JP2010134344A - Adaptive signal processing system and transfer characteristic setting method for the same - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an adaptive signal processing system and a transfer characteristic setting method for the same, capable of reducing a calculation amount required for calculation of a coefficient update equation of adaptive signal processing, while reducing an amount of a transfer characteristic to be stored. <P>SOLUTION: When the adaptive signal processing is performed in a frequency domain, the transfer characteristic from each speaker to each microphone is calculated for each frequency component, and a transfer gain is calculated by each transfer characteristic in two or more adding terms of the coefficient update equation, and compared for each frequency. When a gain difference is a preset value or less, the adding term with a smaller transfer gain is considered as 0, and the transfer characteristic of the adding term with a larger gain is set in an adaptive control section. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an adaptive signal processing system and a transfer characteristic setting method thereof, and in particular, sets a transfer characteristic from each control sound source (speaker) to each microphone in an adaptive control unit, and updates the coefficient including the transfer characteristic in the adaptive control unit The present invention relates to an adaptive signal processing system that determines a coefficient of an adaptive filter so that error signal power is minimized using an equation, and a transfer characteristic setting method thereof.

Active noise control devices using an adaptive control unit (Filtered-x-LMS algorithm, Single frequency Adaptive Notch Filter algorithm, etc.) have become widespread. In an active noise control apparatus, in general, active muffling is realized by an adaptive control unit that automatically operates so that the power of noise at a microphone installation position (observation point) is minimized.
FIG. 11 is a block diagram of a conventional active noise control apparatus that employs a Filtered-x-LMS algorithm (Patent Document 1). The reference signal generator 1 generates and outputs a reference signal x correlated with the noise to be silenced. To do. The adaptive control device 2 receives the reference signal x generated from the reference signal generator 1, and combines the noise NZ and the cancellation sound CN at the noise cancellation position (observation point, for example, near the driver's ear) in the passenger compartment. A sound signal is input as an error signal e, and adaptive signal processing is performed so that the power of the error signal is minimized, and a noise cancellation signal y is output. The adaptive control device 2 convolves the transmission characteristic of the canceling sound transmission system from the speaker to the noise cancellation point into the reference signal x by convolution of the adaptive signal processing unit ADP, the adaptive filter ADF having a digital filter configuration, and the reference signal x. A signal processing filter SFT to be created is provided. The cancel speaker 3 receives the adaptive filter output signal and radiates the noise canceling sound CN. The error microphone 4 is arranged at the noise canceling point, detects the synthesized sound of the noise NZ and the canceling sound CN, and outputs the synthesized sound signal. The error signal e is input to the adaptive signal processing unit ADP of the noise cancellation controller 2.

  The adaptive signal processing unit ADP receives the error signal e at the noise cancellation point and the signal processing reference signal input via the signal processing filter SFT, and uses these signals to cancel the noise at the noise cancellation point. Processing is performed to determine the coefficient of the adaptive filter ADF. For example, the adaptive signal processing unit ADP determines the coefficient of the adaptive filter ADF so that the power of the error signal e input from the error microphone 4 is minimized according to a well-known Filtered-x-LMS (Least Mean Square) adaptive algorithm. The adaptive filter ADF performs a digital filter process on the reference signal x according to the coefficient determined by the adaptive signal processing unit ADP and outputs a noise cancellation signal y. The reference signal x must be a signal having a high correlation with the noise NZ to be deleted, and a sound having no correlation with the reference signal is not deleted.

FIG. 12 is a block diagram of a transfer characteristic identification unit for identifying the acoustic transfer characteristic C from the speaker to the noise cancellation point set in the signal processing filter SFT in the noise cancellation controller 2, and the same parts as those in FIG. It is attached.
When the white noise generating unit 5 is activated, the white noise is input to the adaptive control device 6 and subjected to filtering processing by the adaptive filter ADF ′. Both the routes are radiated from the speaker 3 to the vehicle interior acoustic space and detected by the microphone 4. . The calculation unit 7 subtracts the output signal of the adaptive filter ADF ′ from the detection signal from the microphone 4 and feeds back the difference as an error signal e to the adaptive control device 6. The adaptive signal processor ADP ′ performs adaptive control by the LMS algorithm so as to minimize the power of the error signal e, and determines and sets the coefficient of the adaptive filter ADF ′. Thereafter, the above adaptive control is repeated, and finally, the output of the adaptive filter ADF ′ and the microphone output become equal, whereby the acoustic transfer characteristic C ′ from the speaker 3 to the microphone 4 is set in the adaptive filter ADF ′. After the identification of the acoustic transfer characteristic C ′ is completed, the coefficient (transfer characteristic) of the adaptive filter ADF ′ is copied to the signal processing filter SFT of FIG.

