CN102147458B - Method and device for estimating direction of arrival (DOA) of broadband sound source - Google Patents

Abstract

The invention relates to a method and device for estimating the direction of arrival (DOA) of a broadband sound source. The method comprises the following steps: selecting one section of sound data X(n)=[x1(n),...,xP(n)]T from sound data collected from a microphone array, wherein n is a subset of Z*; evenly dividing the selected sound data X(n) into L frames; carrying out N-point quick Fourier transform on each frame of sound data Xl (n); presenting the frequency domain of the obtained sound data as follows: Xl(k)=[xl1(k),...,xlP(k)]T, wherein k=1,..., N/2; then, calculating the modulus square value eta (k) of a coherence function between two microphones in the microphone array; randomly selecting the modulus square values of q coherence functions from the eta (k); and estimating the DOA only by using a frequency band corresponding to the modulus squares of the q coherence functions. According to the method and device, the computation amount of an algorithm is greatly lowered on the premise of ensuring that the precision is hardly lowered.

Description

A kind of Wave arrival direction estimating method and device thereof for broadband sound source

Technical field

The present invention relates to the Bearing technical field, particularly a kind of Wave arrival direction estimating method and device thereof for broadband sound source.

Background technology

Bearing has important application in fields such as sonar, radar, navigation and wireless sensor networks.And sensor array is the major way of Bearing.

In the Bearing technology, an important technological problems is how broadband sound source to be positioned.For this respect a lot of algorithms has been arranged.Broadband MUSIC algorithm be first with broadband signal by being fourier transformed on the frequency domain, then each frequency band is processed with the MUSIC method of arrowband, the direction that at last result of each frequency band is comprehensively obtained sound source is estimated.Be incoherent subspace method (ISM, Incoherent Sub-space Method).This class methods operand is little, but low precision, and can only process noncoherent signal; Maximum likelihood (ML, Maximum Likelihood) algorithm for estimating is the match class algorithm of representative, and is good gradual with it, applicability flexibly, and computation complexity that can be low has obtained extensive concern.For broadband signal, the people such as Yao professor Kung have proposed AML (the Approximated Maximum Likelihood) DOA estimation algorithm for broadband sound source.This class arithmetic accuracy is high, can process coherent signal, but operand is very large.Simultaneously, can also process with the method for focussing matrix for the coherent signal source.At first needing to utilize low complexity algorithm that Sounnd source direction is carried out one estimates, by this angle configuration focussing matrix of estimating, by focussing matrix on the signal focus to of the different frequency bands centre frequency, then it is processed as a narrow band signal, the shortcoming of this method is to know the angle of estimating of sound source, and the selection of estimating angle is very large for the Algorithm Performance impact.

Summary of the invention

The object of the invention is to, at the operand that guarantees greatly to reduce in the situation that precision does not reduce substantially algorithm.

For achieving the above object, a kind of Wave arrival direction estimating method for broadband sound source is proposed, the method concrete steps comprise:

Step 1): from the voice data that microphone array gathers, choose one section voice data X (n)=[x ₁(n) ..., x _P(n)] ^T, wherein, P represents the number of microphone in the microphone array, P 〉=2; This section voice data X (n) evenly is divided into the L frame, the data X of every frame _l(n) length is N, and the data length of X (n) is N * L, l=1 ..., L; N ∈ Z ^*

To each frame voice data X _l(n) do the conversion of N point quick Fourier, obtain the frequency domain representation X of voice data _l(k)=[x _L1(k) ..., x _LP(k)] ^T, k=1 wherein, 2 ..., N/2;

Step 2): according to described step 1 X that) obtains _l(k) calculate the mould square η (k) of coherence function between two microphones in the microphone array according to following formula (1);

$\mathrm{\η}\left(k\right)={\left|{\mathrm{\γ}}_{i,j}\left(k\right)\right|}^2=\frac{{\left|{\mathrm{\Φ}}_{i,j}\left(k\right)\right|}^2}{{\mathrm{\Φ}}_{i,i}\left(k\right){\mathrm{\Φ}}_{j,j}\left(k\right)}---\left(1\right)$

Wherein, k=1,2 ..., N/2; I, j ∈ [1,2 ..., P]; Φ _{I, j}(k) be the cross-power spectrum of i microphone and j microphone, Φ _{I, i}(k) be the auto-power spectrum of i microphone,

Φ _{J, j}(k) be the auto-power spectrum of j microphone,

Step 3): according to described step 2 η (k) that) obtains select at random q coherence function mould square;

The voice data X of square correspondence of q the coherence function mould of step 4): according to described step 3) selecting _l(k) carry out direction of arrival and estimate, wherein, k ∈ [k ₁..., k _q], k _qExpression band number, and [k ₁..., k _q] ∈ [1 ..., N/2].

