation

Be optimum decoded result.The computation optimization flow process as shown in Figure 3.Optimized Algorithm can be selected calculus of finite differences, genetic algorithm, conjugate gradient method, simulated annealing or tabu search algorithm etc. for use.

The foregoing description is a preferred implementation of the present invention; but embodiments of the present invention are not restricted to the described embodiments; other any do not deviate from change, the modification done under spirit of the present invention and the principle, substitutes, combination, simplify; all should be the substitute mode of equivalence, be included within protection scope of the present invention.

Claims (5)

1.Ambisonic the optimization of sound reproduction system decodes is improved one's methods, and it is characterized in that, may further comprise the steps:

Step 1:, obtain the velocity r of decision low-frequency range and high band virtual sound image alignment quality respectively according to the acoustic image positioning principle _VWith acoustic energy flow vector r _E

Step 2:, calculate the loudness control parameter V of virtual sound image at low-frequency range and high band according to the acoustic image positioning principle _LiAnd V _Hi

Step 3: ideally, the virtual sound image that is produced during Ambisonic system sound reproduction, with the counterparty to true acoustic image identical on direction, loudness and definition.Set up direction optimization aim function Afit, loudness optimization aim function Lfit and resolution optimization target function Mfit thus respectively;

Step 4: the optimization aim function C fit that sets up virtual directivity of microphone control;

Step 5: set up complex optimum target function Tfit;

Step 6: in conjunction with the position of loud speaker in the Ambisonic sound reproduction system, adopt optimized Algorithm, by the complex optimum target function Tfit that interative computation determines to have adaptive optimal control degree value, its pairing spheric harmonic function weight coefficient group

Be optimum decoded result.

2. the optimization of Ambisonic sound reproduction system decodes according to claim 1 is improved one's methods, and it is characterized in that, the described direction optimization aim of step 3 function Afit, loudness optimization aim function Lfit and resolution optimization target function Mfit are respectively:

$\mathrm{Afit}=\underset{k=1}{\overset{6}{\mathrm{\&Sigma;}}}\left(\sqrt{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}\frac{{({\mathrm{\&alpha;}}_{\mathrm{ki}}-{\mathrm{\&beta;}}_{\mathrm{ki}})}^2}{n}}\right)$

$\mathrm{Lfit}=\sqrt{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}\frac{{(1-V_{\mathrm{Li}}/V_{L0})}^2}{n}}+\sqrt{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}\frac{{(1-V_{\mathrm{Hi}}/V_{H0})}^2}{n}}$

$\mathrm{Mfit}=\sqrt{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}\frac{{(1-r_{\mathrm{Vi}})}^2}{n}}+\sqrt{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}\frac{{(1-r_{\mathrm{Ei}})}^2}{n}}$

Wherein, n represents that virtual sound image is at space different directions hits.

3. the optimization of Ambisonic sound reproduction system decodes according to claim 1 is improved one's methods, and it is characterized in that, the described complex optimum target function of step 5 Tfit is:

Tfit＝b ₁·Afit+b ₂·Lfit+b ₃·Mfit+b ₄·Cfit

Wherein, b _j(j=1,2,3,4) are different optimization aim function shared weights in the complex optimum target function, and 0≤b _j≤ 1.

4. the optimization of Ambisonic sound reproduction system decodes according to claim 1 is improved one's methods, and it is characterized in that, the optimization aim function C fit of the virtual directivity of microphone control that step 4 is set up is:

(1) at θ ₁＜θ _Aim＜θ ₂And under ° situation of Δ φ≤180:

$\mathrm{Cfit}=\frac{\underset{-180\&le;\mathrm{\&theta;}\&le;180}{\mathrm{\&Sigma;}}\left(\underset{-90\&le;\mathrm{\&phi;}\&le;90}{\mathrm{\&Sigma;}}{r}_M(\mathrm{\&theta;},\mathrm{\&phi;})\right)}{\underset{{\mathrm{\&theta;}}_1\&le;\mathrm{\&theta;}\&le;{\mathrm{\&theta;}}_2}{\mathrm{\&Sigma;}}\left(\underset{{\mathrm{\&phi;}}_1\&le;\mathrm{\&phi;}\&le;{\mathrm{\&phi;}}_2}{\mathrm{\&Sigma;}}{r}_M(\mathrm{\&theta;},\mathrm{\&phi;})\right)}-1$

(2) at θ ₁＜θ _Aim＜θ ₂And under ° situation of Δ φ≤180:

$\mathrm{Cfit}=\frac{\underset{-180\&le;\mathrm{\&theta;}\&le;180}{\mathrm{\&Sigma;}}\left(\underset{-90\&le;\mathrm{\&phi;}\&le;90}{\mathrm{\&Sigma;}}{r}_M(\mathrm{\&theta;},\mathrm{\&phi;})\right)}{\underset{{\mathrm{\&theta;}}_1\&le;\mathrm{\&theta;}\&le;{\mathrm{\&theta;}}_2}{\mathrm{\&Sigma;}}\left(\underset{-90\&le;\mathrm{\&phi;}\&le;90}{\mathrm{\&Sigma;}}{r}_M(\mathrm{\&theta;},\mathrm{\&phi;})\right)\text{+}\underset{{\mathrm{\&theta;}}_2<\mathrm{\&theta;}\&le;180}{\underset{-180\&le;\mathrm{\&theta;}<{\mathrm{\&theta;}}_1,}{\mathrm{\&Sigma;}}}\left(\underset{{\mathrm{\&phi;}}_2\&le;\mathrm{\&phi;}<90}{\underset{-90<\mathrm{\&phi;}\&le;{\mathrm{\&phi;}}_1,}{\mathrm{\&Sigma;}}}{r}_M(\mathrm{\&theta;},\mathrm{\&phi;})\right)}-1$

(3) at θ _Aim＜θ ₁Or θ _Aim＞θ ₂And under ° situation of Δ φ≤180:

$\mathrm{Cfit}=\frac{\underset{-180\&le;\mathrm{\&theta;}\&le;180}{\mathrm{\&Sigma;}}\left(\underset{-90\&le;\mathrm{\&phi;}\&le;90}{\mathrm{\&Sigma;}}{r}_M(\mathrm{\&theta;},\mathrm{\&phi;})\right)}{\underset{{\mathrm{\&theta;}}_2<\mathrm{\&theta;}\&le;180}{\underset{-180\&le;\mathrm{\&theta;}<{\mathrm{\&theta;}}_1,}{\mathrm{\&Sigma;}}}\left(\underset{{\mathrm{\&phi;}}_1\&le;\mathrm{\&phi;}\&le;{\mathrm{\&phi;}}_2}{\mathrm{\&Sigma;}}{r}_M(\mathrm{\&theta;},\mathrm{\&phi;})\right)}-1$

(4) at θ _Aim＜θ ₁Or θ _Aim＞θ ₂And under ° situation of Δ φ≤180:

$\mathrm{Cfit}=\frac{\underset{-180\&le;\mathrm{\&theta;}\&le;180}{\mathrm{\&Sigma;}}\left(\underset{-90\&le;\mathrm{\&phi;}\&le;90}{\mathrm{\&Sigma;}}{r}_M(\mathrm{\&theta;},\mathrm{\&phi;})\right)}{\underset{{\mathrm{\&theta;}}_2<{\mathrm{\&theta;}}_i\&le;180}{\underset{-180\&le;\mathrm{\&theta;}<{\mathrm{\&theta;}}_1,}{\mathrm{\&Sigma;}}}\left(\underset{-90\&le;\mathrm{\&phi;}\&le;90}{\mathrm{\&Sigma;}}{r}_M(\mathrm{\&theta;},\mathrm{\&phi;})\right)+\underset{{\mathrm{\&theta;}}_1\&le;\mathrm{\&theta;}<{\mathrm{\&theta;}}_2}{\mathrm{\&Sigma;}}\left(\underset{{\mathrm{\&phi;}}_2\&le;\mathrm{\&phi;}<90}{\underset{-90<\mathrm{\&phi;}\&le;{\mathrm{\&phi;}}_1,}{\mathrm{\&Sigma;}}}{r}_M(\mathrm{\&theta;},\mathrm{\&phi;})\right)}-1$

r _M(θ, φ) be all loud speakers at the azimuth for (θ, the absolute value of the signal level value of closing on direction φ):

$r_M(\mathrm{\&theta;},\mathrm{\&phi;})=\left|\underset{j=1}{\overset{k}{\mathrm{\&Sigma;}}}{g}_j\right|=\left|\underset{j=1}{\overset{k}{\mathrm{\&Sigma;}}}(\underset{0\&le;m\&le;M,0\&le;n\&le;m,\mathrm{\&sigma;}=\&PlusMinus;1}{\mathrm{\&Sigma;}}{c}_{m,n,j}^{\mathrm{\&sigma;}}\&CenterDot;Y_{m,n}^{\mathrm{\&sigma;}}(\mathrm{\&theta;},\mathrm{\&phi;}))\right|$

Wherein, the k representative constitutes the number of loudspeakers of loudspeaker array, g _jBe the recurrent signal of loud speaker, Function Y is a spheric harmonic function,

Mixed stocker array for spheric harmonic function Y.

5. the optimization of Ambisonic sound reproduction system decodes according to claim 1 is improved one's methods, and it is characterized in that, the described optimized Algorithm of step 6 can adopt calculus of finite differences, genetic algorithm, conjugate gradient method, simulated annealing or tabu search algorithm.

Images