により、周波数毎に現時刻ｋ＋１の適応フィルタ係数W_M(f_j，k+1)(j=1〜F)を1時刻前の適応フィルタ係数W_M(f_j，k) (j=1〜F)を用いて更新することで消音を実現している。
特許第３４１９８７８号 The active noise control device in FIG. 11 is an example using a time domain reference signal and an error signal. The active noise control device converts these time domain signals into frequency domain signals and performs adaptive signal processing in the frequency domain. There is. FIG. 13 is a block diagram of an active noise control apparatus that performs adaptive signal processing in the frequency domain, and the same components as those in FIG. 11 are denoted by the same reference numerals.
The FFT unit 8a converts the reference signal x in the time domain into a signal X (Xf ₁ , Xf ₂ ,... Xf _F ) in the frequency domain (frequency f ₁ , f ₂ ,... _{F F} ), and the FFT unit 8b. Converts the error signal e into a frequency domain signal E (Ef ₁ , Ef ₂ ,... Ef _F ), and the adaptive control unit 9 uses an coefficient update equation for each frequency component to show an adaptive filter coefficient W (Wf ₁ , Wf ₂ ,... Wf _F ), and the IFFT unit 8 c converts the signal output from the adaptive filter for each frequency component into a time domain signal and inputs it to the speaker 3. The LMS algorithm of the adaptive controller 9 is basically

Thus, the adaptive filter coefficient W _M (f _j , k + 1) (j = 1 to F) at the current time k + 1 is changed to the adaptive filter coefficient W _M (f _j , k) (j = 1 to Silence is realized by updating with F).
Patent No. 3419888

By the way, in recent years, automobiles are becoming lighter and more fuel-efficient due to environmental requirements, and accordingly, further reduction in the price of active noise control devices and the development of various types of vehicles are demanded. However, as the number of sound sources (speakers) for muting and the number of microphones L increase, the number of adaptive control units and the number of transfer characteristics C 'increase, which increases the amount of computation and storage of the transfer characteristics. This is a cause of cost increase.
In addition, when deploying multiple types of active noise control devices, there is a demand to introduce one active noise control device directly into multiple vehicle types in order to achieve mass merit in manufacturing costs. Since the amount of storage of C ′ increases by the number of models, a larger memory is required, which also increases costs.
From the above, an object of the present invention is to reduce the amount of calculation required for the calculation of the coefficient update formula.
Another object of the present invention is to reduce the memory size required by reducing the amount of transfer characteristic C ′ to be stored.

The present invention sets the transfer characteristic from each speaker to each microphone in the adaptive control unit, and uses the coefficient update formula including the transfer characteristic in the adaptive control unit so that the error signal power is minimized. Is an adaptive signal processing system for determining a transmission characteristic and a transfer characteristic setting method thereof.
Transfer characteristic setting method In the transfer characteristic setting method of the present invention, a transfer characteristic from each speaker to each microphone is calculated, and a transfer gain is obtained from each transfer characteristic in two or more addition terms of the coefficient update equation and compared. Step, when the gain difference is less than or equal to the set value, the addition term with the smaller transfer gain is regarded as 0, and only the transfer characteristic of the larger addition term is set in the adaptive control unit.
The reference signal and the error signal input to the adaptive control unit are converted into a frequency domain signal, and the adaptive control unit determines a coefficient of the adaptive filter for each frequency component using a coefficient update formula, and each adaptive filter for each frequency. When the signal output from the signal is converted into a time-domain signal and input to the speaker, the transfer characteristic from each speaker to each microphone is obtained for each frequency component, and two or more addition terms in the coefficient update equation are calculated for each frequency component. If the gain difference is less than or equal to the set value, the transfer term with the smaller transfer gain is regarded as 0 and only the transfer characteristic with the larger addition term is set in the adaptive control unit. To do.

Adaptive signal processing system An adaptive signal processing system according to the present invention receives an input of a reference signal and an error signal, an adaptive control unit for adaptive signal processing, and a signal output from an adaptive filter of the adaptive control unit 2 The above speaker, one or more microphones that detect acoustic signals output from each speaker, a transfer characteristic calculation unit that calculates transfer characteristics from each speaker to each microphone, and two or more addition terms in the coefficient update equation The comparison unit that obtains and compares the transfer gain from each transfer characteristic in FIG. 5 and the gain difference is equal to or smaller than the set value, the addition term with the smaller transfer gain is regarded as 0, and only the transfer characteristic with the larger addition term is applied. A transfer characteristic setting unit to be set in the control unit is provided.
The adaptive signal processing system of the present invention further includes an FFT unit that converts the reference signal into a frequency domain signal, an FFT unit that converts the error signal into a frequency domain signal, and a signal output from the adaptive filter for each frequency component. The IFFT unit converts the signal into a time domain signal and inputs it to the speaker, the transfer characteristic calculation unit calculates the transfer characteristic from each speaker to each microphone for each frequency component, and the comparison unit calculates the coefficient for each frequency. Transfer gain is obtained from each transfer characteristic in two or more addition terms in the update formula and compared. When the gain difference is less than the set value, the transfer characteristic setting unit regards the addition term with the smaller transfer gain as 0 and the larger one Only the transfer characteristic of the addition term is set in the adaptive control unit.