Described step 1) also comprises: preserve the continuous L frame voice data that described microphone array gathers till current; And preserve from behind the up-to-date 1 frame voice data of described microphone array, with the earliest 1 frame voice data deletion in the continuous L frame voice data of preserving.

Described step 2) in, two microphones are two microphones of space length maximum in the microphone array.

Described step 2) in, between two microphones continuously the data of (L-1) frame ask for the mould square of coherence function as data buffer storage according to N data of the current acquisition L frame individual data of the N that deposits * (L-1) of easing up.

Described step 4) in, according to described step 3) the L frame voice data X of square correspondence of q coherence function mould selecting _L(k) carry out direction of arrival and estimate, wherein, k ∈ [k ₁..., k _q], k _qExpression band number, and [k ₁..., k _q] ∈ [1 ..., N/2].

Described step 3) η (k) size that obtains in, more described step 2) is selected a front q η (k) value, i.e. η (k successively from the beginning of maximum ₁) 〉=η (k ₂) 〉=... 〉=η (k _q) 〉=η (k _m), wherein, k _qThe expression band number, [k ₁..., k _q] ∈ [1,2 ...., N/2],

For achieving the above object, propose a kind of direction of arrival estimation unit for broadband sound source, it is characterized in that this device comprises: the mould of Fourier transform module, coherence function square module and direction of arrival estimation module;

Described Fourier transform module comprises: data selection unit, data framing unit and Fourier transform unit; Be used for described data selection unit from the voice data that microphone array gathers, choose one section voice data X (n)=[x ₁(n) ..., x _P(n)] ^T, the voice data X (n) that described data selection unit is selected in described data framing unit evenly is divided into the L frame, and described Fourier transform unit is to each frame voice data X _l(n) do the conversion of N point quick Fourier, obtain the frequency domain representation X of voice data _l(k)=[x _L1(k) ..., x _LP(k)] ^TAnd the voice data of described frequency domain is exported to the mould square module of described coherence function; Wherein, P represents the number of microphone in the microphone array, P 〉=2; The data X of every frame _l(n) length is N, l=1 ..., L, the data length of X (n) is N * L; N ∈ Z ^*K=1,2 ..., N/2;

The mould of described coherence function square module is used for the X that obtains according to described Fourier transform module _l(k) calculate the mould square η (k) of coherence function between two microphones in the microphone array according to formula (2);

$\mathrm{\η}\left(k\right)={\left|{\mathrm{\γ}}_{i,j}\left(k\right)\right|}^2=\frac{{\left|{\mathrm{\Φ}}_{i,j}\left(k\right)\right|}^2}{{\mathrm{\Φ}}_{i,i}\left(k\right){\mathrm{\Φ}}_{j,j}\left(k\right)}---\left(2\right)$

Wherein, k=1,2 ..., N/2; I, j ∈ [1,2 ..., P]; Φ _{I, j}(k) be the cross-power spectrum of i microphone and j microphone,

Φ _{I, i}(k) be the auto-power spectrum of i microphone,

Φ _{J, j}(k) be the auto-power spectrum of j microphone,

Described direction of arrival estimation module, the mould that is used for the coherence function that obtains according to the mould square module from described coherence function square is selected the voice data X of square correspondence of q coherence function mould at random _l(k) carry out direction of arrival and estimate, wherein, k ∈ [k ₁..., k _q], k _qExpression band number, and [k ₁..., k _q] ∈ [1 ..., N/2].

In the mould of the described coherence function square module, two microphones are two microphones of space length maximum in the microphone array.

Described device also comprises the voice data cache module; Described voice data cache module be used for to be preserved the continuous L frame voice data that microphone array gathers till current; And preserve and to receive from behind the up-to-date 1 frame voice data of microphone array, with the earliest 1 frame voice data deletion in the continuous L frame voice data of preserving; And export the voice data of buffer memory to described Fourier transform module.

Between two microphones that the mould of described coherence function square module will be chosen continuously the data of (L-1) frame ask for the mould square of coherence function as data buffer storage according to N data of the current acquisition L frame individual data of the N that deposits * (L-1) of easing up.

The L frame voice data X of square correspondence of q the coherence function mould that described direction of arrival estimation module obtains according to described comparison module _L(k) carry out direction of arrival and estimate, wherein, k ∈ [k ₁..., k _q], k _qExpression band number, and [k ₁..., k _q] ∈ [1 ..., N/2].

This device also comprises a comparison module, for η (k) size that the mould square module according to more described coherence function obtains, selects successively a front q η (k) value, i.e. η (k from the beginning of maximum ₁) 〉=η (k ₂) 〉=... 〉=η (k _q) 〉=η (k _m), wherein, k _qThe expression band number, [k ₁..., k _q] ∈ [1,2 ...., N/2],

Square export the mould of the coherence function picked out to described direction of arrival estimation module.

The invention has the advantages that the present invention is computation reduction greatly.Any algorithm that this shortcut calculation of while is discussed applicable to the background technology part.Take the AML algorithm as example, because the operand of AML algorithm is maximum, can draw in the suitable situation of directional effect by checking, greatly reduce operand.

Description of drawings

Fig. 1 is the space structure synoptic diagram of microphone array;

Fig. 2 is a kind of Wave arrival direction estimating method process flow diagram for broadband sound source that the present invention proposes;

Fig. 3 be in the l-G simulation test in the white noise situation embodiment of the invention scheme and Whole frequency band scheme carry out the as a result comparison diagram that direction of arrival is estimated;

Fig. 4 be in the l-G simulation test in the coloured noise situation embodiment of the invention scheme and Whole frequency band scheme carry out the as a result comparison diagram that direction of arrival is estimated;

Fig. 5 is that embodiment of the invention scheme and Whole frequency band scheme are carried out the as a result comparison diagram that direction of arrival is estimated in the field test.

Embodiment

The present invention is further described below in conjunction with drawings and Examples.

Be provided with R omnidirectional (omni-directional) microphone, p microphone receives signal and is:

$x_p\left(n\right)=\underset{m=1}{\overset{M}{\mathrm{\Σ}}}{s}_0^{\left(m\right)}(n-t_p^{\left(m\right)})+w_p\left(n\right)---\left(1\right)$

Wherein: n=0 ..., L-1, L are a frame signal sampled point number; P=1 ..., R, R are the number of microphone in the receiving array; M=1 ..., M, M are sound source quantity (M＜R). Be source signal, Be the time delay that represents with sampling number.w _p(n) be noise component in the array received signal.Wherein, time delay

Geometric relationship according to formation and target sound source position obtains.For the unit of the R under far field condition circle battle array, namely R microphone equidistantly is arranged on the circumference, and the time delay that represents with sampling number of p microphone is:

$t_p^{\left(m\right)}=\frac{f_{s}r}{c}\cos (\frac{2\mathrm{\π}(p-1)}{R}-{\mathrm{\θ}}_m)---\left(2\right)$

Wherein, f _sBe the signal sampling frequency, r is circle battle array radius, and c is the velocity of wave in the isotropic medium, θ _mIt is the position angle, far field in m source.

Do to received signal the FFT conversion that N is ordered, get the frequency domain signal model:

X(k)＝D(Θ，k)S ₀(k)+η(k) (3)

In the formula, k=0 ..., N/2; Frequency domain array output signal: X (k)=[X ₁(k) ..., X _R(k)] ^TThe frequency spectrum of M sound source:

Direction matrix (Steering Matrix) is: D (Θ, k)=[d ⁽¹⁾(k) ..., d ^(M)(k)]; Direction vector (Steering Vector) is:

η (k) is noise spectrum.

Noise spectrum vector η (k) is that variance is L σ ²Zero-mean plural number white Gaussian noise.Be any independent identically distributed stochastic variable even notice w (n), according to central limit theorem, all can know η (k) Gaussian distributed by inference.

Definition S (k)=D (k) S ₀(k) (4)

Obtain NR/2 dimension space-time-frequency vector:

X＝G(θ)+ξ ₁ (5)

Wherein, G (θ)=[S (0) ^T..., S (N/2) ^T] ^T, and have:

$R_{\mathrm{\ξ}}=E\left[{\mathrm{\ξ\ξ}}^H\right]=L\left[\begin{array}{cccc}{\mathrm{\σ}}^2{I}_R& & & 0\\ & {\mathrm{\σ}}^2{I}_R& & \\ & & L& \\ 0& & & {\mathrm{\σ}}^2{I}_R\end{array}\right]---\left(6\right)$

Wherein, superscript ^HRepresent the conjugate transpose of complex matrix.

Suppose that initial unknown parameter space is:

$\mathrm{\Θ}={[{\mathrm{\θ}}_s^T,S_0^{{\left(1\right)}^T},...,S_0^{{\left(M\right)}^T}]}^T---\left(7\right)$

Wherein, source side parallactic angle

The source frequency spectrum

So maximum likelihood function is arranged:

$f\left(\mathrm{\Θ}\right)=\frac{1}{{\left(2\mathrm{\π}\right)}^{\mathrm{NR}/2/2}{\left|R_{\mathrm{\ξ}}\right|}^{1/2}}\exp \{-\frac{1}{2}{[X-G\left(\mathrm{\Θ}\right)]}^H{R_{\mathrm{\ξ}}}^{-1}[X-G\left(\mathrm{\Θ}\right)\left]\right\}---\left(8\right)$