According to the present invention, the transfer characteristic from each speaker to each microphone is calculated, the transfer gain is calculated from each transfer characteristic in the addition term of two or more of the coefficient update formula, and compared, and when the gain difference is less than the set value, Since the addition term with the smaller transfer gain is regarded as 0 and only the transfer characteristic of the larger addition term is set in the adaptive control unit, the addition term in the coefficient update equation can be reduced. For this reason, the number of operations can be reduced, and high-speed processing has become possible. In addition, since it is not necessary to set the transfer characteristic of the addition term having a smaller transfer gain in the adaptive control unit, it is possible to reduce the amount of memory used to store these transfer characteristics.
Also, when the time domain signal is converted to the frequency domain signal and adaptive signal processing is performed in the frequency domain, the transfer characteristic from each speaker to each microphone is obtained for each frequency component, and the coefficient update equation is calculated for each frequency component. Transfer gains are obtained and compared using transfer characteristics in two or more addition terms, and if the gain difference is less than or equal to the set value, the addition term with the smaller transfer gain is regarded as 0, and the transfer characteristics of the larger addition term Therefore, if the number of frequency components is set to F (for example, 200), the number of operations when the present invention is applied to adaptive signal processing in the time domain is smaller than the amount of reduction in the number of operations. F times can be reduced, enabling high-speed processing. In addition, the number of transfer characteristics set in the adaptive control unit can be significantly reduced, thereby reducing the amount of memory used for storing these transfer characteristics and reducing costs.
Further, even when parameters for specifying transfer characteristics for each vehicle type are stored, the number of stored transfer characteristics can be reduced, and the memory size can be reduced.

(A) Outline of the present invention The present invention compares the transfer gains determined from the transfer characteristics in the two or more addition terms of the adaptive filter coefficient update equation, and if the gain difference is less than the set value, the smaller gain is added. By considering the term as 0, the amount of transfer characteristics to be used is reduced, the calculation of the addition term is unnecessary, the memory amount and the computation amount are reduced, and the cost is reduced.
Hereinafter, the principle of the present invention will be described using simple examples.
Fig. 1 shows an example of the vehicle interior acoustic system of the active noise control system. Two sound source (speakers) SP1 and SP2 are installed at appropriate locations in the vehicle interior CSP (front and rear in the vehicle interior). In this example, two microphones MIC1 and MIC2 are provided at the cancellation point. FIG. 2 is a rewritten version of the vehicle interior acoustic system of FIG. Each microphone acoustic signal generated from the speaker SP1 via the transmission characteristic C _11, C ₂₁ MIC1, reaches the MIC2, each acoustic signal generated from the speaker SP2 via the transfer characteristics C _12, C ₂₂ microphone MIC1, MIC2 Will be detected. Each microphone MIC1, MIC2 also detects the noise d to be silenced. STF and STB are front and rear seats.

となる。なお、Mは消音用音源数、Lはマイクロホン数である。 FIG. 3 is a configuration diagram of an active noise control system including two sound-reducing sound sources (speakers), two microphones, and two adaptive control units, and is a case where adaptive signal processing is performed in the frequency domain. The active noise control system includes an FFT unit 11 that converts a reference signal x into a frequency domain signal X, and an FFT that converts error signals e ₁ and e ₂ output from the microphones MC1 and MIC2 into frequency domain signals E ₁ and E _2. Units 12 ₁ , 12 ₂ , two adaptive control units 13 ₁ , 13 ₂ that perform adaptive signal processing in the frequency domain, frequency domain signals output from the adaptive filters of the adaptive control units 13 ₁ , 13 ₂ for each frequency component An IFFT section 14 ₁ , 14 ₂ for converting Y ₁ , Y ₂ to time domain signals y ₁ , y ₂ and outputting them, and a vehicle interior acoustic system 15 to which the signals y ₁ , y ₂ are input are provided. The vehicle interior acoustic system 15 is shown by modeling the vehicle interior acoustic system of FIG. 1, SP1 and SP2 are speakers, MIC1 and MIC2 are microphones, Cij is an acoustic transfer characteristic from each speaker to each microphone, CB1 and CB2 Is a synthesis part.
For the sound source and the microphone is two by two, an acoustic transfer system between them are four systems of C ₁₁ -C _22. The coefficient update formulas of the adaptive filter based on the LMS algorithm of the adaptive control units 13 ₁ and 13 ₂ are respectively

It becomes. Note that M is the number of sound sources for muting and L is the number of microphones.