This likelihood function of logarithmetics also omits constant term and gets:

$L\left(\mathrm{\Θ}\right)=-\frac{{\left|\right|X-G\left(\mathrm{\Θ}\right)\left|\right|}^2}{R_{\mathrm{\ξ}}}---\left(9\right)$

Can obtain optimization criterion:

$\underset{\mathrm{\Θ}}{\max }L\left(\mathrm{\Θ}\right)=\underset{\mathrm{\Θ}}{\min }\underset{k=0}{\overset{N/2}{\mathrm{\Σ}}}\frac{{\left|\right|X\left(k\right)-D\left(k\right)S_0\left(k\right)\left|\right|}^2}{R_{\mathrm{\ξ}}\left(k\right)}$

$=\underset{\mathrm{\Θ}}{\min }\underset{k=0}{\overset{N/2}{\mathrm{\Σ}}}\frac{{\left|\right|X\left(k\right)-D\left(k\right)S_0\left(k\right)\left|\right|}^2}{{\mathrm{\σ}}^2}---\left(10\right)$

This is equivalent to all k function, finds Again because σ ²Be constant, f (k) can be abbreviated as:

f(k)＝||X(k)-D(k)S ₀(k)|| ² (11)

The minimizing source signal vector S that f (k) is got ₀(k) certain satisfied Therefore for any source being positioned with minimum residual error, the estimation of source vector is given:

Wherein,

It is the pseudo inverse matrix (pseudo-inverse) of direction matrix D (k).The definition vertical plan The direction of arrival of AML algorithm is estimated and can be obtained by following formula so.

${\mathrm{\θ}}_s=\underset{{\mathrm{\θ}}_s}{\max }\underset{k=0}{\overset{N/2}{\mathrm{\Σ}}}{\left|\right|P(k,{\mathrm{\θ}}_s)X\left(k\right)\left|\right|}^2---\left(13\right)$

Although voice signal is the broadband, signal generally all can not be white, and generally all there are some harmonics structures in it, and namely signal mainly concentrates on some frequency range; Or because noise is coloured, and the signal of some frequency band has been fallen into oblivion by noise fully, therefore when calculating, the information on these frequency bands is redundant, even can affect last result.So it is necessary adopting certain method to choose some frequency band, among the present invention program, with the standard of the coherence function on the frequency band as the selection frequency band.

Coherence function (coherence) is used for describing the correlativity between two groups of data.Among the present invention program, coherence function is defined as:

${\mathrm{\γ}}_{\mathrm{xy}}=\frac{G_{\mathrm{xy}}\left(f\right)}{\sqrt{G_{\mathrm{xx}}\left(f\right)G_{\mathrm{yy}}\left(f\right)}}---\left(14\right)$

Wherein, x represents two groups of different data, G with y _Xy(f) be the cross-spectrum of x and y, G _Xx(f) and G _Yy(f) be respectively the certainly spectrum of x and y.G _xy(f)＝E{X(f)Y ^*(f)}，G _xx(f)＝E{X(f)X ^*(f)}，G _yy(f)＝E{Y(f)Y ^*(f)}。

Amplitude square coherence function (MSC) is coherence function γ _XyMould square:

${\left|{\mathrm{\γ}}_{\mathrm{xy}}\left(f\right)\right|}^2=\frac{{\left|G_{\mathrm{xy}}\left(f\right)\right|}^2}{G_{\mathrm{xx}}\left(f\right)G_{\mathrm{yy}}\left(f\right)}---\left(15\right)$

Thus, visible coherence function refers to all is computing on the frequency domain.

The value of coherence function is between 0 to 1, that is:

0≤|γ _xy| ²≤1 (16)

Represent that when value is 0 the frequency spectrum on frequency range x and the y is fully uncorrelated, value is the frequency spectrum complete dependence on 1 expression frequency range x and the y.

In specific situation, have: signal to noise ratio (S/N ratio)

This shows that MSC can be used for the size of the signal to noise ratio (S/N ratio) on certain frequency band of gauge signal.When | γ _Xy(k) | ²When larger, show that signal useful on k the frequency band is more; Otherwise, illustrate that then noise accounts for major part on the k frequency band.

The space structure synoptic diagram of microphone array, as shown in Figure 1.Use the microphone array of cross formation shown in Figure 1 to carry out the collection of wind noise cancellation signal, 12 pieces of B﹠amp of circle 101 expressions; K 4189 type standard microphones, and be furnished with the spherical sponges fan housing that diameter is 10cm.102 expressions are used for fixing the support of these microphones.The aperture of upper three 4 element array of cross battle array is respectively 0.2 meter, 0.6 meter, and 1.2 meters.Adopt SONY SIR-1000 type 16 Channel Synchronous digital audio tapes, sample frequency is 48kHz.During data analysis, by low-pass filtering, normalization, down-sampledly analyze to 1kHz.

Use measured data, the spatial coherence of statistic array wind noise.Adopt following formula to calculate the interchannel average correlation coefficient:

$\stackrel{\‾}{\mathrm{\γ}}\left(k\right)=\frac{2}{R(R-1)}\underset{i=1}{\overset{R}{\mathrm{\Σ}}}\underset{j=i+1}{\overset{R}{\mathrm{\Σ}}}\left|{\mathrm{\γ}}_{\mathrm{ij}}\left(k\right)\right|---\left(17\right)$

Wherein, k=0 ..., N/2; R is port number, γ _Ij(k) be the cross-correlation coefficient of i passage and j passage,

Be average correlation coefficient.

Table 1 wind noise interchannel average correlation coefficient

Table 1 explanation, different from the spatial coherence of low frequency wind noise, the wind noise interchannel related coefficient at medium-high frequency place is little, and spatial coherence is faint, and the array wind noise can think that substantially interchannel is separate.

The present invention program is after having gathered voice data, according to coefficient of coherence voice data is screened, filter out the voice data that carries the more frequency range of useful signal, and abandon the voice data of other frequency ranges, so, just reduce direction of arrival and estimated required data volume, thereby reduced the operand that follow-up direction of arrival is estimated.

A kind of Wave arrival direction estimating method process flow diagram for broadband sound source that the present invention proposes; As shown in Figure 2, comprise the steps:

Step 201: from the voice data that the microphone array that is comprised of P microphone gathers, select one section voice data x (n)=[x ₁(n) ..., x _P(n)] ^T, wherein, n ∈ Z ^*

Step 202: selected voice data x (n) evenly is divided into L section x _l(n), l=1 ..., L, every section data x _l(n) length is N, and the data data length of x (n) is N * L.

Step 203: to each section voice data x _l(n) do N point quick Fourier conversion (FFT), the frequency domain representation that obtains voice data is X _l(k)=[X _L1(k) ..., X _LP(k)] ^T, wherein, k=1,2 ..., N/2.

Step 204: square (MSC) that calculates the coherence function mould between two microphones in the microphone array

Wherein, i, j ∈ [1,2 ..., P], Φ _{I, j}(k) be the cross-power spectrum of i microphone and j microphone, Φ _{I, i}(k) and Φ _{J, j}(k) be respectively the auto-power spectrum of i microphone and j microphone:

${\mathrm{\Φ}}_{i,i}\left(k\right)=\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{\left|X_{\mathrm{li}}\left(k\right)\right|}^2;$ ${\mathrm{\Φ}}_{j,j}\left(k\right)=\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{\left|X_{\mathrm{lj}}\left(k\right)\right|}^2;$ ${\mathrm{\Φ}}_{i,j}\left(k\right)=\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{X}_{\mathrm{li}}\left(k\right)X_{\mathrm{lj}}^{*}\left(k\right).$

Preferably, described two microphones can be two microphones of space length maximum in the microphone array.

Step 205: from square η (k) of coherence function mould, pick out the square value of larger q coherence function mould, i.e. η (k ₁) 〉=η (k ₂) 〉=... 〉=η (k _q) 〉=η (k _m), wherein, k _qBe band number, [k ₁..., k _q] ∈ [1,2 ...., N/2],

Step 206: the voice data X that uses square correspondence of q the coherence function mould of picking out _l(k) carrying out direction of arrival estimates.Wherein, k ∈ [k ₁..., k _q]; Take the AML algorithm as example, then the computing formula of angle by (18) as can be known,

${\mathrm{\θ}}_s=\underset{{\mathrm{\θ}}_s}{\max }\underset{k=k_1}{\overset{k_q}{\mathrm{\Σ}}}{\left|\right|P(k,{\mathrm{\θ}}_s)X_l\left(k\right)\left|\right|}^2---\left(18\right)$

Because the calculating of coherence function needs more data, namely N * L is larger, therefore may affect the real-time of method.In the calculating of coherence function, before having preserved continuously the data of (L-1) frame as data buffer storage, according to N data of current acquisition ease up deposit N * (L-1) individual data are asked for MSC.And obtain after the new frame data at every turn, all abandon the data of former frame, upgrade whole data buffer storage.Be x such as the total data in the system _l(n), l=1 ..., L, wherein x _L(n) be frame data of current acquisition, and x _l(n), l=1 ..., L-1 is former data.Use x _l(n), l=1 ..., L calculates MSC, and only uses x _L(n) carry out angle estimation.