になる。
適応信号処理部２１は、周波数f₁，f₂，・・・f_F毎に上式の演算を実行し、周波数毎の適応フィルタの係数W₁(f₁,k+1)、W₁(f₂,k+1)・・・W₁(f_F,k+1)を計算し、適応フィルタ２２はW₁(f₁,k+1)、W₁(f₂,k+1)・・・W₁(f_F,k+1)を周波数領域に変換した参照信号X₁(f₁)、X₁(f₂)・・・X₁(f_F)に乗算し、乗算結果を図３のIFFT部１４₁に入力する。 Figure 4 is a block diagram of an adaptive control unit 13 ₁ has a structure that updates the adaptive filter coefficients according to expression (2a). (2a) Formula Each frequencies f _1, f _2, rewritten every · · · f _F and the following coefficient update equation

become.
The adaptive signal processing unit 21 performs the above calculation for each of the frequencies f ₁ , f ₂ ,... F _F, and performs adaptive filter coefficients W ₁ (f ₁ , k + 1), W ₁ ( f ₂ , k + 1)... W ₁ (f _F , k + 1) is calculated, and the adaptive filter 22 uses W ₁ (f ₁ , k + 1), W ₁ (f ₂ , k + 1). _{·· W 1 (f F, k} + 1) reference signal X ₁ converted into the frequency domain _{(f 1), X 1 (} f 2) multiplying the ··· X ₁ (f _F), FIG multiplication result 3 is input to IFFT section 14 ₁ .

In the adaptive signal processing unit 21, 31 _{1 to} 31 _F are complex conjugate arithmetic units that calculate and output the complex conjugates of the reference signals X ₁ (f ₁ ), X ₁ (f ₂ )... X ₁ (f _F ). , 32 _{1 to} 32 _F are delay units for outputting the previous adaptive filter coefficients W ₁ (f ₁ , k), W ₁ (f ₂ , k)... W ₁ (f _F , k) with a predetermined time delay. , 33 _{1 to} 33 _F represent the transfer characteristics C ′ ₁₁ (f ₁ ) to C ′ ₁₁ (f _F ) for each frequency from the speaker SP1 to the microphone MIC1 as reference signals X ₁ (f ₁ ) and X ₁ (f ₂ ). · · · X ₁ complex conjugate signal X ₁ ^{* (f} ₁₎ of (f _F) ~X ₁ ^* multiplication unit for multiplying the ^{_{(f F), 34 1 ~34}} F is for each frequency from the loudspeaker SP1 to the microphone MIC2 Transfer characteristics C ′ ₂₁ (f ₁ ) to C ′ ₂₁ (f _F ) are referred to as reference signals X ₁ (f ₁ ), X ₁ (f ₂ )... X ₁ (f _F ) complex conjugate signal X ₁ ^* ( f ₁ ) to X ₁ ^* (f _F ) are multiplied by multiplication units, 35 _{1 to} 35 _F are output signals of the multiplication units 33 _{1 to} 33 _F and error calculation signals E ₁ (f ₁ ) to E ₁ converted into the frequency domain. (f _F) Multiplying unit for multiplying, 36 ₁ ~ 36 _F multiplication unit for multiplying the multiplication unit 34 ₁ to 34C _F and the output signal from the miscalculation signal converted into the frequency domain, _{E 2 (f 1) ~E 2} (f F), 37 1 ˜37 _F is an adder that performs the addition operation in [] in the second term on the right-hand side of the expressions (3-1) to (3-F), and 38 _{1 to} 38 _F are step size parameters of 2 μ to the output signal of the adder. Multipliers 39 _{1 to} 39 _F for multiplying are adders for adding the first and second terms on the right side of the equations (3-1) to (3-F). In the adaptive filter 22, 30 ₁ to 30 _F are filter coefficients W ₁ (f ₁ , k + 1), W ₁ (f ₂ , k + 1)... W ₁ (f _F , k +) for each frequency. It is a multiplier that multiplies reference signals X ₁ (f ₁ ), X ₁ (f ₂ )... X ₁ (f _F ) obtained by converting 1) into the frequency domain.

になる。 The above is the configuration of the adaptive control unit 13 ₁ , but the adaptive control unit 13 ₂ has the same configuration. Incidentally, (2b) expression each frequencies f _1, f _2, rewritten every · · · f _F and the following coefficient update equation

become.

  As can be seen from FIG. 4 and the equations (2a) to (2b), the number of control filter W increases as the number of speakers as the sound source for mute increases, and the calculation related to the transfer characteristics inside the adaptive control unit due to the increase in microphones. It can be seen that the amount of the transfer characteristic increases and the amount of calculation and the amount of memory increase. In particular, since F = 200, the calculation amount and the memory amount increase. In addition, the transmission characteristics of the acoustic transmission system change as the vehicle model to which the noise control system is installed changes, so that a single noise control system can handle multiple vehicle models. The amount of computation and the amount of memory are significant.