The present invention also proposes a kind of device of estimating for the direction of arrival of broadband sound source, comprising:

Fourier transform module is used for the voice data of the microphone array collection of P microphone composition is carried out fast fourier transform, obtains the frequency domain representation of voice data, and described frequency domain voice data is exported to the mould square module of coherence function;

The mould of coherence function square module be used for to be calculated the mould square of the coherence function of the voice data between two microphones of microphone array, and square exports the mould of the coherence function that calculates to comparison module;

Comparison module is used for square picking out a larger q value from the mould of coherence function;

The direction of arrival estimation module, the frequency band of voice data that is used for the mould square correspondence of the coherence function picked out according to comparison module carries out direction of arrival and estimates.

Preferably, the mould of described coherence function square module be used for to be calculated the mould square of the coherence function of voice data between microphone array middle distance two microphones farthest, and square exports the mould of the coherence function that calculates to comparison module.

Preferably, described Fourier transform module comprises:

Data selection unit, being used for from the voice data selection length of the microphone array collection that is comprised of P microphone is that one section voice data of N * L is as pending voice data x (n)=[x ₁(n) ..., x _P(n)] ^T, wherein, n ∈ Z ^*

The data framing unit is used for the selected pending voice data x of described data selection unit (n) is divided into the L section, and wherein arbitrary section voice data is x _l(n), l=1 ..., L, and the voice data after the segmentation exported; And

Fourier transformation unit is used for each section voice data x to the output of described data framing unit _l(n) do the conversion of N point quick Fourier, the frequency domain representation that obtains voice data is X _l(k)=[X _L1(k) ..., X _LP(k)] ^T, wherein, k=1,2 ..., N/2.

Preferably, the mould of coherence function square module is calculated the mould of the coherence function between i microphone of microphone array and j the microphone and square is:

Wherein, i, j ∈ [1,2 ..., P], Φ _{I, j}(k) be the cross-power spectrum of i microphone and j microphone, Φ _{I, i}(k) and Φ _{J, j}(k) be respectively the auto-power spectrum of i microphone and j microphone.

Preferably, described comparison module is picked out the mould square value of a larger q coherence function, i.e. η (k from the mould square η (k) of N/2 coherence function ₁) 〉=η (k ₂) 〉=... 〉=η (k _q), k _qThe expression band number, k=1,2 ..., N/2, [k ₁..., k _q] ∈ [1,2 ...., N/2].

Preferably, the frequency band of the voice data of the mould square correspondence of the coherence function picked out according to comparison module of described DOA estimation module carries out direction of arrival and estimates.

Preferably, this device further comprises:

The voice data cache module is used for preserving the continuous L frame voice data that described microphone array gathers till current; And after receiving the 1 frame voice data up-to-date from described microphone array, with the earliest 1 frame voice data deletion in the continuous L frame voice data of preserving.

The capable fast fourier transform of voice data that described Fourier transform module is used for the voice data cache module is stored obtains the frequency domain representation of voice data, and described frequency domain voice data is exported to the mould square module of coherence function.

Below result by emulation experiment and experiment on the spot the present invention program's technique effect is described.In l-G simulation test, the microphone group pattern is the circle battle array, and array number is 4, and the radius of circle battle array is 0.2m, and sound source is white noise, and number is 1, and incident angle is 70 °.The AML algorithm of use Whole frequency band and the AML arithmetic result that frequency band is chosen contrast respectively as shown in Figure 3 and Figure 4, and wherein, Fig. 3 is the situation of white noise, and Fig. 4 is the situation of coloured noise.Wherein coloured noise is to be produced by the FIR wave filter of white noise by 50 rank.

As can be seen from Figure 3, when background was the condition of white noise, the AML algorithm of Whole frequency band was better than the AML algorithm that frequency band is chosen under the low signal-to-noise ratio; The deviation of these two kinds of algorithms is close under the high s/n ratio all trends towards zero.As can be seen from Figure 4, when background is coloured noise, it has not been optimum that low signal-to-noise ratio uses the AML algorithm of Whole frequency band, as can be seen from the figure under different signal to noise ratio (S/N ratio)s, should choose different frequency band number, this is that signal to noise ratio (S/N ratio) can be very large on some frequency band because background is coloured noise, and signal to noise ratio (S/N ratio) can be very little on some frequency band, therefore choose too much frequency band and can make the low frequency band of these signal to noise ratio (S/N ratio)s also participate in computing, thereby the deviation of algorithm is increased; Under the high s/n ratio, the deviation of two kinds of algorithms also all levels off to 0.

In the actual experiment, array is the circle battle array in the open air, and array number is 4, and circle battle array radius is 0.2m, and sound source is that tank moves in a circle around array.Experimental result as shown in Figure 5, icon " * " is the result of Whole frequency band search, icon " o " is the result that 20 frequency bands only getting the coefficient of coherence maximum calculate gained, as can be seen from the figure, both bearing accuracies are substantially the same, but after using coefficient of coherence to select frequency band, and operand is original 1/15.

Through the above description of the embodiments, those skilled in the art can be well understood to the present invention and can realize by the mode that software adds essential hardware platform, can certainly all implement by hardware, but the former is better embodiment in a lot of situation.Based on such understanding, technical scheme of the present invention is to can embodying with the form of software product in whole or in part that background technology contributes, this computer software product can be stored in the storage medium, such as ROM/RAM, magnetic disc, CD etc., comprise that some instructions are with so that computer equipment is carried out the described method of some part of each embodiment of the present invention or embodiment.Computer equipment is such as personal computer, server, perhaps network equipment etc.

It should be noted last that above embodiment is only unrestricted in order to technical scheme of the present invention to be described.Although with reference to embodiment the present invention is had been described in detail, those of ordinary skill in the art is to be understood that, technical scheme of the present invention is made amendment or is equal to replacement, do not break away from the spirit and scope of technical solution of the present invention, it all should be encompassed in the middle of the claim scope of the present invention.

Claims (12)

1. Wave arrival direction estimating method for broadband sound source, the method concrete steps comprise:

Step 1): from the voice data that microphone array gathers, choose one section voice data X (n)=[x ₁(n) ..., x _P(n)] ^T, wherein, P represents the number of microphone in the microphone array, P 〉=2; This section voice data X (n) evenly is divided into the L frame, the data X of every frame _l(n) length is N, and the data length of X (n) is N * L, l=1 ..., L; N ∈ Z ^*

To each frame voice data X _l(n) do the conversion of N point quick Fourier, obtain the frequency domain representation X of voice data _l(k)=[x _L1(k) ..., x _LP(k)] ^T, k=1 wherein, 2 ..., N/2;

Step 2): the X that obtains according to described step 1) _l(k) calculate the mould square η (k) of coherence function between two microphones in the microphone array according to following formula (1);

$\mathrm{\η}\left(k\right)={\left|{\mathrm{\γ}}_{i,j\left(k\right)}\right|}^2=\frac{{\left|{\mathrm{\Φ}}_{i,j}\left(k\right)\right|}^2}{{\mathrm{\Φ}}_{i,i}\left(k\right){\mathrm{\Φ}}_{j,j}\left(k\right)}---\left(1\right)$

Wherein, k=1,2 ..., N/2; I, j ∈ [1,2 ..., P]; Φ _{I, j}(k) be the cross-power spectrum of i microphone and j microphone,

Φ _{I, i}(k) be the auto-power spectrum of i microphone, ${\mathrm{\Φ}}_{i,i}\left(k\right)=\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{|x_{\mathrm{li}}\left(k\right)|}^2;$ Φ _{J, j}(k) be the auto-power spectrum of j microphone, ${\mathrm{\Φ}}_{j,j}\left(k\right)=\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{\left|x_{\mathrm{lj}}\right|}^2;$

Step 3): according to described step 2) η (k) that obtains from the beginning of maximum select successively q coherence function mould square;

Step 4): the voice data X of square correspondence of q the coherence function mould of selecting according to described step 3) _l(k) carry out direction of arrival and estimate, wherein, k ∈ [k ₁..., k _q], k _qExpression band number, and [k ₁..., k _q] ∈ [1 ..., N/2].

2. the Wave arrival direction estimating method for broadband sound source according to claim 1 is characterized in that, described step 1) also comprises: preserve the continuous L frame voice data that described microphone array gathers till current; And preserve from behind the up-to-date 1 frame voice data of described microphone array, with the earliest 1 frame voice data deletion in the continuous L frame voice data of preserving.

3. the Wave arrival direction estimating method for broadband sound source according to claim 1 is characterized in that, described step 2) in, two microphones are two microphones of space length maximum in the microphone array.

4. the Wave arrival direction estimating method for broadband sound source according to claim 2, it is characterized in that, described step 2) in, between two microphones continuously the data of (L-1) frame ask for the mould square of coherence function as data buffer storage according to N data of the current acquisition L frame individual data of the N that deposits * (L-1) of easing up.

5. the Wave arrival direction estimating method for broadband sound source according to claim 4 is characterized in that, in the described step 4), and the L frame voice data X of square correspondence of q the coherence function mould of selecting according to described step 3) _L(k) carry out direction of arrival and estimate, wherein, k ∈ [k ₁..., k _q], k _qExpression band number, and [k ₁..., k _q] ∈ [1 ..., N/2].