により表現できる。 Therefore, in the present invention, paying attention to the addition formula of the second term on the right side of the formula (2a), the transfer gain is obtained from the transfer characteristics included in each addition term, and compared,
Gain of transfer characteristic C ′ ₁₁ (f _j ) >> If gain of transfer characteristic C ′ ₂₁ (f _j ), then C ′ ₂₁ (f _j ) X ^* (f _j , k) E ₂ (f _j , Since the term of k) is much smaller than C ′ ₁₁ (f _j ) X ^* (f _j , k) E ₁ (f _j , k), C ′ ₂₁ (f _j ) X ^* (f _j , k) E ₂ (f _j , k) is regarded as 0 and ignored. The same applies to the reverse case.
In equation (2b), the gain of transfer characteristic C ′ ₁₂ (f _j ) >> the gain of transfer characteristic C ′ ₂₂ (f _j ), then C ′ ₂₂ (f) X ^* (f _j , k) E ₂ (f _j, k) term C of the ^{_{'12 (f j) X *}} (f j, k) E 1 (f j, k) from very small compared _{to, C' 22 (f j)} X * (f _j , k) E ₂ (f _j , k) is regarded as 0 and ignored. The same applies to the reverse case.
The transfer characteristic C _ij ′ (f _j ) is expressed in the form of a complex number a + bj, and the following equation is used by using the gain G and the phase difference θ.

Can be expressed by

As described above, by ignoring the gain a small section is regarded as 0, the adaptive control unit 13 ₁ of the configuration shown in FIG. 4, as shown in FIG. That is, it is possible to delete a part of a coefficient or a circuit that can be ignored as indicated by a dotted line. However, in FIG.
The gain of C ′ ₁₁ (f ₂ ) >> the gain of C ′ ₂₁ (f ₂ ) and the gain of C ′ ₂₁ (f _F ) >> the gain of C ′ ₁₁ (f _F ).
As described above, according to the present invention, when measuring the acoustic transmission system (transmission characteristics) of a vehicle equipped with a product, the gain of the measured transmission characteristics C ′ ₁₁ (f _j ), the gain of the transmission characteristics C ′ ₂₁ (f _j ), and the transmission The gain of characteristic C ′ ₁₂ (f _j ) and the gain of transfer characteristic C ′ ₂₂ (f _j ) are compared. If the difference is equal to or greater than the set value v [dB] (for example, 20 dB or greater), the smaller transmission Delete a characteristic.

(B) Embodiment FIG. 6 is a block diagram of an embodiment of the present invention for determining an acoustic transmission system (transfer characteristics). The same reference numerals are given to the same parts as those in FIG.
When the white noise oscillator 53 is activated with the switches 51 and 52 shown in the figure, the white noise is input to the adaptive control device 54 and subjected to filtering processing by the adaptive filter ADF, and the vehicle interior acoustic space CSP is obtained from the speaker SP1. And detected by the microphone MIC1. The computing unit 55 subtracts the output signal of the adaptive filter ADF from the detection signal from the microphone MIC1, and feeds back the difference as an error signal e to the adaptive control device 54. The adaptive signal processing unit ADP performs adaptive control by the LMS algorithm so as to minimize the power of the error signal e, and determines and sets the coefficient of the adaptive filter ADF. Thereafter, the above adaptive control is repeated, and finally, the output of the adaptive filter ADF and the microphone output become equal, whereby the acoustic transfer characteristic C ₁₁ ′ from the speaker SP1 to the microphone MIC1 is set in the adaptive filter ADF. Thereafter, the FFT calculation unit 56 takes in a predetermined number (F) of transfer characteristics C ₁₁ ′ from the adaptive filter ADF at a predetermined sampling rate, and performs FFT processing to transfer characteristics C _{11 of F} frequencies f _{1 to} f _F. ′ (F ₁ ) to C ₁₁ ′ (f _F ) are calculated and stored in the C ₁₁ characteristic unit 57 ₁ of the memory 57.

As described above, when the identification process of the transfer characteristics C ₁₁ ′ (f ₁ ) to C ₁₁ ′ (f _F ) is completed, the switch 52 is switched to the dotted line side, and F frequencies f _{1 to} f _F are changed by the same process. transfer characteristic _{C 21 '(f 1) ~C} 21' calculates the (f _F) is stored in C ₂₁ characteristic unit 57 ₂ of the memory 57. Thereafter, the switches 51 and 52 are switched, and similarly, transfer characteristics C ₁₂ ′ (f ₁ ) to C ₁₂ ′ (f _F ) and C ₂₂ ′ (f ₁ ) to C _{22 of F} frequencies f _{1 to} f _F , respectively. '(F _F ) is calculated and stored in the C ₁₂ characteristic section 57 ₃ and the C ₂₂ characteristic section 57 ₄ of the memory 57. FIG. 7 is an explanatory diagram of the storage state of the transfer characteristic in the memory 57.
When the identification process of all transfer characteristics is completed, the first characteristic selection unit 58 transfers the transfer characteristic C ₁₁ ′ (f _j ) (j = 1) of the first addition term in the addition expression of the second term on the right side of the equation (2a). The gain of 1 to F) is compared with the gain of the transfer characteristic C ₂₁ ′ (f _j ) (j = 1 to F) of the second addition term. And
If the gain of C ₁₁ ′ (f _j ) >> C ₂₁ ′ (f _j ) (if there is a difference greater than or equal to the set level VdB), only the transfer characteristic C ₁₁ ′ (f _j ) is stored in C of the memory 60. Save to ₁₁ characteristic unit 60 _1. on the other hand,
If the gain of C ₁₁ ′ (f _j ) << C ₂₁ ′ (f _j ) (if there is a difference equal to or higher than the set level VdB), only the transfer characteristic C ₂₁ ′ (f _j ) is stored in C of the memory 60. Save ₂₁ characteristic unit 60 ₂ stores both unless the difference over the set level VdB the gain C ₁₁ characteristic unit 60 _1, C ₂₁ characteristic unit 60 ₂ of the memory 60.