6. the Wave arrival direction estimating method for broadband sound source according to claim 1, it is characterized in that in the described step 3), more described step 2) in η (k) size that obtains, select successively a front q η (k) value, i.e. η (k from the beginning of maximum ₁) 〉=η (k ₂) 〉=... 〉=η (k _q) 〉=η (k _m), wherein, k _qThe expression band number, [k ₁..., k _q∈ [1,2 ...., N/2],

7. the direction of arrival estimation unit for broadband sound source is characterized in that this device comprises: the mould of Fourier transform module, coherence function square module and direction of arrival estimation module;

Described Fourier transform module comprises: data selection unit, data framing unit and Fourier transform unit; Described data selection unit is chosen one section voice data X (n)=[x from the voice data that microphone array gathers ₁(n) ..., x _P(n)] ^T, the voice data X (n) that described data selection unit is selected in described data framing unit evenly is divided into the L frame, and described Fourier transform unit is to each frame voice data X _l(n) do the conversion of N point quick Fourier, obtain the frequency domain representation X of voice data _l(k)=[x _L1(k) ..., x _LP(k)] ^TAnd described frequency domain voice data is exported to the mould square module of described coherence function; Wherein, P represents the number of microphone in the microphone array, P 〉=2; The data X of every frame _l(n) length is N, l=1 ..., L, the data length of X (n) is N * L; N ∈ Z ^*K=1,2 ..., N/2;

The mould of described coherence function square module is used for the X that obtains according to described Fourier transform module _l(k) calculate the mould square η (k) of coherence function between two microphones in the microphone array according to formula (2);

$\mathrm{\η}\left(k\right)={\left|{\mathrm{\γ}}_{i,j\left(k\right)}\right|}^2=\frac{{\left|{\mathrm{\Φ}}_{i,j}\left(k\right)\right|}^2}{{\mathrm{\Φ}}_{i,i}\left(k\right){\mathrm{\Φ}}_{j,j}\left(k\right)}---\left(2\right)$

Wherein, k=1,2 ..., N/2; I, j ∈ [1,2 ..., P]; Φ _{I, j}(k) be the cross-power spectrum of i microphone and j microphone,

Φ _{I, i}(k) be the auto-power spectrum of i microphone,

${\mathrm{\Φ}}_{i,i}\left(k\right)=\underset{l=1}{\overset{L}{\mathrm{\Σ}}}|{{X_{\mathrm{lj}}\left(k\right)|}_{;}}^2$ Φ _{J, j}(k) be the auto-power spectrum of j microphone, ${\mathrm{\Φ}}_{j,j}\left(k\right)=\underset{l=1}{\overset{L}{\mathrm{\Σ}}}{\left|X_{\mathrm{lj}}\left(k\right)\right|}^2;$

Described direction of arrival estimation module, the mould square that is used for the coherence function that obtains according to the mould square module from described coherence function is selected the voice data X of square correspondence of q coherence function mould successively from the beginning of maximum _l(k) carry out direction of arrival and estimate, wherein, k ∈ [k ₁..., k _q], k _qExpression band number, and [k ₁..., k _q] ∈ [1 ..., N/2].

8. the direction of arrival estimation unit for broadband sound source according to claim 7 is characterized in that, in the mould of the described coherence function square module, two microphones are two microphones of space length maximum in the microphone array.

9. the direction of arrival estimation unit for broadband sound source according to claim 7 is characterized in that described device also comprises the voice data cache module;

Described voice data cache module be used for to be preserved the continuous L frame voice data that microphone array gathers till current; And preserve and to receive from behind the up-to-date 1 frame voice data of microphone array, with the earliest 1 frame voice data deletion in the continuous L frame voice data of preserving; And export the voice data of buffer memory to described Fourier transform module.

10. the direction of arrival estimation unit for broadband sound source according to claim 9, it is characterized in that, between two microphones that the mould of described coherence function square module will be chosen continuously the data of (L-1) frame ask for the mould square of coherence function as data buffer storage according to N data of the current acquisition L frame individual data of the N that deposits * (L-1) of easing up.

11. the direction of arrival estimation unit for broadband sound source according to claim 10 is characterized in that, the L frame voice data X of square correspondence of q the coherence function mould that described direction of arrival estimation module obtains according to comparison module _L(k) carry out direction of arrival and estimate, wherein, k ∈ [k ₁..., k _q].

12. the direction of arrival estimation unit for broadband sound source according to claim 7, it is characterized in that, this device also comprises a comparison module, be used for η (k) size that the mould square module according to more described coherence function obtains, select successively a front q η (k) value, i.e. η (k from the beginning of maximum ₁) 〉=η (k ₂) 〉=... 〉=η (k _q) 〉=η (k _m), wherein, k _qThe expression band number, [k ₁..., k _q] ∈ [1,2 ...., N/2],

Square export the mould of the coherence function picked out to described direction of arrival estimation module.

Images