When the above processing is completed for the transfer characteristics of _F total frequencies f _{1 to} f _F , the second characteristic selection unit 59 transmits the first addition term in the addition expression of the second term on the right side of equation (2b). The gain of the characteristic C ₁₂ ′ (f _j ) (j = 1 to F) is compared with the gain of the transfer characteristic C ₂₂ ′ (f _j ) (j = 1 to F) of the second addition term. And
If the gain of C ₁₂ ′ (f _j ) >> C ₂₂ ′ (f _j ) (if there is a difference equal to or higher than the set level VdB), only the transfer characteristic C ₁₂ ′ (f _j ) is stored in C of the memory 60. Save to ₁₂ characteristic unit 60 _3. on the other hand,
If the gain of C ₁₂ ′ (f _j ) << C ₂₂ ′ (f _j ) (if there is a difference equal to or higher than the set level VdB), only transfer characteristic C ₂₂ ′ (f _j ) is stored in C of memory 60. and stored in ₂₂ characteristic unit 60 ₄ stores both unless the difference over the set level VdB the gain C ₁₂ characteristic unit 60 _3, C ₂₂ characteristic portion 60 ₄ of the memory 60.

FIG. 8 is an explanatory diagram of the storage state of the transfer characteristic in the memory 60.
Gain of C ₁₁ ′ (f ₁ ) >> Gain of C ₂₁ ′ (f ₁ )
Gain of C ₁₁ ′ (f ₂ ) << Gain of C ₂₁ ′ (f ₂ )
There is no difference between the gain of C ₁₁ ′ (f ₃ ) and the gain of C ₂₁ ′ (f ₃ ) above the set level.
The gain of C ₁₁ ′ (f _F ) >> the gain of C ₂₁ ′ (f _F ), and
Gain of C ₁₂ ′ (f ₁ ) << Gain of C ₂₂ ′ (f ₁ )
Gain of C ₁₂ ′ (f ₂ ) << Gain of C ₂₂ ′ (f ₂ )
Gain of C ₁₂ ′ (f ₃ ) >> Gain of C ₂₂ ′ (f ₃ )
This is the case when the gain of C ₁₂ ′ (f _F ) and the gain of C ₂₂ ′ (f _F ) are not different from the set level.
When the selection of the transfer characteristic is completed as described above, the transfer characteristics C ₁₁ ′ (f _j ) and C ₂₁ ′ (f _j ) (j = 1 to F) stored in the memory 60 are changed to the adaptive control unit 13 in FIG. is set to _1, also the transfer characteristic _{C 12 '(f j),} C 22' (f j) (j = 1~F) is set to the adaptive control unit 13 _2. As a result, the adaptive signal processing units of the adaptive control unit 13 ₁ and the adaptive control unit 13 ₂ determine the coefficients of the adaptive filter by the equations (2a) and (2b) using the set transfer characteristics, respectively, and perform noise cancellation control. Can be executed.

(C) Modifications The above is a case where the present invention is applied to an active noise control system. However, an adaptive equalization control system that obtains a desired transfer characteristic and sets it in a filter and outputs audio sound through the filter is also described. Can be applied.
The vehicle interior is a sealed narrow space. Therefore, since reflection occurs in a short time and sound waves interfere with each other, the propagation characteristics up to the listening point are very complicated. In addition, since music or the like is being listened to in an asymmetrical place, the propagation characteristics from the left and right speakers are also greatly different. There is a demand for an audio device that eliminates the adverse effects of the vehicle interior and improves the acoustic characteristics in the vehicle interior, and the adaptive equalization system includes amplitude and phase characteristics at multiple points (control points) in the reproduction space. Control to achieve desired characteristics.

FIG. 9 is a basic configuration diagram of an adaptive equalization system using one speaker and one microphone. 71 is an audio source (tuner, tape deck, CD player, etc.) for outputting an audio signal x, and 72 is a target. A response characteristic (impulse response characteristic) H is set, an audio signal x is input and a target response setting unit that outputs a target signal d; 74 is a microphone that detects sound at a listening position (observation point) in the vehicle interior acoustic space; 75 is an arithmetic unit for calculating an error e between the detected audio signal z and the target signal d output from the target response setting unit 72, and 76 is an adaptation for generating the signal y so that the power of the error e is minimized. A control device 77 is a speaker (control sound source) that radiates sound corresponding to the signal y to the vehicle interior acoustic space 78. The target response characteristic H of the target response setting unit 72 is set to a characteristic corresponding to the sound field space to be reproduced.
The adaptive control device 76 receives the audio signal x as a reference signal and also receives the error signal e output from the arithmetic unit 75, and performs adaptive signal processing so that the power of the error signal is minimized. The signal y is output. The adaptive control device 76 includes an adaptive signal processing unit ADP that performs adaptive signal processing according to the LMS algorithm, and an adaptive filter ADF having an FIR type digital filter configuration.
The adaptive signal processing unit ADP receives the error signal e and the reference signal x at the listening position, and uses these signals so that the audio signal z at the listening position becomes equal to the target signal d by the coefficient updating formula of the formula (1). Determine the coefficients of the adaptive filter ADF. The adaptive filter ADF performs digital filter processing on the audio signal x according to the coefficient determined by the adaptive signal processing unit ADP and outputs a signal y. Therefore, if the coefficients of the adaptive filter ADF converge so that the power of the error signal e is minimized by adaptive signal processing, the audio signal z becomes equal to the target signal d at the listening position. Therefore, by separating the target response setting unit 72, it is possible to listen to a sound equivalent to the sound heard in an ideal space.

に示す係数更新式を用いて適応信号処理を行って、第１、第２聴取位置におけるオーディオ信号z1、z2がそれぞれ目標信号d1、d2と等しくなる適応フィルタの係数を決定する。しかる後、目標応答設定部８６，８７を切り離すことにより、第１、第２聴取位置において理想的な空間で音を聴取したのと同等の音の聴取が可能となる。 FIG. 10 is a block diagram of an adaptive equalization system using two speakers and two microphones. Each speaker SP1, SP2 and microphones MIC1, MIC2 are arranged at appropriate positions in the vehicle interior as in FIG. The acoustic transmission system is as shown in FIG.
The adaptive equalization system includes a music source 80, an FFT unit 81 that converts a music signal x as a reference signal into a signal X in the frequency domain, and error signals output from the microphones MC1 and MIC2 (audio signals z1 and z2 at the listening position and target Error signals of signals d1 and d2) FFT units 82 ₁ and 82 ₂ for converting e ₁ and e ₂ into frequency domain signals E ₁ and E ₂ , two adaptive control units 83 ₁ for performing adaptive signal processing in the frequency domain, 83 _{2, 1} adaptive control unit 83 for each frequency component, 83 signal Y ₁ in the frequency domain output from the adaptive filter _2, Y ₂ a time signal y ₁ region, IFFT section is converted into y ₂ outputs 84 ₁ , 84 ₂ , interior acoustic system 85 to which signals y ₁ and y ₂ are input, and a first target response setting unit in which target response characteristics H ₁ are set at the first listening position (position where microphone MIC ₁ is installed). 86, the target response H ₂ is set is at the second listening position (installation position of the microphone MIC2) The second has a target response setting unit that. The vehicle interior acoustic system 15 is shown by modeling the vehicle interior acoustic system of FIG. 1 or FIG. 2, SP1 and SP2 are speakers, MIC1 and MIC2 are microphones, Cij is an acoustic transfer characteristic from each speaker to each microphone, CB1 and CB2 are synthesis units.
Adaptive control unit 83 _1, 83 _2, like the case of the active noise control system of Figure 3, respectively, the following equation

The adaptive signal processing is performed using the coefficient update equation shown in FIG. 6 to determine the adaptive filter coefficients at which the audio signals z1 and z2 at the first and second listening positions are equal to the target signals d1 and d2, respectively. Thereafter, by separating the target response setting units 86 and 87, it is possible to listen to the sound equivalent to the sound heard in the ideal space at the first and second listening positions.

As described above, also in the adaptive equalization system, the transmission characteristics C ′ ₁₁ (f _j ), C ′ ₂₁ (f _j ), C ′ ₁₂ (f _j ), and C ′ ₂₂ (f _j ) are assigned to the adaptive control unit 83 ₁ , It is necessary to set to 83 ₂ . As can be seen from the equations (6a) to (6b), the number of adaptive control units increases as the number of speakers increases, and the calculation related to the transfer characteristics inside the adaptive control unit and the transfer characteristic amount increase as the number of microphones increases. As a result, the calculation amount and the memory amount increase. Therefore, the transfer characteristics C ′ ₁₁ (f _j ), C ′ ₂₁ (f _j ), C ′ ₁₂ (f _j ), and C ′ ₂₂ (f _j ) used for the adaptive signal processing are already shown in FIG. Set by. In this way, the adaptive equalization control system can achieve the same effects as the active noise control system.
The above embodiment is an example of a system using two speakers and two microphones. However, the present invention is also applied to a noise canceling system and adaptive equalization control system using one or more speakers and two or more microphones. Can be applied.

It is an example of the vehicle interior acoustic system of an active noise control system. It is explanatory drawing which rewritten the vehicle interior acoustic system of FIG. 1 intelligibly. It is a block diagram of an active noise control system provided with two sound source for mute (speaker), two microphones, and two adaptive control units. It is the conventional block diagram of an adaptive control part. It is a block diagram of this invention of an adaptive control part. It is an Example block diagram of this invention which determines an acoustic transmission system (transfer characteristic). It is a 1st preservation | save state explanatory drawing of the transfer characteristic in memory. It is a 2nd preservation | save state explanatory drawing of the transfer characteristic in memory. 1 is a basic configuration diagram of an adaptive equalization system using one speaker and one microphone. FIG. It is a block diagram of an adaptive equalization system using two speakers and two microphones. It is a block diagram of the conventional active noise control apparatus which employ | adopted Filtered-x-LMS algorithm. It is a block diagram of the transmission characteristic identification part which identifies the acoustic transmission characteristic from a speaker to a noise cancellation point. It is a block diagram of the active noise control apparatus which performs adaptive signal processing in a frequency domain.

Explanation of symbols

51, 52 Switch 53 White noise oscillator 54 Adaptive control device 55 Calculation unit 56 FFT calculation unit 57 Memory 58 First characteristic selection unit 59 Second characteristic selection unit 60 Memory

Claims (4)

An adaptive control unit that receives a reference signal and an error signal and performs adaptive signal processing, one or more speakers that receive signals output from the adaptive filter of the adaptive control unit, and an acoustic signal that is output from each speaker Two or more microphones, and the transfer characteristic from each speaker to each microphone is set in the adaptive control unit so that the error signal power can be minimized by using a coefficient updating formula including the transfer characteristic in the adaptive control unit. In a transfer characteristic setting method in an adaptive signal processing system for determining a coefficient of the adaptive filter,
Find the transfer characteristics from each speaker to each microphone,
Transfer gain is obtained from each transfer characteristic in two or more addition terms of the coefficient update formula, and compared,
When the gain difference is less than or equal to the set value, the addition term with the smaller transfer gain is regarded as 0, and the transfer characteristic in the larger addition term is set in the adaptive control unit.
A transfer characteristic setting method characterized by the above. The reference signal and the error signal input to the adaptive control unit are converted into a frequency domain signal, and the adaptive control unit determines an adaptive filter coefficient for each frequency component using a coefficient update formula, and each adaptive filter for each frequency. When the signal output from is converted to a time domain signal and input to the speaker,
For each frequency component, find the transfer characteristics from each speaker to each microphone,
For each frequency component, a transfer gain is obtained from each transfer characteristic in two or more addition terms of the coefficient update formula, and compared,
When the gain difference is less than or equal to the set value, the addition term with the smaller transfer gain is regarded as 0, and the transfer characteristic of the larger addition term is set in the adaptive control unit.
The transfer characteristic setting method according to claim 1. An adaptive control unit that receives a reference signal and an error signal and performs adaptive signal processing, one or more speakers that receive signals output from the adaptive filter of the adaptive control unit, and an acoustic signal that is output from each speaker Two or more microphones, and the transfer characteristic from each speaker to each microphone is set in the adaptive control unit so that the error signal power can be minimized by using a coefficient updating formula including the transfer characteristic in the adaptive control unit. In an adaptive signal processing system for determining coefficients of the adaptive filter,
A transfer characteristic calculator for calculating transfer characteristics from each speaker to each microphone;
A comparison unit that obtains a transfer gain from each transfer characteristic in the two or more addition terms of the coefficient update formula and compares them;
When the gain difference is equal to or smaller than the set value, the transfer term having a smaller transfer gain is regarded as 0, and the transfer characteristic in the larger addition term is set in the adaptive control unit.
An adaptive signal processing system comprising: The adaptive signal processing system of claim 3, further comprising:
FFT section for converting the reference signal into a frequency domain signal,
FFT section for converting the error signal into a frequency domain signal,
An IFFT unit that converts the signal output from the adaptive filter for each frequency component into a signal in the time domain and inputs the signal to the speaker,
The transfer characteristic calculation unit calculates a transfer characteristic from each speaker to each microphone for each frequency component, and the comparison unit obtains a transfer gain from each transfer characteristic in two or more addition terms of the coefficient update formula for each frequency and compares them. When the gain difference is equal to or smaller than the set value, the transfer characteristic setting unit regards the addition term with the smaller transfer gain as 0 and sets the transfer characteristic in the larger addition term to the adaptive control unit.
And an adaptive signal processing system.

Images